Habeas Litigation in Federal Courts Vanderbilt Study 2007

• Topics: Sentencing, Habeas Corpus

Download original document:
Brief thumbnail

This text is machine-read, and may contain errors. Check the original document to verify accuracy.

Final Technical Report:
Habeas Litigation in U.S. District Courts
An empirical study of habeas corpus cases filed by state prisoners under the
Antiterrorism and Effective Death Penalty Act of 1996

by
Nancy J. King, J.D.
Lee S. & Charles A. Speir Professor of Law
Vanderbilt University Law School
Fred L. Cheesman II, Ph.D.
National Center for State Courts
Brian J. Ostrom, Ph.D.
National Center for State Courts

V"'NOERllILT
VI\NOERBllT UNIVERSITY

Law

School

This project was supported by Award No. 2006-IJ-CX-0020, awarded by the National
Institute of Justice, Office of Justice Programs, United States Department of Justice. The
opinions, findings, and conclusions or recommendations expressed in this publication are
those of the authors and do not necessarily reflect the views of the Department of Justice.

Final Technical Report: Habeas Litigation in U.S. District Courts

ACKNOWLEDGMENTS

This project could not have been conducted without the help of many talented people who contributed
their time and expertise.
We would first like to express our gratitude to the members of our Advisory Committee: Mr. Paul
Bottei, Assistant Federal Public Defender, Middle District of Tennessee; Ms. Gena A. Bunn, Chief, Capital
Litigation, Office of the Attorney General of Texas; Mr. Ronald Eisenberg, Deputy District Attorney, Office
of the Philadelphia Attorney General; the Honorable Frances C. Gull, Allen Superior Court, Allen County,
Indiana; Dr. Roger A. Hanson, Hanson & Associates; the Honorable Elizabeth A. Jenkins, U.S. Magistrate
Judge, Middle District of Florida; Attorney Mark E. Olive; and the Honorable Roger Vinson, U.S. District
Court for the Northern District of Florida. The Advisory Committee did not write or approve the report, and
any findings or conclusions contained in this report do not necessarily reflect the views of the Committee.
The Committee's role in this research was, however, vital. Each of its members spent days providing us with
candid and helpful feedback. After assisting us to determine what information would be most important to
collect, they helped us interpret and supplement our initial findings. We are also grateful for the time and
effort of Dr. Emery Lee and Professor Jon Gould of the Federal Judicial Center, who contributed their
insights at our second Committee meeting.
We were fortunate to have the generous support of the Vanderbilt Law School, including its Cecil
D. Branstetter Litigation & Dispute Resolution Program, which provided funding for the data collection
phase of this study. The assistance of more than a dozen talented Vanderbilt law students made this project
possible. Ms. Jillean Dubatowka and Mr. Gray Proctor deserve special thanks for working on the project for
two years. The complicated task of managing dozens of changing data files from several locations around
the country was made possible by the able oversight of Mr. Jason Bradley, Senior Systems Administrator at
Vanderbilt Law School. Ms. Marita Bush and Mr. Jason Hunt provided excellent administrative support for
the project. Numerous faculty colleagues at Vanderbilt Law School and Vanderbilt University provided
invaluable assistance at various stages of this project, including Professors Joni Hersch, Tracey George,
Chris Guthrie, Suzanna Sherry, Edward Rubin, Margaret Blair, John Goldberg, Susan Kay, Stefanie
Lindquist, and Richard Nagareda.
Special thanks are due Ms. Amy Spencer-Westercamp and Mr. Neil LaFountain at the National Center
for State Courts for their tireless efforts as part of the project team.
The support of the National Institute of Justice included not only its funding for completion of the
project, but also the advice and assistance provided by Dr. Linda Truitt and others at the U.S. Department of
Justice. For this assistance we are deeply appreciative.
Finally, we would like to thank the many federal judges and all of the pro se and death penalty law
clerks, attorneys, and court staff around the country who were so helpful to us in providing case
documentation and information about case processing.

Nancy J. King
Fred L. Cheesman II
Brian J. Ostrom

Final Technical Report: Habeas Litigation in U.S. District Courts

TABLE OF CONTENTS
I.

INTRODUCTION

A.
B.
C.
D.

Federal habeas review and the current statutory scheme................................................................... 7
Research goals of the study ............................................................................................................... 9
Prior research
.............................................................................................................................. 13
Study design and methodology ......................................................................................................... 14

II.

DESCRIPTIVE FINDINGS

Petitioner demographics .................................................................................................................... 18
1.
Petitioner race/ethnicity ...................................................................................................... 18
2.
Petitioner gender ................................................................................................................. 19

State proceedings
1.
Conviction offense, number of counts, sentence................................................................. 19
2.
Trial or plea conviction ...................................................................................................... 20
3.
Review in state court........................................................................................................... 20
4.
Time between state judgment and federal filing ................................................................. 21

Representation of petitioner in federal court
1.
In forma pauperis rulings ................................................................................................... 22
2.
Petitioner's counsel.............................................................................................................. 23

The petition
1.
Timing of first petitions ...................................................................................................... 23
2.
Timing of counseled petitions............................................................................................. 25

Type of proceeding challenged
1.
Challenges to decisions other than criminal judgments ..................................................... 26
2.
Sentence challenges ........................................................................................................... 27

Claims raised
1.
Number of claims per case.................................................................................................. 27
2.
Counsel-related error........................................................................................................... 28
3.
Evidentiary rulings.............................................................................................................. 29
4.
New evidence of innocence, insufficient evidence of guilt................................................. 29
5.
Judge or jury bias, selection, misconduct, or instructions................................................... 30
6.
Prosecutor comments; lost, undisclosed, or false evidence................................................. 30
7.
Guilty plea and negotiation challenges ............................................................................... 30
8.
Competence, presence, public, speedy trial ........................................................................ 30
9.
Double jeopardy and appeal-related claims ........................................................................ 31

Intermediate orders
1.
Transfers to another district ................................................................................................ 31
2.
Stays of execution ............................................................................................................... 31
3.
Stays for exhaustion ............................................................................................................ 31

Litigation steps
1.
Amended petitions .............................................................................................................. 34
2.
Answers, motions, replies ................................................................................................... 34
3.
State records........................................................................................................................ 35
4.
Discovery and evidentiary hearings .................................................................................... 35
5.
Magistrate judge use ........................................................................................................... 36

Final Technical Report: Habeas Litigation in U.S. District Courts

Processing time
1.
Pending (not terminated) cases .......................................................................................... 36
2.
Number of docket entries .................................................................................................... 38
3.
Days to termination from first docket entry ........................................................................ 38
4.
Days to termination from the filing of the petition ............................................................ 41
5.
Length of case in days, including pending cases ................................................................ 41
6.
Length of case excluding stayed time ................................................................................ 45

Non-merits dispositions
1.
All claims dismissed without reaching merits..................................................................... 45
2.
Statute of limitations dismissals.......................................................................................... 46
3.
Successive petitions dismissals ........................................................................................... 47
4.
Waiver of habeas review, settlement................................................................................... 47
5.
Voluntary dismissals ........................................................................................................... 47
6.
All claims dismissed as unexhausted .................................................................................. 47
7.
Dismissals for state procedural default ............................................................................... 48
8.
Dismissed as moot or lacking custody ................................................................................ 49
9.
Dismissed as barred by Teague .......................................................................................... 49

Merits dispositions
1.
Standard of review applied.................................................................................................. 50
2.
Harmless error ................................................................................................................ 50
3.
Grants of relief .................................................................................................................... 51

Appeals
1.
Notices of appeal................................................................................................................. 53
2.
Rulings on certificates of appealability............................................................................... 53

III.

COMPARATIVE FINDINGS

A.
B.
C.

Comparison with pre-AEDPA studies, non-capital cases ................................................................. 54
Comparison with pre-AEDPA studies, capital cases......................................................................... 60
Comparison of non-capital with capital cases ................................................................................... 62

IV.

EXPLANATORY FINDINGS

A.
B.
C.
D.

Capital cases: time from state judgment to federal filing ................................................................. 65
Non-capital case processing time ...................................................................................................... 68
Capital case processing time ............................................................................................................ 77
Factors associated with likelihood of relief in capital cases.............................................................. 86

APPENDICES

A.
B.
C.
D.
E.

Statistical tables .............................................................................................................................. 91
Time-barred capital cases.................................................................................................................. 109
Writs granted
.............................................................................................................................. 115
List of district names and district numbers........................................................................................ 125
List of variables collected.................................................................................................................. 126

Final Technical Report: Habeas Litigation in U.S. District Courts

LIST OF TABLES
(C) designates capital cases; (NC) designates non-capital cases
Table 1. (C) Number and percentage of cases in capital case sample, by district .........................................................17
Table 2. (C) Days from state judgment to federal filing, by state..................................................................................21
Table 3. (C) Days from state judgment to federal filing, by grant.................................................................................21
Table 4. (NC) Days from state judgment to federal filing, selected states ....................................................................22
Table 5. (C) Percentage of cases with claim type information, states with over 20 cases ............................................26
Table 6. (C) Percentage of cases stayed, by district.......................................................................................................32
Table 7. (C) Percentage of non-transferred cases stayed, by whether case terminated...................................................33
Table 8. (C) Percentage of cases with discovery or evidentiary hearing, by district .....................................................35
Table 9. (C) Cases with at least one claim time-barred, by state ...................................................................................46
Table 10. (C) Standard of review applied, by district.....................................................................................................50
Table 11. (C) Cases granted relief, by district ................................................................................................................51
Table 12. (C) Percentage cases granted, by whether case includes claim type...............................................................51
Table 13. (NC) Comparison of findings with findings of pre-AEDPA studies .............................................................56
Table 14. (C) Comparison of findings with findings of pre-AEDPA studies ................................................................61
Table 15. Summary of findings, capital and non-capital cases compared ......................................................................64
Table 16. (C) Regression of interval in days between state judgment and federal filing (logged) ................................67
Table 17. (NC) Regression of disposition time in days (logged), for terminated cases, by clerk and caseload .............70
Table 18. (NC) Regression of disposition time in days (logged), for terminated cases, by circuit and state..................71
Table 19. (NC) Regression of disposition time in days (logged), for terminated cases, by district................................72
Table 20. (C) Regression of days for disposition (logged), terminated cases................................................................80
Table 21. (C) Cox regression models, change in probability of termination over time .................................................82
Table 22. (C) Results of probit analysis of likelihood of grant .....................................................................................87
Appendix A.
Table 23. (C) Regression of days for disposition (logged), terminated cases..................................................................91
Table 24. (NC) Cox regression probability of termination over time, by circuit and state..............................................93
Table 25. (NC) Cox regression probability of termination over time, by clerk ...............................................................95
Table 26. (NC) Cox regression probability of termination over time, by caseload ........................................................96
Table 27. (NC) Regression results, terminated cases by circuit, no claims model. .........................................................97
Table 28. (NC) Regression results, terminated cases by circuit, claims model ..............................................................98
Table 29. (NC) Regression results, terminated cases by circuit, no claims model .........................................................99
Table 30. (NC) Regression results, terminated cases, by circuit, claims model .............................................................100
Table 31. (NC) Regression results, terminated cases, by clerk, no claims model ..........................................................101
Table 32. (NC) Regression results, terminated cases, by clerk, claims model ...............................................................102
Table 33. (NC) Regression results, terminated cases, by caseload, no claims model......................................................103
Table 34. (NC) Regression results, terminated cases, by caseload, claims model...........................................................104
Table 35. (NC) Regression results, terminated cases, by district, claims model, with and without caseload..................105
Table 36. (NC) Regression of disposition time, by district, no claims model, with and without caseload......................107
Table 37. (C) Cox regression probability of termination over time, by district, with and without caseload ...................108

Final Technical Report: Habeas Litigation in U.S. District Courts

LIST OF FIGURES
(C) designates capital cases; (NC) designates non-capital cases

Figure 1. (NC) Percentage of cases in sample, by state..................................................................................................16
Figure 2. (C) Days from first docket entry to petition, by district .................................................................................24
Figure 3. (C) Days from first docket entry to counseled petition, by district ................................................................25
Figure 4. (C) Claims per case, by district ......................................................................................................................28
Figure 5. (C) Percentage of cases with at least one claim of innocence, by district ......................................................29
Figure 6. (C) Length of stays for exhaustion, by district ...............................................................................................32
Figure 7. (C) Length of stays for exhaustion, including pending cases, by district .......................................................33
Figure 8. (C) Percentage of cases pending as of November 27, 2006, by district .........................................................36
Figure 9. (NC) Percentage of cases pending, by districts with over 20 cases.................................................................37
Figure 10. (C) Docket entries per case, by district ........................................................................................................38
Figure 11. (C) Disposition time ......................................................................................................................................40
Figure 12. (C) Disposition time, by district .....................................................................................................................40
Figure 13. (NC) Disposition time ...................................................................................................................................41
Figure 14. (C) Length of case in days, pending cases included ....................................................................................42
Figure 15. (C) Length of case in days, pending cases included, by district...................................................................42
Figure 16. (NC) Length of case in days, pending cases included ....................................................................................43
Figure 17. (NC) Length of case in days, pending cases included, by district ................................................................44
Figure 18. (C) Length of case in days, less stayed time, by district ..............................................................................45
Figure 19. (C) Percentage of cases with a defaulted claim, by district ..........................................................................48
Figure 20. (C) Percentage of cases with a Teague- barred claim, by district .................................................................49
Figure 21. (NC) Probability of not reaching termination, over time, in days ...............................................................68
Figure 22. (C) Length of case in days, by district, box chart.........................................................................................77

Final Technical Report: Habeas Litigation in U.S. District Courts

PART I. INTRODUCTION_________________________________________________________

The purpose of this study is to provide empirical information about habeas corpus cases filed by
state prisoners in U.S District Courts under the Antiterrorism and Effective Death Penalty Act of 1996
(“AEDPA”). This information is for policymakers who design or assess changes in habeas law, for litigants
and courts who address the scope and meaning of the habeas statutes, and for researchers who seek
information concerning the processing of habeas petitions in federal courts.
This introduction first describes federal habeas review and the current statutory scheme. It then
outlines five categories of empirical information about habeas review that are examined in this study.
Finally, this introduction summarizes the basic features of study design and methodology.

Federal habeas review and the current statutory scheme

The writ of habeas corpus is a remedy regulated by statute and available in federal court to persons
“in custody in violation of the Constitution . . . .” 1 Most state prisoners who seek federal habeas relief
challenge their state criminal convictions or sentences in federal court under 28 U.S.C. § 2254, after they
have already lost their attempts to secure relief in state court through whatever appeal and post-conviction
proceedings are provided to them by state law. When a federal court grants a writ of habeas corpus, it
orders the state court to release the prisoner, or to repeat the trial, sentencing, or other proceeding that led to
the prisoner’s custody. This order can come years after the prisoner’s original conviction and sentence, and
may be based on new grounds or on claims that the state courts had previously concluded did not warrant
relief. It is not surprising, then, that habeas review has long been a controversial component of the
relationship between the federal courts and state governments.
The current habeas provisions are a legislative response to several decades of change in the federal
oversight of state criminal proceedings. In the 1950s, 1960s, and early 1970s, the U.S. Supreme Court
interpreted the Due Process Clause of the Fourteenth Amendment to guarantee to state criminal defendants
many of the procedural protections that had previously been enjoyed only by defendants in federal criminal
proceedings. At the same time, the Court expanded the scope of the writ, allowing more opportunities for
state prisoners to obtain relief in federal court when actions of state police, prosecutors, and judges violated
their constitutionally protected rights. In combination, these two trends produced an explosion in habeas
filings. 2
By the 1980s, the Supreme Court's decisions began to narrow access to habeas relief for state
prisoners. For example, the Court limited the circumstances under which a judge could grant the writ based
on a claim that the state prisoner had failed to raise properly in state court, a defense to habeas relief known
1

28 U.S.C. § 2241. The statutory provisions governing habeas relief for both federal and state prisoners, 28 U.S.C. §§
2241- 2264, were supplemented by the Rules Governing Section 2254 Cases in the United States District Courts and
Rules Governing Section 2255 Cases in the United States District Courts, effective February 1, 1977.
2
See, e.g., David L. Shapiro, Federal Habeas Corpus: A Study in Massachusetts, 87 HARV. L. REV. 321, 321-24 (1973)
(collecting authority); Report to the Attorney General on Federal Habeas Corpus Review of State Judgments, 22 U.
MICH. J. L. REF. 901, 946-47 (1988-89) (number of state prisoner applications grew from about 1000 in 1961 to nearly
10,000 in 1987); Christopher E. Smith, Judicial Policy Making and Habeas Corpus Reform, 7 CRIM. J. POLICY REV. 91,
96 (1995) (filings went from less than 900 in 1960 to over 13,000 in 1993); Fred L. Cheesman II, Roger A. Hanson, and
Brian J. Ostrom, A Tale of Two Laws: The U.S. Congress Confronts Habeas Corpus Petitions and Section 1983
Lawsuits, 22 LAW & POLICY 89, 93 (2000) (graph showing how growth in habeas petitions outpaced increases in prison
population).

Final Technical Report: Habeas Litigation in U.S. District Courts

as "state procedural default." 3 In addition, the Court held in Teague v. Lane 4 that a federal court must
evaluate a state court decision based upon the federal law that existed at the time that decision was made;
federal courts could not grant the writ because of a state court's failure to apply a later interpretation of the
Constitution that was more favorable to a criminal defendant. 5
For a majority of the members of Congress in the early 1990s, the Court’s decisions did not
adequately address growing concerns about federal court interference with the finality of state criminal
judgments and about delay in the processing of habeas cases. After considering various legislative
proposals for years, Congress passed the Antiterrorism and Effective Death Penalty Act (“AEDPA”) in
1996, limiting federal habeas review. 6 Although more than a decade has elapsed since the enactment of
AEDPA, no research has examined the processing of cases under the Act's provisions.
As its title suggests, AEDPA included provisions that address death penalty cases in particular.
Responding to what a House report characterized as "acute problems of unnecessary delay and abuse in
capital cases," 7 the Act included new time periods during which a federal court must conclude its review of
a habeas petition that has been filed by a state prisoner sentenced to death. 8 These special disposition
deadlines for capital habeas cases have yet to be enforced. The statute conditions their application upon a
prior finding that the state has met certain standards for the provision of competent counsel to capital
defendants in state post-conviction proceedings. No state has yet met the specified standards. In 2006 this
portion of the statute was amended. The amendments shifted to the Attorney General of the United States
the authority to determine if a state has established a program adequate to trigger the disposition deadlines. 9
They also changed the disposition deadline itself so that district courts have 450 days to dispose of a capital
habeas case from start to finish.
The remaining provisions of the Act govern all habeas cases, non-capital as well as capital. Included
were the following five changes:
*
The Act established a statute of limitations for filing. Under the Act, a state prisoner must seek
relief in federal court within one year of the conclusion of either the direct appeal of his state judgment or the
expiration of time for seeking such appeal. 28 U.S.C. § 2244(d)(1)(A). If the prisoner promptly seeks
collateral review of his judgment in state court before coming to federal court, the limitations period will be

See Coleman v. Thompson, 501 U.S. 722 (1991); Schlup v. Delo, 513 U.S. 298 (1995).
489 U.S. 288 (1989).
5
See also Stone v. Powell, 428 U.S. 465 (1976) (habeas review unavailable to state prisoners who alleged that their
custody was the product of unreasonable searches or seizures); Brecht v. Abrahamson, 507 U.S. 619 (1993) (standard of
review for harmlessness on habeas); John H. Blume, AEDPA: The “Hype” and the “Bite,” 91 CORNELL L. REV. 259,
266-68 (2006) (summary of 10 decisions between 1976 and 1995, characterized as “cutbacks on habeas corpus”).
6
Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), Pub. L. No. 104-132, 110 Stat. 1214 (1996).
See, e.g., S. REP. NO. 104-23 (1995); Ad Hoc Committee on Federal Habeas Corpus in Capital Cases, Report on Habeas
Corpus in Capital Cases, 45 CRIM. L. RPTR. 3239 (BNA) (Sept. 27, 1989) ("Powell Report") (proposing statutory
revisions imposing filing time limits and reporting finding that 80% of the time spent in collateral review of death
penalty cases occurs outside of state proceedings); Habeas Corpus Legislation, Hearings on H.R. 4737, H.R. 1090, H.R.
1953, and H.R. 3584, Before the Subcommittee on Courts, Intellectual Property, and the Administration of Justice of the
H. Committee on the Judiciary, 101st Congress, No. 145 (May 24 and June 6, 1990).
7
H. CONF. REP. 104-518, at 111 (1996). See also Alex Kozinski & Sean Gallagher, Death: The Ultimate Run-On
Sentence, 46 CASE W. RES. L. REV. 1 (1995-1996).
8
28 U.S.C. §§ 2261-2266.
9
28 U.S.C. § 2266(b)(1)(A), effective March 9, 2006.
4

Final Technical Report: Habeas Litigation in U.S. District Courts

tolled until that collateral review in the state courts has been concluded. 10 Prior to the Act, federal courts
were authorized by Rule 9 of the Rules Governing Section 2254 cases to dismiss a habeas petition that was
filed long after the conviction and sentence, but only if the court concluded that 1) the petitioner previously
knew or should have known of the existence of the grounds raised in the petition, and 2) the delay resulted in
the state being prejudiced in its ability to respond to the petition. 11
*
The Act authorized federal judges to deny on the merits any unexhausted claim. An unexhausted
claim is a claim that the petitioner failed to present to the state courts for decision before including it in a
federal petition. 28 U.S.C. § 2254(b)(2). Prior law required federal courts to dismiss an unexhausted claim
without prejudice. The petitioner then could litigate that claim in state court, and subsequently file anew in
federal court.
*
Habeas petitions are not resolved by trial. If fact-finding is required, the court may hold an
evidentiary hearing. The Act prohibited a federal judge from holding an evidentiary hearing in a habeas
case when the petitioner had failed to develop the facts in state court unless the facts supporting the claim
would establish the petitioner’s innocence of the underlying offense, and the claim relies on either 1) a new
rule of criminal procedure that the Supreme Court has decided must apply retroactively, or 2) factual
information that the petitioner could not have discovered earlier. 28 U.S.C. § 2254(e). 12 Restrictions on
evidentiary hearings before AEDPA were not nearly as exacting. 13
*
The AEDPA restricted the circumstances under which a federal court is permitted to entertain a
successive petition. It requires prior authorization from the court of appeals, and bars entirely federal court
review of any claim that a petitioner had included in a prior petition. 28 U.S.C. § 2244(b).
*
The Act also mandated a new standard of review for evaluating state court applications of
constitutional law and determinations of fact. The new standards require federal courts to give greater
deference to state decisions. The Act prohibits federal judges from granting relief for any claim adjudicated
on the merits in state court unless the state decision rejecting the claim is 1) “contrary to, or involved an
unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the
United States,” or is 2) “based on an unreasonable determination of the facts in light of the evidence presented
in the State court proceeding.” 28 U.S.C. § 2254(d). 14

Research goals of the study

________________________________________________

"The adjudication of habeas corpus applications lies at perhaps the lowest visibility level of any of the processes of the
federal district courts.” 15

Since the enactment of AEDPA, its provisions have been continually litigated. Each year, more than
18,000 cases, or one out of every 14 civil cases filed in federal district courts, are filed by state prisoners
10

The statute provides that the prisoner's effort to secure state collateral review must be "properly filed" before it will
stop the clock on the one-year period for filing in federal court. 28 U.S.C. § 2244(d)(2). See also Evans v. Chavis, 546
U.S. 189 (2006); Carey v. Saffold, 536 U.S. 214 (2002).
11
See Day v. McDonough, 547 U.S. 198, n.1 (2006); Fed. R. § 2254 Cases 9, Committee Note (1976).
12
Williams v. Taylor, 529 U.S. 362 (2000). When there has been no such failure to develop facts in state court under §
2254(e), the district court must exercise its discretion in deciding whether to grant an evidentiary hearing. If the existing
"record refutes the applicant's factual allegations or otherwise precludes habeas relief, a district court is not required to
hold an evidentiary hearing." Schriro v. Landrigan, 127 S. Ct. 1933, 1940 (2007).
13
See Keeney v. Tamayo-Reyes, 504 U.S. 1 (1992).
14
Williams v. Taylor, 529 U.S. 420 (2000).
15
Shapiro, supra note 2, at 337.

Final Technical Report: Habeas Litigation in U.S. District Courts

seeking habeas corpus relief, and more than 6000 of these cases reach the courts of appeals. 16 During each
term, the Supreme Court devotes a considerable portion of its limited docket to the application or
interpretation of AEDPA. Habeas litigation is also a recurring topic of legislative attention. 17
Despite the salience of contemporary habeas litigation to policy makers, little empirical information
about it exists. Diverse groups including the United States Judicial Conference, 18 the Conference of Chief
Justices, the Conference of State Court Administrators, 19 and the American Bar Association, 20 have
recognized the need for empirical study of litigation under AEDPA to assist courts and legislators as they
continue to craft and evaluate changes in habeas law. The empirical questions and assumptions about habeas
litigation that have been raised in debates about the appropriate scope of habeas review provide the starting
point for the research questions in this study. These issues fall into five general categories:
1) Time before filing. Concern about undue delay in seeking habeas relief led to the provision in
AEDPA establishing a statute of limitations for filing. The limitations provision remains
controversial. It is not known whether it has made any difference in the length of time that elapses
before federal filing nor how frequently it is applied to prevent review of claims on the merits.
2) Claims for relief. One feature of several current as well as proposed procedural rules for habeas
review is differential treatment based on some showing of actual innocence by the petitioner. 21 No
information is available, however, concerning the frequency or success of these claims, either as an
independent basis for relief, 22 or as a means of avoiding a procedural hurdle. 23 It is also unknown
how often petitioners challenge only their sentences and not their convictions.
16

Administrative Office of the Courts, Judicial Business of the U.S. Courts 2006, Tables B1-A, C-2.
For recent debates over amendments to the habeas statutes, see Legislative Hearing on H.R. 3035, the "Streamlined
Procedures Act of 2005," before the Subcommittee on Crime, Terrorism, and Homeland Security, November 10, 2005,
http://judiciary.house.gov, hereinafter House Hearings; Hearing on “Habeas Corpus Proceedings and Issues of Actual
Innocence,” July 13, 2005, before the Committee on the Judiciary of the U.S. Senate,
http://judiciary.senate.gov/hearing.cfm?id=1569 (hereinafter “Senate Hearings”).
18
Judicial Conference of the United States, Leonidas Ralph Mecham, Secretary, Letter to Honorable Arlen Specter,
Chairman, Committee on the Judiciary, United States Senate, dated September 26, 2005 (hereinafter Judicial Conference
Letter).
19
Conference of Chief Justices and the Conference of State Court Administrators, Joint Resolution 16 “In Support of
Gathering Further Information Concerning the Effects of the Anti-Terrorism and Effective Death Penalty Act of 1996 to
Determine Whether Amendments are Needed, adopted August 3, 2005, access at
http://ccj.ncsc.dni.us/CriminalAdultResolutions/resol16EffectsAntiTerrorismEffectiveDeathPenaltyActof1996.html.
20
See also Anne M. Voigts, Narrowing the Eye of the Needle: Procedural Default, Habeas Reform, and Claims of
Ineffective Assistance of Counsel, 99 COLUMB. L. REV. 1103, 1109 (1999) (noting “few empirical studies . . . despite the
fact that many of the most hotly-debated issues in the debate over habeas involve empirical questions”).
21
See S. 1088, 109th Cong. (2005)(proposing special rule for innocence claims based on DNA); Brian M. Hoffstadt, How
Congress Might Redesign a Leaner, Cleaner Writ of Habeas Corpus, 49 DUKE L. J. 947 (1999-2000) (listing specific
claims that should be cognizable in habeas cases and others that should not). In recent years, exonerations have
generated for some a heightened concern about the ability of the criminal justice system, including habeas review, to
prevent and remedy inaccurate convictions. See Joshua Marquis, The Myth of Innocence, 95 J. CRIM. L. & CRIMINOLOGY
501 (2005); Samuel R. Gross et al., Exonerations in the United States 1989 Through 2003, 94 J. CRIM. L. &
CRIMINOLOGY 523 (2005).
22
See Herrera v. Collins, 506 U.S. 390 (1993).
23
See, e.g., Senate Hearings, supra note 17. Compare Thomas Dolgenos, Testimony before the Committee on the
Judiciary Subcommittee on Crime, Terrorism, and Homeland Security, House of Representatives, November 10, 2005, at
3 (“Almost every habeas petitioner claims that he is innocent.” If the “innocence standard is set too low . . . courts will
be deluged with dubious claims . . . .”) with Seth Waxman, Testimony before the Committee on the Judiciary
Subcommittee on Crime, Terrorism, and Homeland Security, House of Representatives, November 10, 2005, at 6 (“It is
often only after discovery is granted, or a procedurally defaulted claim is heard, that evidence of innocence emerges.”)
17

Final Technical Report: Habeas Litigation in U.S. District Courts

3) The operation of defenses.
Statute of limitations. The ongoing controversies about the limitations provision would benefit from
accurate information about how much time is elapsing between the state judgment and the federal
filing, which cases take longer to get to federal court, how often cases are dismissed as time-barred,
how many cases involve a ruling that tolls the statute of limitations, and for what reasons. 24
Exhaustion of claims in state court. The exhaustion requirement has resulted in several divided
Supreme Court decisions, 25 as well as proposed amendments in recent bills. However lawmakers
have as yet no up-to-date information about how often petitions include unexhausted claims or what
courts presently do with those claims. 26 It is not known how often courts use a “stay and abeyance”
procedure to allow a petitioner to pursue unexhausted claims in state court without risking a late
filing, 27 or the length of such stays.
Successive petitions. Among the most controversial provisions in AEDPA are those barring
petitioners from returning to federal court with new constitutional claims, but there is no information
available about how many repeat petitions are turned away or what claims are not reviewed as a
result.
Procedural default. Some consider the default defense too generous to petitioners, others consider it
too restrictive. 28 No information exists about how often the defense is employed to bar review, nor
how many defaulted claims are addressed because the judge has concluded that the state procedure

(http://judiciary.house.gov/hearings.aspx?ID=128). See also Nicholas Berg, Turning a Blind Eye to Innocence – The
Legacy of Herrera v. Collins, 42 AM. CRIM. L. REV. 121 (2005) (claiming that based on a review of published decisions
since Herrera, at least 173 petitioners have raised bare-innocence claims, and only a handful received relief).
24
See, e.g., Pace v. DiGugliemo, 544 U.S. 408 (2005) (declining to decide whether “equitable tolling” is permissible);
Lawrence v. Florida, 127 S. Ct. 1079 (2007) (same, but assuming the availability of equitable tolling for purposes of
decision, also finding “attorney miscalculation is simply not sufficient to warrant equitable tolling, particularly in the
postconviction context where prisoners have no constitutional right to counsel”). Compare H.R. 3035, 109th Cong.
(2005) (bill that would bar equitable tolling) with Limin Zheng, Actual Innocence as a Gateway Through the Statute of
Limitations Bar on the Filing of Federal Habeas Corpus Petitions, 90 CAL. L. REV. 2103 (2002) (supporting an actual
innocence exception to the statute of limitations, referencing pre-AEDPA empirical research). See also Blume, supra
note 5, at 289 n. 143 ("Due to the large number of unpublished district court orders dismissing habeas petitions as
untimely, many of which are not appealed, it is impossible to say with precision how many petitions have been deemed
untimely. However, the number is definitely in the thousands."); Bryan A. Stevenson, Confronting Mass Imprisonment
and Restoring Fairness to Collateral Review of Criminal Cases, 41 HARV. C.R.-C.L. L. REV. 339, 349, 358 (2006)
(stating AEDPA's statute of limitations "has barred thousands of prisoners from review of their constitutional claims
because, without counsel, they could not timely file their pleadings," and that "[r]eform is absolutely critical if
meaningful remedies are going to reduce the number of innocent and wrongly convicted people currently in prison").
25
E.g., Rhines v Weber, 544 U.S. 269 (2005); Lawrence v. Florida, 127 S. Ct. 1079 (2007).
26
E.g., Duncan v. Walker, 533 U.S. 167 (2001) (Breyer, J., dissenting) (citing Roger A. Hanson & Henry W.K. Daley,
Federal Habeas Corpus Review: Challenging State Court Criminal Convictions, Washington, D.C., U.S. Department of
Justice, Bureau of Justice Statistics (Sept. 1995 NCJ 155504), http://www.ojp.usdoj.gov/bjs/pub/pdf/fhcrcscc.pdf).
27
For a sampling of references to these issues, see Rhines v Weber, 544 U.S. 269 (2005); Pace v. DiGuglielmo, 544 U.S.
408, 429 (2005) (Stevens, J., dissenting) (stay and abeyance procedure will result in “a flood of protective filings in the
federal district courts”); Duncan, 533 U.S. at 192 (2001) (Breyer, J., dissenting) (stay and abeyance will ameliorate
unfairness but will add to the burdens on district courts); Dolgenos Testimony, supra note 23, at 8 (“many such stay
orders [are] issued . . . ensuring years of new delays”).
28
See, e.g., House Hearings, supra note 18; Stevenson, supra note 24, at 350 (stating that "in death penalty cases, the
great majority of substantive claims alleging constitutional violations . . . are procedurally barred").

Final Technical Report: Habeas Litigation in U.S. District Courts

isnot “adequate” or “independent,” 29 that the petitioner has established “cause” and “prejudice,” or
that enforcing default would produce a “miscarriage of justice.” 30
4) Time for processing. The 1996 Act was prompted in part by frustration over delays that
accompanied the review of state criminal judgments in federal court, particularly in capital cases. 31
Delay in the processing of habeas cases was again the topic of attention in the most recent round of
legislative proposals. 32 The only information available concerning processing times after the
enactment of AEDPA is that reported by the Judicial Conference of the United States. In the summer
of 2005, it reported to the Senate Judiciary Committee that the median time from filing to disposition
was six months for non-capital habeas cases in district courts in 2004, but was more than 24 months
in capital cases. As of 2004, the Judicial Conference reported, about 46% of state capital habeas
petitions had been pending for more than three years. 33
5) Merits review and case outcomes. Perhaps no empirical question about the habeas remedy is as
important as how many cases end in a grant of relief for the petitioner. Surprisingly, the answer after
AEDPA is unknown. Nor is it known what proportion of cases involve evidentiary hearings, what
percentage of claims are addressed on their merits, or how often the new deferential review of 28
U.S.C. § 2254(d) is being applied. 34 Without this information, lawmakers and courts have little basis
on which to evaluate efforts to expand or restrict the scope of habeas review.

See Dolgenos Testimony, supra note 23, at 10-11. For an overview of state procedural default, see WAYNE R. LA
FAVE, JEROLD H. ISRAEL & NANCY J. KING, 6 CRIMINAL PROCEDURE § 28.4 (2d ed. 2004).
30
No research exists to assist courts in testing assumptions about the use of procedural default such as “habeas corpus
petitions that advance a substantial claim of actual innocence are extremely rare,” Schlup v. Delo, 513 U.S. 298 (1995).
In addition to these defenses, we examined the application of Teague v. Lane, 489 U.S. 288 (1989), to bar review, and
also whether habeas review had been waived as part of a plea agreement. See Nancy J. King & Michael E. O’Neill,
Appeal Waivers and the Future of Sentencing Policy, 55 DUKE L. J. 209 (2005); Anup Malani, Habeas Settlements, 92
VA. L. REV. 1 (2006).
31
“One of the statute's purposes is to ‘reduce delays in the execution of state and federal criminal sentences, particularly
in capital cases.’” Rhines v. Weber, 544 U.S. 269, 276 (2005) (quoting Woodford v. Garceau, 538 U.S. 202, 206 (2003).
See also Stephen J. Spurr, The future of capital punishment: determinants of the time from death sentence to execution,
22 INT’L REV. OF ECON. 1 (2002) (analyzing review time through 1997 for death row inmates, noting “a widely held
belief that the lower federal courts . . . are responsible for a substantial portion of the delay in death penalty litigation”).
This remains a key concern behind recent legislative proposals. See Lisa M. Seghetti & Nathan James, Federal Habeas
Corpus Relief: Background, Legislation, and Issues, Congressional Research Service (February 1, 2006).
http://www.opencrs.com/rpts/RL33259_20060201.pdf.
32
See HR 3035, 109th Cong. (2005).
33
Judicial Conference Letter, supra note 18.
34
E.g., Schriro v. Landrigan, 127 S. Ct. 1933, 1954 (2007) (Stevens, J., dissenting) ("habeas cases requiring evidentiary
hearings have been ‘few in number’”); Carol S. Steiker & Jordan M. Steiker, A Tale of Two Nations: Implementation of
the Death Penalty in "Executing" Versus "Symbolic" States in the United States, 84 TEX. L. REV. 1869, 1902 (2006)
(claiming district courts in Ninth Circuit are 50% more likely than those in Fifth Circuit to hold evidentiary hearings);
Stevenson, supra note 24, at 350 (claiming “most prisoners’ complaints about wrongful convictions, illegal sentences,
and other errors for which there is a constitutional remedy are never addressed on the merits”).

Final Technical Report: Habeas Litigation in U.S. District Courts

Prior research

Post-AEDPA: The only empirical information concerning habeas litigation under AEDPA presently
available is in the annual civil case datasets compiled and maintained by the Administrative Office of the
United States Courts (AO), available at www.icpsr.umich.edu. These datasets include, for each case, filing
and termination dates, disposition information, whether the petitioner had counsel, and if the petitioner
received in forma pauperis status. Two previous studies used the AO data to investigate filing rates in habeas
cases before and after AEDPA, 35 but they did not investigate what happened to these cases after they were
filed. Although the variables in the AO data report four types of dismissals and judgments at various stages,
few of these options resemble what actually happens in habeas cases. 36 Moreover, information about which
party prevailed is missing for most habeas cases. 37 The values for filing and termination dates do provide a
rough baseline for predictions of processing time. They indicate that most non-capital cases take less than a
year to complete, while a large proportion of capital cases take more than three years to complete. 38 A new
comprehensive study of processing time in capital habeas cases by the Federal Judicial Center is underway,
but has not yet been completed.
Pre-AEDPA: We know more about the processing of habeas cases prior to AEDPA. Using data
collected from non-capital cases that were decided in the early 1990s, two studies provide a baseline against
which the findings of this study can be compared. Hanson and Daley examined over 2000 habeas cases
terminated in the year 1992 under pre-AEDPA law, collecting data from court documents at courthouses in 18
districts in nine states. 39 They reported disposition time, offense and sentence, representation, frequency and
disposition of claim types, type of disposition, and reasons for dismissal. Flango examined over 1600 cases
that were terminated in 1990 and 1992 in eight districts from four states, and reported type of offense,
sentence, and method of conviction, representation, claims raised, prior petitions, rates of relief, and reasons
for dismissal. 40
For capital habeas cases prior to AEDPA, two studies provide some baseline information. In 1995 the
Federal Judicial Center reported on disposition times in district courts for capital habeas cases which had been
35

Fred L. Cheesman II, Brian J. Ostrom & Roger A. Hanson, A Tale of Two Laws Revisited: Investigating the Impact
of the Prisoner Litigation Reform Act and The Antiterrorism and Effective Death Penalty Act (2004), Williamsburg, VA,
National Center for State Courts, NIJ 2001-IJ-CX-0013,
http://www.ncsconline.org/WC/Publications/Res_PreCiv_TwoLawsRevPub.pdf; John Scalia, Prisoner Petitions Filed in
U.S. District Courts, 2000, with Trends 1980-2000, Washington, D.C., U.S. Department of Justice, Office of Justice
Programs (Jan. 2002 NCJ 189430), http://www.ojp.usdoj.gov/bjs/pub/pdf/ppfusd00.pdf. See also Cheesman et al., supra
note 2.
36
Administrative Office of the United States Courts, Civil Statistical Reporting Guide, Version 2.1 (1999). For example,
the variable "procedural progress" appears to be of little if any value in habeas cases. We examined the AO data for the
25 capital cases in our sample that we found had been terminated after evidentiary hearings. In the AO data, 7 of these
cases were missing any value for this variable, another seven indicated "judgment on a motion," three had "no court
action," three were "order entered," or "order decided," another three “other,” and one marked "pretrial conference."
37
For this reason, it is risky to attempt to draw conclusions about habeas litigation in the district courts based on the AO
data. See Blume, supra note 5, at 284-85 (noting rates of success in courts of appeals, but also detailing reliability
problems with district court data).
38
Judicial Conference Letter, supra note 18.
39
Hanson & Daley, supra note 26.
40
Victor E. Flango & Patricia McKenna, Federal Habeas Corpus Review of State Court Convictions, 31 CAL. W. L.
REV. 327 (1994-1995); VICTOR E. FLANGO, HABEAS CORPUS IN STATE AND FEDERAL COURTS (1994) (National Center
for State Courts, SJI-92-14M-B-055). Studies conducted even earlier included Richard Faust, Tina J. Rubenstein & Larry
W. Yackle, The Great Writ in Action: Empirical Light on the Federal Habeas Corpus Debate, 18 N.Y.U. REV. L. &
SOC. CHANGE 637 (1990-1991); Paul H. Robinson, An Empirical Study of Federal Habeas Corpus, Review of State
Court Judgments, Washington D.C.: U.S. Department of Justice, Federal Justice Research Program Office for
Improvements in the Administration of Justice, JADAG-79-C-0002 (July 1979).

Final Technical Report: Habeas Litigation in U.S. District Courts

terminated between July 1988 to September 1994, measuring from file date to termination date. 41 Examining
all 500 terminated petitions (excluding 264 cases that were still pending), the mean disposition time was 15
months, the median was nine months. Among districts with ten or more capital petitions, AL-N and AR-E
averaged the longest processing time, with TX-S and FL-N the shortest. First petitions averaged 17 months
for disposition time. Sixty-nine of the 413 death row inmates represented in the disposition data (17%) filed
more than one petition. Among the 634 death row inmates represented in the 764 pending and terminated
cases, 87 had filed more than one petition (14%).
A team of six researchers reported in 2004 their examination of federal habeas review of cases filed by
petitioners who were sentenced to death in 1990 or earlier and who had completed federal review by 1995. 42
Examining case-level information from published decisions, the authors evaluated the effect of various factors
on the probability of relief in the federal courts. For most cases this meant the final outcome at the court of
appeals level. For cases receiving relief in federal courts, invalid instructions were the most frequent reason
given for overturning a death sentence, followed by the denial of the effective assistance of counsel, and then
prosecutor and police misconduct (including Brady and Batson claims).Whether the defendant’s lawyer was
from out of state was significantly associated with relief, as was the existence of an evidentiary hearing on
state collateral review, the presence of more aggravating factors at sentencing, and the granting of a
evidentiary hearing in federal court. The number of claims raised was also significant. A greater number of
claims lowered the probability of relief. This study also found that, on average, cases in which the
petitioner’s conviction or sentence was invalidated in federal court took two years longer than cases in which
the petitioner was denied relief. 43
D.

Study design and methodology

Information sought.
Descriptive. Because this is the first empirical examination of the application of AEDPA, the most
important contribution of the present study may be to provide a thorough catalogue of descriptive information
about habeas litigation. For each of the five general features of habeas review outlined above, the study
collected detailed information for both capital and non-capital cases. Part II reports these descriptive findings,
as well as additional descriptive information about other features of habeas litigation that were collected in
order to conduct the comparative and explanatory analyses summarized below.
41

Scott Gilbert & Patricia Lombard, A Report to the Conference of Chief Circuit Judges and Circuit Executives: An
Analysis of Disposition Times for Capital Habeas Corpus Petitions, Federal Judicial Center (Sept. 1, 1995).
42
Jeffrey Fagan, James S. Liebman, Valerie West, Andrew Gelman, Alexander Kiss & Garth Davies, Getting to Death:
Fairness and Efficiency in the Processing and Conclusion of Death Penalty Cases after Furman, Final Technical Report,
Dept. of Justice Document No. 203935, Award Number 2000-IJ-CX-0035 (Feb. 2004). No cases with sentences later
than 1990 had yet been reviewed by federal courts at the time of the study. Of the 596 federal habeas outcomes
examined, only 14 cases involved sentences imposed in 1987 or later. Id. at 40, Table 11. See also Andrew Gelman,
James S. Liebman & Valerie West, A Broken System: The Persistent Patterns of Reversals of Death Sentences in the
United States, 1 J. OF EMPIRICAL LEG. STUD. 209-261 (Issue 2, July 2004) (reporting on same study, 240 invalidations in
598 federal cases); James S. Liebman, Jeffrey Fagan, Andrew Gelman, Valerie West, Garth Davies & Alexander Kiss, A
Broken System, Part II: Why There Is So Much Error in Capital Cases, and What Can Be Done About It, Feb. 11, 2002.
43
See also Barry Latzer & James Cauthen, Justice Delayed? Time Consumption in Capital Appeals: A Multistate Study
(March 2007 NCJ 217555, 2004-IJ-CX-0005), http://www.ncjrs.gov. The authors examined processing time for state
direct appeals of capital cases in 14 states. Processing time was not related to how many staff attorneys supported the
court, and caseload was only slightly influential. Cases took significantly less time when the appellate court affirmed
rather than reversed the trial court’s capital judgment, when the state had a single level of appellate review, or when the
state had a legal rule directed to increasing appellate efficiency in capital cases. Spurr, supra note 31, studied time from
sentence to removal from death row through 1997, but did not separate out time spent in state court from time in federal
court. Spurr predicted that after AEDPA, the overall time for processing capital cases through execution would continue
to decline.

Final Technical Report: Habeas Litigation in U.S. District Courts

Comparative. Part III of this report first builds upon research prior to AEDPA to examine the
differences between pre-AEDPA and post-AEDPA processing of habeas cases in order to identify the
possible effects of the 1996 amendments. Second, because so much of habeas policy is driven by capital cases
(although capital cases make up only a tiny portion of all habeas filings), the report also examines the
differences between the processing of capital and non-capital habeas cases.
Explanatory. The study reports in Part IV additional explanatory analyses of the data directed at these
particular questions: 1) For capital cases, what features are associated with longer periods before filing? 2)
For both capital and non-capital cases, what features are associated with longer processing time? and 3) For
capital cases, what features are associated with the likelihood of relief?
Sampling strategy.
We elected to examine cases that had been filed no earlier than 2000, rather than select a sample of
cases terminated during a specified period. This strategy has three advantages. First, it limited the cases in
the study to cases processed after the Supreme Court had settled some of the more fundamental questions
about the application of AEDPA that had divided lower courts through 1999. Second, it provided a
consistent cohort of observations for predicting the likelihood of termination and the likelihood that the writ
would be granted. Third, it facilitated the collection of much of the data from the internet, as explained
below. To select cases, we first merged the AO data sets reporting district court civil terminations from 2000
through 2005 and civil cases pending in district court in 2005. This merged set was then reduced to only
those cases that were coded by the AO as general or capital habeas cases that had federal question as the basis
for jurisdiction. This excluded cases suits against federal authorities.
As the sole method of data collection for non-capital cases, we elected to examine documents posted on
PACER, the on-line filing system of the federal courts. Availability of documents on PACER for cases that were
started in 2002 or earlier was insufficient, so we limited our sample to cases begun in 2003 and 2004. 44 Because
capital cases take several years to complete, however, the capital case sample required that cases started prior to
2002 be included, cases for which PACER would not provide adequate information. Therefore, capital case data
was collected from original case documents at courthouses and federal archives, as well as from online research.
Non-capital sample. To construct the non-capital case sample, we selected at random from the merged set
described above, 7.5% of the general state habeas cases which had been filed in calendar year 2003 (with docket
numbers beginning “03”), and 7.5% of those started in 2004. We selected cases by docket number rather than file
date, because we learned that the file date recorded in the AO data was often later, sometimes much later, than the
date the case began. From this group, we deleted duplicate cases with the same docket sheet, cases that turned out
to be mislabeled as state non-capital habeas cases, 45 and cases that despite their docket numbers were actually
started prior to 2003 or after 2004. Our final sample numbered 2384 non-capital cases, approximately 6.5% of all
36,745 non-capital 2003-2004 habeas filings by state prisoners during the sample period. The geographic
distribution of cases in the sample is consistent with the geographic distribution of the 36,745 cases. The number
of cases from California, Texas, Florida, and New York reflect the large prison populations in those states. See
Figure 1.

Research has only recently used the PACER method as an increasing number of districts have posted more case
documents on line. E.g., Gillian Hadfield, Where Have All the Trials Gone? Settlements, Non-trial Adjudications and
Statistical Artifacts in the Changing Disposition of Federal Civil Cases, 1 J. OF EMPIRICAL LEG. STUD. 705 (2004).
45
Mislabeled cases included those filed by prisoners challenging federal detention, a death sentence, or prison
conditions.

Final Technical Report: Habeas Litigation in U.S. District Courts

Figure 1. (NC) 46 Percentage of cases in sample, by state.
California
Texas
Florida
New York
Pennsylvania
Michigan
Indiana
Virginia
Georgia
Louisiana
Missouri
Alabama
Mississippi
Illinois
Oklahoma
North Carolina
Ohio

State Containing District

South Carolina
Tennessee
Wisconsin
New Jersey
Colorado
Oregon
Washington
Nevada
Arizona
Maryland
Iowa
Arkansas
Minnesota
Kansas
Kentucky
Massachuetts
New Hampshire
Connecticut
Montana
West Virginia
Delaware
Idaho
Utah
North Dakota
Wyoming
Nebraska
Maine
South Dakota
Puerto Rico
Rhode Island
Vermont
0

Percentage of Sample

Capital case sample. Because of the need to read documents in capital cases on site, it was not feasible to
code a random nationwide sample. Instead, the capital case sample was drawn from districts with the highest
volume of capital habeas filings. According to the annual reports of the AO, in only thirteen districts were 20 or
more capital habeas cases filed during the years 2001-04 (the four most recent years’ figures available at the time
of study design): TX-S, TX-E, TX-N, TX-W, PA-E, OH-N, OH-S, CA-C, AZ, NV, AL-N, FL-M, and OK-W.
46

Throughout the report, Table and Figure headings use (C) to designate capital cases and (NC) to designate non-capital
cases.

Final Technical Report: Habeas Litigation in U.S. District Courts

These 13 districts were selected as the basis for the capital case study. Cases from these districts comprised over
60% of the capital habeas cases filed nationwide from 2001 through 2004. We then selected from the merged set
of all habeas cases described earlier all of state capital habeas cases started in these 13 districts during 2000, 2001,
and 2002. To those, we added dozens of capital habeas cases started in the same years in those same districts that
were not included in the AO data sets, but which were found by searching PACER. 47 Deleted were duplicates of
other cases already in the sample; two next-friend actions; a few cases that were found when coded to be noncapital cases or Section 1983 cases; one case where the judge applied pre-AEDPA law; and numerous cases
which began prior to 2000, despite a docket number starting with “00”. 48
The final sample was 368 capital cases, representing filings by 348 death row inmates. The sample
includes more than half of the capital habeas cases filed nationwide during the period, spans six federal circuits
and nine states from all regions of the country. 49 See Table 1. Over 14% of the cases are from the Ninth Circuit
(from California, Arizona, and Nevada), a circuit that has been particularly salient in policy debates. More than
half of the cases in our capital case sample are from the state of Texas. As a result, we have attempted at various
points to indicate when the results in Texas differ from the results in other locations.
Table 1. (C) Number and percentage of cases in capital case sample, by district.

District
TX-S
TX-N
TX-W
OH-N
TX-E
OH-S
AL-N
AZ
NV
PA-E
OK-W
FL-M
CA-C

N
87
45
33
29
27
21
20
20
20
19
18
16
13

Percentage of sample
23.6
12.2
9.0
7.9
7.3
5.7
5.4
5.4
5.4
5.2
4.9
4.3
3.5

Total

368

100.0 %

These additional cases were either "mc" cases which had not been incorporated into the docket sheet of a subsequent
"cv" case (the AO data did not include cases designated "mc"), or "cv" cases which did not appear in the AO data. These
additional cases were located using the following PACER searches: 1) a search of the 13 districts for any case designated
casetype = 535 using PACER file date between January 1, 2000 and December 31, 2002; 2) a search of the 13 districts
for any case by a party having the same name as the petitioners in the cases we had found so far; and 3) a search in the
three districts that use mc cases regularly (TX-S, OH-S, PA-E) for cases designated 535 with PACER file date between
January 1, 2003 and December 31, 2003.
48
We also deleted the last 11 of 13 cases which had been filed in a short period by one mentally unstable petitioner in
FL-M; and the last 6 of 9 cases filed by another mentally unstable petitioner in the TX-E. The filings in these discarded
cases were all summarily rejected by the court due to their vexatious nature. Because of the large number of these cases,
keeping them in the sample would have skewed the results for these particular districts. We did retain in the sample
these petitioners' first filings and initial subsequent filings to which the courts applied a successive petition analysis.
49
It is not clear whether or how excluding districts with few capital cases influences the study’s findings. Taking
processing time as one example, a district with fewer capital cases may have more resources to process these cases more
quickly, or, alternatively, the judges and litigants in a district with fewer capital cases may take much longer with one of
these when it does come along than do judges and litigants in districts where they occur more frequently.

Final Technical Report: Habeas Litigation in U.S. District Courts

Data collection and variables. Initial data collection was completed in the spring and summer of 2006
by a closely supervised team of Vanderbilt law students, each of whom had had prior coursework in criminal
procedure and prior work experience either in the criminal justice system or in empirical research about the
federal courts. 50 All non-capital cases that had been coded as pending were recoded in late October 2006, and all
capital cases coded as pending were recoded at the end of November 2006.
Data collected for each case included: demographic information about the prisoner; the nature of the state
decision challenged; information about the petitioner’s state offense and sentence; proceedings in state courts;
petitions; amendments to petitions; in forma pauperis and representation status; post-petition pleadings (answers,
motions, and replies); magistrate judge involvement; discovery; evidentiary hearings; stays; rulings on certificates
of appealability; number of claims raised; and number of docket entries. The timing of various procedural events
was tracked with over a dozen separate date variables. Information on each individual claim was collected,
including claim type (about 100 separate claim types were tracked and then later aggregated into a smaller
number of claim-type variables), the application of each of six different defenses, type of disposition, and reason
for disposition for each claim. A list of variables collected appears in Appendix E.
PART II. DESCRIPTIVE FINDINGS

Petitioner demographics

PETITIONER RACE/ETHNICITY
•

•

CAPITAL CASES. For all but one of the 368 capital cases, the race/ethnicity of the petitioner was
available either from court documents or internet resources (including WESTLAW, LEXIS, and
state corrections websites). 51
o 41% (151) of capital cases were filed by white prisoners, 59% (216) were filed by prisoners who
were identified in court documents or state corrections websites as Latino or Hispanic, AfricanAmerican, Asian, or Native American.
o The proportion of white death row inmates in our sample (41%) is lower than that on death row
nationwide on January 1, 2003, which was 45%. This is probably because the percentage of
whites on death row in Texas, which dominates our capital sample, has been lower than the
national percentage. For most variables, race/ethnicity was not associated with significant
differences. For example, the same percentage of white petitioners received relief as non-whites.
NON-CAPITAL CASES. Data collection for the thousands of non-capital cases was limited to
documents that could be viewed through PACER. Those documents typically did not include
information about the race of the petitioner, so race information was not available for 97% of the
non-capital cases. Inferences based on the tiny proportion of our sample where such information

After intensive training in January and February of 2006, a pilot study was completed, using 12 capital cases filed in
districts other than the 13 districts in our study sample, and 100 non-capital cases randomly selected from those filed in
2002 nationwide (a year prior to our sample period). The pilot study included duplicate coding of the same cases, and
allowed troubleshooting of issues related to coding, PACER access, courthouse access, technical communication, data
merging, conversion, and analysis. The results of the pilot study were discussed at the first meeting of the Advisory
Committee, after which the codebook was finalized. Among the other steps taken to maximize reliability, capital coding
commenced after non-capital coding, and was completed by just four individuals: Professor King and three of the most
seasoned coders.
51
Race/ethnicity of the victim was not collected, but would be a sensible variable to research and add to the capital data
in the future.

Final Technical Report: Habeas Litigation in U.S. District Courts

did appear (as in cases where it was mentioned in connection with a jury selection challenge, for
example) would be inappropriate, so we do not include race in the analyses of non-capital cases.
PETITIONER GENDER
•

CAPITAL CASES. In 1.1% (4) of the 368 capital cases, the petitioner was a woman. The
proportion of female to male death row inmates in our sample was three to 346 or 0.86% (one
woman filed two cases). At the beginning of our sample period on January 1, 2000, 1.4% of capital
prisoners nationwide were female. 52

•

NON-CAPITAL CASES. 3.8% (90) of the non-capital cases in our sample were filed by women
prisoners. This roughly approximates the percentage of women serving relatively lengthy sentences
in the nation's state prisons at the end of 2003. At that time 4.4% of all violent offenders were
women (5.7% of those convicted of murder; 4.0% robbery). 53 In comparisons, cases filed by female
prisoners generally did not differ from those filed by males across most measures.

State proceedings
NOTE ON STATE INFORMATION FOR NON-CAPITAL CASES: The information
about state proceedings in non-capital cases reported below should be considered with
caution. In just over half the cases in our non-capital case sample, information about
many features of the underlying state criminal proceedings was available from PACER.
State information was generally not available in cases in which no documents were
accessible from the docket sheets. As a result, the availability of information about state
proceedings was very low in some districts, and high in others. In many additional cases
the documents filed in federal court could be accessed from PACER but did not contain
information about the petitioner's state proceedings. For example, state information was
missing in a higher percentage of the cases in which prisoners challenged an
administrative decision other than the underlying state criminal judgment, such as a
disciplinary proceeding, as well as in more abbreviated cases, such as those voluntarily
withdrawn, terminated by transfer to another district, or dismissed as successive.

CONVICTION OFFENSE, NUMBER OF COUNTS, SENTENCE
•

•

CAPITAL CASES. 155 (42%) of the capital cases were filed by petitioners who were convicted of
another crime in addition to capital murder. 104 of these cases involved three or more convictions.
The most frequent second conviction for those capital petitioners convicted of more than one
offense was another murder (72), followed by robbery (26), kidnapping (10), and rape (10). Murder
and robbery topped the list of third convictions as well, followed by burglary and kidnapping.
NON-CAPITAL CASES. Of the 63.4% (1512) cases with conviction information,
o 27 (1.8%) had been convicted of capital murder but not sentenced to death. For 372 (24.6%)
non-capital murder was the most serious offense of conviction. In another 26 cases the most
serious offense of conviction was manslaughter. Altogether 28.2 % of the prisoners with
conviction information had some sort of homicide as their most serious offense of conviction.

Tracy L. Snell, BJS Bulletin, Capital Punishment 1999 (Dec. 2000 NCJ 184795),
http://www.ojp.usdoj.gov/bjs/pub/pdf/cp99.pdf.
53
Paige M. Harrison & Ellen J. Beck, BJS Bulletin, Prisoners in 2005 (Nov. 2006 NCJ 215092),
http://www.ojp.usdoj.gov/bjs/pub/pdf/p05.pdf.

Final Technical Report: Habeas Litigation in U.S. District Courts

Other than murder, the crime most frequently coded as the most serious offense of conviction
was sexual assault (15.4%). Other common crimes for petitioners in our sample were robbery
(13.6 %) and drug offenses (13.0%). Assault (9.6), burglary (5.2) kidnapping (3.6), property
offenses (3.7), other felonies (3.3), weapons (1.5), and arson (.4) accounted for the remainder.
Five cases had a misdemeanor as the most serious offense of conviction. 37 cases (2.4%)
involved custody without a conviction.
Information on the number of counts of conviction was available for 62.3% (1511) cases. Of
these, 772 (51.1%) were convicted of more than one offense and 428 (28.3%) were convicted of
three or more counts of conviction.
60% of the non-capital cases had information on the type of sentence imposed in state court. Of
those 27.7% were serving life sentences. Of the remainder, the average sentence being served
was 20 years. Only 12% of those with a term of years were serving five years or less, 25% were
filed by prisoners sentenced to 30 years or more.

TRIAL OR PLEA CONVICTION
•

CAPITAL CASES. 95% of the capital petitioners had had a jury trial; seven had bench trials, nine
pleaded guilty to capital murder and one entered a nolo plea. The 10 petitioners who did not go to
trial
o raised fewer claims on average (22 v. 28);
o took less time to get to federal court (average 5.0 yrs compared to 7.4 yrs), but stayed longer;
o were more likely to be still pending in federal court (40%, four of the 10) compared to 25.5% of
those tried; and
o did not raise any innocence claims, compared to 11.2% of cases tried.

•

NON-CAPITAL CASES. For non-capital cases, information on what sort of plea or trial led to the
petitioner's conviction was available in 1294 (54.2%) of the cases. Of these
o 64.9% (840) involved petitioners who had gone to trial. Of those tried, 86% (726) had jury trials;
6.5% (55) had bench trials; for 7% (59) the type of trial was not available.
o 32.2% (417) of the 1294 cases involved convictions by guilty or nolo pleas. Of the pleas, 11%
(46) were pleas of nolo contendre.
o Compared to cases filed by prisoners who went to trial, cases filed by plea-convicted prisoners
took less time to resolve in federal court using all measures, and were less likely to be
stayed for exhaustion or less likely to remain pending;
were more likely to involve a challenge to an administrative decision, a plea or plea
negotiations, or to sentence alone;
were more likely to be dismissed as barred by the statute of limitations, and more likely
to be dismissed without reaching the merits; and
were less likely to raise innocence or insufficiency of evidence.

REVIEW IN STATE COURT
•

CAPITAL CASES.
o All capital petitioners in our sample had appealed their state judgments prior to starting their
federal cases. All but three of these had pursued state post-conviction review. (For an additional
three cases, this information was unavailable.)
o Of the 87% (320) cases with information about representation on state post-conviction review,
98% (315) of these had counsel; two did not.

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES.
o Of the 1333 or 55.9% cases with information about whether a direct appeal had been filed,
78.6% had appealed. Of the 1274 or 53.4% with information on whether the petitioner had
pursued state post-conviction review, 73.0% (931) had done so.
o Information on whether the non-capital petitioners had counsel during state post-conviction
review proceedings was available from PACER in too few cases to report.

TIME BETWEEN STATE JUDGMENT 54 AND FEDERAL FILING
•

CAPITAL CASES.
o The interval between state judgment and federal filing averaged 7.4 years, with a median of 6.5
years, and ranged from a minimum of 30 days to a maximum of 20.6 years. One in ten cases
took more than 13.8 years to reach federal court. Another one in ten cases took three years or
less.
o The period varied between states. See Table 2. Petitioners from OK-W reached federal court
sooner, and more consistently sooner than petitioners from other states, in an average of 3.8
years. Alabama death row prisoners reached federal court an average of 12.2 years following
their state judgments. 55
Table 2. (C) Days from state judgment to federal filing, by state. 56
State containing district
Alabama
Florida
Pennsylvania
California
Nevada
Ohio
Arizona
Texas
Oklahoma
Total

Average period in days
4467
4070
4001
3967
3357
2875
2834
2180
1386
2693

# of cases
20
16
19
12
19
49
20
192
18
365

Std. deviation
1156
1343
1447
1528
2004
1554
1112
1303
389
1565

Cases ending in a grant of relief took longer to get to federal court than those in which there was
either no termination or no relief. See Table 3. Cases dismissed as successive petitions took an
average of 11.2 years, compared to an average 6.5 years for non-successive petitions.
Table 3. (C) Days from state judgment to federal filing, by grant.

Writ not granted
Claim granted
Total

Mean days/yrs

Std. dev.

Median days/yrs

2670/7.3
2923/8.0
2693/7.4

332
33
365

1556
1661
1565

2343/6.4
2846/7.8
2380/6.5

The date of state judgment was calculated as the later of any of the following dates when available: date of conviction,
date of sentence, or the date the judgment was entered.
55
For a detailed explanation of factors contributing to the longer period for state court review of death sentences in
California as compared to Texas, see Steiker & Steiker, supra note 34, at 1875-1901.
56
Three cases lacked information concerning dates of state proceedings.

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES. The date of the petitioner's state judgment, conviction or sentence was
available from PACER in 54.5% (1299) of the non-capital cases (57% of the non-transferred cases).
Of those:
o An average of 6.3 years elapsed from the prisoner's state judgment to the filing of the federal
habeas case, ranging from 32 days to 35 years, and varied by state. See Table 4. The median
period was 5.7 years.
o Prisoners convicted of murder averaged an 8.0-year interval between state judgment and federal
filing.
o Prisoners challenging only sentencing reached federal court 9.6 years after their judgments;
those challenging conviction alone, 5.1 years
o Prisoners raising a challenge to a revocation decision filed their petitions an average of 10.7
years after their state judgments, compared to 5.8 for others
o Cases barred by the statute of limitations averaged an interval of 7.8 years, while cases not
barred averaged an interval of 5.6 years
o The interval for cases with at least one defaulted claim averaged 5.1 years; for cases with no
defaulted claims the average was 6.5 years
o Cases with all claims unexhausted reached federal court in an average of 4.0 years after state
judgment
o Cases in which the defendant had appealed his judgment took less time to get to federal court
than those with no appeals, while state post-conviction proceedings were associated with longer
intervals
Table 4. (NC) Days from state judgment to federal filing, selected states. 57

State containing district
Massachusetts
Arkansas
Louisiana
Florida
Michigan
North Carolina
Missouri
Nevada
Ohio

Mean in years
12.4
7.5
7.3
7.1
6.8
5.0
4.7
4.1
3.5

# cases
with info
8
12
54
127
83
28
47
15
28

% cases w/ info
73
75
79
71
83
72
80
79
72

Std. dev.
3760
2631
2290
2529
2086
2163
995
539
937

Representation of petitioner in federal court

IN FORMA PAUPERIS RULINGS
•

CAPITAL CASES.
o 86% of the cases included an IFP motion, and 89% of these were granted.
o IFP motion practice varied considerably between districts. In CA-C, 12 of 13 cases had no IFP
motions docketed; in OK-W, 14 of 18 IFP motions were denied. The percentage of motions
granted elsewhere ranged from 56% (FL-M) to 85% (AZ, NV).

Limited to states with more than 10 cases in the study, and with filing period information in more than 70% of those
cases. Among the states not represented in this table because less than 70% of the cases from the state included
information about judgment date are four of the six states with the highest volume of habeas filings: California (17.6% of
total sample); Texas (14.3% of sample); New York (6.9% of sample); and Pennsylvania (4.9% of sample).

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES. 56.1% included an IFP motion; only 62% of these were granted. In 2.4%
cases, an IFP motion was granted by the court of appeals.

PETITIONER'S COUNSEL
•

CAPITAL CASES.
o Only 26 of 368 cases involved petitioners who remained pro se. Of these 26 cases, 21 were
dismissed without reaching the merits on any claim, and nine lacked any petition.
o 12.5% of the petitioners had counsel when they first filed. Of these 46 cases, 18 included
retained or volunteer (not appointed) attorneys, 18 were appointed, and for 10 cases attorney
type was unclear. Most of the petitioners in PA-E had counsel at the outset, as a result of an
arrangement with the defender office to represent capital petitioners in that district. AL-N, the
only state in our sample that does not provide counsel for state post-conviction review of death
cases, had the largest proportion of cases with counsel who had not been appointed by the court.
State law in Texas required state post-conviction counsel to file a motion for appointment of
counsel in federal court for the capital petitioner, so this motion typically was the first document
filed in capital habeas cases from Texas.
o The court appointed counsel in 300 (81.3%) cases after the case commenced.
o Substitution of petitioner’s counsel was docketed in 54 (14.6%) of the capital cases. This, too,
varied by district. Nearly half (10 of 21) of the cases in OH-S had substitutions; only two of 45
cases in TX-N did.

•

NON-CAPITAL CASES.
o 95% (2271) of the petitioners were pro se at the beginning of the case. Of the 114 petitioners
with a lawyer at the beginning of the case, half of the attorneys were retained, four were
appointed, the source of counsel in the remaining cases with counsel could not be determined.
o 74 of those who started pro se received counsel at some point in the case. 58 Overall, 92.3%
(2202) of the cases involved no petitioner’s counsel.
o Although representation was rare in most districts, in OR an attorney was appointed in 86.4% of
the cases (19 of 22).
o In comparisons, representation was associated with longer processing times.
o Of cases with attorneys, 36 involved the substitution of petitioner’s counsel during the case.
This appeared to be associated with longer processing times.

The petition

TIMING OF FIRST PETITIONS
•

CAPITAL CASES.
o 19 cases were terminated with no petition ever filed. In 115 (32.9%) of the cases in which a
petition was filed, the petition commenced the case. Overall, the interval between the start of the
case and the petition averaged 5.3 months; the median interval was three months.
o The time before a petition was filed varied between districts. See Figure 2. Cases from CA-C
took on average 16 months before any petition was filed, compared to cases from PA-E and AL-

For example, Rule 8 of the Rules Governing Section 2254 Cases provides for the appointment of counsel in a habeas
case by a state prisoner if an evidentiary hearing is ordered. Rule 6 authorizes the appointment of counsel if the court
grants leave to employ discovery devices and the participation of counsel is deemed “necessary for effective utilization
of [such] procedures.” See also 18 U.S.C. § 3006A(a)(2)(B); 28 U.S.C. § 2254(h).

Final Technical Report: Habeas Litigation in U.S. District Courts

N where all cases began with petitions. This is probably related to differences among districts in
the process for appointing counsel in these cases, and the availability of counsel, with some
districts relying upon either federal or state defender offices primarily or exclusively, others
upon panel attorneys.

Mean Days between case beginning and filing of petition

Figure 2. (C) Days from first docket entry to petition, by district. 59
500.00

400.00

300.00

200.00

100.00

0.00
70

District

•

NON-CAPITAL CASES.
o In 71 cases (3.6% of the 1986 cases terminated other than by transfer to another district), no
petition was ever filed.
o Of the 2314 non-capital cases in which a petition was filed and the case was not transferred, in
92.7% (2145) the petition commenced the case.
o Among the cases that involved time between the beginning of the case and the filing of a
petition, the longest interval was almost two years. However, in only 1% of the cases did more
than two months elapse between the beginning of the case and the filing of the petition.

Districts numbers are those assigned in the Administrative Office of the Courts data sets, as follows:
13
PA-E
26
AL-N
3A
FL-M
39
TX-N
40
TX-E
41
TX-S
42
TX-W
47
OH-N
48
OH-S
70
AZ
73
CA-C
78
NV
87
OK-W

Final Technical Report: Habeas Litigation in U.S. District Courts

TIMING OF COUNSELED PETITIONS
•

CAPITAL CASES. 88.9% of the cases in the sample included a counseled petition.
o In most of the districts in the study, a motion for appointment of counsel was filed and granted
prior to the preparation and filing of a counseled petition.
o The average interval between the beginning of the case and the filing of a counseled petition was
8.2 months.
o The interval varied between districts. See Figure 3. The longest time period was in NV, where
on average, nearly three years passed between the beginning of the case and the filing of a
counseled petition. Most of this time in NV fell between the first petition and the counseled
petition, while in other districts, like CA-C, where the first petition was also the counseled
petition, the bulk of time elapsed prior to the filing of any petition. Compare Figure 2.

Mean Days between beginning of case to filing of counseled
petition

Figure 3. (C) Days from first docket entry to counseled petition, by district.

1,000.00

800.00

600.00

400.00

200.00

0.00
26

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

•

NON-CAPITAL CASES. The number of non-capital cases with counsel was small (188) and the
number with counseled petitions was even smaller – only 6% (137) cases out of our sample of 2384.
o 111 (81%) of these 137 counseled petitions were filed the day the case began.
o The average interval from commencement of case to the filing of a counseled petition was about
two months, but that is attributable to some particularly lengthy intervals between the
commencement of a case and the filing of the counseled petition, the longest of which lasted
more than two years (858 days).
o Only three of the seven cases that received a grant included a counseled petition.

Final Technical Report: Habeas Litigation in U.S. District Courts

Type of proceeding challenged

NOTE: NON-CAPITAL CASE CLAIM INFORMATION. Information about claims, and often about
defenses, was available only in cases with documents linkable through the docket sheet from PACER, and
only when those documents discussed claims. Information was available in 64% of the cases in the sample
(1521). The unavailability of claims information was distributed among states and districts. Table 5
presents availability in states with over 20 cases in the sample.
Table 5. (NC) Percentage of cases with claim type information, states with over 20 cases.

State

# with
claims
info

% w/
claims
info

total
cases
from
state

Ohio
Missouri
Louisiana
Alabama
Indiana
Texas
Michigan
Florida
Oklahoma
S. Carolina
New Jersey
Oregon

35
52
58
45
72
277
87
137
32
24
17
14

89.7%
88.1%
85.3%
81.8%
81.8%
81.0%
79.8%
76.5%
72.7%
70.6%
65.4%
63.6%

39
59
68
55
88
342
109
179
44
34
26
22

State
Illinois
N. Carolina
Georgia
Washington
Pennsylvania
California
New York
Tennessee
Colorado
Wisconsin
Mississippi
Virginia

# w/
claims
info

% w/
claims
info

total
cases
from
state

31
23
44
12
67
233
90
13
7
10
15
11

63.3%
59.0%
57.9%
57.1%
56.8%
55.6%
54.5%
39.4%
30.4%
30.3%
28.3%
12.6%

49
39
76
21
118
419
165
33
23
33
53
87

CHALLENGES TO DECISIONS OTHER THAN CRIMINAL JUDGMENTS
•

CAPITAL CASES. Five of the capital petitions challenged only the constitutionality of the method
of execution or the petitioner’s competency for execution, not the state judgment. All of these cases
were in Texas, none received relief, and two were dismissed as successive.

•

NON-CAPITAL CASES. Information about what proceeding was being challenged was available
in 1837 (77.1%) of our non-capital sample. The following percentages are of these 1837 cases,
unless otherwise noted.
o 19.3% (356), or nearly one in every five cases with information about what proceeding was
challenged, did not challenge the constitutionality of the underlying state judgment at all.
o 29 (1.5%) cases, filed in 15 different states, involved constitutional challenges to pretrial
custody by state officials. 60
o 327 (17.8%) alleged a constitutional flaw in a post-commitment administrative proceeding. The
administrative challenges are particularly concentrated in two states, Texas and Indiana.
Administrative challenges constitute 34.8% (111 of 342) cases from the Texas districts and
60.7% (54 of 89) from Indiana. Of the districts with over 20 cases in the sample, frequent
administrative challenges also appear in CA-E (21.7%), PA-M (32.1%), and FL-N (21.2%).

Among the challenges to pretrial custody that may be considered under 28 U.S.C. § 2241 by habeas courts before the
state prosecution runs its course is a claim that the prisoner is being put in jeopardy twice for the same offense. E.g.,
Walck v. Edmondson, 472 F.3d 1227 (10th Cir. 2007).

Final Technical Report: Habeas Litigation in U.S. District Courts

203 (13.3%) of all cases with claims information included at least one claim objecting to a
disciplinary decision; 131 (8.6%) included at least one objection to a decision concerning parole
or probation revocation or release.
Challenges to prison disciplinary decisions were routine in Indiana, where 53.8% of all habeas
cases with claims information raised such a challenge. Challenges to parole revocation and
release decisions were more evenly distributed among states.

SENTENCE CHALLENGES
•

CAPITAL CASES.
o 21 cases (5.1%) raised challenges to the sentence alone. All of those cases were in Texas. At
least eight cases challenged the conviction alone.
o 29.0% of the capital cases challenged the sufficiency of evidence supporting a sentencing factor.
This challenge was most frequent in NV, OK, OH, and AZ.
o 40.4% raised a challenge to a ruling excluding or admitting evidence at sentencing, a claim that
was raised in only 26.9% of the cases in Texas.
o Apprendi/Ring claims were raised in 54 cases and were most common in Arizona where 75% (15
of the 20 cases) were Ring-related. 79 cases raised a claim under either Roper, Atkins, or Ring.
Of the 25 cases raising Roper or Atkins, 15 were from Texas (8% of the 193 capital cases from
that state), and six (12% of the 50 cases) from Ohio.

•

NON-CAPITAL CASES.
o Of the 1521 cases with claims information available, at least 21% (325) raised challenges to the
sentence alone. Another 271 (17.8%) challenged only the conviction. The remaining cases with
claims information included claims that could have been directed at either sentence or
conviction.
o 16 (1%) cases challenged evidentiary rulings at sentencing.
o 57 (3.7%) alleged insufficient proof of a sentencing enhancement. An Apprendi violation was
alleged in 41 (2.7%) of the cases. One in four of the Apprendi claims was raised by a petitioner
in California. 61

Claims raised

NUMBER OF CLAIMS RAISED PER CASE
•

CAPITAL CASES. Capital petitions included more claims than non-capital petitioners.
o The average number of claims was 28; the median, 18.
o Two cases had more than 200 claims. In 5% of the capital cases, the petitioner raised more than
88 claims. 62
o 90% raised 63 or fewer claims, 21% raised five or fewer claims and 30 cases raised one claim.
o Fewer claims were raised in Texas than in other states. In CA-C, petitions included 80 claims on
average, nearly six times the average number in Texas districts. 63 See Figure 4.

In Cunningham v. California, 127 S. Ct. 856 (2007), the Court invalidated California’s sentencing scheme as violating
Apprendi/Blakely. Less than 40% of all 2384 cases in the sample were commenced after Blakely was handed down in
June 2004.
62
Multiple claims of the same type directed at different facts were eliminated after coding where necessary to reduce the
data size to 88 claim fields per case. At least one claim of each different claim type was retained. The number of
individual claims originally coded was retained as a separate variable.

Final Technical Report: Habeas Litigation in U.S. District Courts

Figure 4. (C) Claims per case, by district.

Mean Number of claims per dataset

100

0
39

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

•

NON-CAPITAL CASES. Non-transferred, terminated cases with claims information available
averaged four claims per case, with a median of three claims.
o Just one claim was raised in 28.7% of the 1521 cases for which we had claims information.
o Five or fewer claims appeared in 77.3% of the 1521 cases with claims information that were
pending or terminated other than by transfer.
o One case had 131 claims, all the rest had 49 or fewer.

COUNSEL-RELATED ERROR
•

CAPITAL CASES.
o 81.0% (299) of the capital cases included at least one claim alleging the ineffective assistance of
counsel (hereinafter “IAC”). 123 of these cases raised at least one IAC claim regarding appellate
counsel.
29% of the Texas cases lacked a single IAC claim compared to only 8.0% of cases from
districts outside of Texas.
Three of the 69 cases without an IAC claim received relief on another ground, five
remained pending, 47 were dismissed without reaching the merits, including two
transfers and eight dismissed voluntarily.
30 of the 33 grant cases (91%) included an IAC claim.
o Right to counsel claims that were not based on ineffective assistance, including the denial of
counsel altogether, Ake claims, interference with counsel, and denial of the right to selfrepresentation, were raised in 86 or 23.3% of the capital cases.

•

NON-CAPITAL CASES.
o 768 or 50.4% of the 1521 non-capital cases with claims information included at least one claim
of ineffective assistance, including 10% raising ineffectiveness of appellate counsel.
o 57 (3.7% of 1521) included at least one right to counsel claim other than IAC.
o Of the seven grant cases, two included an IAC claim.
o Of the 59 cases alleging new evidence of innocence, 75% also raised IAC.

See also Steiker & Steiker, supra note 34, at p. 1877 (noting that on direct appeal of capital cases in Texas, briefs are
limited to 125 pages, while California rules allow 280 pages).

Final Technical Report: Habeas Litigation in U.S. District Courts

EVIDENTIARY RULINGS
•

CAPITAL CASES.
o 169 (45.8%) included at least one challenge to a ruling admitting or excluding evidence at the
guilt phase (other than a confession or statement, and other than a ruling based on the Fourth
Amendment).
o The use of the petitioner’s confession or statement at guilt or sentencing was challenged in 104
(18.2%) cases.
o Search and seizure issues were raised in 45 (12.2%) of cases (including both free-standing
Fourth Amendment claims and Kimmelman claims).

•

NON-CAPITAL CASES.
o 301 (19.8% of 1521) included at least one challenge to a guilt-phase ruling admitting or
excluding evidence, other than a confession or statement and other than a ruling based on the
Fourth Amendment.
o The use of the petitioner’s confession or statement at guilt or sentencing was challenged in 82
cases (5.4% of 1521).
o Search and seizure issues were raised in 131 cases (8.6% of 1521), including both free standing
Fourth Amendment claims and Kimmelman claims.

NEW EVIDENCE OF INNOCENCE, INSUFFICIENT EVIDENCE OF GUILT
CAPITAL CASES.
o 94 cases (25.5%) included a claim of insufficient evidence of guilt. Ten of the 33 grant cases
included such a claim.
o A free-standing claim of innocence (defined as new evidence of innocence of the offense of
conviction -- either DNA or non-DNA) was raised in 40 (10.8%) cases. None of these were filed
in PA-E, but this claim appeared in over 30% of cases filed in CA-C. See Figure 5.
o Of the 33 cases granted relief, five included such a claim. One of the 40 cases containing this
kind of claim was dismissed as successive, two were time-barred.
Figure 5. (C) Percentage of cases with at least one claim of innocence, by district.
40.0%

% > 0 At least one claim of innocence?

•

30.0%

20.0%

10.0%

0.0%
13

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES. Of the 1521 non-capital cases with claims information,
o More than 18.9% (288) included a claim of insufficient evidence of guilt. Two of these cases
received relief.
o A free-standing claim of innocence (defined as new evidence of innocence of the offense of
conviction -- either DNA or non-DNA) was raised in only 59 (3.9% of 1521) cases. None
received a grant of relief on any claim.

JUDGE OR JURY BIAS, SELECTION, MISCONDUCT, OR INSTRUCTIONS
•

CAPITAL CASES. Almost all capital petitioners were convicted by a jury.
o Jury instructions and judicial comments to the jury were challenged in 68.3% of the cases.
o Jury selection or bias was challenged in 161, 43.6%.
o Jury misconduct or tampering issues were raised in 69, 18.7%.
o Judicial bias at any stage was alleged in 55, 14.9%.

•

NON-CAPITAL CASES. Of the 1521 non-capital cases with claims information
o Jury instructions and judicial comments were challenged in 221 (14.5%) cases.
o Some aspect of jury selection or jury bias was challenged in 90 (5.9 %).
o Jury misconduct or tampering was raised in 35 cases (2.3%).
o Judicial bias at any stage of the process was protested in 45 cases (3.0%).

PROSECUTOR COMMENTS; LOST, UNDISCLOSED, OR FALSE EVIDENCE
•

CAPITAL CASES.
o 177 (48.0%) raised at least one claim challenging the constitutionality of conviction or sentence
based on improper comments by the prosecutor.
o 159 (43.1%) raised a claim that the state had lost evidence, failed to disclose evidence, or
presented false evidence.

•

NON-CAPITAL CASES.
o 154 (10.1% of 1521) raised a claim of improper comments by the prosecutor at trial or
sentencing.
o 197 (13.0%) raised a claim that the state lost evidence, failed to disclose evidence, or presented
false evidence.

GUILTY PLEA AND NEGOTIATION CHALLENGES
•

CAPITAL CASES. Only 10 capital petitioners pleaded guilty to capital murder. A total of 15 cases
(4.0%) included either a claim of ineffective assistance of counsel regarding a plea or plea
negotiations, or a claim challenging the validity of the plea itself.

•

NON-CAPITAL CASES. Of the 1521 non-capital cases with claims information, a claim of
ineffective assistance of counsel regarding a plea or plea negotiations, or a claim challenging the
validity of the plea itself was raised in 224 (14.8% of 1521) cases.

COMPETENCE, PRESENCE AT TRIAL, PUBLIC, OR SPEEDY TRIAL
•

CAPITAL CASES.
o The right to a public or speedy trial, delay in review proceedings or in charging, or the denial of
the right of the defendant to be present at trail was raised in 44 (11.9%) cases.

Final Technical Report: Habeas Litigation in U.S. District Courts

o
•

Competency of the petitioner was an issue in 49 (13.3%).

NON-CAPITAL CASES. Of 1521 non-capital cases with claims information
o Presence, speedy, or public trial claims were raised in 81 (5.3%).
o Competency of the petitioner was an issue in 27 or 1.8%.

DOUBLE JEOPARDY AND APPEAL-RELATED CLAIMS
•

CAPITAL CASES.
o 14 or 3.7% of cases raised a double jeopardy issue.
o 67 (18.2%) included a challenge based on the denial or delay of appeal or trial transcript.

•

NON-CAPITAL CASES. Of 1521 non-capital cases with claims information,
o 84 (5.5%) raised a claim of double jeopardy.
o 81 (5.3%) included a claim of denial or delay of an appeal or trial transcript.

Intermediate orders

TRANSFERS TO ANOTHER DISTRICT
•

CAPITAL CASES. Of the 368 capital cases, only six (1.6%) ended by transfer to another district.
These six cases averaged 8.8 months from start to finish.

•

NON-CAPITAL CASES. 200 or (8.4%) cases ended by transfer to another district. These cases
tended to be terminated early; 157 of the 200 had been transferred by the fifth docket entry. Some
districts had more transferred cases than others. Among the districts with more than 10 cases in the
sample: 40% of the cases filed in GA-M and over 30% in CA-S were transferred to another district.
In California cases, judges indicated that it is general practice there to transfer habeas actions to the
district in which the petitioner was convicted.

STAYS OF EXECUTION
•

If an execution date had been set by the state, a petitioner could seek a stay of that execution from
the federal court. Only 22.2% of the capital cases in our sample included a docket entry staying a
petitioner’s execution. While such stays were routine in AZ (90%), and in PA-E and CA-C (each
69%), none of the cases from FL-M were stayed, and most of the other districts afforded stays in
fewer than 10% of the cases.

STAYS FOR STATE EXHAUSTION
•

CAPITAL CASES. 63 (17%) cases included at least one stayed period allowing the petitioner to go
back to state court and exhaust claims. 64
o only 19 of these 63 stayed cases had been terminated in federal court by the end of November
2006, the rest were still pending.
o Stayed cases had on average a longer period of time before reaching federal court (7.8 years
compared to 7.3 years), but the median period for stayed and non-stayed cases was the same.

A check in May 2007 of docket sheets revealed that in some of these cases that were pending but not stayed in
November 2006, stays subsequently had been imposed.

Final Technical Report: Habeas Litigation in U.S. District Courts

Stay practices vary widely by district. See Table 6. In CA-C, most cases are stayed for state
exhaustion. Stays are the exception in Texas districts. 65

Table 6. (C) Percentage of cases stayed, by district.
District
CA-C
FL-M
PA-E
OK-W
NV
OH-S
AZ
TX-W
OH-N
AL-N
TX-S
TX-E
TX-N

Percentage cases stayed in district
53.8
43.8
36.8
33.3
30.0
28.6
15.0
12.1
10.3
10.0
9.2
7.1
4.4

Total number cases from district
13
16
19
18
20
21
20
33
29
20
87
28
45

Mean Days betwen
stay impose and
lifted

Figure 6. (C) Length of stays for exhaustion, by district. 66
1,000.00
800.00
600.00
400.00
200.00
0.00
13

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

Including the stays from the pending cases, the average stay for exhaustion in a capital case has
lasted nearly two years (23 months; median = 22 months) so far. See Figure 6.

Some of the variance in the use of stays among courts could be related to circuit case law. Decisions from the Third,
Sixth, and Ninth Circuits encouraged the use of the stay device during the relevant period, Crews v. Horn, 360 F.3d 146,
151-154 (3d Cir. 2004) (when confronted with a petition containing an unexhausted claim, the federal court should stay
the action whenever dismissal would expose the petitioner to the limitations period of the habeas statute); Palmer v.
Carlton, 276 F.3d 777, 781 (6th Cir. 2002); Kelly v. Small, 315 F.3d 1063, 1070 (9th Cir. 2003). The Capital Habeas
Case Management and Budgeting Plan for the Ninth Circuit, the circuit containing three of the districts with the longest
disposition times and the most cases pending, states that holding the case in abeyance is a "nearly universal practice."
The plan is available on line at www.cacd.uscourts.gov, under documents. Decisions from the Fifth Circuit upheld the
district court's discretion to refuse a stay, or endorsed alternative dispositions of an unexhausted claim, such as finding
the claim defaulted or denying it on the merits. E.g., Brewer v. Johnson, 139 F.3d 491 (5th Cir. 1998). Circuit law
would not explain, however, the difference between FL-M with 44% of its cases stayed and AL-N with only 10% of its
cases stayed.
66
The length of a stay was measured until lifted, or until the end of November 2006 when pending capital cases were last
coded.

Final Technical Report: Habeas Litigation in U.S. District Courts

In 12 cases (in seven districts) the court entered a second stay for state exhaustion (four cases),
or a stay pending a federal decision in the court of appeals or U.S. Supreme Court (seven cases),
or a stay for settlement negotiations. These extra stays 67 averaged 1.4 years (17.3 months) in
length. One case from Houston was stayed three times, each time for a different federal
decision, and was still pending when last coded. Cases stayed were more likely to remain
pending than cases with no stay. See Table 7.

Table 7. (C) Percentage of non-transferred cases stayed, by whether case terminated.

Case stayed?
Not stayed
Stayed

•

Number/% of stayed cases
pending
51/17%
44/70%

Total cases
305
63

Number/% of stayed cases
terminated
254/83%
19/30%

NON-CAPITAL CASES. In the non-capital sample, only 57 (2.6% of 2184 non-transferred cases)
were stayed for exhaustion.
o The average time for stays varied by district. See Figure 7. Including cases that were still stayed,
at the end of October 2006, the average stay was 13.1 months. The shortest stay was just over
two months (62 days). The longest that a non-capital case had been stayed was 2.25 years (820
days).

Mean Days betwen stay impose and lifted

Figure 7. (NC) Average length of stays for exhaustion, including pending cases, by district.
1,000.00

800.00

600.00

400.00

200.00

0.00

Tennessee - Middle

Wisconsin - Eastern

Nevada

Michigan - Eastern

Connecticut

Rhode Island

California - Southern

Pennsylvania - Western

New Hampshire

California - Northern

Oregon

Georgia - Northern

California - Eastern

California - Central

New York - Southern

New York - Eastern

Indiana - Northern

Tennessee - Western

Ohio - Southern

Kansas

Pennsylvania - Middle

We did not rely entirely on the terminology used on the docket sheet to distinguish stays from other temporary
closures or suspensions. Not counted as stays were time periods designated on docket sheets as “stays” or “closures” for
“administrative” or “statistical” purposes, if the case was stayed or closed for a reason other than awaiting another
court’s decision. For example, some cases were “closed” or “stayed” in order to give the parties time to prepare
pleadings or to give the federal judge time to decide matters in the federal habeas case itself. These time periods are
functionally continuances, not stays. Although an interval for settlement negotiations in one case was probably more
like a continuance than a stay, we included it with the stays.

Final Technical Report: Habeas Litigation in U.S. District Courts

o
o

Of the stayed cases, 19 were terminated, 38 remained pending.
Relatively few districts granted stays in non-capital cases; about half were from the Ninth Circuit
(all but five from California), and another 35% were from either the Second or the Sixth Circuits.
Of the 26 cases stayed in the Ninth Circuit, 19 were still pending, but less than half of the cases
stayed in the Second or Sixth Circuit were still pending. Of the districts with over 20 cases in our
sample, no cases were stayed in Texas, compared to other districts where stays were much more
frequent, e.g., 11.0% of the cases in NY-E, and 22.2% of the cases in CA-C. The stays imposed
in districts from MI, NY, and CA accounted for 63% of all stayed cases.

Litigation steps

AMENDED PETITIONS
•

CAPITAL CASES. Amended petitions were filed in 35.2 % of the cases. (These included the first
counseled petition when the original petition was filed prior to the appointment of counsel.)
o In all districts except those in the Ninth Circuit – AZ, NV, and CA-C – cases without amended
petitions were more frequent than those with amended petitions. In AZ 85% of the cases had
amended petitions.

•

NON-CAPITAL CASES. Only 11.8% (281) included an amended petition. 68
o Amended petitions were filed in 20% of cases with an attorney.
o Over 18% of the cases in the Ninth Circuit included amended petitions. In CA-E, FL-N, and MSS (districts with over 20 cases in the sample) more than 30% of the cases included amended
petitions.

ANSWERS, MOTIONS, REPLIES
•

CAPITAL CASES.
o 306 cases included either an answer (also termed “response” to a show cause order or a “return”)
or a dispositive motion (motion for summary judgment or motion to dismiss) by the state (82.9%
of all cases and 90.9% of cases with amended petitions). In 85% (260) of these cases the
petitioner filed a reply or a “traverse.”
o Comparisons suggest that presence of these pleadings are unrelated to whether the case remained
pending or received relief. However, NV, the district with the second highest percentage of
pending cases, also had the lowest percentage of state pleadings (only nine of 20 cases). On
average, several years elapsed prior to the filing of petitions in this district, so in many of the NV
cases, state pleadings had yet to be filed.

•

NON-CAPITAL CASES. Only 58.2% of non-capital cases included either an answer (response or
return) or a motion to dismiss filed by the state. Over 40% of non-capital cases apparently were
resolved without any such pleading.
o 89.4% of the still pending cases included an answer or dispositive motion.
o Among the states with over 20 cases in our sample, some had a much higher percentage of cases
with these pleadings, e.g., over 70% of the cases from OH, OR, and MO.
o In 66.3% (920) of the cases where the state filed an answer or dispositive motion, the petitioner
filed some sort of reply to the state’s pleading.

Because amended petitions often follow the completion of a stay for exhaustion, we expect that in at least some of the
several dozen cases that had been stayed and remained pending amended petitions are yet to be filed.

Final Technical Report: Habeas Litigation in U.S. District Courts

STATE RECORDS
•

CAPITAL CASES. State records were entered separately on the docket sheet in 66% of all cases
and in over 75% of the cases in TX, OH, and AZ. Of the 306 cases in which the state had filed
either an answer or a dispositive motion, 77% also included the state record. The state record had
been filed in only 14% of the 62 cases showing no answer or dispositive motion from the state. Of
pending cases, 44% lacked state records compared to 29% of terminated cases. The record may yet
be filed in many of these still pending cases.

•

NON-CAPITAL CASES. In only 16.9 % (404) of the non-capital cases did a notation on the federal
court docket sheet indicate that the state record had been filed. This appears to vary by state, with
some states responding in fewer cases, or attaching exhibits instead of filing the record in federal
court. 69

DISCOVERY AND EVIDENTIARY HEARINGS
•

CAPITAL CASES.
o 12.5% of the cases included a deposition or a mental or physical examination.
o 9.5% (35) or about one in 10 non-transferred cases had been granted an evidentiary hearing by
the end of November 2006. Of the terminated, non-transferred cases in which a petition was
filed, the percentage is 9.7% (24 of 248). 11% (10) of the 95 cases still pending had been
granted an evidentiary hearing. Evidentiary hearings may yet be ordered in additional pending
cases that were stayed, or that were at early stages of litigation when coding closed.
o Ten of the 35 cases with hearings received relief (30.3% of all grant cases; grants made up
41.7% of the 24 terminated cases with petitions and hearings). Relief was ordered in only 23
(10%) of the 229 terminated cases with petitions but no hearings.
o Some districts granted evidentiary hearings more frequently, ranging from nearly four of every
10 cases in OH-S to none of the 33 cases in TX-W. See Table 8.
o 14 of the 35 evidentiary hearings were held by a magistrate judge.
Table 8. (C) Percentage of cases with discovery or evidentiary hearing, by district.

District
OH-S
PA-E
OH-N
OK-W
AZ
TX-S
CA-C
TX-E
TX-N
AL-N
NV
FL-M
TX-W
Average/Total

Percentage with
evidentiary hearings
38
16
14
11
10
9
8
7
7
5
5
0
0
9.5

Percentage with
deposition or exam
57
11
31
6
25
3
23
7
7
0
10
0
12
1.5

Total cases from district
21
19
29
18
20
87
13
27
45
20
20
16
33
368

The record of state proceedings may be filed only after any stay for exhaustion is completed. Several cases in our
sample were still stayed for exhaustion when coding closed.

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES. Only six (0.3% of 2184 cases pending or terminated without transfer)
included a deposition or examination. Only five of the 1986 cases terminated without transfer were
provided with an evidentiary hearing, a rate of 0.3%, or one of every 397 cases terminated. Of all
cases, including those still pending, nine included an evidentiary hearing, a rate of one of every 243
cases or (0.4%).

MAGISTRATE JUDGE USE
•

CAPITAL CASES.
o
Of the 35 capital cases with evidentiary hearings, 14 were before a magistrate judge.
o
In 10.8% cases magistrate judges issued a report and recommendation ("R&R") relating to
the disposition of any claim. Nearly one third of the cases in TX-E and OH-S included
R&Rs. Of the 95 pending cases, many had been referred to magistrate judges, but had not
yet reached the point where an R&R for disposition would have been filed.

•

NON-CAPITAL CASES.
o
About half of the cases included an R&R for disposition.
o
Of the nine evidentiary hearings, four were before a magistrate judge.

Processing time

PENDING (NOT TERMINATED) CASES
CAPITAL CASES. 95 or 26% of the cases were unresolved as of the end of November 2006. Of the
pending cases 67.7% had been stayed, 10.5% had been granted evidentiary hearings (versus 9.2% of
the terminated cases); 27% had counsel substituted (versus 10.2% for terminated cases); seven had
not yet filed a counseled petition (33 of terminated cases lacked counseled petitions). More than
half of the cases in five districts remained pending. See Figure 8.
Figure 8. (C) Percentage of cases pending as of end of November 2006, by district.
0.80

Percentage of cases still pending as of Nov 2006

•

0.60

0.40

0.20

0.00
41

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES. Of the non-capital cases started in 2003 and 2004, 8.3% (198) were still
pending at the end of October 2006 (9% of non-transferred cases).
o Pending for an average of 30 months, these cases more frequently involved an amended
petition, state pleading, stay, or a petitioner with counsel.
o Of districts with more than 10 cases, NV had 44% pending, OK-N 38%. Of districts
with over 20 cases in the sample, the four districts in NY were among the nine districts
with the most cases unresolved. See Figure 9.
Figure 9. (NC) Percentage of cases pending, by districts with over 20 cases.
Pennsylvania - Eastern
Louisiana-East
Texas - Northern
Indiana - Northern
California - Southern
Pennsylvania - Middle
Pennsylvania - Western
Georgia - Northern
Colorado
Louisiana - Western

Federal District Court

Michigan - Western
Florida - Northern
Texas - Western
Mississippi - Southern
Alabama - Northern
California - Central
Florida - Middle
New Jersey
Wisconsin - Eastern
Ohio - Southern
Michigan - Eastern
New York - Eastern
Ohio - Northern
California - Northern
Oregon
New York - Southern
New York - Western
Missouri - Eastern
California - Eastern
New York - Northern
0.00

0.10

0.20

0.30

0.40

Percentage Cases pending in Oct 2006

0.50

Final Technical Report: Habeas Litigation in U.S. District Courts

NUMBER OF DOCKET ENTRIES 70
•

CAPITAL CASES. The number of docket entries is one measure of the complexity of a case and
the amount of activity in the case generated by the litigants, indicating separate motions, pleadings,
events, etc. The average number of docket entries was 55, median 44. The amount of docket
activity in the districts from the Ninth Circuit, Ohio, and Oklahoma exceeded that of other districts.
See Figure 10. Arizona cases on average included far more activity than those in Texas, for
example.
Figure 10. (C) Docket entries per case, by district.

Mean ‘Total no of docket entries’

120

100

0
40

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

•

NON-CAPITAL CASES. The average number of docket entries was 18, the median 15. Compared
to capital cases, there was less variation in the number of docket entries by location. However, the
Ninth Circuit cases included the greatest number of docket entries, averaging 23 compared to an
average of 16 for the other circuits. Cases including a magistrate judge's report and
recommendation averaged 21 docket entries compared to an average of 15 for other cases.

DAYS TO TERMINATION FROM FIRST DOCKET ENTRY
Our variable BEG2TERM is the interval in days between the date of the first docket entry and the
date of termination in district court. Date of termination is the later of either the date of the order
terminating the case (including an order to transfer, dismiss, grant, or deny all claims in the petition) or the
date of the order denying a motion to reconsider that order. 71 For capital cases we also checked for start
dates not reflected in the “cv” (civil case) docket sheet. This occurred if the capital case was started as a
miscellaneous case (usually with a motion for the appointment of counsel, designated with an "mc" instead
of a “cv” in the docket number) and the “mc” case was closed at the same time that a new case by the same
petitioner was started with a new number (usually upon filing of the petition, with the new case designated
70

Measured as of date last coded. Non-capital cases that remained pending when first coded in the spring of 2006 were
recoded in October 2007; capital cases that remained pending in the summer of 2006 were recoded in November 2006.
71
See Latzer & Cauthen, supra note 43 (measuring processing time to end with motion for reconsideration ruling).

Final Technical Report: Habeas Litigation in U.S. District Courts

"cv"). In this situation the two cases were counted as a single case and the start date for the mc case was
entered as the start date. 72
We coded as separate cases two filings by a single capital petitioner with different docket numbers,
if the first case terminated by dismissal without prejudice and the second case was commenced with a new
docket number. Some districts would not dismiss in this situation but instead would stay the case, then
reopen it, on the same docket sheet. This we would count as one case regardless of whether the docket sheet
indicated that during the stay the case had been "stayed," "closed for statistical purposes," or
"administratively closed" during the period in which exhaustion was taking place. In several of the districts
that we examined, capital cases that were still open but stayed were designated as "closed" or "terminated,"
either in the AO data for the case or on the docket sheet. These districts included PA-E, TX-N, TX-S, TXW, NV, and FL-M. We coded any case that was stayed, without disposition of all claims, as stayed and still
pending.
•

CAPITAL CASES.
o Average time to resolve a non-transferred terminated case was 28.7 months, median 27.1
months. The longest terminated case took 6.8 years. Eight cases ended in one day or less, all in
Texas, and all but one involved a successive petition. The six cases ending in transfer to another
district took an average 8.8 months from start to transfer order, the 14 successive petitions
averaged 50 days.
o 10% of the 267 cases terminated without transfer were ended in about three months or less. Most
of these were expedited dispositions other than successive petitions. See Figure 11. These
expedited cases were included in the descriptive information reported below and in the
comparisons with pre-AEDPA processing times, to be conservative. It is not known whether
pre-AEDPA studies included or excluded capital cases that were terminated quickly before a
petition was filed, by voluntary dismissal or otherwise. Judicial conference reports of
termination times were based on all capital cases, presumably including those terminated
quickly, before a petition was filed. Because information about only those cases fully litigated
may be of interest to some readers of this report, the explanatory analyses in Part IV were
conducted and reported both with and without the 10% shortest cases.
o Grants took longer (3.4 years or 41 months) than denials or dismissals.
o Average time to termination after at least one claim was considered on the merits was 2.7 years,
compared to the average time to termination where no claims were considered on the merits.
That average was about one and a half years.

The results for the Central District of California are probably understated. There were two separate death row inmates
who began a case with one number, which was closed at some point during the time in which counsel was being
appointed or discovery was underway. Then each petitioner started another case with another number after a short break.
To be conservative, because of the short break between the two cases, rather than count these as one case, we counted the
first segment as a separate case that closed, without a petition being filed. Had we coded these as continuous cases,
instead of showing 13 cases with four of those terminating, the district would show instead 11 cases two of which had
terminated, and the processing time for capital cases in this district would have been much longer. Also artificially
shrinking the average time for disposition for this district is the fact that one of the terminated cases was a petition
dismissed as incomprehensible within seven days of filing.

Final Technical Report: Habeas Litigation in U.S. District Courts

Figure 11. (C) Disposition time. 73

Number of cases

5
Mean =923.73
Std. Dev. =548.108
N =245
0
0

500

1,000

1,500

2,000

2,500

Disposition time in days

The average time to resolve terminated cases varied between districts, ranging from 1.8 years in
FL-M to 4.4 years in OH-S. See Figure 12.

Figure 12. (C) Disposition time, by district. 74

3A
41
40
78

District

39
42
70
47
26
73
13
87
48
0

200

400

600

800

1000

1200

1400

Average days for disposition

73
74

Excluding cases dismissed as successive or transferred.
Tranferred cases excluded.

1600

1800

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES. The average overall processing time for all terminated, non-transferred
cases in our sample was 9.5 months, with a median of 7.1 months. The longest terminated case
lasted 3.7 years (1337 days), the 32 shortest cases were resolved on the same day they began. 90%
of all cases (transferred cases included) were resolved within 22 months. Figure 13 shows that
unlike the capital cases, most non-capital cases were ended early.
Figure 13. (NC) Disposition time. 75
300

Frequency

200

100

Mean =289.41
Std. Dev. =270.404
N =1,973
0
0.00

250.00

500.00

750.00

1000.00

1250.00

Days between beginning and termination of federal case

DAYS TO TERMINATION FROM THE FILING OF THE PETITION
•

CAPITAL CASES. Subtracting from disposition time the period before any petition was filed
reduced the average time until termination from 2.4 years to 2 years (740 days). Of terminated
cases, the longest was 6.5 years, the shortest 255 days.

•

NON-CAPITAL CASES. Measuring disposition time from the petition forward reduced the
average processing time to 9.4 months from 9.5, and the median time to 7.0 from 7.1 months.

LENGTH OF CASE IN DAYS, INCLUDING PENDING CASES
Length of case is the time in days between the date of the first docket entry and either 1) the date the docket
sheet was last coded in late 2006 (if the case was pending), or 2) the date the case was terminated in district
court (if it was terminated). This is a better measure of how long these cases are taking because so many of
them remain pending.
•

CAPITAL CASES. Including pending cases raised the average time from 2.4 years to 3.1 years or
37.9 months. Median time was about 3 years or 36.5 months. Figure 14.

Transferred cases excluded.

Final Technical Report: Habeas Litigation in U.S. District Courts

Figure 14. (C) Length of case in days, pending cases included. 76
Normal
40

Frequency

Mean =1152.06
Std. Dev. =697.741
N =362
0
0.00

500.00

1000.00

1500.00

2000.00

2500.00

3000.00

Length of case, including pending cases

Pending cases averaged 5.3 years (1935 days), the longest was 6.9 years and the shortest 3.9
years.
Using this measure of processing time, the three districts in the Ninth Circuit along with the OHS showed the highest averages among the 13 districts in the sample, and the four Texas districts
showed the shortest average processing times. See Figure 15.

Figure 15. (C) Length of case in days, pending cases included, by district. 77

Mean Length of case, including pending cases

2,000.00

1,500.00

1,000.00

500.00

0.00
41

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers
76
77

Transferred cases excluded.
Transferred cases excluded.

Final Technical Report: Habeas Litigation in U.S. District Courts
•

NON-CAPITAL CASES. With the pending cases included, the average and median times for
processing non-transferred cases jumped to 11.5 and 8.1 months, respectively. The longest noncapital case had been in district court for 3.8 years. See Figure 16. Generally, California and New
York districts took longer to complete these cases than many other districts. See Figure17.
Figure 16. (NC) Length of case in days, pending cases included. 78

300

Frequency

200

100

Mean =351.48
Std. Dev. =323.137
N =2,092
0
0.00

250.00

500.00

750.00

1000.00

1250.00

Length of case, including pending cases

Transferred cases excluded.

Final Technical Report: Habeas Litigation in U.S. District Courts

Figure 17. (NC) Length of case in days, including pending cases included , by district.
Oklahoma - Northern
New York - Western
Rhode Island
Oklahoma - Eastern
Nevada
Missouri - Eastern
New York - Northern
Oregon
New York - Southern
California - Eastern
Connecticut
New Jersey
Illinois - Southern
Ohio - Southern
Delaware
New York - Eastern
California - Northern
West Virginia - Northern
Alabama - Middle
Utah
Alabama - Northern
Alabama - Southern
New Hampshire
Michigan - Eastern
Tennessee - Western
Florida - Northern
Massachusetts
Florida - Middle
Tennessee - Middle
Kentucky - Eastern
Pennsylvania - Western
California - Central
Louisiana - Western

Federal District Court

Arizona
Ohio - Northern
Mississippi - Northern
Louisiana-East
California - Southern
Nebraska
Wisconsin - Eastern
Pennsylvania - Middle
Pennsylvania - Eastern
Mississippi - Southern
Iowa - Southern
Florida - Southern
South Dakota
Kentucky - Western
North Carolina - Western
Texas - Western
Kansas
Colorado
Michigan - Western
Georgia - Middle
Montana
Arkansas - Eastern
Texas - Eastern
Washington - Western
Illinois - Central
Washington - Eastern
Oklahoma - Western
Vermont
North Carolina -Eastern
Illinois - Northern
Minnesota
Texas - Northern
Texas - Southern
Idaho
Georgia - Northern
Iowa - Northern
Indiana - Northern
South Carolina
Virginia - Eastern
Arkansas - Western
Virginia - Western
North Carolina - Middle
Missouri - Western
West Virginia - Southern
North Dakota
Maryland
Tennessee - Eastern
Georgia - Southern
Indiana - Southern
Maine
Puerto Rico
Louisiana - Middle
Wyoming
Wisconsin - Western

0.00

200.00

400.00

600.00

800.00

Mean Length of case in days, including pending, excluding transferred
cases

Final Technical Report: Habeas Litigation in U.S. District Courts

LENGTH OF CASE EXCLUDING STAYED TIME
•

CAPITAL CASES. On average, after subtracting stayed periods the average time for processing all
cases dropped from 3.1 years to 2.7 years (999 days). Subtracting stayed time, the average time that
pending cases had been pending was reduced from 5.3 years to 4.2 years (1516 days). Excluding
stayed time did not change the relative average speed among the districts. The same five districts
(Texas plus FL-M) remained on the low end, and the same six districts (Ninth Circuit, OH-S, PA-E,
and OK-W) remained on the high end. AL-N and OH-N remained in the middle of the pack. See
Figure 18.

Mean Days between beginning and termination of federal
case minus stay time, including pending cases

Figure 18. (C) Length of case in days, less stayed time, by district.

2,000.00

1,500.00

1,000.00

500.00

0.00
3A

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

•

NON-CAPITAL CASES. Subtracting the stays for exhaustion made little difference in disposition
time for non-capital cases, the median length remained the same at 8.1 months, the average dropped
from 11.5 to 11.2.

Non-merits dispositions

ALL CLAIMS DISMISSED WITHOUT REACHING MERITS
•

CAPITAL CASES. Of the 234 cases terminated without relief or transfer (170), 72.6% included at
least one claim considered on the merits. Non-merits dispositions included voluntary dismissal,
dismissal as moot, time-barred, defaulted, or successive. Each of these are examined in more detail
below. Fifteen cases were terminated before any petition was filed.

•

NON-CAPITAL CASES. 631 of the 1311 cases terminated other than by transfer or grant with
claims information (42%) had at least one claim denied on the merits.

Final Technical Report: Habeas Litigation in U.S. District Courts

STATUTE OF LIMITATIONS DISMISSALS
•

CAPITAL CASES. The limitations defense was employed infrequently.
o The limitations period for 32% of the petitioners began to run in 2000 or later. 79
o Only 11 cases included a ruling that a claim was time-barred (3.0% of all 368 and 4.1% of 267
terminated, non-transferred cases, and 4.7% of the 235 cases terminated without relief or
transfer). See Table 9, showing time-barred cases by district. The procedural history of each of
these cases is summarized in Appendix B.
o Twelve of the capital cases included a ruling that one or more claims was not barred by that
statute of limitations. Of these, 10 cases included a judicial finding of equitable tolling. 80 All but
two were from Texas. Several of the equitable tolling decisions found fault with state habeas
counsel. Of the cases in which the limitations period was equitable tolled, 10 were terminated
and none of those received relief. Two remained pending. None included an evidentiary hearing.
Table 9. (C) Cases with at least one claim time-barred, by state.
State
Texas
Alabama
Ohio
Florida
Pennsylvania
Arizona
California
Nevada
Oklahoma
Total

•

# cases time-barred
5
3
2
1
0
0
0
0
0
11

# cases terminated from state
179
16
31
11
8
4
4
5
15
273

Total # cases
192
20
50
16
19
20
13
20
18
368

NON-CAPITAL CASES. The one-year limitations period barred more than one in five non-capital
cases terminated without transfer.
o In 432 cases, 18.1% of all cases and 21.8% of 1986 non-transferred terminated cases, included a
ruling that one or more claims was barred by the statute of limitations. Of the 1158 cases
terminated other than by transfer that had claims information, the statute of limitations barred
261, or 22.5%.
o Of the 432 time-barred habeas filers, 22 or 5.1% had counsel.
o Only 8.4% of the cases raising challenges to an administrative decision were barred by the
statute of limitations, a much smaller percentage than the portion of other cases held time-barred.
This may illustrate the challenge of applying AEDPA’s limitations provisions to cases that do
not attack the underlying state conviction.
o In at least 30 cases, the judge ruled that one or more claims was not barred by that statute of
limitations. In 16 of these cases the judge concluded that the period was properly tolled for state
proceedings. Twelve of these cases based the rejection of the state’s defense on equitable tolling,
one found that the state had waived the time bar. Of the 30 cases in which the judge rejected the
time-bar, seven petitioners had attorneys, none were granted relief.

The start date for the limitations periods was collected only in capital cases.
Our coding would have missed cases with equitable tolling if 1) the district court had dismissed an earlier case by the
same petitioner with the understanding that the statute would be equitably tolled until the petitioner returned to federal
court, or that the post-exhaustion petition would “relate-back” to the original filing, and 2) this arrangement did not
appear in the record of the post-exhaustion case coded. For cases using this process (decided prior to the Court’s
decision in Rhines v. Weber, 544 U.S. 269 (2005)), see RANDY HERTZ & JAMES LIEBMAN, 1 FEDERAL HABEAS CORPUS
PRACTICE AND PROCEDURE § 5.2b n. 106 (5th ed. 2005).
80

Final Technical Report: Habeas Litigation in U.S. District Courts

SUCCESSIVE PETITION DISMISSALS 81
•

CAPITAL CASES. Only 14 of the 368 cases (3.8%) were found to be successive petitions by
district courts, or 5.2% of 267 cases terminated without transfer, and 6.2% of the cases terminated
without relief or transfer. Only one of the 14 cases included a claim of actual innocence. A finding
of successiveness was rare or absent, except in TX-E where 21% of the cases there (six of 28) were
found to be successive.

•

NON-CAPITAL CASES. 138 cases (6.9% of the 1986 non-transferred terminated cases) were
rejected as successive. These include 72 of the 1311 non-transferred terminated cases with claims
information (5.5%) (8 of these cases included a claim of actual innocence), and 64 of 676 nontransferred terminated cases without claims information (9.5%).

WAIVER OF HABEAS REVIEW, SETTLEMENT
•

We looked for evidence in the record that defendants were waiving habeas review by agreement.
None of the 10 capital cases in which the petitioner had admitted guilt included a discussion of
whether the petitioner had waived habeas review by agreement. Only one of the non-capital cases
included a ruling on whether the petitioner had waived habeas review in his plea agreement. An
unknown proportion of the voluntarily dismissed cases or cases dismissed as moot (see below) were
ended by settlement. Evidence of settlement would not necessarily appear in the documents that
were filed in the cases or available on PACER.

VOLUNTARY DISMISSALS
•

CAPITAL CASES. In 31 cases (11.6% of 267 terminated, non-transferred cases) all claims were
dismissed voluntarily. Nearly 39% (five of 13) cases in CA-C were voluntarily dismissed, while
none of the 45 cases in TX-N were voluntarily dismissed. Only two of the 40 voluntarily dismissed
cases involved a ruling that every claim was unexhausted.

•

NON-CAPITAL CASES. All claims were voluntarily dismissed in 63 non-capital cases, or 3.2% of
the 1986 cases terminated without transfer.

ALL CLAIMS DISMISSED AS UNEXHAUSTED
•

CAPITAL CASES. In only nine cases (2.4% of 368, 3.4% of 267 terminated non-transferred cases)
were all claims dismissed as unexhausted. 82

•

NON-CAPITAL CASES. In 216 cases (5.7% of all cases, 10.9% of terminated, non-transferred
cases) all claims were dismissed as unexhausted. Among the 1311 non-transferred terminated cases
with claims information, 9.8 (129) cases had all claims dismissed as unexhausted; 12.9 % (87 of
676) non-transferred terminated cases with no claims information were dismissed for this reason.

This reports only district court determinations that a petition is successive and that it lacked authorization from the
court of appeals. Tracking the number (or nature) of petitions turned away as successive in the courts of appeals was
beyond the scope of this study.
82
These figures include only cases in which the judge ruled each claim was not exhausted and do not include dismissals
of mixed petitions, that is, petitions including exhausted and unexhausted claims. Nor does this figure include voluntary
dismissals by the petitioner, some of which may be sought by the petitioner so that he can exhaust his claims in state
court.

Final Technical Report: Habeas Litigation in U.S. District Courts

DISMISSALS FOR STATE PROCEDURAL DEFAULT
•

CAPITAL CASES.
o At least 156 cases (42.2% of 368 cases; 53.3% of terminated, non-transferred cases) included a
ruling that at least one claim was barred by procedural default. These included cases with other
claims still pending, and cases in which the judge reached the merits of the claim despite finding
it defaulted. 83 In 11 (11.6%) of pending cases, at least one claim was dismissed as defaulted.
The Ninth Circuit protocol for district courts includes a separate phase for consideration of
procedural default 84 and it appeared that many of the cases in CA-C and NV had not reached that
phase. See Figure 19.
o The procedural default defense was rejected for at least one claim in 58 cases (16% of 368). In
only one case was the reason for rejection stated as “miscarriage of justice,” reflecting a concern
about actual innocence. Several cases included a finding that the state procedural rule was not
"adequate and independent," or a statement that the judge had found it more efficient to skip
default analysis and reach the merits.

% > 0 At least one claim barred as defaulted?

Figure 19. (C) Percentage of cases with a defaulted claim, by district.

60.0%

40.0%

20.0%

0.0%
73

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

•

NON-CAPITAL CASES. 265 (13.3% of 1986 non-transferred terminated cases, and 19.4% of nontransferred terminated cases with claims information) included a ruling that a claim was defaulted.
This includes cases with other claims pending and cases in which the defaulted claim was also
addressed on the merits. 37 cases (1.7% of non-transferred cases) included a ruling rejecting the
defense of procedural default. In many of these, the reason stated for reaching the merits was

This number is underinclusive of the total number of cases involving at least one ruling of default, since in some cases
if the judge noted as alternative reasons for dismissing a given claim that the claim failed on its merits and was either
time-barred or unexhausted; the third reason for dismissal – default – would not be included in the analysis. A merits
denial – even as an alternative ground - was always coded as the first reason for disposition, and statute of limitations as
the second reason. Additional reasons for dismissal of a given claim were recorded in text fields so that they could be
added later as a new variable if necessary.
84
See Capital Habeas Case Management and Budgeting Plan for the Ninth Circuit, supra note 65. For a collection of
various approaches adopted by district courts in handling motions and answers and records in these cases and a
disucssion of the relative merits of requiring all issues and merits to be briefed at once or seriatim, see Ebert v. Clarke,
320 F. Supp. 2d 903 (D. Neb. 2004).

Final Technical Report: Habeas Litigation in U.S. District Courts

judicial economy. In no case was a defaulted claim addressed because that the court found that the
failure to review the claim would result in a miscarriage of justice.
DISMISSED AS MOOT OR LACKING CUSTODY
•

CAPITAL CASES. At least eight cases involved a ruling that a claim was moot, including three
cases in which the petitioner had died in prison. Several of the other cases with claims dismissed as
moot involved a grant on another claim.

•

NON-CAPITAL CASES. At least 23 cases 85 were dismissed as moot because the petitioner had
died, had been released, or had received relief in state court. Seven of these cases were from Texas.
Another 30 cases were dismissed because the person was not in custody as a result of the challenged
state action. (35% of these cases were from California.) None of these moot or no custody cases
involved a claim of innocence.

DISMISSED AS BARRED BY TEAGUE
•

CAPITAL CASES. 64 (23.9% of 267 terminated, non-transferred cases) involved a ruling that one
or more claims was barred by Teague v. Lane. 86 Almost all (89%) of these rulings were in Texas.
See Figure 20.

% > 0 At least one claim with defense of Teague ruling by DJ
bef app?

Figure 20. (C) Percentage of cases with a Teague-barred claim, by district.
50.0%

40.0%

30.0%

20.0%

10.0%

0.0%
47

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

•

NON-CAPITAL CASES. Only eight non-capital cases included a ruling that any claim was
Teague-barred. In practice, it is possible that Teague may be raised by states’ attorneys in
connection with claims related to only capital sentencing proceedings, or that it may be subsumed
under the new standard of review in § 2254(d).

This is the number of cases in which the reason for disposition of the first claim was coded as moot. Another three
cases included moot as an alternative reason for dismissing the first claim. It is possible that even when a first claim was
not moot, additional claims were moot. For example, if the state had granted relief from sentence this would render
sentencing claims moot but may leave conviction claims as live controversies.
86
See supra notes 4 & 30.

Final Technical Report: Habeas Litigation in U.S. District Courts

Merits dispositions

STANDARD OF REVIEW APPLIED
•

CAPITAL CASES.
o Of the cases including merits review of at least one claim, 27% applied to at least one claim a
standard other than the § 2254(d) “unreasonable application of” or “contrary to” Supreme Court
precedent.
o Cases from Ohio and Oklahoma were more likely than cases from Texas to include the
application of a standard other than that specified in Section 2254(d). See Table 10.
Table 10. (C) Standard of review applied, by district.

District
OH-S
OK-W
OH-N
AL-N
PA-E
AZ
FL-M
TX-E
TX-S
TX-W
TX-N
CA-C
NV
Total

•

total cases
from district
21
18
29
20
19
20
16
27
87
33
45
13
20
368

cases terminated
without transfer
7
14
23
16
8
4
10
25
82
29
40
4
5
267

# cases with
merits
review
4
11
20
10
7
3
7
18
60
25
37
1
3
206

# cases
applying
non-§ 2254 std
4
8
14
5
3
1
2
5
10
3
0
0
0
55

% of cases
applying non§ 2254 std
100%
73%
70%
50%
43%
33%
29%
28%
17%
12%
0%
0%
0%
27%

NON-CAPITAL CASES. In 105 cases, judges applied a standard for reviewing the state decision
other than the deferential standard set out in § 2254(d). About 16.6% of the 631 cases with at least
one claim were denied on the merits. Alternative standards were used more frequently among cases
attacking administrative decisions and not the underlying state judgment. This may reflect the
challenges of applying AEDPA’s provisions to this kind of claim.

HARMLESS ERROR
•

CAPITAL CASES. Only four of the 267 cases terminated without transfer involved a finding that
an error was not harmless. Each of these ended in a grant of relief. Three of these cases were from
Texas, one from Ohio.

•

NON-CAPITAL CASES. In only four cases, all of them granting the writ, did the judge find an
error was not harmless. Most of the grants, see below and Appendix C, were based on grounds that
were not subject to harmless error analysis.

Final Technical Report: Habeas Litigation in U.S. District Courts

GRANTS OF RELIEF
•

CAPITAL CASES. All capital cases that received relief are described in Appendix C. Of all 368
cases, 33 received relief (this is 12.4% of 267 terminated non-transferred cases). 87
o
Judges in PA-E granted relief in six of eight terminated non-transferred cases, while judges
in TX-S granted relief in six of 80 terminated non-transferred cases. See Table 11. Outside
of Texas, the grant rate was 17.9%.
o
23 of the 33 cases receiving relief involved relief from sentence alone, not from conviction.
o
Relief was granted in 17% of the cases that involved an application of a standard other than
that specified in § 2254(d) to one or more claims, compared to 11% of the cases in which all
claims were reviewed using the deferential standard of § 2254(d).
Table 11. (C) Cases granted relief, by district.

District
PA-E
AZ
OH-S
OH-N
TX-W
AL-N
TX-N
TX-E
TX-S
OK-W
FL-M
CA-C
NV
Total

Number of
cases granted
6
2
2
4
4
2
4
2
6
1
0
0
0
33

Percentage of terminated,
non-transferred cases
receiving a grant
75%
50%
29%
17%
14%
13%
10%
8%
7%
7%
0%
0%
0%
12%

Number of cases
terminated w/o transfer
8
4
7
23
29
16
40
25
82
14
10
4
5
267

Table 12. (C) Percentage cases granted, by whether case includes claim type.

Type of Claim
Roper/Adkins/Ring
judicial bias
IAC on appeal
claim of innocence
jury selection
improper argument
search and seizure
insufficient ev. guilt
ev. error guilt phase
Brady-related
sentence claim only
any IAC claim
IAC for sentencing
87

Percentage of cases raising claim
that were granted relief
24.1
23.1
20.3
20.0
18.1
18.0
16.7
16.1
15.9
15.3
15.0
14.4
11.6

Percentage of cases not
raising claim granted relief
21.4
10.9
9.0
11.2
8.3
8.0
11.5
10.8
9.6
9.8
21.0
3.6
3.9

Rechecking in May 2007 the docket sheets of capital cases pending in November 2006 revealed that only three
additional cases had ended, two of them in grants. Recalculated, the grant rate would be 13.0% (35 of 270).

Final Technical Report: Habeas Litigation in U.S. District Courts

•

Table 12 shows differences in grant rate by claim type. For example, less than 4% of the
cases that did not raise any claim of ineffective assistance of counsel received relief
compared to 14% of those that did raise such a claim.

NON-CAPITAL CASES. Of the 2,384 cases in our random sample, only seven ended in a grant of
relief for the petitioner. See Appendix C. This is a overall grant rate of 0.29%, or one out of every
341 cases filed. 88 As a percentage of the 1986 non-transferred, terminated cases, the grant rate is
0.35%, or only one in every 284 cases terminated without transfer. 89
o The average disposition time for the grant cases was 14.5 months compared to 9.5 months for
other non-transferred terminated cases
o The average time between state judgment and the beginning of federal habeas for the seven grant
cases was shorter. It was 2.8 years compared to a 6.3 year-average for other cases.
o All granted cases included orders that the prisoner be released or retried. Five of the granted
cases involved challenges to a conviction, four were challenges to judgments after jury trials, one
was a challenge to a guilty-plea conviction. One grant case was a pretrial challenge to the state’s
effort to retry the petitioner after a successful appeal; the federal court barred the state from
proceeding. The final grant case was a challenge to a post-trial hearing; the federal court ordered
a new hearing be conducted.
o Two cases involved conviction for murder, one case each involved attempted murder, assault,
robbery, or drugs. Two of the petitioners receiving relief were sentenced to life, the others to
terms of imprisonment ranging from six to 40 years.
o None of the cases ending in a grant raised a claim of innocence, only one grant was based on
ineffective assistance.
o None of the cases ending in a grant included an order of discovery, an order granting an
evidentiary hearing, or a stay for exhaustion.
o Three of the seven prisoners receiving relief had retained attorneys.

Given the size of the sample, this suggests that the grant rate nationwide for all non-capital cases filed is no higher
than 0.51% (one in every 196 cases) and no lower than 0.07% (one in every 1429 cases). Std. dev. For GRANT =
.05412; Std. Error of the Mean for GRANT = (STD. Dev.) / (Sq. Rt. (N)) = .05412/ (SQ. RT (2,384)) = .0011084; t
(.025, 2280) = 1.96. Confidence Interval = t (.025, 2280)* (Std. Error of the Mean for GRANT) = (1.96) * (.0011084) =
.0021725. Upper limit on GRANT Rate = Mean value for GRANT+ Confidence Interval = .0029 + .0021725 =
.0050725. Lower limit on GRANT Rate = Mean value for GRANT+ Confidence Interval = .0029 - .0021725 =
.0007274.
89
Rechecking in June 2007 the docket sheets for non-capital cases that were pending in October 2006 revealed that 69
additional cases had been terminated, only one of them (a jury bias claim) ending in a grant. This moved the grant rate
up marginally to 0.34% instead of 0.29%.

Final Technical Report: Habeas Litigation in U.S. District Courts

Appeals

NOTICES OF APPEAL
•

CAPITAL CASES. Of the 274 terminated capital cases, 70.8% included a notice of appeal filed by
either party. A notice of appeal was more common in cases with at least one claim denied on the
merits (91.4%) than in cases dismissed without reaching the merits. Of the 25 terminated cases
raising a claim of innocence, 21 (84%) had a notice of appeal. A notice of appeal had been filed in
24 of the 33 grant cases (73%). (An unknown number of cases were coded immediately after
termination, prior to the expiration of the period allowed for filing a notice of appeal; a notice of
appeal may have been subsequently entered in these cases.)

•

NON-CAPITAL CASES. Of terminations other than transfer, 693 (34.8%) had a notice of appeal.
(An unknown number of cases were coded immediately after termination, prior to the expiration of
the period allowed for filing a notice of appeal; a notice of appeal may have been subsequently
entered in these cases.)

RULINGS ON CERTIFICATE OF APPEALABILITY (“CoA”)
•

CAPITAL CASES.
o
District courts reached a decision on a CoA in 193 cases, 133 of them from Texas. Of the
60 cases filed outside of Texas, judges granted a CoA in 70%. Texas district courts granted
a CoA on any claim in 21% of the capital cases concluded.
o
The median time between termination and a CoA ruling by the district judge in these 193
cases was 32 days. There was great variation among the districts, ranging from a few days
to several months. Texas cases averaged 27 days from judgment to a CoA ruling; cases in
other states averaged 42 days.

•

NON-CAPITAL CASES.
o
Of the terminated non-capital cases, 737 included a CoA ruling. In 711 (96%) of these, a
CoA was denied on all claims; in only 26 (3.5%) did the judge grant a CoA on at least one
claim.
o
The median interval from final order to a ruling on a CoA was just under one month (26
days), the average was more than five weeks (38 days). In 90% of the cases with a CoA
ruling, that ruling was entered 84 days or less after the judgment. The longest period
extended over two years (792 days). In nearly 37% of these cases the CoA ruling and case
termination were simultaneous. 90

In an unknown proportion of these cases with simultaneous rulings, the judge had entered the initial final
order disposing of claims prior to ruling on a certificate of appealability, and then later denied a motion for
reconsideration. In these cases, the date on which the reconsideration was denied was entered as both the
termination date and the date of the CoA ruling.

Final Technical Report: Habeas Litigation in U.S. District Courts

PART III. COMPARATIVE FINDINGS

Comparison with pre-AEDPA studies, non-capital cases

Comparing the findings of this study with the findings of earlier research about non-capital habeas
cases reveals that AEDPA may have changed some important aspects of habeas processing, but that other
aspects remain the same. The comparisons appear in table form in Table 13 and are highlighted below.
More serious offenders. Compared to before AEDPA, the information available suggested that a
greater proportion of habeas cases were filed by petitioners convicted of homicide or sentenced to life
imprisonment. Non-capital habeas filers continue to be predominantly those inmates serving lengthy
sentences, sentences long enough for them to still be “in custody” by the time they have been through direct
appeal and post-conviction proceedings in state court. The state exhaustion process can take several years.
Most prisoners who are convicted of felonies have served their sentences by the time their state appeals have
been completed. 91 Only 1.1% of those sentenced to prison in 1998 were sentenced to life, 92 yet cases filed by
those serving life sentences made up 28% of all habeas cases that were filed in 2003 and 2004, up from 21%
of habeas filings before AEDPA. 93 About 12.1% of state prisoners in 2005 were serving sentences for
homicide, 94 yet more than 28% of cases with conviction information in our non-capital study were filed by
prisoners convicted of homicide, up from 23% before AEDPA.
There could be several explanations for the increased proportion of filers who are serving life
sentences or who have been sentenced for homicide. First, the proportion of prisoners serving life sentences
or sentences for homicide in state prisons has increased somewhat post-AEDPA. Life sentences increased
from 0.7% of all felony sentences of incarceration in 1992 to 1.1% in 1998. 95 In 1996 10% of inmates in state
prisons were serving sentences for homicide, compared to more than 12% ten years later. 96
Second, state appeal and post-conviction procedures could be taking longer to complete after AEDPA,
so that fewer prisoners with shorter sentences remain in custody (and thus be eligible to challenge their state
judgment in federal court) by the time state review processes are completed. 97 Additional research would be
91

According to an annual report based on large urban counties, only 32% of convicted felons were sentenced to prison.
Of those sentenced to prison, the median sentence was less than three years, the average sentence was less than five
years, and only 1% received life sentences. Thomas H. Cohen & Brian A. Reaves, Felony Defendants in Large Urban
Counties, 2002 (Feb. 2006 NCJ 210818), http://www.ojp.usdoj.gov/bjs/abstract/fdluc02.htm.
92
See Matthew R. Durose, David J. Levin & Patrick A. Langan, BJS Bulletin, Felony Sentences in State Courts, 1998, at
4 (Oct. 2001 NCJ 190103), http://www.ojp.usdoj.gov/bjs/pub/pdf/fssc98.pdf .
93
We used the same measure for "life sentence" as Hanson & Daley, supra note 26, that is, any sentence including life
with or without parole, as well as life with any term of years. Statistics about conviction types and sentences imposed are
based on just over half of our cases where this information was available. See pages 19-21 supra. It is possible that the
addition of the missing information would produce a different finding, including the finding that the proportion of habeas
filers who are serving life sentences is the same or even lower than it was before the passage of AEDPA.
94
See Harrison & Beck, supra note 53.
95
See Patrick A. Langan & Helen A. Graziadei, BJS Bulletin, Felony Sentences in State Courts, 1992 (Jan. 1995 NCJ
151167), http://www.ojp.usdoj.gov/bjs/abstract/felsent.htm; Durose et al., supra note 92.
96
See Christopher J. Mumola & Allen J. Beck, BJS Bulletin, Prisoners in 1996 (June 1997 NCJ 164619),
http://www.ojp.gov/bjs/abstract/p96.htm, at Table 13; Harrison & Beck, supra note 53.
97
This would be expected if between the mid-1980s and the mid-1990s there had been an increase in either the claims
raised per state criminal appeal, or in the criminal caseload of state courts generally. It is also possible that the
parole release rate for life prisoners has also declined. See Woods v. Marshall, 183 Fed. Appx. 620 (9th Cir. 2006)

Final Technical Report: Habeas Litigation in U.S. District Courts

required to examine these hypotheses. If further research confirms that this increase is an ongoing trend,
prisoners serving life sentences could soon account for a third of all filings in non-capital cases.
Over 92% of non-capital petitioners filed and litigated their habeas cases without counsel, basically
the same proportion that did so prior to AEDPA. This was expected. The law regarding the provision of
counsel in federal post-conviction proceedings has remained unchanged.
No reduction in filing time. AEDPA's statute of limitations for filings was designed to reduce the
delay between the imposition of judgment in state court and the review of that judgment in federal habeas
corpus proceedings. Instead, the average period has increased from about five years before AEDPA 98 to over
six years for the cases in this study filed in 2003 and 2004. 99 Even considering only the cases that were not
time-barred by the new statute of limitations, the average filing period is 5.6 years, an increase of half a year
from the average filing period before AEDPA. Cases that challenge post-conviction administrative decisions
that revoked or denied good time or parole also tend, on average, to reach federal court later than cases that
challenge the underlying state criminal judgment. Setting these administrative challenges aside and looking
only at cases challenging state judgments that were not deemed to be barred by the statute of limitations, the
average filing period is 5.1 years, still slightly longer than the pre-AEDPA average for all cases.
Two developments could explain this expansion of the filing period, despite AEDPA's one-year filing
deadline. First, the one-year limitations period is tolled for “properly filed” direct and collateral state review
proceedings. The increase in the pre-filing interval could be the result of an increase in the average time for
direct and collateral review of state criminal judgments in state court over the period between the pre-AEDPA
study and this study. Empirical information about the time for state post-conviction review is not yet available
to test this hypothesis. A second explanation for an increase in the average time that elapses between state
judgment and federal filing could be an increase in the proportion of non-capital habeas cases that challenge
not the underlying state judgment but a later, post-conviction administrative decision. Unfortunately, there is
no information on the proportion of habeas filings prior to AEDPA that challenged a good time or parole
revocation or denial, so testing this hypothesis would also require additional research.

(noting "the statistics Woods adduces evidence a sharp decline in the number of life prisoners granted parole"). We also
considered whether the rise in the proportion of all non-capital cases filed by lifers might be related to an increase in the
proportion of cases challenging the administration of sentence (disciplinary hearings, parole release) rather than the
underlying conviction, if those who are sentenced to life with the possibility of release have more opportunities to file
these types of challenges than do prisoners serving shorter sentences. But we found that of the cases filed by prisoners
serving a term of years other than life, 13% challenged the administration of sentence compared to only 6% of prisoners
revealed to be serving life sentences. The data does not distinguish between life with and without possibility of parole.
98
A study of cases filed in six districts in the mid-1970s found that "most petitioners filed [their federal habeas cases]
within a few years of conviction."Allen et al., Federal Habeas Corpus and Its Reform: An Empirical Analysis, 13
RUTGERS L. J. 675, 704 (1982). A later study of cases that were terminated in 1992 from 18 districts found that the time
period had stretched to five years. Hanson & Daley, supra note 26.
99
This is based on the 57% of non-transferred cases in our non-capital cases that included information on state
conviction, sentence, or judgment date. See page 21 supra. It is possible that, excluding the cases that are missing, this
information inflates the average filing period nationwide, but we believe any bias would cut the other way. Two of the
states with the longest periods for stays for exhaustion (an indication of lengthy state post-conviction processing) are
California and New York. Over 55% of California cases lack state conviction, sentence, or judgment date information,
53% of cases from New York lack this information. Together these two states account for more than 29% of all cases
that are missing this information. California also has a relatively high percentage of administrative challenges. These
tend to be filed later than challenges to the state criminal judgment.

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 13. (NC) Comparison of findings with findings of pre-AEDPA studies.
Feature
Sample

Conviction
Sentence
Filing time
Counsel
Claim type

Claim No.

Disposition
Type and
Reason

Disposition
Timing
(omitting
transfer
cases )

Sample from how many districts
Sample size – number of cases
Sample strategy – how cases identified
% cases w/ homicide as most serious conviction
% tried by jury
% cases with life sentences imposed in state cour
Median sentence for petitioners with term of year
Average time from judgment to federal filing
% cases with pro se petitioners
% cases in which court appointed counsel
% raising ineffective assistance of counsel
% raising prosecutorial misconduct
% raising jury instruction error
% raising Fourth Amendment violation
% raising double jeopardy violation
% claim of invalid plea
% raising insufficient evidence
% cases with one claim
% cases with 2 claims
% cases with 3 claims
% cases with 4 or more claims
% dismissed w/o reaching the merits
% granted relief
% all claims dismissed for failure to exhaust
% dismissed because successive (same claim)
% dismissed as abuse of the writ (new claim)
% dismissed as time-barred
% defaulted
Median time beginning to termination
10% quickest lasted
10% slowest lasted
25% quickest lasted
25% slowest lasted
Average time to resolve on the merits
Average time to resolve without reaching merits
Average time to resolve case with single claim
Average time to resolve case with two claims
Average time to resolve any case with a grant
Average time to resolve any case with no grant
Average time in federal court (pending cases
included, measured through October 2006)

* Of the 64% non-transferred cases with claims information available

100

Of the 66% of non-transferred cases with this information.
Of the 57% of non-transferred cases with this information.
102
Of the 60% of the non-transferred cases with this information.
103
Of the 57% of the non-transferred cases with this information.
101

H&D
18
1976 cases
1992 terminations
23
n/a
21
n/a
4.9 years
93
4
25% claims
6 (claims)
n/a
5 (claims)
n/a
n/a
n/a
31
26
30
11
65 % claims
1 (claims)
57 (claims)
3 (claims)
5 (claims)
n/a
12 (claims)
6 months
29 days or less
> 761 days
83 days or less
> 379 days
477 days
268 days
211
270
n/a
n/a
n/a

F&M
8
3346 cases
filings 1990-2
22
66%petitioners
16-24 years
91
n/a
45%petitioners
16 (petitioners)
10 (petitioners)
14 (petitioners)
4 (petitioners)
8 (petitioners)
14 (petitioners)
27
27
21
25
1.04(petitioners
32
2.0
2.6
n/a
6.2 (cases)
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a

This study
All districts
2384 cases
filings 2003-4
28 100
65% cases 101
28 102
20 years
6.3 years 103
92
3
50% cases*
23 (cases)*
15 (cases)*
9 (cases)*
6 (cases)*
15 (cases)*
19 (cases)*
29*
16*
14*
41*
42 *
0.35 (cases)
11
6.8* combined
22
13* (cases)
7.1 months
25 days or less
> 679 days
76 days or less
>412 days
398
204
203
323
440 (14 mos)
289 (9.5 mos)
11.5 mos

Final Technical Report: Habeas Litigation in U.S. District Courts

More claims raised per case. The proportion of non-capital cases with four or more claims has
greatly increased. Prior to AEDPA, between 11% and 25% of cases included four or more claims. Over 41%
now include four or more claims. There are at least two possible explanations for this jump in the number of
petitions with four or more claims. The first is that AEDPA's limits on successive petitions may be prompting
more petitioners to bring all of their allegations of error at once in a single petition. Second, it seems that
many prisoners simply list all the claims raised earlier in state post-conviction proceedings. If the number of
claims that are raised in state court is increasing, this may result in an increase in the number of federal claims
as well.
The types of challenges that prisoners raise to their underlying convictions and sentences did not
change much after AEDPA. About half of all petitioners raise an ineffective assistance claim, a proportion up
only slightly from before AEDPA, and there is a slight increase in the proportion of cases raising
prosecutorial misconduct, instructional error, invalid plea, double jeopardy, and insufficient evidence of guilt.
Many assume that habeas corpus as applied to state prisoners is only a remedy for addressing
constitutional flaws in a state defendant's conviction or sentence. However, a sizeable portion of the habeas
corpus cases filed in federal court challenged something else entirely: prison administrative decisions, parole
and probation revocation proceedings, or pretrial detention. 104 Without better comparative information, we
cannot know whether the frequency of these administrative challenges changed after AEDPA. The study
suggests that administrative challenges make up a majority of habeas filings in at least one district, the
Southern District of Indiana, and a large portion of the filings in several other districts. 105
One in five cases dismissed as time-barred. Of the cases terminated in our study (excluding those
ended by transfer), more than one of every five filings (22%) was dismissed as time-barred under AEDPA's
new statute of limitations. It is not known how many cases prior to AEDPA were dismissed under Rule 9(a)
for delay in filing. That information was not collected by the pre-AEDPA studies available to us as
comparisons. Because the rule was much less stringent than the new limitations period, it was probably a
much lower percentage.
Fewer cases dismissed for exhaustion. A smaller proportion of dismissals after the enactment of
AEDPA were for failure to exhaust (11% of cases compared to between 32% and 57% before AEDPA). But
one should not conclude from this that a greater proportion of petitioners under AEDPA are exhausting their
claims before filing in federal court. Unexhausted claims may be dismissed after AEDPA for other reasons,
including 1) return to state court would be barred by state law, so the claim is dismissed as procedurally
defaulted; 2) the claim lacked merit and so is denied under AEDPA’s provision that allows denial on the
merits of unexhausted claims; or 3) the unexhausted claim was filed late, and is dismissed as time-barred.

104

Two studies evaluating habeas cases filed in the 1970s found that a large portion of these cases involved challenges to
civil commitment, parole revocations, and pretrial restraint. See Shapiro, supra note 2, at 329-30 (noting that as many as
40% of the cases challenged something other than the validity of the petitioner's conviction), id. at 321 ("It is not correct
to assume that the problem of federal habeas corpus for state prisoners is exclusively a problem involving attacks on
state court convictions in criminal cases. A substantial number of other challenges to allegedly illegal detention must
also be considered in appraising the merits of the federal habeas corpus jurisdiction."); Karen M. Allen, Nathan A.
Schactman & David R. Wilson, Federal Habeas Corpus and its Reform: An Empirical Analysis, 13 RUTGERS L. J. 675,
675 (1981-1982). One of the two studies conducted immediately prior to AEDPA noted that only 3% of the cases raised
challenges to prison decisions impacting good time credit, Flango & McKenna, supra note 40, and did not discuss
parole-related challenges. Hanson & Daley, supra note 26, did not mention challenges to administrative decisions.
105
Unlike many district forms and the AO's model petition form that ask only about the criminal judgment, the forms
used in two districts with a high proportion of cases challenging something other than the state criminal judgment (IN-S
and TX-S) specified which proceeding was challenged: the criminal judgment, a disciplinary hearing, or a revocation
proceeding.

Final Technical Report: Habeas Litigation in U.S. District Courts

Predictions that the post-AEDPA practice of “stay and abeyance” would add to the burden of district
courts 106 appear not to have been realized for non-capital cases, at least not by the time these cases were
coded in 2006. By 2003 and 2004 (the time during which the non-capital cases in this study were filed), most
circuits had endorsed the practice as a way of reconciling the state exhaustion requirement with AEDPA’s
filing deadline. Yet we found that stays were relatively rare in non-capital cases (less than 3% of cases), and
concentrated in districts located in the Second, Sixth, and Ninth Circuits.
Procedural default unchanged. Default rules were not changed by AEDPA, and default appears to
be applied by district courts at roughly the same rate as it was before AEDPA.
Successive petition dismissals in district courts unchanged. District courts are rejecting habeas
filings as successive petitions today at about the same rate that second or successive petitions were rejected by
district courts prior to AEDPA. AEDPA now requires a petitioner to obtain permission from the court of
appeals before filing any second or successive petition, however. It is possible that many additional
successive petitions are being turned away by the courts of appeals. Our findings reflect only those cases in
which the petitioner came first to the district court before applying for permission to file a successive petition
in the court of appeals. Without an evaluation of how many applications to the courts of appeals were
rejected as successive (a task beyond the scope of this study), the overall effect of AEDPA's restrictions on
successive petitions remains undetermined. 107
The only study to examine the habeas filing rate per prisoner before and after the enactment of
AEDPA found that rate in 2000 was no lower than it was before the passage of the successive petition bar in
AEDPA. 108 A higher filing rate, despite vigorous enforcement of the successive petition bar by both district
and appellate courts, could be related to any number of factors. They include a greater proportion of
prisoners imprisoned for multiple judgments (each of them subject to a separate petition); an increase in the
number of challenges filed by prisoners to administrative decisions concerning release or good time,
objections that would not be considered successive to the prisoner's prior petition challenging the underlying
state court judgment; or a response to particular decisions of the Supreme Court providing new grounds for
objection that also would not be considered successive (more likely in capital cases). Additional research
would be required to test these various hypotheses.
More merits dispositions. The study also suggests that even with the dismissal of late petitions,
federal courts are reaching the merits of claims in a larger percentage of cases than they did before AEDPA.
Fewer writs granted. The rate at which the writ is granted by district courts has dropped. Prior to
AEDPA, one in 100 cases resulted in relief. Our nationwide random sample of 2384 filings (1986 terminated,
non-transferred cases) included only seven cases in which relief was granted, a rate of one in 284 cases. 109
106

See supra note 27.
Nor do we know how many of our district court cases were initiated only after a court of appeals had granted
permission to file a successive petition. The litigation in the court of appeals would not necessarily be mentioned in the
documents available from PACER in the district court file.
108
See Cheesman et al., supra note 2; Cheesman et al., supra note 35 (using filing statistics and state-wide prison
population figures from 1992 through December 2000, finding no decrease in the number of state habeas cases filed in
district courts when compared to state prison populations after AEDPA).
109
Shortly before this final report was submitted in July 2007, an additional check as of June of 2007 of the cases in the
sample that had been pending revealed that 69 additional cases had been resolved, but only one had resulted in a grant of
the writ. This was a case from LA-W, where the court found that the jury was biased due to the inclusion of a juror who
failed to reveal that he was closely connected to the victim and her family and that his own father had been the victim of
a prior offense by the defendant. The additional grant brings the grant rate up only slightly to one in every 257 cases
instead of our initial finding of one in every 284 cases. These numbers are so small that confirmation of an overall
decrease in relief would require additional research using a larger sample.
107

Final Technical Report: Habeas Litigation in U.S. District Courts

AEDPA contains several provisions that could contribute to a lower grant rate, including: 1) greater
restrictions on evidentiary hearings; 2) limitations barring review of claims that might otherwise be found
meritorious (successive petitions, statute of limitations); and 3) a more deferential standard of review for state
decisions of both fact and law. Without also examining appellate outcomes, it is not possible to know whether
AEDPA has reduced overall relief rates for non-capital petitioners.
It is also possible that the consideration of only district court orders that grant the writ may
underestimate the role of habeas litigation in changing the terms of a prisoner's custody. Favorable outcomes
for petitioners that would not show up as a grant could include the state's agreement, without court order, to
reduce a sentence or conduct a resentencing, to vacate a conviction or retry a petitioner, to recalculate a
release date, to reverse, dismiss, or redo a state administrative proceeding such as a disciplinary or revocation
hearing, or even to commence or complete a proceeding that for one reason or another had stalled out. 110 If
this occurs, the resolution could appear as a case in which every claim was dismissed, either because the
petitioner sought a voluntary dismissal, or because the claims in the petition were found to be moot.
Longer processing time per case. Overall disposition time per case has increased on average since
AEDPA. The quickest cases are disposed of even faster, but the 25% slowest now take a month longer.
Taking into account pending cases, these non-capital cases average at least a year in federal court before they
are completed. To the extent that AEDPA’s provisions were designed to streamline the overall processing of
individual cases, that result has not been achieved.
It is possible that measured per prisoner, rather than per case, time in federal court has been or will yet
be reduced as a result of AEDPA. If the successive petition bar provides a greater deterrent to the filing of
multiple petitions than that provided by pre-AEDPA law, or faster resolution of later filings, the result may be
more time per case but less time per prisoner. However, information about the processing time per prisoner is
unavailable for comparison before or after AEDPA. 111 This study, for example, sampled individual cases.
We did not attempt to collect information on the number of habeas cases filed challenging a single judgment
by a sample of state prisoners over the life of their confinement, or the total time for resolving all of those
cases. It is not known, for example, what proportion of state prisoners before or after AEDPA filed one or
more habeas petitions. Nor is it known whether multiple filers attacked the same or different state judgments
or whether they attacked separate administrative rulings denying or revoking good time credits or parole,
attacks that would not necessarily be affected by the limits on successive petitions. 112
Mention should be made of another source of comparison for this study. In 2003 Judge Weinstein took on 500
non-capital habeas cases that had been pending in the Eastern District of New York for up to six years. (These 500 cases
were not included in our study, as all of them had been filed prior to January 1, 2003.) The concentrated disposition of
500 cases provides a statistical snapshot of post-AEDPA habeas proceesing by one judge in one federal district.
Compared to courts nationwide processing cases filed in 2003-2004, Judge Weinstein dismissed fewer cases as
successive or time-barred and addressed more cases on the merits. He granted relief in 9 of the 494 cases terminated
without transfer, a grant rate of 2.0%. This is nearly six times higher than the grant rate of 0.35% in our nationwide
sample (7 of 1986 terminated, non-transferred cases). With one exception, the grants were conviction-related: three
counsel claims, two jury instruction errors, and a smattering of other trial related claims under Brady, Chambers, and the
Confrontation Clause. Of the 500 cases, 8.9% were "administratively closed" - stayed for exhaustion or unripe, a rate
more than twice the rate of stays for exhaustion that we found in non-capital cases nationwide. In re Habeas Corpus
Cases, 298 F.Supp.2d 303 (E.D.N.Y. 2003).
110
See, e.g., Shapiro, supra note 2, at 340-42 ("there are a substantial number of cases -- difficult to count because they
are not often disclosed by the face of the record -- where habeas corpus proves a useful remedy even though the matter
prayed for is not granted. For the most part, these are cases in which state processes have for one reason or another been
derailed and federal assistance has been helpful in getting them back on the track . . . . "). We also found several
examples in our cases where a state appeared to be prompted to act after the filing of a petition in federal court.
111
This information was not available in either of the pre-AEDPA studies that we used as comparisons.
112
See note 108, supra.

Final Technical Report: Habeas Litigation in U.S. District Courts

Fewer evidentiary hearings. It is not likely that greater processing time for non-capital cases after
AEDPA is the result of an increase in the use of evidentiary hearings. The pre-AEDPA studies did not report
the frequency of evidentiary hearings, but one federal report using the AO data noted that 1.1% of all habeas
cases received an evidentiary hearing, 113 a rate that is higher, not lower, than the rate after AEDPA.

Comparison with pre-AEDPA studies, capital cases

Empirical information about capital cases prior to AEDPA is more limited than information about
non-capital cases, but some comparisons are possible. Table 14 includes a summary of comparative
information for capital cases.
Filing times. It takes on average 7.4 years after sentencing for the death row inmates in our study to
file in federal court; an average of 6.5 years for them to file their the first petitions. No comparative
information about filing periods is available from studies of capital cases prior to AEDPA.
Few cases are time-barred. The new filing deadline appears to have been the basis for few
dismissals in capital cases. Only 4.1% of the filings in our sample were dismissed for this reason.
One in six cases is stayed. Stays were entered in 17% of the capital cases for the petitioner to exhaust
claims in state court. This practice was prompted by AEDPA’s bar against statute of limitations.
Fewer evidentiary hearings. After AEDPA, most capital habeas cases continue to be concluded
without evidentiary hearings in district court, and the percentage of cases with evidentiary hearings appears to
have dropped. One study of published capital cases concluded in the federal courts of appeals by 1995
reported that 19% received an evidentiary hearing in federal court. Among the cases in our sample, only
9.5% received an evidentiary hearing. It is likely that evidentiary hearings will be granted in at least some of
the 85 cases in the sample that remained pending and that had not had an evidentiary hearing, but hearings
would have to be ordered in 35 or 42% of those 85 cases to reach an overall rate for the sample near the 19%
pre-AEDPA figure.
Longer processing time per case. Each capital habeas filing appears to be taking at least twice as
long to finish, on average, than prior to AEDPA. The Federal Judicial Center study prior to AEDPA found
that the average disposition time for a capital habeas case involving a first petition was 15 months,
significantly shorter than the average disposition times that we found in this study. We found averages of 29
months for the disposition of terminated capital cases, 30.4 months for non-transferred first petitions, and 37.3
months so far for all cases including those still pending. It is not known whether AEDPA has had any effect
on total processing time for all habeas challenges filed by a given death row inmate. No information about
processing time per prisoner (as opposed to per filing) is presently is available for comparison before or after
AEDPA.
Given how long capital habeas cases presently take to resolve, the statutory 450-day time limit for
resolving capital habeas cases from states that may qualify for expedited review under AEDPA will pose a
challenge for courts. In the districts that we examined, the average processing time for capital cases is well
over two and a half times that long. Not one of the 13 districts in this study has completed its capital habeas
cases in less than 500 days on average, even excluding stayed time.
113

Federal Courts Study Committee, Working Papers and Subcommittee Reports, June 1, 1990, at 482 (cited in Schriro
v. Landrigan, 127 S. Ct. 1933, 1954 (2007) (Stevens, J., dissenting)). This figure should be considered with some
skepticism if it is based on the "procedural progress" variable in the AO data for these cases, see note 36 supra.

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 14. (C) Comparison of findings with findings of pre-AEDPA studies.
Feature compared
Sample from how many courts

Fagan, Liebman et al.

FJC – 1995

This study

all districts & circuits

all districts

13 districts

500

368

Sample size – number of cases
Sample

Data collected from
Barred by statute of
limitations
Cases stayed for
exhaustion
Evidentiary
hearing
Disposition time

Grant rate
Higher probability of
relief if

598
Published terminations in
Sample strategy – how cases identified both courts of appeals
and district courts
Pending cases included?
no
published decisions

AO terminations AO + PACER
filings in district
district court
no

yes

PACER

% of cases with time-barred claims
% of cases with unexhausted claims,
stayed while petitioner returns to state
court
% of cases including evidentiary hearin
19%
in federal court
Av time in district ct, terminated,
non-transferred cases
Av time in district ct, terminated,
non-transferred first petitions
% terminated, non-transferred
40%
first petitions receiving relief
federal evidentiary hearing held, lower
yes
number of claims raised

4.1%
17.0%
9.5%
15 months

29 months

17 months

30 months
12.4%
yes

Fewer writs granted. The study suggests that fewer death row inmates are receiving relief in federal
district court after AEDPA. About one in eight or 12.4% of 267 terminated capital cases that filed in 2000,
2001, and 2002 received relief. This is much lower than the 40% grant rate reported by Fagan et al. for the
much older capital cases that had already made it through both the federal district and appellate courts by
1995. 114 Excluding the 14 cases dismissed as second petitions and looking only at first petitions, as Fagan et
al. did, brings the grant rate in this study up to 13.0%. Given that cases ending in a grant of relief take longer
to complete on average than cases in which relief is denied, we expect that the grant rate for the 95 cases still
pending will be at least 13%. But the overall grant rate for the sample is not likely to approach the 40%
reported by Fagan et al. It would be surprising if by the time all of the cases are completed at the court of
appeals level, the existing 33 grants expand to 139 (40% of the non-transferred, non-successive petitions in
our sample), even with 95 cases pending.

114

Unlike this study, which included cases filed in district court, the study included only cases for which review had
been completed in federal courts. See Fagan et al., supra note 42, at 30 ("For a federal habeas case to have “finally
reviewed” a capital verdict within the study period, all of the following events must have occurred in the case within that
period: (1) a United States District Court must have denied habeas corpus relief from the capital judgment, thereby
approving the judgment, or [have] granted habeas relief from the capital judgment (either the conviction or sentence) on
one or more grounds; (2) if an appeal was timely filed, a United States Court of Appeals must have approved or reversed
. . . ; and (3) if certiorari review was timely filed, the United States Supreme Court must have either (a) denied review or
(b) granted review and affirmed or reversed . . . "). 38% of the 568 death sentences imposed between 1973 and 1990 that
had completed review in the federal courts were overturned. Id. See also Gelman et al., supra note 42 (reporting 40%
figure for the same study).

Final Technical Report: Habeas Litigation in U.S. District Courts

Comparison of non-capital cases with capital cases

Capital habeas cases are quite different from non-capital habeas cases, although both types of cases
are governed by the same statutory provisions. The single most important difference is that all but 7% of
death row filers have counsel to assist them in seeking federal habeas relief, while all but 7% of non-capital
prisoners proceed pro se. A summary of other comparative findings appears in Table 15.
Filing period. It takes on average one extra year for death row inmates to reach federal court
compared to non-capital petitioners, that is, 7.4 years compared to 6.3 years if all cases are considered, 6.5
years compared to 5.6 years if time-barred cases are excluded.
Number and type of claims. Capital petitioners raised on average seven times as many claims as
non-capital petitioners. Compared to petitions prepared by capital counsel, sometimes well over 100 pages
long, non-capital petitioners generally filled in the habeas form used in the individual district. Many such
forms have space for only three or four claims, so that to raise additional claims required the prisoner to
append additional sheets.
More than eight of every 10 capital petitions included a claim of ineffective assistance of counsel,
compared to only half of the non-capital petitions. Claims of innocence, insufficient evidence, and Brady
violations were more frequent in capital cases than in non-capital petitions. Petitions challenging the sentence
alone and not the conviction were more common in non-capital cases than in capital cases.
Non-capital petitions included a surprisingly large group of challenges to administrative decisions.
Eighteen percent of non-capital petitioners challenged a state administrative decision that had occurred after
state judgment (disciplinary proceeding, revocation, parole denial, etc.). Less than 2% of capital cases
challenged only the execution method or competency for execution, and not the underlying judgment.
Statute of limitations. Five times as many non-capital cases were considered time-barred than capital
cases. The greater frequency of time-barred cases for non-capital prisoners is expected given that unlike death
row inmates in most states, non-capital habeas filers navigate the post-conviction process and its deadlines
without counsel. 115 Although the statute of limitations issue may yet be unresolved in some of the still
pending capital cases, the rate at which courts explicitly rejected the limitations defense was similar for both
capital and non-capital cases. It is 1.5% of non-transferred non-capital cases compared to 1.9 % of nontransferred capital cases.
Exhaustion and procedural default. In over 10% of the non-capital cases, all claims were dismissed
because of the failure to exhaust, compared to less than 4% of capital cases. Stays for exhaustion were nearly
seven times more likely in capital cases than in non-capital cases. Capital cases were put on hold for about
two years on average compared to the average one-year stay in non-capital cases. Procedural default is over
four times more likely to be the basis for dismissing a claim in capital cases than in non-capital cases.
Teague. Dismissal due to non-retroactivity was also more prevalent in capital cases. A reference to
Teague was almost entirely absent in judicial rulings in non-capital cases, although nearly one in four capital
cases included a claim dismissed as Teague-barred. In Texas, this may be the result of the custom and
115

See, e.g., Lawrence v. Florida, 127 S. Ct. 1079 (2007) (Ginsburg, J., dissenting) (“An attorney, of course, is better
equipped than a pro se petitioner to clear procedural hurdles, including shortened timelines.”); Stevenson, supra note 24,
at 350 (claiming reason for untimely filings is lack of counsel); Jessica Feierman, "The Power of the Pen": Jailhouse
Lawyers, Literacy, and Civic Engagement, 41 HARV. C.R.-C.L. L. REV. 369, 379 (stating that the statute of limitations
"has an especially adverse impact on prisoners with low literacy levels" who may not understand the time limit or "may
have difficulty concluding their research and filing their petitions before it expires").

Final Technical Report: Habeas Litigation in U.S. District Courts

practice of the capital case lawyers in the office of the attorney general (but not those defending non-capital
cases) to emphasize the Teague defense.
Litigation steps. Depositions, examinations, and evidentiary hearings were rare in non-capital cases.
In about one of every 10 capital cases (including those still pending) the judge had ordered an evidentiary
hearing, and an even larger proportion included an order of discovery. Amended petitions were three times as
frequent in capital cases.
Disposition time. Capital cases in our sample have taken on average at least three and a half times
longer to complete than non-capital cases, including pending cases (37.3 months on average in federal court
compared to 10.6 months for non-capital cases). This difference is likely to be even greater once all of the
still pending cases are completed.
Merits review. Overall, about 28% of the capital cases were dismissed without reaching the merits of
any claim compared to 42% of non-capital cases. This study did not track how many cases applied each of
what are arguably four different approaches to harmless error review in habeas proceedings, 116 but did track
when a court explicitly found that an error was not harmless. Only four non-capital cases and four capital
cases found an error not harmless, all ending in a grant. Most of the 33 granted capital cases involved claims
not subject to harmless error analysis, such as the Eighth Amendment ban on executing those who are
mentally retarded or who committed their offenses before turning 18, ineffective assistance of counsel, or the
failure to disclose exculpatory evidence.
Grants. The rate at which petitions are granted in capital cases is 35 times higher than the rate in noncapital cases, a difference that is likely to increase as pending capital cases in our study are resolved.

116

Jeffrey S. Jacobi, Note, Mostly Harmless: An Analysis of Post-AEDPA Federal Habeas Corpus Review of State
Harmless Error Determinations, 105 MICH. L. REV. 805 (2007). Shortly before this writing, the Supreme Court settled
in part a dispute in the lower federal courts about how and when to apply the concept of harmless error in federal habeas
review of state decisions. Fry v. Pliler, 127 S. Ct. 2321 (2007) (holding that regardless of whether the state court
recognized an error and reviewed it for harmlessness, a federal court evaluating a habeas claim must assess the impact of
any error using the "substantial and injurious effect" standard of Brecht v. Abrahamson, 507 U.S. 619 (1993)).

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 15. Summary of findings, capital and non-capital cases compared.
Sample
Capital cases
Total cases in sample
368
Sample cases started in which years
2000, 2001, 2002
Districts included in sample
13
% of cases in time period represented
100%
Filing period
Average /median time state conviction to fed filing
7.4/6.5
Petitioners and representation
% cases filed by male/white
98.9/40.9
% cases petitioner without counsel
7.1
% cases filed by petitioners with multiple convictions
42.0
% cases filed by petitioners who pleaded guilty or nolo
2.7
Petitions, claims raised
% cases with no petitions filed
4.6
Average/median number claims per case
28/18
% raising claim of ineffective assistance of counsel
81.0
% raising a claim of innocence of conviction
10.8
% raising claim of false, lost, or undisclosed evidence
43.1
% challenging prison or revocation decision/execution
1.4
% challenging sentence/sentencing proceeding only
5.1%
% challenging sufficiency of evidence of guilt
25.5
Litigation
% cases with amended petitions
35.2
% answer or motion to dismiss filed by state
82.9
% cases including a stay for exhaustion
17.0
Average/median length of stay for exhaustion (in months)
23
% cases with discovery ordered
12.5
% cases with evidentiary hearing held
9.5
Average/median # of docket entries
55/44
Disposition type and reason
% cases transferred to another district
1.6
% cases with all claims voluntarily dismissed**
11.6
% cases dismissed as time-barred**
4.1
% cases dismissed as successive**
5.6
% cases all claims dismissed as unexhausted**
3.4
% cases including at least one procedurally defaulted claim**
53.3
% cases including at least one Teague-barred claim**
23.9
% cases dismissed or denied, w/o reaching merits
27.7 (64/234)
% terminated cases dismissed w/o reaching merits **
24.0 (64/267)
% granted relief on any claim**
12.4 (33/267)
% of grants on sentence only
69.7
Disposition time (in months)
% still pending in federal court
25.7 (95/368)
Average/median time to resolve non-transferred terminated cases 28.8/28/.7
Average/median time in fed ct non-transferred (includes pending) 37.3/36.4
* Of non-capital cases with claims information available – 1521 (63.8%)
** Of non-transferred terminated cases (267 capital cases; 1986 non-capital cases)

117

Non-capital cases
2384
2003, 2004
87 (nationwide)
6.5%
6.3/ 5.7 117
96.2/unavailable
92.3
51.1
32.2
3.0
4/3*
50.4*
3.9*
13.0*
17.8 118
12.9*
18.9*
11.8
58.2
2.6
13.1
0.3
0.4
18/15
8.4
3.2
21.7
6.8
10.9
13.3
0.4
42.0 (631/1311) 119
31.8 (631/1986)
0.35 (7/1986)
0.0
8.3 (198/2384)
9.5/7.1
10.6/7.3

Of 1299 cases with information on state dates (54.4% of all cases).
Of 1837 (77%) cases with information on type of proceeding challenged.
119
1311 of the 1979 non-transferred, terminated non-capital cases ending other than in grant had claims information.
118

Final Technical Report: Habeas Litigation in U.S. District Courts

PART IV. EXPLANATORY ANALYSES

Capital cases, time from state judgment to federal filing

Analysis design

Given the variation between states in appellate review times for capital cases, 120 variation in the time
before the federal filing was expected. See Table 2. To examine whether some of this variation could be
explained by factors other than state-specific law or practice, we used regression analysis.
The features expected to be associated with longer filing times included: conviction by trial not plea
(presumably it would take state courts longer to review a trial-based conviction and sentence than it would
take to review a plea-based case); dismissal of the federal petition as successive (presumably second petitions
generally come later than first petitions); a greater number of claims in the federal petition (suggesting that the
state courts had more claims to deal with as well); the presence of various specified claims likely to take
longer in state court (competency proceedings, for example); dismissal of federal claims as filed too late;
filing of the state record in federal court (suggesting the state proceedings were complete); and the presence of
an order staying execution of the death sentence (as opposed to cases where the state has not set an execution
date, which may be due to ongoing state proceedings). A finding by the federal court that the petitioner had
failed to properly raise, develop, or exhaust a claim in state court may indicate a more abbreviated state
review process, so we expected default, exhaustion, hearings, and stays could be associated with shorter filing
times.
2.

Results

The analysis accounted for about 31% of the variation in filing time. The results are summarized in
Table 16. Of the variables considered, the identity of the individual state was the most influential (using
Oklahoma, with the shortest average filing period, as a comparison). 121 Five other points deserve notice.
First: As expected, petitioners convicted by trial reached federal court later, 60.8% more days later, on
average, than those convicted by plea. This is possibly related to the additional challenges that can be raised
after a trial-based conviction.
Second: Filing intervals generally did not vary by the number of claims raised in federal court, the
closest measure available for the number of claims raised in state court, nor by type of claim, with one
exception. Petitioners who alleged unconstitutional delay before trial or appeal, or a violation of their own or
the public's right to be present at trial, filed sooner than petitioners who did not include such a claim.
Third: Cases eventually dismissed as successive were indeed filed later than first petitions, but cases
dismissed as time-barred were not filed any later than cases that were not time-barred. 122 This suggests that
AEDPA’s statute of limitations is not barring review in cases that take the longest time to reach federal court.
120

Latzer & Cauthen, supra note 43.
The addition of other variables may have improved the explanatory power of our models, but were unavailable to us
in this study, which was primarily limited to information from the federal district court case. The number of claims
raised in state court, the presence and type of post-conviction counsel for the petitioner, and the presence or absence of
rigid filing deadlines for state post-conviction proceedings would all be useful additions to future examinations of filing
time in capital habeas cases.
122
The 11 capital cases barred by the statute of limitations are all detailed in Appendix B.
121

Final Technical Report: Habeas Litigation in U.S. District Courts

Rather the cases with the longest time periods before filing include intervals for which the limitations period
is being tolled. 123
Fourth: Whether the case was later stayed for exhaustion 124 was not significantly related to the amount
of time it took to get to federal court either. This suggests that cases in which the federal court ultimately
decides that the petitioner should be allowed to return to state court to exhaust claims are reaching federal
court no sooner than cases in which a stay is not ordered. Cases in which the federal court found any claim
defaulted were filed 13.2% additional days later, not earlier, than cases with no defaulted claims. The
presence of a stay of execution was not related to the length of time before filing.
Fifth: The identity of the individual state remained the one factor that accounted for the most variation,
even after controlling for case-level factors. Among the nine states examined in our study, petitioners in
Oklahoma and Texas reached federal court the fastest, with no significant difference between filing times in
the two states once other factors were taken into account. Petitions from Florida were filed on average 164%
days later than the petitions from Oklahoma. Alabama's filing periods averaged 147% longer; 125
Pennsylvania's 150% longer. California and Nevada petitioners began their federal cases after a time lag that
was 112% longer than those from Oklahoma inmates, on average. Arizona capital petitioners reached federal
court about 64% later, and Ohio petitioners filed their federal cases after a period that was 49% longer than
the filing period for Oklahoma petitioners.
Notably, Latzer and Cauthen examined the time to complete state direct appeals in capital cases, in
the four states, Florida, Arizona, Texas, and Ohio, and found that the relative times for state appeals were in
almost the opposite order from the order we found for relative periods from state judgment to federal filing.
They found that Florida took the least time on direct appeal; we found that Florida cases were the slowest to
reach federal court. Their study indicated that Texas was the slowest of the four states for completing direct
appeals in capital cases; we found that Texas cases reach federal court more quickly than cases from all but
one of the other states in the sample. This inverse relationship between state appeal time and total filing time
for cases from these four states suggests that the second stage of state court review of capital cases following
direct appeal prior to federal habeas litigation, that is, namely, state post-conviction review, can have a
dramatic impact on the time period that elapses between the time a prisoner is sentenced to death and the time
that prisoner seeks review of his judgment in federal court. 126

123

See Steiker & Steiker, supra note 34, at 1925 (arguing that the one-year statute of limitations will not curb delay in
the review of capital cases because it is tolled for the state review process, which accounts for most of the delay).
124
In another model, we substituted for this variable the variable indicating cases in which every claim had been
dismissed for exhaustion. After controlling for other factors, neither the presence of a stay nor the later decision of the
federal court to dismiss to allow exhaustion appeared to be related to the length time period that elapsed prior to filing.
125
Alabama, unlike the other states, does not provide counsel to indigent death row prisoners for their state postconviction proceedings. See, e.g., Leonard Post, On Their Own, NAT'L L. J., December 1, 2003, at 1.
126
See also Steiker & Steiker, supra note 34, at 1888 (reporting that in Texas "postconviction proceedings start
expeditiously," and "frequently [do] not involve any extensive factual investigation or discovery" as compared to cases
from California).

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 16. (C) Regression of interval in days between state judgment and federal filing (logged). 127

Feature investigated

% difference ^

coeff

(r.s.e.)

Federal court held evidentiary hearing

-0.063

(0.076)

Federal court ordered discovery

0.005

(0.081)

Federal court stayed death sentence

0.005

(0.099)

Federal claims stayed for exhaustion

-0.100

(0.083)

Number of claims raised in federal petition

0.001

(0.001)

Petitioner convicted by trial (not plea)

60.8

0.475*

(0.236)

At least one claim alleging denial of public trial, right to presence, or
delayed trial or appeal

-21.7

-0.245*

(0.105)

0.107

(0.120)

At least one claim dismissed by federal court as time-barred
At least one claim dismissed by federal court as procedurally defaulted

13.2

0.124+

(0.071)

Federal court dismisses petition as successive

95.2

0.669***

(0.142)

State of Pennsylvania

150.4

0.918***

(0.156)

State of Alabama

147.2

0.905***

(0.120)

State of Florida

164.1

0.971***

(0.138)

0.121

(0.110)

State of Texas
State of Ohio

49.0

0.399*

(0.141)

State of Arizona

63.6

0.492*

(0.175)

State of California

112.5

0.754**

(0.318)

State of Nevada

111.9

0.751***

(0.165)

Constant (Oklahoma)

6.831**

(0.331)

Observations (cases ending in transfer excluded)
R-squared
Robust standard errors in parentheses
+ significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%

362
0.31

^For example, a case that is ultimately dismissed as successive by the federal court takes about twice as long to reach federal court (a

95.2% increase) than a case that is not dismissed as successive, controlling for all other factors. Calculated for significant associations
only.

127

Variables not shown that had no significant association with variance in time before beginning of federal case are:
petitioner race; state record filed in federal court; jury instruction claim; improper admission or exclusion of evidence
claim; jury selection misconduct or bias claim; counsel related-claim other than ineffective assistance of counsel; double
jeopardy claim; plea-related claim; ineffective assistance of counsel claim; claim of innocence of guilt; claim of
insufficient evidence of guilt; competency (in state court) claim; denial or delay of appeal or transcript claim; insufficient
evidence of sentencing factor claim; Roper, Atkins, or Ring claim.

Final Technical Report: Habeas Litigation in U.S. District Courts

B. Non-capital case processing time
1.

Analysis design

Federal habeas cases filed by state prisoners who are not on death row may be resolved on the same
day they are filed or many years later.
Figure 21. (NC) Probability of not reaching termination over time, in days.

0.00

0.25

0.50

0.75

1.00

Kaplan-Meier survival estimate

500

1000

1500

analysis time

Figure 16 (Part II), and Figure 21, above, show that about 25% of the cases are terminated after about
90 days, 50% after 200 days and 75% after about 530 days. It appears that all of the cases should be
terminated after about 1500 days. To examine possible explanations for this variation in the time for the
disposition of these cases, we used two separate analyses. For terminated cases, we used multivariate
regression to estimate the association of specific features of habeas litigation with variations in processing
time. The dependent variable for this analysis was the log transformation of the number of days that elapsed
from the beginning of the case to its first termination in district court. Second, we used Cox regression
analysis to investigate the effect over time of individual features of habeas litigation on the likelihood of
terminating. 128
The features expected to be associated with longer processing time included: a petitioner’s attorney,
presence of amended petition, presence of answer or dispositive motion by the state, whether the state record
was filed, use of a magistrate judge for a report and recommendation on the resolution of any claim, record
notation of a prior petition that had been dismissed without prejudice, number of claims in the petition,
whether the case challenged conviction and not the sentence alone, presence of various specified claims, type
of proceeding challenged (criminal judgment v. administrative proceeding such as disciplinary hearing or
128

We excluded cases ending in transfer from both sets of non-capital case analyses, and eliminated from the set of
cases evaluated five cases that diagnostic tests revealed were outliers and that upon inspection were discovered to have
ben miscoded. For the district regressions, we omitted cases from districts with fewer than 9 observations.

Final Technical Report: Habeas Litigation in U.S. District Courts

parole revocation), merits review of any claim, application of a standard other than the deferential standard in
Section 2254(d), notice of appeal, and whether a certificate of appealability was granted.
Features expected to be associated with shorter processing times included whether the case involved a
non-merits dismissal, the application of procedural default to bar any claim, application of statute of
limitations to bar any claim, application of successive petition bar, and whether every claim was dismissed
voluntarily or as unexhausted. To examine whether the novelty of a female petitioner prompted speedier
disposition than in cases with male petitioners, we included petitioner gender in the analysis. 129
Finally, we explored the impact of district, circuit, and state to see if the differences in average
processing times between locations remained after controlling for other features. 130 District specific
caseload 131 and staffing levels were also examined as alternative specifications.
For each different specification (controlling separately for districts, circuit, state, clerk staffing, and
caseload) we ran the regression analysis of termination time in two ways. The first model included all cases
but no claims-related variables (for which many cases were missing information). The second model
included claims-related variables, but compared only cases for which claims information was available.
Tables 17, 18, and 19 summarize the results of the 10 different models for terminated cases. Non-significant
variables are excluded from the tables, but appear in the margin. 132
The Cox regression analyses included basically the same features as those included in the regression
analyses of terminated cases, but examined the effect of those features on the likelihood of termination rather
than on disposition time. The Cox regression analyses omitted variables that would be present only in
terminated cases, such as merits review of any claim, application of the successive petition bar, application of
a standard of review other than that specified in Section 2254, or the presence of a notice of appeal. Tables
showing the results for the analyses of non-capital cases appear in Appendix A.

129

Race information was missing in too many cases to include in the analyses. There were three additional features we
had expected to be related to processing time, but could not include because of their rarity: whether the case included
discovery, whether the case included an evidentiary hearing, and whether the case included a grant of relief.
130
In initial regression models examining termination times for non-capital cases, we included as a variable the number
of docket entries for each case and found it was highly associated with variations in processing time. We omitted this
variable from our final models because we were interested in what might account for that variance in docket entries. In
other words, the number of docket entries functioned as an alternative dependent variable to processing time.
131
Caseload was derived from the AO Tables posted on the AO’s website. The value for each district is the average of,
for each year 2003, 2004, and 2005, the annual state prisoner habeas filings per district (Table C-3A) divided by the the
number of authorized judgeships listed for each district (Table X-IA).
132
Not significantly associated with any difference in processing time when compared to the control locations in any
model were the following: the states of OK, AR, TN, MD, WA (compared to IN); First Circuit (compared to the
Seventh); and the districts of MD, NC-M; VA-W, AL-M, AL-S, MS-N, FL-S, GA-S, AR-E, MN, TN-E, TN-M, WA-W,
OK-W (compared to IN-S).
Also not significantly associated with either delayed or expedited disposition times were any of the following
claim types: claims raising constitutional errors in revocation proceedings; prosecutorial misconduct claims; claims of
lost, destroyed, undisclosed, or false evidence; jury instruction claims, appeal related claims, claims of insufficient
evidence of guilt or sentencing factors, counsel claims other than ineffective assistance, claims of judicial bias, illegal
confession claims, Apprendi claims, ineffective assistance claims, innocence claims, or plea-related claims.
Disposition time was also not related to the following factors after other factors were controlled: the year the
case began, petitioner gender, whether the case challenged sentence alone, whether the petition was voluntarily
dismissed, whether the state record was filed in the federal court, the standard of review applied, whether a notice of
appeal was filed by either party, or whether the district court rejected a certificate of appealability.

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 17. (NC) Regression of disposition time in days (logged), for terminated cases, by clerk and caseload.

Model:
Observ's
Prob > F
R-squared
Constant
amended petition
answer, mo to dism
magistrate j. R&R
no. of claims
petitioner had atty
competency claim
jury misconduct
speed, public, pres.
evidentiary claim
prison discipline
all claims unexhaus
time-barred
defaulted claim
successive pet
other dismissal
Clerk
Caseload

Comparing by average number of pro se law clerks
NO CLAIMS VARIABLES
CLAIMS VARIABLES
INCLUDED
All terminated cases
included
Only cases w/ claims info
1827
1076^
0.000
0.000
0.538
0.578
[%diff] coef
r.s.e.
[%diff] coef
r.s.e.
3.928 *** 0.113
3.944 ***
0.161
[45.1] 0.372 *** 0.057
[29.6] 0.259 ***
0.069
[393.3] 1.596 *** 0.050
[420.7] 1.650 ***
0.079
[34.0] 0.293 *** 0.041
[34.4] 0.296 ***
0.051
[15.6] 0.145 *** 0.023
[9.6] 0.092 *
0.045
[49.0] 0.399 *** 0.076
[22.6] 0.204 *
0.088
[38.7] 0.327 +
0.175
[-30.6] -0.366 **
0.137
[23.6] 0.212 *
0.092
[11.3] 0.107 +
0.060
[-26.7] -0.310 *
0.137
[-21.7] -0.244 *
0.111
[-11.5] -0.122 +
0.066
0.092
0.062
-0.203
0.140
-0.146
0.121
-0.002
0.006
0.008
0.007

comparing by averaged non-cap habeas caseload per judge
NO CLAIMS VARIABLES
CLAIMS VARIABLES
INCLUDED
All terminated cases
included
Only cases w/ claims info
1827
1076^
0.000
0.000
0.545
0.585
[%diff] coef
r.s.e.
[%diff] coef
r.s.e.
3.759 *** 0.115
3.789 ***
0.160
[40.4] 0.339
*** 0.056
[25.6] 0.228 ***
0.068
[405.8]1.621 *** 0.050
[430.7] 1.669 ***
0.079
[28.5] 0.251 *** 0.041
[29.7] 0.260 ***
0.051
[16.9] 0.155 *** 0.023
[9.7] 0.093 *
0.045
[47.8] 0.391 *** 0.076
[23.6] 0.212 *
0.086
[36.9] 0.314 +
0.172
[-32.1] -0.387 **
0.085
[20.6] 0.187 *
0.095
[11.7] 0.111 +
0.060
[-26.2] -0.304 *
0.137
[-20.3] -0.227 *
0.111
-0.088
0.067
[12.3] 0.116 +
0.060
-0.181
0.140
-0.150
0.117
[.00.4] 0.004

+ significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%
^ cases with information on claims only

***

0.001

[0.4] 0.004

***

0.001

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 18. (NC) Regression of disposition time in days (logged), for terminated cases, by circuit and state.
model:
Observ's
Prob > F
R-squared
am. petition
answer or motion
Mag. j. R & R
no. of claims
pet. Had atty.
competency claim
jury miscon. claim
speedy, public trial
evidence claim
prison discipline
all clms unexhaust
time-barred
defaulted claim
successive pet.
other dismissal
Circuit 2
Circuit 3
Circuit 4
Circuit 5
Circuit 6
Circuit 9
Circuit 10
Circuit 11
MA
NJ
PA
NC
SC
VA
AL
FL
LA
MS
TX
GA
MI
OH
IL
NY
MO
CA
OR
KS
constant -

Circuit Comparisons
Claims
variables
1162^
0.000
0.599
% diff
coef
r.s.e.

No claims
variables
1827
0.000
0.562
% diff coef
[38.1] 0.323
[394.8]1.599
[26.7] 0.237
[17.4] 0.160
[39.1] 0.330

***
***
***
***
***

0.055
0.049
0.043
0.023
0.076

[160.4]0.957
[65.5] 0.504
[36.5] 0.311
[41.6] 0.389
[65.0] 0.501
[11.4] 0.573
[15.4] 0.562
[39.7] 0.334

***
***
**
***
***
***
***
***

0.106
0.120
0.099
0.096
0.105
0.097
0.144
0.098

circuit 7

[46.2] 0.233
[403.2] 1.617
[25.2] 0.225
[10.7]0.102
[22.9] 0.206
[35.2] 0.302
[-27.6] -0.323
[21.2]0.192
[10.4]0.099
[-25.2]0.290
[-22.2] -0.251
[-10.5] -0.111
[16.7] 0.155
[-23.5] -0.268
[-20.9] -0.235
[138.5]0.869
[10.1]0.531
[27.8]0.245
[49.9]0.405
[62.1]0.483
[64.2]0.496
[68.4]0.521
[38.7]0.327

circuit 7

^ cases
with
information
on claims
only.
r.s.e.
***
***
***
*
*
+
*
*
+
*
*
+
**
*
*
***
***
+
***
***
***
*
**

0.063
0.074
0.053
0.043
0.083
0.162
0.130
0.082
0.057
0.129
0.104
0.063
0.056
0.138
0.113
0.127
0.135
0.128
0.117
0.122
0.125
0.216
0.122

No claims
variables
1827
0.000
0.579
% diff
coef
[35.3] 0.302
[398.3] 1.606
[29.2] 0.256
[16.6] 0.154
[34.2] 0.294

[14.5]0.557
[205.9]1.118
[84.4]0.612
[.51.7]0.417
[66.4]0.509
[88.9]0.636
[89.8]0.641
[.65.7]0.505
[133.1]0.849
[81.5]0.596
[69.0]0.525
[60.0]0.470
[101.8]0.702
[121.4]0.795
[48.0]0.392
[216.5]1.152
[14.0]0.554
[109.2]0.738
[202.8]1.108
[182.1]1.037
Indiana

***
***
***
***
***

*
***
***
*
***
***
***
***
***
***
***
**
***
***
*
***
***
***
***
***

^ includes cases with information on claims only.
+ significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%

State Comparisons
Claims
variables
1128^
0.000
0.617
r.s.e.
% diff
coef
0.055
[26.2] 0.233
0.050
[410.4]1.630
0.047
[35.1] 0.301
0.024
[10.1] 0.096
0.082
[17.6] 0.162
0.240
[-30.6]-0.366
[22.8]0.205
[11.4]0.108
-0.161
[-21.6]-0.243
[-12.4]-0.132
[13.5]0.127
[-21.7]-0.244
[-19.6]-0.218

0.271
0.211
0.147
0.166
0.154
0.131
0.159
0.127
0.155
0.167
0.122
0.149
0.128
0.155
0.169
0.127
0.167
0.122
0.169
0.302

[104.6]0.716
[286.9]1.353
[83.9]0.609
[63.2]0.490
[46.7]0.383
[83.7]0.608
[ 70.6]0.534
[15.6]0.563
[132.8]0.845
[143.5]0.890
[69.9]0.532
[11.1]0.537
[95.4]0.670
[105.2]0.719
[83.1]0.605
[203.1]1.109
[68.5]0.522
[102.0]0.703
[208.6]1.127
[214.0]1.319
Indiana

^ cases with
information
on claims
only.
r.s.e.
***
0.06
***
0.08
***
0.06
*
0.04
+
0.088
0.171
**
0.131
*
0.082
+
0.056
0.130
*
0.107
*
0.062
*
0.057
+
0.142
+
0.115

*
***
***
**
*
***
**
***
***
***
***
**
***
***
***
***
**
***
***
***

0.310
0.214
0.152
0.156
0.176
0.181
0.184
0.141
0.179
0.212
0.130
0.170
0.145
0.179
0.184
0.140
0.179
0.142
0.212
0.385

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 19. (NC) Regression of disposition time in days (logged), for terminated cases, by district.
model:
Observ's
Prob > F
R-squared

No claims variables

% difference
amended petition
answer or mo. to dismiss
magistrate judge R&R
number of claims in petition
petitioner had attorney
jury misconduct claim
presence, speedy, public trial
evidence ruling claim
all claims unexhausted
time-barred
defaulted claim
Successive
other dismissal
NJ
MA
PA- E
PA- M
LA- W
TX- N
FL- M
AL-N
GA- N
LA-E
TX- E
TX-S
TX- W
MI- E
MI- W
OH- N
IL- N
MO- E
NY- E
CA- N
CA- E
CA- C
CA- S
OR
NY- S
KS
_constant (IN-S)

[35.4]
[397.3]
[42.5]
[19.2]
[26.1]

[18.2]
[58.4]
[34.5]
[121.6]
[216.4]
[41.3]
[68.0]
[102.1]
[55.4]
[52.0]
[48.2]
[48.5]
[70.5]
[14.5]
[19.2]
[17.2]
[18.7]
[22.4]
[15.4]
[18.9]
[16.5]
[36.5]
[68.2]
[191.5]
[214.9]
[158.1]

Claims variables
1813
0.000
0.605
r.s.e.

coeff.
0.303
1.604
0.354
0.176
0.232

1.023
0.460
0.297
0.769
1.152
0.346
0.519
0.704
0.441
0.419
0.394
0.396
0.534
0.713
0.416
0.524
0.312
1.029
0.886
1.098
0.985
0.311
0.520
1.070
1.147
0.948
3.397

***
***
***
***
**

***
+
+
***
***
**
***
***
**
**
*
**
**
***
**
***
+
***
***
***
***
*
**
***
***
**
***

0.054
0.053
0.061
0.024
0.082

0.211
0.280
0.172
0.240
0.225
0.157
0.150
0.185
0.162
0.159
0.163
0.143
0.168
0.135
0.176
0.159
0.172
0.202
0.143
0.173
0.153
0.130
0.193
0.175
0.162
0.306
0.158

1076^
0.000
0.642
.r.s.e.

% difference coeff
[24.5]
[423.3]
[51.6]
[11.4]
[10.7]
[-31.5]
[22.6]
[13.5]
[-20.1]
[-11.5]
[13.8]
[-13.9]
[-14.3]
[276.2]
[108.5]
[53.1]
[87.8]
[234.0]
[67.4]
[106.1]
[85.5]
[75.4]
[68.5]
[61.1]
[63.4]
[69.5]
[133.7]
[45.1]
[60.7]
[85.3]
[191.8]
[173.7]
[252.9]
[184.3]
[53.0]
[59.8]
[233.3
[227.7]
[281.5]

0.219
1.655
0.416
0.108
0.102
-0.378
0.204
0.127
- 0.225
- 0.122
0.129
- 0.150
- 0.154
1.325
0.735
0.426
0.630
1.206
0.515
0.723
0.618
0.562
0.522
0.477
0.491
0.528
0.849
0.372
0.474
0.620
1.071
1.007
1.261
1.047
0.425
0.469
1.204
1.187
1.339
3.225

^ includes cases with information on claims only.
+ significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%

***
***
***
*
**
*
*
*
*
*

***
*
*
*
***
**
***
**
**
**
*
**
**
***
+
*
**
***
***
***
***
*
*
***
***
***
***

0.068
0.078
0.081
0.044
0.093
0.134
0.088
0.057
0.106
0.063
0.059
0.140
0.117
0.230
0.334
0.184
0.266
0.292
0.188
0.182
0.232
0.196
0.199
0.195
0.172
0.202
0.160
0.222
0.196
0.200
0.237
0.174
0.379
0.191
0.169
0.223
0.231
0.205
0.391
0.227

Final Technical Report: Habeas Litigation in U.S. District Courts

Results

State pleadings and amended petitions. When the state files an answer, return, or motion to dismiss,
disposition time for terminated cases jumps by about 400%, compared to the cases lacking these pleadings.
This is expected. Cases in which the federal court dismisses the case before the state files an answer, return,
or motion to dismiss should not take as long. The filing of an amended petition also adds to the length of a
case, increasing processing time between 26% and 45%.
The Tables in Appendix A reporting Cox regression results examining both pending and terminated
cases show that when comparing cases early on, the presence of a state pleading decreased the likelihood of
termination by about 99%. Among older cases, however, the filing of such a state pleading increases the
probability of termination. The filing of an amended petition decreases the probability of termination
between 24% and 87%. The effect of this feature only varied by time over the life of the case when no claims
variables were included.
Magistrate judge. The presence of a report and recommendation regarding disposition from a
magistrate judge appears to be associated with significantly longer processing times, accounting for an
approximate 10% to 49% increase in the number of days for disposition for terminated cases. 133 Reference to
magistrate judges was suspected as a cause of delay in at least one earlier study. 134 This could add to
processing time by interposing additional steps in the litigation sequence (the report and recommendation, the
objections, and any response to the objections) that would otherwise be absent. On the other hand, the
analysis only tells us that among all cases, controlling for other factors, those with dispositive orders by
magistrate judges are longer than those without. It does not indicate whether the use of magistrate judges in a
given district helps the district judges to dispose of these cases more quickly than they would be able to
without delegating the initial decisions to magistrate judges. It is also possible that the causal relationship is
reversed for this variable, that is, that district judges in the districts that were already taking the longest time
to process these cases are most likely to refer non-capital habeas cases to magistrate judges.
The presence of a report and recommendation for disposition reduced the probability of termination
early on in the case, but over time, for the cases that lasted longer, a report and recommendation made
termination more likely. So for cases that have lasted the longest, those in which a report and recommendation
has been filed are more likely to have been completed than the longer cases in which a report and
recommendation may have yet to be filed.
Representation. As expected, the presence of a petitioner's attorney was associated with more
lengthy proceedings, adding between 11% and 49% more time than in cases without attorneys. The Tables in
Appendix A show that the presence of an attorney functions in the same way as a magistrate judge’s report
and recommendation. It decreases the probability of termination at first, but for longer cases, counseled
petitions were more likely to reach disposition than uncounseled petitions.

133

The study tracked whether a report and recommendation ("R&R") for disposition was filed, and who presided over
any evidentiary hearing held. Cases referred to a magistrate judge which had not yet resulted in either a hearing or an
R&R for disposition would not have been captured by these measures. See Asifa Quraishi, Resource Guide for
Managing Capital Cases, Vol. II: Habeas Corpus Review of Capital Convictions FJC 2004,
http://www.fjc.gov/library/fjc_catalog.nsf, at 8 ("Many magistrate judges handle the counsel appointment and scheduling
orders . . . others continue . . . responding to procedural motions and managing the case by monitoring budgets and
holding status conferences.").
134
Shapiro, supra note 2, at 333.

Final Technical Report: Habeas Litigation in U.S. District Courts

Claims. A larger number of claims in the petition meant longer processing time. Each additional
claim increased the number of days in federal court by 10 to 19 percentage points. Generally, each additional
claim decreased the likelihood of termination between 3% and 7%.
As for types of claims, only a few appeared to be significantly related to the time it took these courts
to complete the case. The exceptions were: claims of violations of the right to be present at trial or receive a
public trial, or to receive a speedy trial or appeal, were associated with lengthening the number of days to
disposition by about 20 to 23 percentage points; a claim of improper exclusion or admission of evidence
added between 10% and 13% more time; and a claim regarding competency at trial added between 35% and
39%. Each of these three claims was also associated with a decrease in the probability of termination in the
claims-specified Cox analyses. The presence of each of these three claims decreased the probability of
termination between 15% and 23%. A similar decrease in probability of termination was associated with
Brady-type claims and jury instruction claims.
Also related to processing time but operating in the other direction to reduce disposition time were the
presence of a claim of jury misconduct, reducing processing days by about 21% to 31%, and a claim
challenging a prison disciplinary proceeding, reducing disposition days by about 26%. These two claims were
also associated with greater likelihood of termination. A misconduct claim increased the probability of
termination by about 53% to 75%. The presence of a disciplinary claim increased the probability of
termination by more than 300 times when controlling for caseload or clerk. Once district or state was
specified, however, the relationship between a claim challenging an administrative disciplinary ruling and
probability of termination was either not significant or was associated with a more modest drop in the
probability of termination. This confirms that these claims are concentrated in certain districts and states. 135
Defenses and reasons for dismissal. Speedier dispositions were indicated for cases in which all
claims were dismissed as unexhausted, as time-barred, or as filed in a successive petition. Cases dismissed
for “other reasons,” such as failure to pay the filing fee, or to provide adequate information on the petition
form, also took less time. The disposition time for dismissal of cases including any claim that was
procedurally defaulted was longer on average than the disposition time for those cases in which no claims
were dismissed for this reason.
Regarding the probability of termination, the presence of at least one claim dismissed as defaulted
functioned like the presence of a magistrate’s report and recommendation, which was the other dispositionrelated (but not disposition-contingent) variable that we tested in the Cox analyses. Its main effect was to
decrease the probability of termination, but it varied by time. The longer cases were more likely to be
terminated if they included at least one claim dismissed for this reason.
Caseload and clerk staffing levels. Districts with the heaviest habeas caseloads 136 generally took
more time with these cases, on average, than did districts with lighter caseloads. With the addition of each
additional non-capital habeas case per judge, disposition time increased by 0.4%. As Table 28 in Appendix A
illustrates, every additional case per judge in any given district also made it 0.85% less likely that a case in
that district would be completed, controlling for other factors.

135

Two variables had a significant relationship to termination times in only one of the two models: the presence of a
search and seizure claim was associated with decreased disposition time, which is expected given that the Supreme Court
held that these claims should be dismissed as non-cognizable, and a reference in the record to a prior petition dismissed
without prejudice increased disposition time. This is also expected, because that usually means that the petitioner was
back in federal court after exhausting his claims.
136
See note 129 supra for the derivation of the continuous caseload variable.

Final Technical Report: Habeas Litigation in U.S. District Courts

Using a continuous variable that indicated the number of pro se law clerks or attorneys per district
averaged over the two years 2003 and 2004, we also investigated the relationship between extra staffing and
disposition time. Because clerks were allocated depending upon pro se caseload, for each additional clerk, we
expected longer disposition time, just as greater caseload increased disposition time. The analysis found that
staffing level was not related to processing time for terminated non-capital cases, but that as expected, cases
were less likely to terminate in districts where additional staff was in place. Every additional law clerk was
associated with a 1.5% decrease in the likelihood of termination for a non-capital case. See Table 27,
Appendix A. Districts that had been assigned more pro se law clerks take even longer than courts assigned
fewer, but each individual district may have taken even longer still with fewer clerks. An analysis of
individual districts over time as clerk staffing levels change may be a useful investigation.
Informal discussions with court staff suggest that efficiencies may be created by allowing court staff
to specialize in particular types of cases (prison discipline habeas cases, parole habeas cases, regular habeas),
and to serve as support to all of the judges in the district on that specific type of case. This promotes expertise,
unlike judge-specific assignments requiring all clerks to master all case types.
Location of court. Even after controlling for other factors, one of the most powerful influences on
processing time was the identity of the district in which the case was filed. In the model using dummy
variables for districts, cases from districts with less than nine observations were excluded. 137 Using as the
comparison the Southern District of Indiana, which is the district with the lowest average disposition time, the
analyses found that many districts were associated with average disposition times that are much longer than
the average for the Southern District of Indiana. For example, disposition times in the Southern District of
New York were 215% to 228% longer; those in the Eastern District of Pennsylvania were 35% to 53% longer.
Splitting up the cases by circuit instead of by district, 138 using the Seventh Circuit as the comparison
circuit, disposition time continued to vary greatly by location even after controlling for other factors. Cases in
districts located in the Second Circuit took 139% to 160% longer for disposition than cases in the Seventh
Circuit. 139 This suggests that something about the circuit other than the features already examined is affecting
processing time.
Cox regressions reported in Appendix A confirmed the influence of location on processing time.
When districts were specified, many were associated with significantly slower progress towards termination
than the Southern District of Indiana, ranging from 42% less likely to terminate (TX-S) to 94% less likely
(NY-N). 140 When circuits were specified, again the Seventh Circuit appeared to be associated with a much
higher probability of termination than all but the Fourth Circuit. Cases filed in the Second Circuit were over
75% less likely to be completed than those in the Seventh Circuit, after controlling for other factors. This
general pattern remained whether or not claims were specified.
Because practices adopted by the attorneys representing different states may have an impact on
processing time in otherwise comparable cases, we also checked for the influence of the individual state.
Some of this variation we attempted to capture with other specific variables, namely whether or not the state
137

The smaller districts were not aggregated into a single variable because of the variation among them in processing
time.
138
Using the Seventh Circuit as the comparison and dummy variables for the remaining ten circuits.
139
Several locations were significant in only one of the two models that tested their association with disposition time.
When significant, each was associated with increased disposition time: the Eighth Circuit (compared to the control
location, the Seventh Circuit); the districts of FL-N, PA-W, NC-E, AZ, NV, CO, NY-W, NY-N, SC, VA-E, MS-S, KYW, OH-S, IN-N, WI-E (compared to the control district, IN-S); and the states of KY, WI, MN, AZ, NV, CO (compared
to the control state, IN).
140
The effect on probability of termination varied over time for five districts.

Final Technical Report: Habeas Litigation in U.S. District Courts

attorney general filed an answer, return, or motion to dismiss, as well as the state's use of affirmative defenses
such as procedural default. Yet the state in which the case is litigated was expected to influence processing
time, both because our coding did not capture other potentially state-specific practices (such as the use of
continuances, the relative time for procuring state records, or the frequency with which state counsel is
substituted), and also because we coded only the federal court’s ultimate resolution of each claim and defense,
and not the arguments raised by the state’s attorney. 141 Examining variance in disposition time using the
fastest state, Indiana, as the comparison, 20 states were associated with significantly longer processing times.
After controlling for other factors, filings in New York and New Jersey, for example, were associated with
more than 200% more days for disposition than filings in Indiana.
Cases in the state of Indiana were also more likely to terminate than cases in the most other states
examined using the Cox regression analysis. 142 Georgia cases, for example, were about 30% to 35% less
likely to terminate, while cases from New York were more than 85% less likely to terminate.
Not every state examined showed significantly slower disposition times or lower probability for
termination. For example, disposition times for cases in Minnesota and Washington were not significantly
different than the time for cases from Indiana. Overall, the most striking differences found with the statespecific analyses were between the shortest most probable terminations in Indiana cases and the longer less
probable dispositions in cases from New York and New Jersey.
As a final test of the independent influence of location on processing time, two dummy variables were
added to the district-specified regression models for terminated cases (both the “claims” and the “no claims”
models) in order to control for the effect of caseload size. One variable was set equal to one if the case was
from a district with a caseload between 18 and 50 habeas cases per judge (medium caseload). The other
variable was set equal to one if the caseload for the district was greater than or equal to 50 habeas cases per
judge, which corresponded to the upper quartile of caseload size. The comparison was any case from a district
with a caseload level of less than 18, roughly the lower quartile of caseload size.
The results of these additional analyses suggest that variations in processing time between districts are
not entirely a function of state habeas caseload. 143 When compared to the analyses that did not take caseload
into account, a district’s relative influence on termination time was generally diminished in districts with the
highest and lowest caseloads with the addition of caseload to the analysis. Indeed, in a few districts with high
habeas caseloads such as MI-W and TX-N, caseload appeared to explain much of the relative variation in
disposition time. After caseload was added to the analysis, these particular districts were no longer associated
with longer disposition times. The analysis suggests that had these districts not had such high caseloads, they
would have processed cases more quickly than IN-S. Most of the districts that showed a significant
141

Processing time may also vary between states for cases stayed for exhaustion, if review by state courts may take
longer in some states than in others. Only a small percentage of non-capital cases in our sample were stayed, however,
and the presence of a stay was independently assessed in one of the models and was not significant. The presence of a
stay was not included in the survival analyses of all cases because of collinearity with the presence of an amended
petition.
142
Oklahoma, Tennessee, and Massachusetts cases did appear more likely to terminate than cases in Indiana in one of the
two state-specified models.
143
In the district-specified model with claims information, cases filed in the highest caseload districts were not
associated with processing times that were significantly longer than those filed in the lowest caseload districts. See
Table 37, Appendix A. But in the model without claims information, cases from districts with the highest caseloads took
almost 50% longer than those from districts with the lowest caseloads. See Table 38, Appendix A. Suprisingly, cases
from districts with medium-sized caseloads ended sooner that the cases from districts with the lowest caseloads in both
models, 23% to 27% sooner. It is possible that the presence of a moderately heavy habeas caseload prompts districts to
develop efficiencies not developed in districts with relatively few habeas cases, and not achievable in the districts
burdened by the highest habeas caseloads.

Final Technical Report: Habeas Litigation in U.S. District Courts

association with additional disposition time in the analyses without caseload continued, however, to show a
significant association with additional disposition time once caseload was taken into account. The addition of
the caseload variable to the analysis actually increased the influence of location on longer processing time for
some districts. See Tables 35 and 36, Appendix A.
C.

Capital case processing time

1.
Analysis design. Termination times varied widely and varied by district. See Figures
15 and 18, above, and 22, below. More than one in every four of the capital cases was still
pending and unresolved at the end of November 2006.
Figure 22. Length of case in days, by district, box chart.

Length of case, including pending cases

3,000

78
2,000

76
6

1,000

367
134
321
296

140
133 308

295

135

0
41

District

See Appendix D or footnote 59 for a list of which districts correspond to the district numbers

One policy issue in particular is brought into sharp relief by these findings. There is a looming conflict
between present practice and the statutory time limits for processing capital habeas cases that may bind
federal courts in the near future. Given the length of time that capital habeas cases presently take to resolve,
the statutory 450-day time limit for resolving capital habeas cases from states that may qualify for expedited
review under AEDPA will pose a challenge for courts. 144 In the districts that we examined, the average
processing time for capital cases is well over two and a half times that long. See Figures 18 and 22. Not one
of the 13 districts in this study has completed its capital habeas cases in less than 500 days on average, even
excluding stayed time. In some of these districts the average processing time after subtracting stayed time
was more than three and a half times as long as 450 days. Examining only terminated cases and excluding
those ended by transfer, only about 26% had been resolved in 450 days or less. Indeed, Arizona may be the
first state to qualify for the expedited review requirements, 145 but the District of Arizona is among the slowest
of the 13 districts in the study in terms of processing these cases, having completed less than a quarter of the
cases filed between 2000 and 2002 by the end of 2006.
To examine possible explanations for the variation in case processing time which would help to
inform debates about appropriate disposition periods for these cases, we used the same two analyses that were

144

See 28 U.S.C. § 2266(b)(1)(A), effective March 9, 2006.
See Spears v. Stewart, 283 F.3d 992 (9th Cir. 2002) (finding that state scheme for providing counsel satifies the
statutory “opt-in” provisions, but was not followed for this petitioner).

145

Final Technical Report: Habeas Litigation in U.S. District Courts

employed for non-capital cases. 146 For terminated cases, we used multivariate regression, with the log
transformation of the number of days that elapsed from beginning of a case to termination as the dependent
variable. The features expected to be associated with longer processing time among terminated cases were the
substitution of one petitioner's attorney for another attorney, the presence of amended petition, the length of
the time period between beginning of case and the filing of a counseled petition, the presence of an
evidentiary hearing, the presence of an order authorizing a deposition or examination, the use of a magistrate
judge for a report and recommendation on the resolution of any claim, a greater number of claims in the
petition, the length of the time period during which the case was stayed for exhaustion, and the presence of 11
specified claim types (listed in Table 20A and footnote 148). The application of procedural default to bar any
claim was expected to be associated with shorter processing times. Petitioner race 147 was included but not
expected to be associated with variance in processing time.
We examined termination times for capital cases using two basic models with two different sets of
additional variables. In the "merit model" we examined whether longer processing times were associated
with denial of any claim on the merits, or with the grant of relief on any claim. In the “reason model” we
substituted variables with more specific measures of the reason for disposition in order to evaluate which, if
any, of these reasons for dismissal were associated with shorter disposition times. Included were the
application of Teague to bar review of any claim; application of statute of limitations to bar any claim;
application of successive petition bar; and whether all claims were dismissed as unexhausted. For both the
“reason” and the “merits” models, we explored alternative location specifications: 1) districts (using as the
comparison FL-M, the district with the fastest average dispositions), 2) capital habeas case load per district, 148
and 3) death penalty law clerk staffing levels. 149
In a separate effort to identify what may distinguish the longest cases from the shorter cases, we also
used Cox’s proportional hazard models to examine whether individual features were associated with a change
in the hazard rate for termination, that is, the likelihood that the case would reach termination by late
November 2006, when we completed our collection of data for these capital cases. For many features, the
assumption of proportional effects over time was not appropriate. In other words, the difference in likelihood
of termination associated with that feature changed the longer a case remained pending. For those variables
that tests indicated had time-varying effects, we estimated the standard Cox model with the addition of an
interaction effect. 150
The features examined for their association with likelihood of termination in the Cox models were the
same as those used in the terminated-case analyses, with the omission of termination-contingent variables.
Omitted were variables assessing whether any claim was granted; application of statute of limitations to bar
any claim; application of successive petition bar; whether all claims were dismissed as unexhausted; dismissal
146

Measure of time was the number of days to termination, or, for pending cases, the log of the number of days from the
beginning of the case until the last week in November 2006, when all pending cases were rechecked to see if they had
been resolved.
147
Too few of the capital cases in our sample were filed by female petitioners to include gender of petitioner in the
analyses. As with the non-capital cases, the number of docket entries was associated with processing time, but we ran
the analyses without this variable. See note 130 supra.
148
This continuous variable was the number of capital habeas cases filed per year in the district, see Judicial Business,
supra note 16, divided by the number of authorized judges per district (provided by the Federal Judicial Center),
averaged for the six years 2000 through 2005. This value ranged from 5.95 to .86 cases per judge. The districts, in order
from heaviest to lightest average caseload, were: OH-S; AZ; NV; OK-W; CA-C; TX-N; TX-E; TX-W; AL-N; TX-S; FLM; PA-E.
149
This continuous variable was the number of death penalty law clerks assigned to each district, provided by the Federal
Judicial Center, averaged over the six years 2001 through 2005, ranging from no clerks in PA-E to 5.5 in CA-C.
150
See Jane M. Box-Steffensmeier & Christopher J.W. Zorn, Duration Models and Proportional Hazards in Political
Science, 45 AM. J. POL. SCI. 972 (2001).

Final Technical Report: Habeas Litigation in U.S. District Courts

of any claim as defaulted; and the use of Teague to bar review of any claim. We explored the alternative
specifications of the district (using as a comparison the district with the greatest proportion of pending cases,
that is AZ); circuit (using as a comparison the Fifth Circuit, the circuit with the fewest cases pending); clerk
staffing; and caseload. Case length was not evenly distributed but instead showed that about 10% percent of
the capital cases ended very quickly (most of these were dismissals not on the merits, which we did not
control for in our analysis of pending as well as terminated cases). Therefore we ran the second analysis both
with and without the shortest 10% of cases.
2.

Results

Terminated case regressions. The results of the regressions for terminated cases are summarized in
Table 20. The analyses were able to explain over half of the variation in processing time for the capital cases
that had reached termination.
As with the non-capital case regression tables, the value in brackets indicates the percentage difference
in length associated with the presence of the feature, as compared to a case without that feature, all other
features held constant. A negative value means that the feature was associated with shorter cases; a positive
value means that the feature was associated with longer cases.
For the continuous variables, that is, days stayed and days to counseled petition, the value inside the
brackets is the percentage difference in processing time for every additional day, all other features held
constant. For caseload and clerk staffing, also continuous variables, the value indicates the difference in
processing time associated with the addition of one more capital habeas case per judge or one more death
penalty clerk or death penalty staff attorney per district.
Cox models with time interactions for non-proportionality, terminated and pending cases. Table
21 presents significant associations between a change in the likelihood of termination and various features.
Changes in the likelihood of termination are indicated by the change in the hazard rate, the first
number by which asterisks appear indicating the degree of significance. If the change in the hazard rate
exceeds one, the feature is associated with shorter processing time, that is, a greater likelihood of termination.
If the change in the hazard rate is less than one, the feature is associated with longer processing time, that is, a
lower likelihood of terminating.
For those features whose effect on the likelihood of termination changed over time, the table also
includes the results for the "tvc” or time varying effect, indicating the change in the likelihood of termination
associated with the feature in question later in the life of the longest lasting cases.

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 20. (C) Regression of days for disposition (logged), terminated cases. (cont. next p.)
comparing by circuit

comparing by district

reason
model
0.276**
(0.106)
[0.318]
0.746**
(0.190)
[1.109]
-0.005
(0.004)

merit
model
0.322*
(0.128)
[0.380]
0.557**
(0.191)
[0.745]
-0.000
(0.005)

days stayed

-0.000
(0.000)

0.001
(0.000)

reason
model
0.263*
(0.111)
[0.301]
0.636***
(0.190)
[0.889]
-0.008+
(0.005)
[-0.008]
0.000
(0.000)

days to
counseled
petition
all claims
unexhausted

0.001*
(0.000)
[0.001]
-0.792
(0.583)

0.001*
(0.001)
[0.001]

any timebarred claim

0.618*
(0.267)
[0.855]
0.345**
(0.127)
[0.412]
-3.150**
(0.635)
[-0.957]
0.628*
(0.252)
[0.874]
0.389*
(0.155)
[0.476]
0.413*
(0.178)
[0.511]
0.148
(0.109)

0.001*
(0.000)
[0.001]
-0.960+
(0.514)
[-0.617]
0.749**
(0.271)
[1.115]
0.451***
(0.133)
[0.570]
-3.336***
(0.563)
[-0.964]
0.592*
(0.255)
[0.808]
0.445**
(0.162)
[0.560]
0.482**
(0.180)
[0.619]
0.175
(0.112)

variable
am. petition

magistrate
R&R
# claims raised

any defaulted
claim
successive
petition
any IAC claim

any
competency
claim
pros.
misconduct or
Brady
jury selec,
miscond. Bias

0.183
(0.176)
[0.201]

0.970**
(0.320)
[1.638]
0.326+
(0.186)
[0.385]
0.239
(0.190)
0.226+
(0.130)

merit
model
0.323*
(0.136)
[0.381]
0.415*
(0.193)
[0.514]
-0.002
(0.006)

by no. of death penalty
law clerks per district
reason
merit
model
model
0.194+
0.251*
(0.102)
(0.126)
[0.214]
[0.285]
0.619***
0.444*
(0.181)
(0.172)
[0.857]
[0.559]
-0.004
-0.000
(0.004)
(0.004)

by capital habeas
caseload per judge
reason
merit
model
model
0.248*
0.297*
(0.102)
(0.124)
[0.281]
[0.346]
0.630**
0.462*
(0.183)
(0.189)
[0.878]
[0.587]
-0.005
-0.000
(0.004)
(0.004)

0.001
(0.000)

0.000
(0.000)

0.001*
(0.001)
[0.001]

0.001*
(0.000)
[0.001]
-0.849
(0.517)

0.239
(0.173)

0.998**
(0.331)
[1.713]
0.357+
(0.187)
[0.429]
0.274
(0.193)

0.382+
(0.228)
[0.465]
0.271*
(0.120)
[0.311]
-3.158***
(0.642)
[-0.957]
0.674**
(0.251)
[0.962]
0.399*
(0.165)
[0.490]
0.494**
(0.178)
[0.639]
0.211+
(0.110)
[0.235]
0.378**
(0.141)
[0.459]

0.001*
(0.000)
[0.001]
0.001*
(0.001)
[0.001]

0.141
(0.171)

1.016**
(0.320)
[1.762]
0.361+
(0.194)
[0.437]
0.343+
(0.191)
[0.409]
0.298*
(0.132)
[0.347]
0.475*
(0.195)
[0.608]
0.554**
(0.175)
[0.740]
0.762**
(0.284)
[1.143]

0.001*
(0.000)
[0.001]
-0.736
(0.596)
0.478*
(0.227)
[0.613]
0.305*
(0.118)
[0.357]
-3.228***
(0.621)
[-0.960]
0.636**
(0.245)
[0.889]
0.327*
(0.158)
[0.387]
0.469**
(0.174)
[0.598]
0.158
(0.107)

0.001*
(0.000)
[0.001]
0.001*
(0.001)
[0.001]

0.158
(0.167)

0.959**
(0.316)
[1.609]
0.286
(0.177)
0.291
(0.183)

0.243+
0.242+
(0.130)
(0.127)
[0.275]
[0.273]
Roper, Atkins,
0.414**
0.494*
0.506***
0.545*
0.370**
0.476*
(0.145)
(0.201)
(0.147)
(0.211)
or Ring claim
(0.140)
(0.194)
[0.513]
[0.659]
[0.725]
[0.448]
[0.610]
was at least
0.432*
0.515**
0.490**
one claim
(0.183)
(0.185)
(0.164)
granted?
[0.540]
[0.674]
[0.632]
at least one
0.806**
0.792**
0.775**
claim denied
(0.282)
(0.289)
(0.279)
on merits?
[1.240]
[1.208]
[1.171]
+ significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%
Robust standard errors in parentheses; percentage difference in length associated with specified variable in brackets

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 20. (C) Regression of days for disposition (logged), terminated cases, (cont. )
comparing by
circuit
reason
model

merit
model

PA-E

AL-N
TX-N
TX-E

TX-S
TX-W
OH-N
OH-S

CA- C

OK-W

AZ
NV
Av # death pen. clerks per
district
Average cap. habeas cases per
judge
3 Circuit
(PA-E only)
6 Circuit
(OH- S & N only)
9 Circuit
(AZ; CA-C; NV only)
10 Circuit
(OK-W only)

0.615*
(0.260)
[0.850]
0.375
(0.230)
0.264
(0.463)
0.818*
(0.336)
[1.266]
-0.029
(0.327)

by no. of death
penalty law clerks
per district
merit
reason
model
model
0.384
0.901+
(0.460)
(0.520)
[1.462]
0.194
0.389
(0.462)
(0.542)
0.171
0.289
(0.444)
(0.491)
0.245
0.901+
(0.456)
(0.528)
[1.462]
-0.016
0.076
(0.426)
(0.454)
-0.102
0.103
(0.453)
(0.491)
0.169
0.588
(0.440)
(0.474)
0.795
1.212*
(0.523)
(0.503)
[2.360]
2.002*
1.972*
(0.899)
(0.804)
[6.404]
[6.185]
0.836+
1.078*
(0.454)
(0.535)
[1.307]
[1.939]
-0.126
0.047
(0.717)
(0.864)
-0.063
-0.077
(0.595)
(0.534)

by capital habeas
caseload per judge

comparing by
district

reason
model

merit
model

reason
model

merit
model

0.175
(0.123)

0.249+
(0.132)
[0.283]
0.165*
(0.073)
[0.179]

0.127
(0.077)

0.411
(0.253)
0.235
(0.227)
0.499
(0.499)
0.804**
(0.270)
[1.234]
0.076
(0.272)

11th Circuit
(FL-M & AL-N only)
Constant
3.698**
4.421**
4.476**
3.462**
4.425**
Fl-M; 5C
(0.458)
(0.538)
(0.433)
(0.390)
(0.382)
Observations
267
267
267
267
267
267
267
R-squared
0.63
0.51
0.53
0.67
0.63
0.51
0.63
+ significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%
Robust standard errors in parentheses; percentage difference in length associated with specified variable in brackets.

3.644**
(0.397)
267
0.50

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 21. (C) Cox regression models, change in probability of termination over time.
Model
LR chi2
Prob > chi2
Observations
Am. petition

caseload
all cases
(23)316.09
0.00000
353
0.701*
(0.111)

caseload
no short
(22)241.28
0.00000
330
0.726*
(0.117)

district
all cases
(35)428.66
0.00000
353
0.686*
(0.116)

(tvc)
atty sub’d

district
no short
(33)339.77
0.00000
330
0.001***
(0.002)
2.674**
(0.760)

0.696+
(0.150)

mag.j. R&R
claim no.
stayed time
rep.pet. time
( tvc)

0.978***
(0.005)
0.997***
(0.000)
0.977***
(0.003)
1.003***
(0.000)

0.978***
(0.005)
0.997***
(0.000)
0.981***
(0.004)
1.003***
(0.001)

0.687***
(0.043)

0.675***
(0.044)

ineff. asst.
(tvc)
caseload
(continuous)
clerk staff

0.979***
(0.006)
0.997***
(0.000)
0.975***
(0.004)
1.003***
(0.001)
0.001***
(0.001)
2.962***
(0.638)

Circuit 3

0.152***
(0.065)
0.479*
(0.138)
0.097***
(0.034)
0.368**
(0.108)
0.537*
(0.149)

Circuit 6
Circuit 9
Circuit 10
Circuit 11
AL-N

0.985**
(0.005)
0.997***
(0.000)
0.976***
(0.004)
1.003***
(0.001)

circuit
all cases
(28)400.16
0.00000
353
0.008**
(0.014)
1.943**
(0.489)
0.618*
(0.141)
0.709+
(0.136)
0.986**
(0.005)
0.997***
(0.000)
0.976***
(0.004)
1.003***
(0.001)
0.004***
(0.005)
2.344***
(0.490)

6.389**
(4.487)
FL-M
9.035**
(6.887)
TX_N
11.379***
(7.637)
TX-E
22.688***
(15.389)
TX-S
36.313***
(24.193)
TX-W
11.611***
(7.916)
OH-N
11.754***
(7.978)
CA-C
4.086+
(3.392)
OK-W
6.611**
(4.684)
Constant
AZ
AZ
5th Circuit
+ significant at 10%; * significant at 5%; ** significant at 1%, *** significant at 0.1%
Robust standard errors in parentheses

circuit
no short
(27)293.23
0.00000
330
0.007*
(0.014)
2.094**
(0.562)
0.665+
(0.154)
0.711+
(0.144)
0.985**
(0.005)
0.997***
(0.000)
0.978***
(0.004)
1.003***
(0.001)

0.137***
(0.059)
0.404**
(0.118)
0.081***
(0.031)
0.348**
(0.107)
0.409**
(0.127)

5.635**
(3.471)
6.487**
(4.334)
6.325**
(3.682)
10.192***
(5.964)
16.649***
(9.510)
7.466**
(4.426)
8.518***
(5.137)

5th Circuit

clerk
all cases
(22)284.29
0.00000
353
0.752+
(0.117)

clerk
no short
(22)203.12
0.00000
330

0.598*
(0.129)

0.613*
(0.134)

0.978***
(0.005)
0.997***
(0.010)
0.978***
(0.003)
1.003***
(0.000)
0.622*
(0.122)

0.977***
(0.005)
0.997***
(0.000)***
0.983***
(0.004)
1.002***
(0.001)

0.795**
(0.059)

0.866+
(0.068)

Final Technical Report: Habeas Litigation in U.S. District Courts

Based on both sets of analyses, several features appeared to be significantly associated with the
variance in disposition time and length of case for capital cases in our sample. 151
Amended petitions. An amended petition lengthened the time for termination of capital cases, adding
between 21% and 38% more days to disposition, controlling for other factors. See Table 20. Table 21,
reporting the results for the Cox analysis of all cases, including those that had not yet been terminated in the
several districts where most of the cases were still pending (AZ, NV, CA-C, OH-S, and PA-E), showed that
later on as cases aged, those with amended petitions were about twice as likely to have reached conclusion
compared to those that lacked an amended petition. This would be consistent with a pattern in which within a
given circuit or district, a case was more likely to terminate once an amended petition was filed, but that in
some of the oldest cases no amended petition had yet been filed. This is the situation in Nevada, for example,
where several cases were still stayed for exhaustion in state court, and where amended petitions are typically
filed following the lifting of the stay for exhaustion.
Magistrate judge reports. Longer disposition times among terminated cases were also associated
with the use of magistrate judges for a report and recommendation on the disposition of claims. The extra
steps required when a district judge refers the case for an initial opinion from a magistrate judge can include
the report and recommendation itself, objections to the report, a reply to the objections, and even, as in the
Southern District of Ohio, a supplemental report and recommendation with another round of objections.
These extra steps led to between 56% and 89% more days before disposition on average for cases in which a
report and recommendation was filed. This relationship between the use of magistrate judges for dispositive
rulings and the length of the case was less evident once all cases (whether terminated or not) were compared.
This is expected because fewer of the longest cases had reached the stage where a magistrate's report and
recommendation regarding disposition of claims would have been filed.
Substitution of counsel. The substitution of a petitioner’s attorney was unrelated to disposition time.
It slowed a case’s progression toward termination only when the district was left unspecified. In other words,
within any given district, the likelihood that a case would terminate does not change if the petitioner secures
new counsel. This is consistent with findings that the districts with fewer terminated cases also had more
counsel substitutions. Indeed, 48% of cases in OH-S and 30% cases in AZ (both districts in which over 60%
of cases had not yet reached conclusion) included a substitution.
Discovery, evidentiary hearings. Analyses of cases that had reached termination showed that
disposition time was not significantly lengthened by a judge's decision to authorize a discovery deposition or
a mental or physical examination, nor by the decision to hold an evidentiary hearing, once other factors were
taken into account. 152 Neither did discovery or an evidentiary hearing have any significant effect on the
likelihood of termination when evaluating terminated and non-terminated cases, controlling for other factors.
Stays for exhaustion. Among terminated cases, the length of a stay for exhaustion had no significant
effect on termination time, once location was specified. In other words, comparing only cases that had been
completed in any given district, the presence or absence of a stay for exhaustion did not make a significant
151

The following factors were not significant in the models of terminated cases: petitioner race, substitution of one
petitioner's attorney for another, any claim Teague-barred, any claim of innocence of guilt, any claim of delayed trial or
appeal or right to be present at trial or right to a public trial, any jury instruction or judicial comment claim, any claim of
insufficient evidence of guilt (reason model had this barely significant when controlling for circuits only, suggesting that
cases including this claim took 28% more days than those without this claim), whether the petitioner challenged
sentencing phase only, any claim of improper admission or exclusion of evidence, whether the court ordered discovery,
and whether the court ordered an evidentiary hearing.
152
Because of small numbers or interactions with other variables, we did not include these features in the analysis of
non-capital disposition time.

Final Technical Report: Habeas Litigation in U.S. District Courts

difference. But stays had a strong effect on the likelihood that a case would be terminated. For every
additional 100 days that a case was stayed, termination was 30% less likely. This is consistent with two
findings: 1) that of the seven districts with the fewest cases resolved, five were also among the six districts
with the longest average stayed periods; 2) only 17% of the cases with stays had reached termination
compared to 71% of those cases without stays.
Time period for preparation of counseled petition. Longer periods of time before a counseled
petition extended disposition times. Among terminated cases, each 10 additional days increased the time for
termination by 1%. Also, in the first few months in any given district and controlling for other factors, a case
was 20% less likely to terminate for every 10 additional days that elapse prior to the filing of a counseled
petition. 153
Number of claims. The greater the number of claims raised in a petition, the less likely the case was
to reach termination. Controlling for other factors, including location, termination was 20% less likely for
every 10 additional claims. This was consistent with our finding that of the 7 districts with the fewest cases
resolved, six of those districts were also among the seven districts with the most claims per petition. For
terminated cases, the number of claims was significant in only one of the six models, and only at the 10%
significance level. This suggests that even though the number of claims does have an effect in delaying
termination generally, for cases that had already been resolved by the end of November 2006, the number of
claims did not make much of a difference in the time for disposition.
Claim type. Among all of the claim types investigated, ineffective assistance of counsel was the only
one associated with both reducing the likelihood of termination 154 and lengthening disposition time. It added
from 81% to 176% more days to complete than in terminated cases without an ineffective assistance claim.
Among terminated cases, some additional claims were associated with longer disposition times: 1) a claim
challenging the petitioner’s competency at trial; 2) a claim of prosecutorial misconduct (improper argument,
Brady violations, presentation of false evidence, destruction of evidence, intimidation of witnesses, etc.), and
3) a claim of error in jury selection, misconduct, or bias. Roper, Atkins, and Ring claims (regarding eligibility
for execution) also added 46% to 73% more days to disposition for terminated cases. Other claims, including
claims of new evidence of innocence, were not significantly related to the variance in disposition time.
Defenses and reasons for dismissal. In our analysis of terminated cases, in any given district and
controlling for other factors, a successive petition took about half the time (96% fewer days) to complete as a
case that was not dismissed as successive. Another potentially speedy disposition, namely dismissal because
of the statute of limitations, was not faster at all. 155 Instead, among the capital cases that had already
terminated, a ruling that a claim was time-barred added 47% to 112% more days to disposition compared to
cases without such a ruling, on average and after controlling for other factors. Procedural default rulings
appear to slow down processing time of terminated cases as well. Cases including at least one claim rejected
153

The relationship between likelihood of termination and the change in the period before a counseled petition was not
constant over the life of these cases. For the cases that last the longest, each additiona1 10 days before the counseled
petition was filed raised the likelihood of termination by about 3%.
154
A case with an IAC claim was initially much less likely to terminate than one that lacked such a claim, a relationship
that did not show up in any of the models once the shortest cases were removed. This suggests that several of the
shortest cases lacked IAC claims. This makes sense, given the findings noted earlier in the report that 29% of the Texas
cases lacked a single IAC claim compared to only 8% of the cases from districts outside of Texas, and that of the 70
cases without an IAC claim, 47 were dismissed without reaching the merits.
155
“AEDPA was supposed to speed things up. Significant new provisions like the time bar, if honestly applied, should
have reduced disposition times, especially for non-capital cases.” Ronald Eisenberg, Testimony before the Committee on
the Judiciary Subcommittee on Crime, Terrorism, and Homeland Security, House of Representatives, November 16,
2005, at 8 (emphasis in original).

Final Technical Report: Habeas Litigation in U.S. District Courts

as defaulted took 30% to 57% longer than cases without such a ruling, holding other factors constant. 156
(These termination-contingent covariates were not included in the analyses investigating the likelihood of
termination.)
Merits review. Terminated cases in which the court granted the writ on any claim took 54% to 74%
more days to complete, a finding consistent with other studies of capital cases which concluded that it takes a
reviewing court longer to disturb than to affirm a state capital conviction or sentence. 157 Controlling for other
factors, any case with at least one claim denied on the merits took more than twice as long on average as other
cases, that is, from 114% to 121% longer. (These termination-contingent covariates were not included in the
analyses investigating the likelihood of termination.)
Caseload. Controlling for caseload, a greater number of capital habeas petitions per judge was
associated with slower and fewer dispositions. Each additional capital habeas case per judge added about
18% more days to disposition time for terminated cases, controlling for factors other than location. Each
additional capital habeas case per judge reduced the likelihood that a case would be terminated by about 32%,
controlling for other factors.
Specialized staff, death penalty law clerks. Because death penalty clerk positions are allocated by
caseload and caseload is associated with longer and fewer dispositions, it was no surprise that an increase in
the number of death penalty law clerks was associated with longer not shorter disposition times, and with a
lower likelihood of termination, after controlling for other factors. Each additional clerk was associated with a
decrease of between 16% and 19% in the likelihood of termination. Again, this does not indicate whether the
disposition times in a district with a relatively high numbers of specialized staff would be longer or shorter
without the extra staff. It only suggests that holding everything but location constant, districts with more
death penalty law clerks are slower in processing capital habeas cases than districts with fewer clerks. An
analysis of individual districts over time as clerk staffing levels change may be a useful investigation.
Location. To various degrees, the identity of the federal court in which the case was filed was
correlated with disposition time and the likelihood of termination, even after controlling for case-specific
features.
Comparing terminated cases only, OK-W was consistently associated with longer average disposition
times, whether comparing districts or circuits, in all four models. Table 20. PA-E was significantly longer as
well, but only in the two models controlling for the reason for disposition. These differences were observed
after controlling for stayed periods, number of claims, dismissal of defaulted claims, use of amended
petitions, and other factors. The small number of cases that had reached termination in the Ninth Circuit were
not significantly longer than the terminated cases in the Fifth Circuit, after controlling for other factors.
Evaluating both terminated and non-terminated cases for likelihood of termination, however, produced
different results. By this measure, OK-W was not the slowest of the thirteen districts. The results of five
different models examining the effect of location on the likelihood of termination are included in Table 21.
Examining location alone without also controlling for caseload showed that AZ, OH-S, NV, and PA-E were
all associated with less progress towards termination than other districts. A capital habeas case started in any
of the thirteen districts other than OH-S, NV, and PA-E was significantly more likely to reach termination

156

In one model, cases in which all claims were dismissed as unexhausted were speedier as well. These cases took 62%
fewer days to complete than other cases, controlling for other factors including district.
157
Fagan et al., supra note 42; Latzer & Cauthen, supra note 43.

Final Technical Report: Habeas Litigation in U.S. District Courts

than a case filed in the District of Arizona, which had the lowest percentage of terminated capital cases in the
sample. 158
Comparing the likelihood of termination by circuit produced similar results. Controlling for other
factors, cases filed outside of Texas were significantly less likely to be completed than those filed in the four
Texas districts within the Fifth Circuit. For example, a case filed in the one district in our sample from the
Third Circuit (PA-E) was 85% less likely to be completed than a case filed in the Fifth Circuit. Cases from
the three districts in the Ninth Circuit (AZ, NV, and CA-C) were 90% less likely to be concluded than cases
from the four Texas districts in the Fifth Circuit.
To examine whether the influence of location remained after controlling for capital habeas caseload,
we added to the Cox regression model that specified districts two dummy variables for caseload (medium
caseload, between 1.5 and three capital cases per judge, and high caseload, that is, more than three cases per
judge). See Table 37. The results suggest that differences in likelihood of terminating in some districts are
somewhat but not completely a function of a higher capital caseload. The presence of a high capital habeas
caseload was independently associated with slowed progress toward termination. Four districts that had been
significantly associated with a higher likelihood of termination than AZ when caseload was not included were
no longer significantly associated with a difference in termination likelihood once caseload was added. The
relative increase in the likelihood of termination associated with two of the Texas districts was diminished,
but remained significant. The increased likelihood of termination in one district, OK-W, was unchanged. One
district that showed no significant association of with a difference in the likelihood of termination in the
district model, PA-E, became significantly associated with a decrease in the likelihood of termination as
compared to AZ once caseload was controlled.
These analyses confirm that the location where the case is filed has a significant impact on how soon it
will be terminated, even after controlling for features associated with disposition time, including number of
claims, stay length, time before the counseled petition is filed, the use of a magistrate judge, petitioner race,
claim type, use of discovery or evidentiary hearing, substitution of counsel, and capital caseload.

Factors associated with likelihood of relief in capital cases

Of the 267 terminated (not transferred) capital cases in our sample, 33 received relief. The grant rate
varied by district, from a high of 75% (six of eight cases) in PA-E to 0% in three districts – FL-M (zero of 10
cases), CA-C (zero of five cases), and NV (zero of six cases). See Table 11, Part II. Some claims may be
associated with relief. See Table 12, Part II. AEDPA contains several provisions that could contribute to a
lower grant rate, including: greater restrictions on evidentiary hearings (if indeed evidentiary hearings are
associated with a higher likelihood of relief); limitations barring merits review of claims that could be
meritorious (successive petitions, statute of limitations); and a more deferential standard of review for state
decisions of fact and law. We tested whether the presence of various features was associated with a lower or
higher probability of relief using an analysis assessing likelihood of grant.

158

Cases from CA-C and OK-W showed an increase in likelihood of termination only when omitting the 10% shortest
cases.

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 22. (C) Results of probit analysis of likelihood of grant (cont. next page).
model controls for >
variable \/
evidentiary hearing held probit coeff
(r.s.e.)
Dprobit coeff
deposition or exam ordered

at least one claim barred as defaulted

default defense rejected on any claim

claim of innocence of capital murder
included in case
any claim of ineffective assistance
included in case
claim of ineffective assistance for
sentencing included in case
presence, public or delay claim
included in case

159

CIRC’S
IACANY

DIST’S
IACANY

STATES
IACANY

CIRCUITS
IACSENT

DIST’ S
IACSENT

STATES
IACSENT

1.051**
(0.374)
[0.211]
0.810*
(0.351)
[0.142]
-0.394
(0.260)
[-0.044]
0.720*
(0.303)
[0.111]
0.667+
(0.370)
[0.109]
0.410
(0.374)
[0.037]

1.318***
(0.387)
[0.291]
0.768*
(0.335)
[0.128]
-0.354
(0.282)
[-0.039]
0.777*
(0.327)
[0.120]
0.694+
(0.382)
[0.113]
0.379
0.391
[0.034]

1.063**
(0.379)
[0.241]
0.731*
(0.349)
[0.141]
-0.376
(0.258)
[-0.050]
0.828**
(0.295)
[0.154]
0.592
(0.380)
[0.108]
0.375
(0.395)
[0.042]

1.095**
(0.377)
[0.210]
0.792*
(0.378)
[0.127]
-0.429
(0.263)
[-0.044]
0.750*
(0.301)
[0.108]
0.760*
(0.367)
[0.121]

1.466***
(0.402)
[0.315]
0.669+
(0.365)
[0.093]
-0.371
(0.292)
[-0.035]
0.790*
(0.328)
[0.108]
0.811*
(0.387)
[0.125]

1.083**
(0.387)
[0.231]
0.685+
(0.366)
[0.119]
-0.446+
(0.265)
[-0.054]
0.860***
(0.294)
[0.149]
0.678+
(0.375)
[0.119]

0.808**
(0.290)
[0.072]
0.619+
(0.375)
[0.092]
-0.512
(0.328)
[-0.041]
0.804**
(0.299)
[0.125]
-0.029**
(0.010)
[-0.003]

0.933**
(0.303)
[0.076]
0.874*
(0.406)
[0.145]
-0.551
(0.353)
[-0.040]
0.701*
(0.335)
[0.096]
-0.036***
(0.011)
[-0.003]

0.842***
(0.297)
[0.089]
0.466
(0.447)
[0.074]
-0.601+
(0.346)
[-0.056]
0.539+
(0.327)
[0.084]
-0.017
(0.011)
[-0.002]

0.387
0.525
0.241
(0.365)
(0.395)
(0.427)
[0.054]
[0.079]
[0.036]
insufficient evidence of sentencing
-0.487
-0.511
-0.609+
(0.334)
(0.358)
(0.359)
factor claim included in case
[-0.044]
[-0.044]
[-0.063]
Roper, Atkins, or Ring claim included
0.720*
0.605+
0.479
(0.298)
(0.332)
(0.321)
in case
[0.116]
[0.090]
[0.079]
number of claims raised
-0.026**
-0.032**
-0.015
(0.010)
(0.011)
(0.010)
[-0.002]
[-0.003]
[-0.003]
Robust standard errors in parenthesis; dprobit coefficients in brackets
+ Significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%

159

To interpret, the number in brackets is the difference in probability of relief associated with the variable: E.g.:
for the model that controls for districts and ineffective assistance at sentencing claims ("DISTRICTS IACSENT"),
1) a case in which an evidentiary hearing was held was 32 percentage points more likely to result in a grant than a case in
which an evidentiary hearing was not held, controlling for other factors.
2) a case in the Northern District of Texas was nine percentage points less likely to result in a grant than a case from the
Eastern District of Pennsylvania, controlling for other factors.
3) Each additional claim raised in a petition was associated with a reduction in the likelihood of relief of 0.3%.

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 22. (C) Results of probit analysis of likelihood of grant (cont.).
model controls for >
CIRCUITS DISTRICTS STATES CIRCUITS DISTRICTS STATES
variable \/
IACANY
IACANY
IACANY IACSENT IACSENT
IACSENT
Circuit 5 (4 districts in TX only)
-2.602***
-2.459***
(0.646)
(0.645)
[-0.517]
[-0.457]
Circuit 6 (OH-N & OH-S only)
-1.857**
-1.769**
(0.637)
(0.648)
[-0.078]
[-0.068]
Circuit 9 (AZ, CA-C, & NV only) -2.261**
-2.109**
(0.719)
(0.728)
[-0.072]
[-0.062]
Circuit 10 (OK-W only)
-3.056***
-2.951***
(0.820)
(0.822)
[-0.073]
[-0.064]
Circuit 11 (FL-M & AL-N only)
-1.969**
-1.902**
(0.689)
(0.701)
[-0.077]
[-0.068]
Alabama Northern District
-1.545*
-1.509+
(0.758)
(0.780)
[-0.051]
[-0.062]
Texas Northern District
-2.643***
-2.502***
(0.738)
(0.750)
[-0.113]
[-0.092]
Texas Eastern District
-2.884***
-2.534***
(0.739)
(0.747)
[-0.090]
[-0.071]
Texas Southern District
-2.983***
-2.950***
(0.704)
(0.707)
[-0.251]
[-0.220]
Texas Western District
-2.369**
-2.202**
(0.764)
(0.765)
[-0.087]
[-0.070]
Ohio Northern District
-1.531*
-1.333*
(0.664)
(0.671)
[-0.066]
[-0.053]
Ohio Southern District
-2.651**
-2.641**
(0.919)
(0.904)
[-0.060]
[-0.050]
Arizona District
-1.772*
-1.471
(0.897)
(0.929)
[-0.056]
[-0.045]
Oklahoma Western District
-3.194***
-3.135**
(0.866)
(0.881)
[-0.074]
[-0.061]
State of Texas (all)
-1.134**
-0.997*
(0.402)
(0.404)
[-0.207]
[-0.161]
Constants: 3 Cir., PA-E, PA
0.389
0.554
-1.116*
0.128
0.151
-1.422**
(0.709)
(0.758)
(0.468)
(0.720)
(0.771)
(0.475)
Observations
266
245^
245^
266
245^
245^
Prob > chi2
0.0000
0.0000
0.0001
0.0000
0.0000
0.0002
+ Significant at 10%; * significant at 5%; ** significant at 1%; *** significant at .10%.
^ FL, CA, and NV cases had no grants. States of OH, AL, and PA were not significant in states comparisons.

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 22 presents significant associations 160 for six models testing the relationship between specified
features and the odds of ending in a grant. The first three models control for whether the case included any
claim of ineffective assistance expected to be associated with a higher likelihood of relief compared to those
cases that lacked ineffective assistance claims. The second three substitute for this variable whether the case
included a claim of ineffective assistance for the sentencing stage. This subset of ineffective assistance claims
appeared to be the primary basis for grants of relief in cases from PA-E. Within each model, we examined the
effect of circuit, district, and state. For descriptive information about each feature, including distribution
among districts, see Part I.
Evidentiary hearings and discovery. Turning first to the features that remained significant regardless
of other specifications, one stands out in particular, that is, whether or not the federal court held an evidentiary
hearing. An evidentiary hearing has a more powerful relationship with the probability of relief than any
variable other than location. It increases the likelihood of relief by 21 to 32 percentage points, after
controlling for other factors. Also increasing the likelihood of relief (by nine to 12 percentage points) was an
order of discovery and (by about 11 to 15 percentage points) the rejection of the defense of procedural default
on any claim. Combining this finding with the analyses of disposition time, these results suggest that neither a
judge’s decision to order a hearing nor an order of discovery is linked to greater delay in resolution of the
capital cases in our sample. Both are associated with an increased likelihood that the federal court will grant
the writ. It is unclear whether evidentiary hearings and discovery are granted because the judge first
determines that a claim is potentially meritorious, or whether the causal relationship operates in the other
direction, with discovery and hearings revealing proof of merit that would otherwise be unavailable.
Claim type. The presence of three types of constitutional claims in the petition also was associated
with a greater likelihood of relief. A Roper, Atkins, or Ring claim raised the likelihood that a case would
receive relief by about 10 percentage points. Nine of the 33 cases involving a grant received relief on this
basis. Another subset of cases included sentences overturned because of ineffective assistance of counsel at
the sentencing stage. After controlling for other factors, the presence of this claim raised the likelihood that
the case would be granted by about 8 percentage points. (Note that an unspecified claim of ineffective
assistance of counsel had no significant association with relief; only when claims alleging failings related to
sentencing were specified did the significant relationship appear.) Finally, a claim raising new evidence of
innocence of guilt was related to a higher likelihood of a grant, raising it by about 11 percentage points.
Actually, in none of the 33 cases receiving relief did the federal court grant relief because it concluded that
the defendant had advanced convincing proof of factual innocence that was not presented in the state trial
court. Instead, it appeared that the presence of an innocence claim operated somehow to make a grant of relief
on a different claim more likely. Six other claim types were examined and none were significant in relation to
whether or not the federal court granted a writ. 161
160

Not significantly associated with a change in the probability of relief were petitioner race (expected to have no
association with likelihood of relief); whether an amended petition was filed (expected to increase likelihood of relief);
the substitution of petitioner's attorney (expected to decrease probability of relief); the presence of a magistrate judge
report and recommendation (expected to decrease the likelihood of relief); a claim challenging the admission or
exclusion of evidence, including search and seizure and confession violations ; a jury selection misconduct or bias claim;
a competency claim; a jury instruction claim; a claim of delay of appeal or transcript; insufficient evidence of guilt
claim; greater number days before filing of counseled petition standard of review (expected to increase likelihood of
relief). We also ran all of these with the additional control of the continuous variable ST2FED, the time interval in days
between state judgment and federal filing, expecting that the longer the period before the federal filing the higher the
probability of relief. It was barely significant at the 10% level for the two state-specified models only, indicating that for
each additional day before filing, a grant became just slightly more likely.
161
A review of the docket sheets of the 95 pending cases as of May 2007 found that since coding closed in November
2006, only three additional cases had been resolved. Of these, two were granted relief. An Atkins claim was the basis for
relief from sentence in a case from AL-N and a Brady claim persuaded the district judge to order relief from both
conviction and sentence for a petitioner in OH-N. The third case, from OH-N, included a claim of actual innocence.

Final Technical Report: Habeas Litigation in U.S. District Courts

Standard of review. A simple comparison of cases that applied the deferential standard of review
with those that did not suggested that AEDPA's deferential standard would reduce the likelihood of a grant.
Once other factors were taken into account in this more detailed analysis, however, the application of a
standard of review other than that specified by 28 U.S.C. § 2254(d) had no significant relationship to the
likelihood of relief. This suggests that once other factors are taken into account, AEDPA’s new standard of
review for state court decisions is not suppressing relief, at least for this set of already terminated capital
cases.
Number of claims and preparation time. Consistent with pre-AEDPA research, we found that relief
is more likely for petitions raising fewer claims. Specifically, for every additional claim raised, the likelihood
of relief declines about 0.3 percentage points. Combining this finding with the analysis of timing, it appears
that additional claims increase the processing time for these cases but do not raise the odds of relief. A
somewhat related finding was that the time that elapses before the filing of a counseled petition had no
relationship to the likelihood of a grant. This suggests that lengthier periods for preparation of the petition do
not independently raise a petitioner’s chances either.
Location. In controlling for district, circuit, and state in this analysis of probability of relief, we used
as our comparison PA-E in the Third Circuit where 6 of 8 terminated cases ended with the court granting the
writ. This grant rate of 75% far exceeded the rate in any other jurisdiction in our sample. See Table 22. Under
all specifications, after all other features were taken into account (including the types of claims raised, the
number of claims whether there was a hearing or discovery, and whether the default defense was rejected on
any claim), location remained significant. Texas cases were associated with likelihood of relief that was
about 7 to 22 percentage points lower than PA-E when comparing districts, 16 to 21 percentage points lower
when comparing states, and 46 to 52 percentage points lower comparing circuits. 162

CONCLUSION
Two general points should be emphasized in closing. First, this new information about habeas
litigation addresses only one level of federal habeas review – the district court. Although litigation in noncapital cases is largely a district court phenomenon, 163 a large portion of capital cases are regularly appealed
to the upper levels of the federal court system. Further research is needed into the appellate review of habeas
cases under AEDPA. Examination of the appellate review of the specific cases in this study would be a
logical next step, but must await the disposition of the significant portion of those cases still unfinished in the
district courts.
Second, any recommendations for policy change are beyond the scope of this particular study. The
research provides a rich source of empirical information with which policy options can be explored.

162

In addition, AL-N and Ohio capital cases were 5 to 8 percentage points less likely than cases in PA-E to end in relief,
whether comparing districts or circuits, but were not significantly different when comparing states.
163
Only a tiny portion of non-capital habeas petitioners are granted a certificate of appealability by the district court, and
the number of such certificates granted by the appellate courts is probably also quite small.

Final Technical Report: Habeas Litigation in U.S. District Courts

Appendix A: Statistical Analyses Tables
Table 23: (NC) Cox regression probability of termination over time, by district.
District Comparisons
Model
Observ's
Log likelihood
LR chi2(18)
Prob > chi2
Covariate
amended petition
answer, mo to dism
magistrate judge R&R
notice of appeal filed
no. of claims
prior petition dismissed w-o prej
petitioner had atty
defaulted claim
jury misconduct
judicial bias
speed, public, pres.
petitioner incompetence
evidentiary claim
improper comment by prosecutor
false, lost, or undisclosed evidence
jury instructions, trial or sentencing, or
judicial comment
denial or delay of appeal or transcript
MA
NJ
PA-E
PA-M
PA-W
NC-E
NH
SC
VA-E
AL-N
AL-M
AL-S
FL-N
LA-W
MS-N
MS-S
TX-N
FL-M
FL-S
GA-S
LA-E
TX-E
TX-S
TX-W
MI-E
MI-W
OH-N
OH-S
CT
TN-M
IL-N
IN-N
WI-E
NY-N

No Claims

Claims

1954
-10892.511
2088.900
0.000
Coef.
-1.821
-8.718
-2.328
-1.618
-0.071
-0.414
-2.545

1246
-6312.006
1390.930
0.000
Std. Err.
0.441
0.544
0.273
0.304
0.009
0.180
0.796

P>z
0
0
0
0
0
0.021
0.001

%Change in
Hazard Rate
-83.81%
-99.98%
-90.25%
-80.17%
-6.89%
-33.91%
-92.15%

-1.177
-1.524
-0.821
-1.061
-1.176
-0.716
-1.477
-0.522
-0.665
-1.297
2.328
-1.189
-0.949
-1.435
-0.605
-1.046
-0.735

0.350
0.270
0.196
0.262
0.255
0.329
0.403
0.223
0.192
0.272
1.064
0.349
0.260
0.273
0.286
0.227
0.175

0.001
0
0
0
0
0.03
0
0.019
0.001
0
0.029
0.001
0
0
0.034
0
0

-69.18%
-78.23%
-56.00%
-65.38%
-69.16%
-51.15%
-77.16%
-40.69%
-48.56%
-72.67%
925.65%
-69.54%
-61.27%
-76.19%
-45.40%
-64.86%
-52.07%

-0.388
-0.521
-0.798
-0.496
-0.630
-0.855
-1.001
-0.903
-1.087
-1.480
-1.750
1.925
-0.469
-0.473
-1.090
-2.304

0.211
0.211
0.229
0.203
0.168
0.215
0.190
0.224
0.274
0.336
0.402
1.155
0.211
0.226
0.273
0.303

0.065
0.014
0
0.015
0
0
0
0
0
0
0
0.096
0.027
0.036
0
0

-32.18%
-40.61%
-54.97%
-39.09%
-46.75%
-57.47%
-63.26%
-59.48%
-66.27%
-77.23%
-82.62%
585.35%
-37.44%
-37.67%
-66.37%
-90.02%

Coef.
-0.280
-8.916
-2.530
-1.655
-0.044
-0.392
-0.338
-4.705
0.560
0.305
-0.260
-0.261
-0.261
-0.004
-0.168

Std. Err.
0.102
0.765
0.363
0.400
0.012
0.221
0.128
0.791
0.218
0.210
0.152
0.259
0.093
0.120
0.102

P>z
0.006
0
0
0
0
0.076
0.009
0
0.01
0.146
0.087
0.313
0.005
0.971
0.099

% Change in
Hazard Rate
-24.40%
-99.99%
-92.03%
-80.90%
-4.34%
-32.44%
-28.67%
-99.09%
75.07%
35.69%
-22.88%
-22.98%
-22.98%
-0.43%
-15.46%

-0.241
0.331
-1.326
-1.442

0.099
0.143
0.404
0.326

0.015
0.02
0.001
0

-21.41%
39.17%
-73.45%
-76.34%

-0.820
0.327
0.012
-55.96%
-0.562
0.368
0.126
-43.02%
District not included in Claims Regression
District not included in Claims Regression
-0.201
0.288
0.487
-18.18%
District not included in Claims Regression
-1.624
0.346
0
-80.29%
2.585
1.185
0.029
1226.58%
-1.422
0.413
0.001
-75.88%
-1.058
0.309
0.001
-65.28%
-1.428
0.318
0
-76.02%
District not included in Claims Regression
-1.616
0.455
0
-80.12%
-0.808
0.211
0
-55.41%
-1.310
0.228
0
-73.02%
-0.469
0.267
0.078
-37.46%
-0.459
0.283
0.105
-36.82%
-0.880
0.267
0.001
-58.51%
0.490
0.728
0.501
63.18%
-0.537
0.202
0.008
-41.58%
-0.727
0.247
0.003
-51.65%
-1.069
0.241
0
-65.66%
-0.043
0.693
0.951
-4.21%
-1.216
0.324
0
-70.36%
-1.789
0.377
0
-83.29%
District not included in Claims Regression
District not included in Claims Regression
-0.605
0.279
0.03
-45.38%
-0.431
0.307
0.16
-35.00%
District not included in Claims Regression
-2.765
0.430
0
-93.70%

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 23: (NC) Cox regression probability of termination over time, by district, cont.
Model
Covariate
MN
MO-E
MO-W
NY-E
AZ
CA-N
CA-S
NV
OR
NY-S
CO
KS
OK-N
NY-W

No Claims
Coef.
Std. Err P>z
.
-0.527
0.307
0.086
-1.879
0.270
0
0.249
0.240
0.3
-1.542
0.204
0
-1.460
0.301
0
-1.763
0.189
0
-0.922
0.274
0.001
-1.955
0.347
0
-1.285
0.294
0
-2.195
0.219
0
-1.223
0.253
0
-1.405
0.293
0
-2.084
0.405
0
-2.274
0.336
0

%Change in Hazard Rate
-40.97%
-84.73%
28.32%
-78.61%
-76.78%
-82.85%
-60.22%
-85.85%
-72.34%
-88.87%
-70.56%
-75.46%
-87.56%
-89.71%

Claims
Coef. Std. Err.
0.095
-2.140
0.524
-1.645
-1.707
-0.966
-2.101
-1.533
-2.527
-1.629
-2.784
-1.974

P>z

% Change in Hazard Rate

0.396
0.809
10.01%
0.310
0
-88.23%
0.294
0.075
68.89%
0.256
0
-80.71%
District not included in Claims Regression
0.340
0
-81.85%
0.336
0.004
-61.95%
0.458
0
-87.77%
0.387
0
-78.40%
0.350
0
-92.01%
District not included in Claims Regression
0.388
0
-80.38%
0.471
0
-93.82%
0.430
0
-86.12%

Time Varying Covariates
amended petition
answer, mo to dism
magistrate judge R&R
notice of appeal filed
petitioner had atty
defaulted claim
AL-M
FL-M
TX-E
TN-M
CA-E
CA-C

0.283
1.387
0.428
0.374
0.354

0.078
0.106
0.050
0.055
0.134

0
0
0
0
0.008

32.77%
300.35%
53.41%
45.42%
42.53%

-0.603
-0.295

0.190
0.075

0.001
0

-45.27%
-25.54%

-0.515
-0.406
-0.246

0.209
0.119
0.062

0.014
0.001
0

-40.23%
-33.36%
-21.82%

1.443
0.465
0.377

0.152
0.065
0.071

0
0
0

323.37%
59.16%
45.79%

0.846
-0.693

0.133
0.206

0
0.001

132.95%
-49.98%

-0.232

0.135
0.084
-20.73%
District not included in Claims Regression
-0.378
0.154
0.014
-31.48%
-0.180
0.089
0.043
-16.44%

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 24: (NC) Cox regression probability of termination over time, by circuit and state.
State comparisons
Model
Observ’s
Log likelihood
LR chi2(18)
Prob > chi2

No Claims

Claims

1954
-11006.961
1860.000
0.000

1246
-6377.741
1259.460
0.000

Covariate

Coef.

Std. Err.

P>z

% Change in
Hazard Rate

amended petition
answer, mo to dism
magistrate j R&R
notice of appeal filed
no. of claims
petition filed?
petitioner had atty
defaulted claim
guilty plea challenged?
jury misconduct
evidentiary claim
false, lost, or undisclosed
evidence
jury instructions, trial or
sentencing, or judicial comment
disciplinary hrg. Challenge
only sentence challenged
circuit1
circuit2
circuit3
circuit5
circuit6
circuit8
circuit9
circuit10
circuit11
MA
NJ
PA
NC
NH
SC
VA
AL
LA
MS
TX
GA
MI
OH
TN
CT
IL
WI
NY
MO
AZ
NV
OR
CO
KS
OK

-0.339
-8.631

0.078
0.542

0
0

-28.76%
-99.98%

-1.651
-0.057

0.301
0.008

0
0

-80.82%
-5.52%

-2.845

0.797

-94.19%

-0.81385
-1.42131
-0.62974
-0.37931
-0.60029
-0.44171
-0.96106
-0.79671
-0.41716
-1.116
-1.403
-0.881
-0.485
-1.365
-0.457
-0.460
-1.130
-1.033
-0.796
-0.586
-0.369
-0.890
-1.178
2.147
-1.555
-0.356
-0.933
-1.857

0.257
0.132
0.132
0.108
0.128
0.153
0.108
0.159
0.113
0.345
0.263
0.161
0.231
0.399
0.215
0.170
0.195
0.190
0.190
0.136
0.180
0.161
0.225
1.158
0.397
0.204
0.266
0.158

-1.335
-1.765
-1.163
-1.099
-1.264
-1.073

0.295
0.342
0.288
0.246
0.287
0.249

0.002
0
0
0
0
0.004
0
0
0
0.001
0
0
0.036
0.001
0.033
0.007
0
0
0
0
0.041
0
0
0.064
0
0.081
0
0

-55.69%
-75.86%
-46.73%
-31.57%
-45.13%
-35.71%
-61.75%
-54.92%
-34.11%
-67.23%
-75.40%
-58.58%
-38.46%
-74.45%
-36.71%
-36.90%
-67.69%
-64.41%
-54.87%
-44.37%
-30.86%
-58.91%
-69.21%
755.87%
-78.87%
-29.94%
-60.66%
-84.39%

0
0
0
0
0
0

-73.67%
-82.88%
-68.75%
-66.69%
-71.76%
-65.78%

P>z

% Change in
Hazard Rate

-0.336
-9.122
-2.425
-1.630
-0.034
-0.364
-0.320
-4.468
-0.165
0.459
-0.227

Std. Err
.
0.100
0.764
0.342
0.391
0.012
0.218
0.126
0.778
0.099
0.212
0.091

0.001
0
0
0
0.004
0.095
0.011
0
0.094
0.03
0.013

-28.50%
-99.99%
-91.15%
-80.41%
-3.34%
-30.48%
-27.39%
-98.85%
-15.21%
58.28%
-20.29%

-0.241

0.101

0.017

-21.41%

Coef.

-0.257
0.150
0.087
-22.68%
0.240
0.137
0.08
27.12%
-0.93642
0.384501 0.015
-60.80%
-1.61942
0.187978
0
-80.20%
-0.7048
0.178933
0
-50.58%
-0.48503
0.139241
0
-38.43%
-0.80451
0.165102
0
-55.27%
0.942884
0.54489 0.084
156.74%
-0.33007
0.437196
0.45
-28.11%
-1.15783
0.234461
0
-68.58%
-0.68861
0.154834
0
-49.77%
0.912
1.270
0.473
148.93%
-1.481
0.313
0
-77.26%
-0.913
0.214
0
-59.89%
-0.534
0.353
0.13
-41.40%
State not included in Claims Regression
-0.161
0.269
0.551
-14.83%
State not included in Claims Regression
1.683
0.901
0.062
438.17%
-0.998
0.220
0
-63.13%
-1.556
0.446
0
-78.90%
-0.591
0.160
0
-44.64%
-0.436
0.264
0.099
-35.34%
-0.999
0.198
0
-63.17%
-1.356
0.262
0
-74.22%
State not included in Claims Regression
State not included in Claims Regression
-0.603
0.264
0.022
-45.30%
State not included in Claims Regression
-1.978
0.209
0
-86.16%
-1.271
0.234
0
-71.96%
State not included in Claims Regression
-1.898
0.447
0
-85.02%
-1.444
0.377
0
-76.41%
State not included in Claims Regression
-1.595
0.377
0
-79.72%
1.759
0.989
0.075
480.52%

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 24: (NC) Cox regression probability of termination over time, by circuit and state, cont.
State comparisons
Model

Covariate

No Claims

Claims

Coef.

Std. Err.

P>z

% Change in
Hazard Rate

0.214
1.294
0.363
0.353
0.313

0.078
0.108
0.048
0.057
0.142

0.006
0
0
0
0.027

23.85%
264.80%
43.73%
42.31%
36.72%

% Change
in Hazard
Rate

Coef.

Std. Err.

P>z

1.491
0.444
0.392

0.151
0.062
0.069

0
0
0

344.06%
55.88%
47.97%

0.809

0.131

124.48%

-0.522
-0.260

0.154
0.084

0.001
0.002

-40.66%
-22.89%

-0.244
-0.561

0.077
0.177

0.002
0.001

-21.63%
-42.91%

Time Varying Covariates
amended petition
answer, mo to dism
magistrate judge R&R
notice of appeal filed
petitioner had atty
defaulted claim
disciplinary hrg. Challenge
circuit8
circuit9
AL
FL
TN
MO
CA
OK

0.28498
-0.12416
-0.427
-0.464
-0.766
-0.977
-1.028
-1.218

0.099371
0.075325
0.123
0.068
0.214
0.104
0.053
0.174

0.004
0.099
0.001
0
0
0
0
0

-24.80%
-11.68%
-34.74%
-37.11%
-53.52%
-62.37%
-64.22%
-70.41%

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 25: (NC) Cox regression probability of termination over time, by clerk.
Model
Observ's
Log likelihood
LR chi2(18)
Prob > chi2
Covariate
amended petition
answer, mo to dism
magistrate judge R&R
notice of appeal filed
no. of claims
petitioner had atty
defaulted claim
jury misconduct
evidentiary claim
false, lost, or
undisclosed evidence
Jury instructions, trial
or sentencing, or
judicial comment
denial or delay of
appeal or transcript
disciplinary claim
only sentence
challenged
average number of
clerks

No Claims

Claims

1954
-11122.610
1628.700
0.000
Coef.
-1.796
-8.316
-2.028
-1.572
-0.278
-3.398

1246
-6472.648
1069.650
0.000
Std. Err.
0.424
0.539
0.243
0.293
0.058
0.805

P>z
0
0
0
0
0
0

% Change in
Hazard Rate
-83.41%
-99.98%
-86.83%
-79.24%
-24.31%
-96.66%

-0.015

0.007

0.033

-1.50%

0.240
1.346
0.397
0.370
0.040
0.492

0.075
0.105
0.045
0.053
0.010
0.135

0.001
0
0
0
0
0

27.11%
284.07%
48.76%
44.76%
4.09%
63.50%

Coef.
-0.447
-9.057
-2.414
-1.522
-0.027
-2.833
-4.350
0.430
-0.148

Std. Err.
0.097
0.762
0.315
0.385
0.011
0.962
0.757
0.208
0.089

P>z
0
0
0
0
0.017
0.003
0
0.039
0.097

% Change in
Hazard Rate
-36.03%
-99.99%
-91.05%
-78.17%
-2.67%
-94.11%
-98.71%
53.72%
-13.77%

-0.196

0.097

0.044

-17.79%

-0.244

0.094

0.01

-21.64%

0.244
1.499

0.138
0.532

0.077
0.005

27.67%
347.79%

1.482
0.458
0.363

0.151
0.057
0.068

0
0
0

340.37%
58.06%
43.80%

0.411
0.798
-0.286

0.160
0.127
0.097

0.01
0
0.003

50.86%
122.19%
-24.85%

0.205

0.089

0.021

22.73%

Time Varying Covariates
amended petition
answer, mo to dism
magistrate j. R&R
notice of appeal filed
no. of claims
petitioner had atty
defaulted claim
disciplinary hrg.
only sentence
challenged

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 26. (NC) Cox regression probability of termination over time, by caseload.
Model
Observ's
Log likelihood
LR chi2(18)
Prob > chi2

1954
-11080.024
1713.880
0.000

Covariate
amended petition

Coef.
-2.017

Std. Err.
0.438

answer, mo to dism

-8.777

magistrate j R&R

No Claims

Claims
1246
-6432.198
1150.550
0.000

% Change in
Hazard Rate
-86.70%

Coef.
-1.447

Std. Err.
0.587

0.537

-99.98%

-9.117

0.761

-99.99%

-2.012

0.242

-86.62%

-2.262

0.315

-89.58%

notice of appeal filed

-1.585

0.294

-79.51%

-1.515

0.388

-78.01%

no. of claims

-0.054

0.008

-5.21%

-0.029

0.011

0.009

-2.90%

petitioner had atty

-2.959

0.791

-94.81%

-2.493

0.967

0.01

-91.73%

defaulted claim

-4.297

0.750

-98.64%

jury misconduct

0.556

0.209

0.008

74.36%

Evidentiary claim

-0.163

0.089

0.068

-15.03%

false, lost, or
undisclosed evidence

-0.201

0.098

0.04

-18.19%

jury instructions, trial
or sentencing, or
judicial comment

-0.208

0.093

0.026

-18.77%

denial or delay of
appeal or transcript

0.227

0.138

0.101

25.44%

Disciplinary hrg.
Challenge

1.514

0.530

0.004

354.41%

only sentence
challenged

-1.084

0.497

0.029

-66.17%

caseload per judge

P>z

P>z
0.014

% Change in
Hazard Rate
-76.48%

-0.008

0.001

-0.85%

-0.010

0.001

-0.95%

amended petition

0.304

0.077

35.57%

0.208

0.101

0.04

23.11%

answer, mo to dism

1.426

0.105

316.08%

1.496

0.150

346.44%

magistrate j R&R

0.401

0.045

49.40%

0.445

0.057

56.02%

notice of appeal filed

0.369

0.053

44.64%

0.362

0.068

43.59%

0.415

0.132

0.002

51.46%

0.355

0.161

0.027

42.64%

defaulted claim

0.777

0.126

117.47%

disciplinary hrg.
challenge

-0.277

0.097

0.004

-24.19%

only sentence
challenged

0.262

0.089

0.003

29.99%

Time Varying Covariates

no. of claims
petitioner had atty

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 27. (NC) Regression results: terminated cases by circuit, no claims model.
LNBEG2TERM

Coef.

Robust Std. Err.

P>t

Beta

PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
lnclaimcount
IACANY
INNANY
PLEAANY
SXSANY
CONFANY
JURSELANY
APPRENANY
JURMISANY
JUDBIASANY
ATRTSANY
INCOMPANY
EVGANY
PMISCANY
BRADYANY
JRYINSANY
APPEALANY
INSUFGANY
INSUFSANY
CNSLANY
DISCANY
REVOCANY
UNEXH
SOLANY
PDANY
SPANY
NOPETFLD
VOLDISM
othdisms
REVSTD
PPDISNOP
REPADDR
COADRULR
SENTONLR
Grant
YEARSTRT
circuit1
circuit2

0.0857096
0.2331176
1.617229
0.0044925
0.2249254
-0.0862639
0.1021543
-0.0316745
-0.0068114
0.0436707
-0.0943017
-0.0791629
0.0390426
-0.0360187
-0.3233891
-0.1116282
0.1921404
0.3022985
0.0988846
0.0296941
0.0014284
0.0459689
-0.0615665
-0.0042953
-0.0676791
-0.1416818
-0.2898041
-0.1226392
-0.2505789
-0.1108252
0.1551538
-0.2676652
-0.0536216
-0.1847394
-0.2347372
-0.0842682
0.2436374
0.2064404
0.1696671
0.0464805
0.277225
0.0691127
0.4547692
0.868604

0.101556
0.062985
0.074222
0.004693
0.053186
0.049238
0.042626
0.061428
0.120723
0.069429
0.072224
0.098253
0.082741
0.140144
0.130311
0.100894
0.08227
0.162055
0.056588
0.065238
0.072202
0.05684
0.101258
0.057252
0.134148
0.11592
0.128645
0.122826
0.103936
0.063119
0.05578
0.13792
0.122496
0.148942
0.113499
0.079443
0.19532
0.082874
0.156711
0.11589
0.403817
0.046965
0.294944
0.126624

0.84
3.7
21.79
0.96
4.23
-1.75
2.4
-0.52
-0.06
0.63
-1.31
-0.81
0.47
-0.26
-2.48
-1.11
2.34
1.87
1.75
0.46
0.02
0.81
-0.61
-0.08
-0.5
-1.22
-2.25
-1
-2.41
-1.76
2.78
-1.94
-0.44
-1.24
-2.07
-1.06
1.25
2.49
1.08
0.4
0.69
1.47
1.54
6.86

0.399
0
0
0.339
0
0.08
0.017
0.606
0.955
0.529
0.192
0.421
0.637
0.797
0.013
0.269
0.02
0.062
0.081
0.649
0.984
0.419
0.543
0.94
0.614
0.222
0.024
0.318
0.016
0.079
0.006
0.053
0.662
0.215
0.039
0.289
0.213
0.013
0.279
0.688
0.493
0.141
0.123
0

0.0130731
0.0624958
0.6002351
0.0189038
0.0936664
-0.0352673
0.0710252
-0.0132482
-0.0010995
0.012912
-0.0220256
-0.0149974
0.0079761
-0.0047007
-0.0422049
-0.0145684
0.0361272
0.0328933
0.0326346
0.0073607
0.0003996
0.0136575
-0.0116637
-0.00142
-0.0106888
-0.0226248
-0.0846287
-0.0290339
-0.0613723
-0.0383739
0.0512049
-0.0510866
-0.0077729
-0.0232843
-0.0441365
-0.0193166
0.0295745
0.0466342
0.018002
0.0162943
0.0151813
0.0287235
0.0333542
0.1608096

circuit3
circuit4
circuit5
circuit6
circuit8
circuit9
circuit10
circuit11

0.531101
0.2451378
0.4050607
0.4831853
0.149645
0.4955391
0.5206957
0.3274696

0.134983
0.127835
0.11657
0.122437
0.14734
0.125069
0.215679
0.122329

3.93
1.92
3.47
3.95
1.02
3.96
2.41
2.68

0
0.055
0.001
0
0.31
0
0.016
0.008

0.1105252
0.0420123
0.1471593
0.1158963
0.0272637
0.1577247
0.0679551
0.1008912

_cons

3.644159

0.183742

19.83

Number of obs
F( 21, 1805)
Prob > F
R-squared
Root MSE

=
=
=
=
=

1827
95.22
0
0.5619
0.83838

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 28. (NC) Regression results terminated cases by circuit, claims model.

LNBEG2TERM

Coef.

Robust
Std. Err.

P>t

Beta

Number of obs
F( 52, 1109)
Prob > F
R-squared
Root MSE

=
=
=
=
=

1162
24.11
0
0.599
0.77483

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 29. (NC) Regression Results, terminated cases, by states, no claims model.
LNBEG2TERM

Coef.

Robust
Std. Err.

P>t

Beta

PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
lnclaimcount
NOPETFLD
PPDISNOP
REPADDR
YEARSTRT
MA
NJ
PA
MD
NC
SC
VA
AL
FL
LA
MS
TX
GA
KY
MI
OH
TN
IL
WI
NY
AR
MN
MO
AZ
CA
NV
OR
WA
CO
KS
OK
_cons

-0.1199486
0.301995
1.606476
0.0028122
0.2556185
0.0161474
0.1535254
-0.159907
0.1498743
0.2939347
0.0676763
0.5572983
1.11763
0.6120354
0.035604
0.4167865
0.5086384
0.6359756
0.6409117
0.5054078
0.848863
0.5962846
0.525095
0.4700992
1.376576
0.7023725
0.7950381
0.1315714
0.3920327
0.6398827
1.151994
0.1629451
0.6136782
0.5538763
1.067955
0.7380147
1.077292
1.107976
0.3817885
0.9635194
1.037138
0.1601835
3.339117

0.1122
0.0548
0.0503
0.0035
0.0473
0.0404
0.0238
0.1061
0.1568
0.0825
0.0389
0.2706
0.2109
0.1466
0.2214
0.1663
0.1541
0.1308
0.1588
0.127
0.1549
0.1673
0.1222
0.1489
0.2992
0.1281
0.1555
0.3045
0.1689
0.2262
0.1266
0.2324
0.2
0.1667
0.2173
0.1222
0.4061
0.1687
0.2356
0.127
0.3016
0.2511
0.1589

-1.07
5.51
31.94
0.79
5.4
0.4
6.46
-1.51
0.96
3.56
1.74
2.06
5.3
4.18
0.16
2.51
3.3
4.86
4.04
3.98
5.48
3.56
4.3
3.16
4.6
5.48
5.11
0.43
2.32
2.83
9.1
0.7
3.07
3.32
4.91
6.04
2.65
6.57
1.62
7.58
3.44
0.64
21.01

0.285
0
0
0.427
0
0.69
0
0.132
0.339
0
0.082
0.04
0
0
0.872
0.012
0.001
0
0
0
0
0
0
0.002
0
0
0
0.666
0.02
0.005
0
0.483
0.002
0.001
0
0
0.008
0
0.105
0
0.001
0.524
0

-0.0176474
0.075382
0.6217925
0.010352
0.1013336
0.0061251
0.1012284
-0.0224523
0.0158534
0.0571569
0.0268286
0.0342428
0.0900536
0.1100508
0.0027151
0.0399447
0.0545966
0.0976245
0.0805906
0.110866
0.1141868
0.074979
0.1524056
0.0632365
0.0765501
0.1219939
0.0840975
0.0103215
0.0444372
0.0554341
0.2120004
0.0108782
0.0425041
0.0703637
0.0739679
0.2228044
0.0631303
0.0869186
0.0264431
0.0834714
0.0743341
0.0114807
.

Number of obs
F( 42, 1784)
Prob > F
R-squared
Root MSE

=
=
=
=
=

1827
53.42
0
0.5791
0.82664

Final Technical Report: Habeas Litigation in U.S. District Courts

100

Table 30. (NC) Regression results, terminated cases, by states, claims model.
LNBEG2TERM
PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
lnclaimcount
IACANY
INNANY
PLEAANY
SXSANY
CONFANY
JURSELANY
APPRENANY
JURMISANY
JUDBIASANY
ATRTSANY
INCOMPANY
EVGANY
PMISCANY
BRADYANY
JRYINSANY
APPEALANY
INSUFGANY
INSUFSANY
CNSLANY
DISCANY
REVOCANY
UNEXH
SOLANY
PDANY
SPANY
NOPETFLD
VOLDISM
othdisms
REVSTD
PPDISNOP
REPADDR
COADRULR
SENTONLR
Grant
YEARSTRT
MA
NJ
PA
NC
SC
VA
AL
FL
LA
MS
TX
GA
MI
OH
IL
NY
MN
MO
AZ
CA

Coef.

Std. Err.

P>t

Beta

0.0558293
0.2328576
1.629848
0.0050365
0.3010578
-0.08834
0.0955119
-0.049511
0.0115366
0.0502477
-0.071792
-0.121677
0.0282389
-0.065389
-0.365919
-0.091225
0.2054117
0.2401485
0.1075813
0.0450068
0.0092947
0.0328426
-0.055943
0.0147497
-0.058537
-0.137844
-0.160647
-0.094575
-0.242805
-0.131797
0.1265654
-0.244424
-0.041381
-0.185075
-0.21844
-0.076926
0.2908349
0.161595
0.1185278
0.0419897
0.4455086
0.0655106
0.7159515
1.35301
0.6091405
0.4902126
0.3833141
0.6077041
0.5341969
0.5629672
0.8446293
0.8897264
0.5319763
0.5370226
0.6696701
0.7186381
0.6054039
1.109245
0.2994255
0.5222709

0.097816
0.064384
0.075995
0.004811
0.058909
0.04934
0.043413
0.062408
0.127249
0.070928
0.074723
0.101781
0.086041
0.127132
0.131203
0.105703
0.081809
0.171316
0.056425
0.06499
0.074509
0.056674
0.104188
0.057464
0.128848
0.114952
0.130375
0.125602
0.106967
0.062162
0.057057
0.142375
0.124796
0.165619
0.115342
0.080768
0.192193
0.088435
0.14972
0.115677
0.341045
0.047186
0.310261
0.213896
0.152286
0.15643
0.176221
0.18079
0.184136
0.140845
0.179239
0.212229
0.129721
0.170412
0.144687
0.179037
0.18363
0.13988
0.189204
0.179209

0.57
3.62
21.45
1.05
5.11
-1.79
2.2
-0.79
0.09
0.71
-0.96
-1.2
0.33
-0.51
-2.79
-0.86
2.51
1.4
1.91
0.69
0.12
0.58
-0.54
0.26
-0.45
-1.2
-1.23
-0.75
-2.27
-2.12
2.22
-1.72
-0.33
-1.12
-1.89
-0.95
1.51
1.83
0.79
0.36
1.31
1.39
2.31
6.33
4
3.13
2.18
3.36
2.9
4
4.71
4.19
4.1
3.15
4.63
4.01
3.3
7.93
1.58
2.91

0.568
0
0
0.295
0
0.074
0.028
0.428
0.928
0.479
0.337
0.232
0.743
0.607
0.005
0.388
0.012
0.161
0.057
0.489
0.901
0.562
0.591
0.797
0.65
0.231
0.218
0.452
0.023
0.034
0.027
0.086
0.74
0.264
0.059
0.341
0.131
0.068
0.429
0.717
0.192
0.165
0.021
0
0
0.002
0.03
0.001
0.004
0
0
0
0
0.002
0
0
0.001
0
0.114
0.004

0.0085052
0.0615843
0.6054855
0.0214355
0.1247565
-0.0360014
0.0660303
-0.0206414
0.0018625
0.0146939
-0.016861
-0.0224072
0.0057931
-0.0086293
-0.0466374
-0.0118349
0.0387388
0.024397
0.035359
0.0111783
0.0025787
0.0097738
-0.0106315
0.0048563
-0.0092415
-0.0220096
-0.047337
-0.022314
-0.0593249
-0.0456452
0.0412292
-0.0464507
-0.0060651
-0.0222696
-0.040876
-0.0176424
0.0357009
0.035759
0.0127185
0.0147456
0.0246777
0.0271472
0.0531122
0.1292315
0.1157621
0.0526019
0.0461232
0.0397941
0.0803676
0.1436677
0.1407351
0.0729044
0.1801467
0.0743442
0.1338159
0.088215
0.0728467
0.2075694
0.0222126
0.0785734

0.7033531

0.142214

4.95

0.212825

(dropped)

Final Technical Report: Habeas Litigation in U.S. District Courts

101

Table 30. (NC) Regression results, terminated cases, by states, claims model, cont.
LNBEG2TERM
OR
WA
KS
OK
_cons
Number of obs
F( 65, 1062)
Prob > F
R-squared
Root MSE

=
=
=
=
=

Coef.

Std. Err.

1.126901
0.0004847
1.318727
0.1589591
3.369873

0.212424
0.233673
0.384684
0.283807
0.202753

5.3
0
3.43
0.56
16.62

P>t

Beta

0
0.998
0.001
0.576
0.

0.0880808
0.000036
0.0978285
0.0141471

1128
21.81
0
0.617
0.76486

Table 31. (NC) Regression results, terminated cases, by clerk, no claims model.
LNBEG2TERM
PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
LNclaimcount
NOPETFLD
PPDISNOP
REPADDR
YEARSTRT
CLERK
_cons
Number of obs
F( 12, 1814)
Prob > F
R-squared
Root MSE

=
=
=
=
=

1827
147.68
0
0.538
0.8588

Coef.

Robust
Std. Err.

-0.097653
0.3717411
1.595757
0.0036965
0.2932186
0.0337478
0.145007
-0.1430992
0.0900567
0.3990412
0.0763421
-0.0019551
3.928384

0.108425
0.057426
0.050403
0.003693
0.040619
0.042206
0.023199
0.099489
0.154812
0.076342
0.040329
0.005818
0.113161

-0.9
6.47
31.66
1
7.22
0.8
6.25
-1.44
0.58
5.23
1.89
-0.34
34.71

P>t
0.368
0
0
0.317
0
0.424
0
0.151
0.561
0
0.059
0.737
0.

Beta
-0.0143672
0.0927916
0.6176433
0.0136073
0.1162393
0.0128013
0.0956117
-0.0200924
0.009526
0.0775954
0.0302639
-0.0053361

Final Technical Report: Habeas Litigation in U.S. District Courts

102

Table 32. (NC) Regression results, terminated cases, by clerk, claims model.
LNBEG2TERM
PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
lnclaimcount
IACANY
INNANY
PLEAANY
SXSANY
CONFANY
JURSELANY
APPRENANY
JURMISANY
JUDBIASANY
ATRTSANY
INCOMPANY
EVGANY
PMISCANY
BRADYANY
JRYINSANY
APPEALANY
INSUFGANY
INSUFSANY
CNSLANY
DISCANY
REVOCANY
UNEXH
SOLANY
PDANY
SPANY
NOPETFLD
VOLDISM
othdisms
REVSTD
PPDISNOP
REPADDR
COADRULR
SENTONLR
Grant
CLERK
_cons
Number of obs
F( 42, 1033)
Prob > F
R-squared
Root MSE
Prob > F

=
=
=
=
=
=

Coef.

Rob.Std. Err.

0.0304954
0.2590784
1.650142
0.0064673
0.2963784
-0.0395104
0.0920809
-0.0551076
0.015531
0.0252935
-0.0913629
-0.1419614
0.0533458
-0.0339829
-0.365707
-0.1206068
0.2124719
0.3271825
0.1067444
0.0418663
0.0292575
0.079071
-0.086673
-0.0277918
-0.0753819
-0.0935749
-0.310204
-0.0646603
-0.2436349
-0.1223287
0.0917131
-0.2027662
-0.0319627
-0.114027
-0.1457136
-0.0928219
0.2989885
0.2035015
0.2652622
-0.0262888
0.3994602
0.0077019
3.944282

0.111555
0.069316
0.078697
0.004379
0.051126
0.052763
0.045281
0.063593
0.133281
0.072214
0.075619
0.101073
0.085889
0.145125
0.136817
0.112181
0.092462
0.175029
0.060162
0.069659
0.076481
0.059999
0.107334
0.06136
0.139066
0.121507
0.137305
0.132095
0.111024
0.065705
0.061575
0.140067
0.120412
0.181279
0.121016
0.085354
0.189179
0.087828
0.167362
0.125336
0.409363
0.007306
0.161489

0.27
3.74
20.97
1.48
5.8
-0.75
2.03
-0.87
0.12
0.35
-1.21
-1.4
0.62
-0.23
-2.67
-1.08
2.3
1.87
1.77
0.6
0.38
1.32
-0.81
-0.45
-0.54
-0.77
-2.26
-0.49
-2.19
-1.86
1.49
-1.45
-0.27
-0.63
-1.2
-1.09
1.58
2.32
1.58
-0.21
0.98
1.05
24.42

1076
24.11
0
0.5784
0.79855
0

P>t
0.785
0
0
0.14
0
0.454
0.042
0.386
0.907
0.726
0.227
0.16
0.535
0.815
0.008
0.283
0.022
0.062
0.076
0.548
0.702
0.188
0.42
0.651
0.588
0.441
0.024
0.625
0.028
0.063
0.137
0.148
0.791
0.529
0.229
0.277
0.114
0.021
0.113
0.834
0.329
0.292
0.

Beta
0.0045511
0.067926
0.6140778
0.0280431
0.1221883
-0.0159747
0.0637207
-0.0228642
0.0024963
0.0073708
-0.0210969
-0.0257219
0.0109906
-0.0044898
-0.0466042
-0.0159346
0.0391657
0.0328635
0.0352355
0.0103105
0.0080683
0.0234073
-0.0161109
-0.0090576
-0.0118355
-0.0150402
-0.0920637
-0.0150051
-0.0597447
-0.0420721
0.0293773
-0.0386126
-0.0045735
-0.0133921
-0.0270854
-0.0212185
0.0366419
0.0450601
0.0282335
-0.0092394
0.0225457
0.021924

Final Technical Report: Habeas Litigation in U.S. District Courts

103

Table 33. (NC) Regression results, terminated cases, by caseload, no claims model.
LNBEG2TERM
PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
LNclaimcount
NOPETFLD
PPDISNOP
REPADDR
YEARSTRT
CASELOAD
_cons
Number of obs
F( 12, 1814)
Prob > F
R-squared
Root MSE

=
=
=
=
=

Coef.

Rob.Std. Err.

-0.083
0.339
1.621
0.003
0.251
0.034
0.155
-0.115
0.099
0.391
0.064
0.004
3.759

0.106653
0.056113
0.05028
0.003536
0.040731
0.041561
0.022906
0.104066
0.150601
0.075827
0.039948
0.00065
0.115164

-0.78
6.04
32.24
0.93
6.16
0.82
6.75
-1.1
0.66
5.16
1.61
5.97
32.64

1827
149.9
0
0.5451
0.85223

P>t
0.437
0
0
0.352
0
0.412
0
0.269
0.512
0
0.107
0
0.

Beta
-0.012187
0.0846212
0.627515
0.0121163
0.0995249
0.0129308
0.1019108
-0.0161458
0.0104515
0.076066
0.0255114
0.086594

Final Technical Report: Habeas Litigation in U.S. District Courts

104

Table 34. (NC) Regression results, terminated cases, by caseload, claims model.
LNBEG2TERM

Coef.

Rob.Std. Err.

P>t

Beta

0.0418
0.2277
1.6685
0.0061
0.2596
-0.0392
0.0933
-0.0365
0.0226
0.0219
-0.086
-0.1438
0.0719
-0.0374
-0.3871
-0.0996
0.1867
0.3137
0.1114
0.0441
0.038
0.0635
-0.0736
-0.0227
-0.0693
-0.1156
-0.3044
-0.0523
-0.2273
-0.0882
0.1162
-0.1805
0.0262
-0.1416
-0.1496
-0.0675
0.3015
0.2118
0.1475
-0.0367
0.3341
0.0038
3.7888

0.106876
0.068403
0.078674
0.004334
0.051232
0.052418
0.044985
0.063609
0.130881
0.071293
0.074269
0.100997
0.08462
0.14404
0.130719
0.110785
0.095349
0.17169
0.059738
0.06968
0.07615
0.059907
0.107574
0.061416
0.137431
0.119162
0.137144
0.131233
0.110647
0.06664
0.059736
0.139698
0.124412
0.173558
0.116731
0.083907
0.184585
0.086172
0.15129
0.125068
0.419918
0.000833
0.160035

0.39
3.33
21.21
1.4
5.07
-0.75
2.07
-0.57
0.17
0.31
-1.16
-1.42
0.85
-0.26
-2.96
-0.9
1.96
1.83
1.86
0.63
0.5
1.06
-0.68
-0.37
-0.5
-0.97
-2.22
-0.4
-2.05
-1.32
1.95
-1.29
0.21
-0.82
-1.28
-0.8
1.63
2.46
0.98
-0.29
0.8
4.6
23.67

0.696
0.001
0
0.161
0
0.455
0.038
0.566
0.863
0.758
0.247
0.155
0.396
0.795
0.003
0.369
0.05
0.068
0.062
0.527
0.618
0.289
0.494
0.712
0.614
0.332
0.027
0.69
0.04
0.186
0.052
0.197
0.833
0.415
0.2
0.422
0.103
0.014
0.33
0.769
0.426
0
0

0.0062354
0.0597019
0.6209159
0.0263925
0.1070425
-0.0158459
0.0645356
-0.0151518
0.0036291
0.0063924
-0.0198581
-0.0260582
0.0148065
-0.0049388
-0.04933
-0.0131559
0.0344223
0.0315134
0.0367753
0.0108528
0.0104872
0.018796
-0.0136767
-0.0073919
-0.0108837
-0.0185801
-0.0903324
-0.0121429
-0.0557358
-0.0303252
0.0372313
-0.034378
0.0037522
-0.0166254
-0.0278094
-0.0154194
0.036952
0.0468921
0.0157027
-0.0129002
0.0188564
0.0893684
.

Number of obs
F( 42, 1033)
Prob > F
R-squared
Root MSE

=
=
=
=
=

1076
24.53
0
0.5846
0.79262

Final Technical Report: Habeas Litigation in U.S. District Courts

105

Table 35. (NC) Regression results, terminated cases, by district, claims model, with and without caseload.
Variable

LNBEG2TERM
PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
lnclaimcount
IACANY
INNANY
PLEAANY
SXSANY
CONFANY
JURSELANY
APPRENANY
JURMISANY
JUDBIASANY
ATRTSANY
INCOMPANY
EVGANY
PMISCANY
BRADYANY
JRYINSANY
APPEALANY
INSUFGANY
INSUFSANY
CNSLANY
DISCANY
REVOCANY
UNEXH
SOLANY
PDANY
SPANY
NOPETFLD
VOLDISM
othdisms
REVSTD
PPDISNOP
REPADDR
COADRULR
SENTONLR
Grant
YEARSTRT

Caseload not included

Coef.
0.132
0.219
1.655
0.004
0.416
-0.030
0.108
-0.048
0.051
0.037
-0.147
-0.142
0.058
-0.045
-0.378
-0.064
0.204
0.174
0.127
0.048
-0.002
0.066
-0.044
0.016
-0.015
-0.139
-0.207
-0.068
-0.225
-0.122
0.129
-0.150
0.028
-0.159
-0.154
-0.126
0.392
0.102
-0.065
0.061
0.457
0.053

Robust
Std.
Err.
0.104
0.068
0.078
0.003
0.081
0.051
0.044
0.063
0.129
0.069
0.075
0.109
0.088
0.126
0.134
0.103
0.088
0.177
0.057
0.068
0.076
0.057
0.108
0.059
0.129
0.110
0.132
0.124
0.106
0.063
0.059
0.140
0.139
0.157
0.117
0.086
0.192
0.093
0.152
0.115
0.390
0.048

Caseload included

P>t
1.270
3.210
21.220
1.350
5.150
-0.580
2.490
-0.760
0.390
0.530
-1.970
-1.300
0.660
-0.360
-2.820
-0.620
2.310
0.980
2.230
0.710
-0.020
1.150
-0.410
0.280
-0.120
-1.270
-1.570
-0.550
-2.110
-1.930
2.200
-1.070
0.200
-1.010
-1.310
-1.470
2.050
1.100
-0.430
0.530
1.170
1.110

0.204
0.001
0.000
0.179
0.000
0.563
0.013
0.450
0.695
0.594
0.049
0.194
0.509
0.719
0.005
0.535
0.021
0.325
0.026
0.478
0.981
0.249
0.684
0.781
0.906
0.206
0.116
0.585
0.035
0.054
0.028
0.285
0.841
0.314
0.190
0.143
0.041
0.273
0.669
0.594
0.241
0.268

Beta
0.020
0.057
0.616
0.018
0.171
-0.012
0.075
-0.020
0.008
0.011
-0.034
-0.026
0.012
-0.006
-0.048
-0.008
0.038
0.018
0.042
0.012
-0.001
0.020
-0.008
0.005
-0.002
-0.022
-0.062
-0.016
-0.055
-0.042
0.041
-0.029
0.004
-0.019
-0.029
-0.029
0.048
0.023
-0.007
0.021
0.026
0.022

medcaseload
hicaseload
MAD
NJD
PAeastD
PAmidD
NCmidD
SCD
VAwestD
ALnorthD
ALmidD
FLnorthD
LAwestD
TXnorthD
FLmidD
FLsouthD
GAnorthD
LAD

0.735
1.325
0.426
0.630
0.328
0.271
(dropped)
0.618
0.379
0.423
1.206
0.515
0.723
0.204
0.562
0.522

0.334
0.230
0.184
0.266
0.198
0.202

2.200
5.760
2.310
2.360
1.660
1.350

0.028
0.000
0.021
0.018
0.097
0.179

0.056
0.129
0.069
0.061
0.027
0.033

0.232
0.298
0.203
0.292
0.188
0.182
0.188
0.196
0.199

2.660
1.270
2.080
4.130
2.740
3.980
1.090
2.870
2.630

0.008
0.204
0.037
0.000
0.006
0.000
0.277
0.004
0.009

0.058
0.038
0.046
0.121
0.117
0.141
0.030
0.070
0.072

Coef.
0.136
0.217
1.647
0.004
0.413
-0.031
0.104
-0.050
0.054
0.034
-0.142
-0.136
0.052
-0.038
-0.396
-0.081
0.213
0.182
0.126
0.055
-0.002
0.066
-0.036
0.020
-0.011
-0.154
-0.191
-0.071
-0.230
-0.132
0.127
-0.156
0.047
-0.159
-0.157
-0.118
0.360
0.106
-0.064
0.056
0.446
0.055
-0.767
[-0.272]
0.592
[0.215]
0.071
0.666
-0.227
0.741
-0.333
0.380
(dropped)
0.725
0.489
0.534
1.310
-0.742
0.829
-0.452
-0.095
-0.135

Robust
Std.
Err.

P>t

Beta

0.104
0.068
0.078
0.003
0.081
0.051
0.043
0.063
0.129
0.069
0.075
0.109
0.088
0.126
0.135
0.101
0.088
0.178
0.057
0.068
0.076
0.057
0.108
0.059
0.129
0.109
0.132
0.124
0.106
0.063
0.059
0.140
0.142
0.157
0.117
0.086
0.187
0.094
0.152
0.115
0.389
0.048

1.320
3.200
21.170
1.340
5.130
-0.600
2.390
-0.800
0.420
0.500
-1.900
-1.250
0.600
-0.300
-2.940
-0.800
2.430
1.030
2.210
0.810
-0.030
1.160
-0.330
0.340
-0.080
-1.410
-1.450
-0.580
-2.170
-2.090
2.170
-1.110
0.330
-1.010
-1.340
-1.360
1.930
1.130
-0.420
0.490
1.150
1.150

0.188
0.001
0.000
0.180
0.000
0.547
0.017
0.424
0.676
0.618
0.057
0.213
0.551
0.766
0.003
0.423
0.015
0.305
0.027
0.418
0.979
0.248
0.738
0.737
0.934
0.159
0.149
0.563
0.031
0.037
0.030
0.268
0.740
0.311
0.180
0.174
0.054
0.260
0.676
0.626
0.252
0.250

0.020
0.057
0.613
0.018
0.170
-0.012
0.072
-0.021
0.009
0.010
-0.033
-0.025
0.011
-0.005
-0.050
-0.011
0.039
0.018
0.042
0.013
-0.001
0.020
-0.007
0.006
-0.002
-0.025
-0.057
-0.017
-0.056
-0.045
0.041
-0.030
0.007
-0.019
-0.029
-0.027
0.044
0.023
-0.007
0.020
0.025
0.023

0.211

-3.630

0.000

-0.317

0.375
0.336
0.233
0.182
0.274
0.202
0.213

1.580
0.210
2.850
-1.250
2.700
-1.650
1.780

0.115
0.832
0.004
0.213
0.007
0.099
0.075

0.215
0.005
0.065
-0.037
0.072
-0.028
0.047

0.243
0.306
0.216
0.299
0.378
0.194
0.190
0.193
0.196

2.980
1.600
2.470
4.380
-1.960
4.260
-2.380
-0.490
-0.690

0.003
0.110
0.014
0.000
0.050
0.000
0.017
0.621
0.490

0.068
0.049
0.058
0.132
-0.169
0.161
-0.068
-0.012
-0.019

Lo, Med,
or Hi
Habeas
Caseload

L
L
L
M
L
M
M
M
M
M
M
H
M
L
L
L

Final Technical Report: Habeas Litigation in U.S. District Courts

106

Table 35. (NC) Regression results, terminated cases, by district, claims model, with and without caseload, cont.
Variable

LNBEG2TERM
TXeastD
TXsouthD
TXwestD
MIeastD
MIwestD
OHnorthD
OHsouthD
ILnorthD
INnorthD
AReastD
MND
MOeastD
MOwestD
NYeastD
CAnorthD
CAeastD
CAcenD
CAsouthD
ORD
NYsouthD
WAwestD
KSD
OKwestD
_cons

Caseload not included
Robust
Std.
Err.
Coef.
0.477
0.195
0.491
0.172
0.528
0.202
0.849
0.160
0.372
0.222
0.474
0.196
0.982
0.297
0.620
0.200
0.508
0.226
-0.050
0.251
0.247
0.209
1.071
0.237
-0.005
0.179
1.007
0.174
1.261
0.379
1.047
0.191
0.425
0.169
0.469
0.223
1.204
0.231
1.187
0.205
-0.066
0.258
1.339
0.391
0.032
0.300
3.225
0.23
Number of obs=
F( 80, 995) =
Prob > F =
R-squared =
Root MSE =

Caseload included

t
2.450
2.860
2.610
5.300
1.680
2.420
3.310
3.090
2.250
-0.200
1.180
4.520
-0.030
5.770
3.320
5.480
2.510
2.100
5.220
5.800
-0.260
3.430
0.110
14.21

P>t
0.014
0.004
0.009
0.000
0.094
0.016
0.001
0.002
0.025
0.842
0.239
0.000
0.979
0.000
0.001
0.000
0.012
0.036
0.000
0.000
0.798
0.001
0.916
0
1076
19.84
0
0.642
0.750

Beta
0.074
0.104
0.076
0.136
0.048
0.046
0.078
0.076
0.049
-0.004
0.019
0.114
-0.001
0.146
0.119
0.158
0.107
0.044
0.096
0.108
-0.005
0.101
0.003

Robust
Std.
Err.
0.204
0.184
0.213
0.173
0.377
0.197
0.304
0.207
0.231
0.260
0.211
0.248
0.185
0.186

Coef.
0.576
0.593
-0.144
0.953
-0.883
-0.186
1.081
-0.044
0.599
0.052
-0.414
1.179
-0.665
1.113
(dropped)
-0.210
0.369
-0.828
0.351
-0.192
0.223
1.306
0.241
1.290
0.215
-0.730
0.260
1.446
0.395
0.134
0.307
3.882
0.213
Number of obs =
F( 81, 994) =
Prob > F =
R-squared =
Root MSE =

t
2.820
3.220
-0.680
5.510
-2.340
-0.940
3.560
-0.210
2.590
0.200
-1.960
4.760
-3.600
5.970
-0.570
-2.360
-0.860
5.420
5.990
-2.810
3.660
0.440
18.190

P>t
0.005
0.001
0.499
0.000
0.019
0.346
0.000
0.832
0.010
0.842
0.050
0.000
0.000
0.000

Beta
0.089
0.125
-0.021
0.153
-0.115
-0.018
0.086
-0.005
0.058
0.004
-0.031
0.126
-0.074
0.162

0.568
0.019
0.389
0.000
0.000
0.005
0.000
0.662
0.000
1076
19.7
0
0.643
0.749

-0.032
-0.209
-0.018
0.104
0.117
-0.055
0.109
0.012
.

Lo, Med,
or Hi
Habeas
Caseload
M
M
L
M
H
L
M
L
M
M
L
M
L
M
H
H
H
L
M
M
L
M
M

Final Technical Report: Habeas Litigation in U.S. District Courts

Table 36. (NC) Regression of disposition time in days (logged), by district, no claims model,
with and without caseload.
Showing
Change in Beta
Only,
Significance indicated by shading
Districts w/o caseload
LNBEG2TERM
Beta
PETSEX
AMENDPET
STPLEAD
RECORD
RR4DISP
NOTAPP
Lnclaimcount
NOPETFLD
PPDISNOP
REPADDR
YEARSTRT
Medcaseload
Hicaseload
MAD
NJD
PaeastD
PamidD
PawestD
MDD
NceastD
NcmidD
SCD
VaeastD
VawestD
AlnorthD
AlmidD
AlsouthD
FlnorthD
LawestD
MsnorthD
MssouthD
TxnorthD
FlmidD
FlsouthD
GanorthD
GasouthD
LAD
TxeastD
TxsouthD
TxwestD
KywestD
MieastD
MiwestD
OhnorthD
OhsouthD
TneastD
TnmidD
IlnorthD
InnorthD
WieastD
NynorthD
AreastD
MND
MoeastD
MowestD
NyeastD
AZD

Number of obs
F( 66, 1746)
Prob > F
R-squared
Root MSE

0.028
0.083
0.041
0.064
0.055
-0.002
0.039
0.003
0.036
0.075

Caseload added
Beta
-0.014
0.075
0.620
0.008
0.141
0.017
0.116
-0.022
0.021
0.046
0.025
[-23.2%] -0.264
[49.5%] 0.402
-0.001
0.044
-0.025
0.079
0.012
-0.039
0.009
-0.032
0.056
0.100

0.057
0.026
0.020
0.029
0.093
0.020
0.063
0.062
0.087
0.018
0.053
-0.008
0.046
0.053
0.071
0.064
0.072
0.098
0.046
0.042

0.072
0.040
0.032
0.045
0.108
0.034
0.082
-0.233
0.118
-0.041
-0.005
0.004
-0.007
0.078
0.103
0.006
0.045
0.123
-0.137
0.003

-0.010
0.019
0.035
0.009
0.050
0.060
0.002
0.032
0.089
-0.005
0.108
0.067

-0.038
-0.006
-0.019
0.030
0.066
-0.054
0.014
-0.002
0.105
-0.050
0.131
0.079

-0.012
0.075
0.621
0.009
0.140
0.021
0.116
-0.024
0.023
0.045
0.026

=1813
=39.23
=0
=0.608
=0.804

Habeas caseload level

L
L
L
M
L
L
L
L
M
M
M
M
M
M
M
M
M
M
H
M
L
L
M
L
M
M
L
L
M
H
L
M
L
L
L
M
M
H
M
L
M
L
M
M

107

Final Technical Report: Habeas Litigation in U.S. District Courts

108

Table 37. (C) Cox regression probability of termination over time, by district, with and without caseload.
No. of subjects
No. of failures
Time at risk
LR chi2(33)
Log likelihood
Prob > chi2
No of obs

Districts without caseload
330.000
236.000
2298.041
340.51
1058.377
=
0
=
330
P>z
% Change
in Hazard
Coef.
Rate
-0.010 0.414
-1.00%
-6.855
0
-99.89%
-0.264 0.259
-23.21%
-0.080 0.729
-7.70%
-0.016 0.003
-1.58%
-0.003
0
-0.27%
-0.024
0
-2.40%
0.013 0.957
1.30%
0.281 0.213
32.41%
0.181
0.5
19.83%
0.003 0.988
0.34%
-0.027 0.872
-2.64%
-0.284 0.108
-24.72%
0.129 0.433
13.77%
0.112 0.513
11.82%
0.070 0.672
7.28%
0.006 0.978
0.65%
-0.366
0.13
-30.64%
-0.290
0.15
-25.15%

_t
PETRACE
AMENDPET
REPSUB
RRREVD
claimcount
STAYTIME
REPETLAG
INNANY
IACANY
ATRTSANY
INCOMPANY
JRYINSANY
INSUFGANY
SXSCONFVGL~Y
PMISCBRADY~Y
JURSELMISB~Y
DISCR
EVH
ROPATRNG
hicaseload
medcaseload
DISTRICT:
PaeastD
0.906 0.211
147.42%
AlnorthD
1.841 0.008
530.37%
FlmidD
2.189 0.004
792.30%
TXnorthD
2.392
0
993.51%
TXeastD
3.109
0
2140.56%
TXsouthD
3.572
0
3459.25%
TXwestD
2.438
0
1045.06%
OHnorthD
2.470
0
1082.50%
OHsouthD
0.859 0.251
136.18%
CAcenD
1.717 0.047
456.59%
OKwestD
1.873 0.008
551.05%
NVD
-0.171 0.843
-15.76%
t
AMENDPET
0.967 0.001
162.90%
REPETLAG
0.003
0
0.32%
^AL-N and FL- M dropped due to collinearity

Districts with caseload
330
236
2298.041
340.51
1058.377
0
330
% Change
in Hazard
Rate
-1.00%
-99.89%
-23.21%
-7.70%
-1.58%
-0.27%
-2.40%
1.30%
32.41%
19.83%
0.34%
-2.64%
-24.72%
13.77%
11.82%
7.28%
0.65%
-30.64%
-25.15%
-88.79%
-29.35%

Caseload
level

L
M
M
L
M
L
M
M
H
M
H
H

Coef.
-0.010
-6.855
-0.264
-0.080
-0.016
-0.003
-0.024
0.013
0.281
0.181
0.003
-0.027
-0.284
0.129
0.112
0.070
0.006
-0.366
-0.290
-2.189
-0.347

Std. Err.
0.012
1.938
0.234
0.231
0.005
0.000
0.005
0.240
0.226
0.268
0.234
0.166
0.176
0.165
0.171
0.166
0.238
0.242
0.201
0.761
0.505

Z
-0.820
-3.540
-1.130
-0.350
-2.950
-5.360
-5.350
0.050
1.240
0.670
0.010
-0.160
-1.610
0.780
0.650
0.420
0.030
-1.510
-1.440
-2.870
-0.690

P>z
0.414
0.000
0.259
0.729
0.003
0.000
0.000
0.957
0.213
0.500
0.988
0.872
0.108
0.433
0.513
0.672
0.978
0.130
0.150
0.004
0.492

[95%
Conf.
-0.034
-10.653
-0.723
-0.534
-0.026
-0.004
-0.033
-0.458
-0.161
-0.345
-0.455
-0.353
-0.630
-0.194
-0.223
-0.255
-0.461
-0.840
-0.684
-3.681
-1.338

-1.283

0.560

-2.290

0.022

-2.380

-0.186

-72.27%

0.551
1.268
1.384
0.597
0.629
0.859
-0.124
1.873
-0.171

0.411
0.431
0.444
0.410
0.384
0.748
0.696
0.708
0.865

1.340
2.940
3.110
1.450
1.640
1.150
-0.180
2.650
-0.200

0.180
0.003
0.002
0.146
0.101
0.251
0.858
0.008
0.843

-0.255
0.423
0.513
-0.207
-0.123
-0.608
-1.488
0.485
-1.867

1.356
2.113
2.254
1.401
1.381
2.326
1.239
3.261
1.524

0.967
0.003

0.284
0.001

3.410
5.000

0.001
0.000

0.410
0.002

1.523
0.004

73.47%
255.44%
298.89%
81.65%
87.59%
136.18%
-11.70%
551.05%
-15.76%
-1.00%
-99.89%
-23.21%

Interval]
0.014
-3.057
0.194
0.374
-0.005
-0.002
-0.015
0.483
0.723
0.707
0.462
0.299
0.062
0.452
0.447
0.396
0.473
0.108
0.105
-0.696
0.643

^
^

Final Technical Report: Habeas Litigation in U.S. District Courts

109

Appendix B. Time-barred capital cases, summarized, by state. 164
ALABAMA
Arthur v. Haley
Thomas D. Arthur was convicted of murder and sentenced to death. His conviction and sentence were
affirmed by the Alabama Court of Criminal Appeals and the Alabama Supreme Court. The Alabama Supreme
Court denied Arthur’s application for rehearing on March 20, 1988. Arthur did not file a petition for review
with the United State Supreme Court. Because a certificate of judgment was issued by the Alabama Court of
Criminal Appeals on April 7, 1998, Arthur had until April 7, 2000, to file a state petition for post-conviction
review. Arthur did not do so, and on September 8, 2000, the Attorney General of the State of Alabama filed a
motion for the court to set a date for Arthur’s execution. On January 25, 2001, Arthur’s pro bono counsel
filed a motion with the Circuit Court of Jefferson County seeking leave to file a state post-conviction petition,
arguing that due to Arthur’s indigence he was unable to obtain counsel to help him file the petition prior to the
deadline. His motion was denied. On March 23, 2001 the Alabama Supreme Court ordered that Arthur be
executed on April 27, 2001.
Arthur then moved the Alabama Supreme Court to stay the death warrant, and that motion was denied. On
April 20, 2001 (20 months after the one-year AEDPA period ended), Arthur filed his first and only petition
for a writ of habeas corpus in the NDAL raising 113 claims of error, including a claim of actual innocence.
Arthur’s attorney also filed a motion for stay of execution, which was granted. The state appealed the district
court’s grant of a stay of execution and the Eleventh Circuit refused to vacate the stay.
The district court then denied Arthur’s petition as being time-barred, finding that there was not a sufficient
showing of actual innocence to trigger equitable tolling. The court also rejected Arthur’s arguments that the
statute of limitations should be tolled because the state failed to provide him with state post-conviction
counsel or legal assistance, prevented him from communicating with people who might have helped him,
failed to notify him of the certificate of judgment that triggered the limitations period, and failed to provide an
adequate law library. Arthur appealed the district court’s decision and the Eleventh Circuit affirmed. 452
F.3d 1234. A petition for writ of certiorari was denied by the Supreme Court on April 16, 2007. Arthur is still
on death row. http://www.doc.alabama.gov/deathrow.asp.
Siebert v. Haley
Daniel Siebert’s conviction for capital murder became final when the U.S. Supreme Court denied his petition
for writ of certiorari on November 5, 1990. Siebert filed a state petition for post-conviction relief on August
25, 1992, which the trial court found to be time-barred while also denying Siebert’s claims on the merits. The
Alabama Court of Criminal Appeals affirmed the denial on December 30, 1999, and the Alabama Supreme
Court denied certiorari on September 15, 2000. Siebert filed his federal petition for a writ of habeas corpus in
the NDAL on September 14, 2001. With representation in federal court by attorneys from the federal
defender's office, Siebert raised 42 claims of error in his petition.
The district court found that the statute of limitations for filing Siebert’s federal petition had expired on April
23, 1997, and that Siebert’s state post-conviction petition did not toll the running of the time. The district
court deferred to the state court’s conclusion that as a matter of state law, the post-conviction petition was not
filed in time. The district court also found that the state court’s filing deadline was clear and well established
and consistently applied. The district court denied equitable tolling and rejected Siebert’s arguments that the
164

Status reported as of March 2007 unless otherwise noted.

Final Technical Report: Habeas Litigation in U.S. District Courts

110

Alabama Court of Criminal Appeals had failed to notify him when it issued the certificate of judgment and
that he was merely attempting to exhaust his state court remedies in order to satisfy AEDPA.
Siebert appealed the district court’s decision and the Eleventh Circuit vacated and remanded, finding that
Alabama’s procedural rule was not firmly established nor was it regularly applied at the time it was applied in
Siebert’s case. The Eleventh Circuit held that Siebert’s post-conviction petition was timely filed and tolled the
time limit for filing a federal petition until September 15, 2000, thus making his September 14, 2001, federal
petition timely filed. The Supreme Court then decided Pace v. DiGuglielmo, which held that a post-conviction
petition rejected by the state courts as untimely filed under state law is not considered “properly filed” within
the meaning of AEDPA’s tolling provision.
On remand, the district court held that Pace overruled the Eleventh Circuit’s decision and once again
dismissed Siebert’s federal petition as time-barred. That decision was appealed and the Eleventh Circuit
reversed and remanded on March 7, 2007, finding that Pace did not address the same question at issue in this
case. The case is still pending in district court. http://www.doc.alabama.gov/deathrow.asp.
Kuenzel v. Hopper
William Ernest Kuenzel was convicted of capital murder on September 23, 1998, and was sentenced to death.
The conviction and sentence were affirmed on appeal and the U.S. Supreme Court denied Kuenzel’s
application for a writ of certiorari on October 7, 1991. Kuenzel filed a petition for post-conviction relief in
state court which was dismissed as untimely on February 18, 1999. Kuenzel appealed that dismissal and the
Alabama Court of Criminal Appeals affirmed the decision on January 28, 2000. While his post-conviction
appeal was still pending in state court, Kuenzel filed a federal petition for a writ of habeas corpus in the
NDAL on February 7, 2000, raising 52 claims of error. Kuenzel was represented in federal court by retained
attorneys.
When Kuenzel filed his federal petition, he also filed a motion asking that AEDPA’s statute of limitations be
tolled until all his state court appeals were exhausted. The district court denied the motion as premature, and
stayed the case until state court proceedings had finished. The Alabama Supreme Court denied certiorari on
July 28, 2000, and the U.S. Supreme Court denied certiorari on January 16, 2001. On February 5, 2001, the
district court lifted the stay and Kuenzel filed an amended petition.
The district court then dismissed Kuenzel’s petition as time-barred, finding that the one-year filing deadline
had ended no later than April 23, 1997, and that Kuenzel’s state post-conviction petition was not “properly
filed” and therefore did not toll the running of the time. The court deferred to the state court’s finding that the
post-conviction petition was untimely. Kuenzel argued that the AEDPA statute of limitations should be
equitably tolled on several grounds. He argued that (1) he did not have counsel from October 1991 to
September 1992, and was unaware of state filing deadlines, (2) the law library he had access to was
inadequate, and (3) that he was merely trying to exhaust his state remedies before filing his habeas petition.
The district court rejected these arguments and also rejected Kuenzel’s assertion that the state court’s ruling
regarding his post-conviction petition was ambiguous and based on a rule not firmly established nor regularly
followed.
Kuenzel appealed and the Eleventh Circuit vacated the district court’s decision and remanded for further
proceedings consistent with Siebert (discussed above). The district court on remand found that Pace was an
intervening controlling authority that excused the district court from following the Eleventh Circuit’s
mandate. The district court held that because the state court had concluded that the post-conviction petition
was untimely, the state proceedings did not toll the AEDPA statute of limitations and Kuenzel’s federal
habeas petition was time-barred. On June 13, 2007, the Eleventh Circuit again reversed and remanded. 488
F.3d 1341. http://www.doc.alabama.gov/deathrow.asp.

Final Technical Report: Habeas Litigation in U.S. District Courts

111

FLORIDA
Downs v. Crosby
Ernest Charles Downs was convicted of capital murder and conspiracy and was sentenced to death, after
being granted a new sentencing hearing, on February 17, 1989. Because Downs was convicted prior to the
enactment of AEDPA, the limitations period did not start running until April 24, 1996. Downs had postconviction proceedings pending in state court until October 18, 1999, which tolled the limitations period.
Downs then filed a state habeas petition in the Florida Supreme Court on October 18, 2000, one day too late
to toll the federal limitations period which had expired on October 17, 2000. After the Florida Supreme Court
concluded its review and denied rehearing on December 3, 2001, Downs filed his federal petition for a writ of
habeas corpus in the MDFL (Jacksonville) on December 12, 2001. The petition raised 29 claims of error and
was prepared by his retained attorneys. Downs subsequently asked the court to dismiss his counsel, which the
court did, and Downs represented himself for the remainder of the case.
Downs argued that the limitations period should be equitably tolled because his counsel was ineffective and
incompetent and because he is innocent of murder and merely guilty of conspiracy. The district court rejected
Downs’s arguments and found that Downs had failed to introduce any newly discovered evidence of
innocence. After rejecting all of Downs’s arguments in favor of equitable tolling, the district court dismissed
his federal habeas petition as untimely. Downs appealed the district court’s decision and was appointed
counsel by the Eleventh Circuit. The appeal is still pending. Florida corrections webpage
OHIO
Keenan v. Bagley
Thomas M. Keenan’s direct appeal of his conviction concluded on October 5, 1998. On March 26, 1999, he
filed a post-conviction petition in Ohio state court, which was denied on the merits on December 10, 1999.
Keenan appealed that decision and the Eighth District Court of Appeals of Ohio dismissed the post-conviction
petition as untimely because it was filed outside the 180-day statutory filing limit. Keenan filed his petition
for a writ of habeas corpus in the NDOH on November 21, 2001. Represented in federal court by appointed
attorneys, Keenan raised 76 claims of error.
The district court found that Keenan’s petition was time-barred. The court rejected his claim that his actual
innocence warranted equitable tolling, deferred to the Ohio court’s determination that Keenan’s state postconviction petition was not timely filed, and concluded that the deadline for filing a petition in the district
court had expired October 5, 1999.
The Sixth Circuit vacated the decision and remanded the case to the district court for an evidentiary hearing,
finding that, based on the record, it was unable to determine whether Keenan’s petition should be equitably
tolled. An evidentiary hearing was held and the district court decided to equitably toll the statute of limitations
and reinstate Keenan’s petition. The case is still pending in the district court. Ohio department of corrections
webpage.
Raglin v. Mitchell
Walter Raglin was convicted of capital murder and sentenced to death. Raglin appealed his conviction and
sentence, and the U.S. Supreme Court denied certiorari on March 1, 1999. Raglin’s state post-conviction
petition was denied in state court. This ruling was affirmed, and the Ohio Supreme Court dismissed Raglin’s
appeal on October 27, 1999. Raglin filed a petition for a writ of habeas corpus in the SDOH on September

Final Technical Report: Habeas Litigation in U.S. District Courts

112

13, 2000, raising 33 claims of error. Raglin later asked that the representation of his appointed attorneys be
withdrawn, and the court granted his request.
An amended petition was filed in federal court on August 3, 2002, with the help of a new attorney, and the
case was stayed pending exhaustion of state court remedies from August 4, 2003, until March 3, 2005. The
amended petition voluntarily withdrew 12 of the original claims and added three new claims.
The district court found that Raglin’s three new claims did not relate back to the filing of the original petition
because they rely on different facts from those pleaded in support of his original claims. The court also
rejected Raglin’s argument that the evidence supporting his new claims was newly discovered and was not
known at the time of his original petition. The court dismissed the new claims as time-barred. Each of
Raglin’s remaining claims were dismissed as unexhausted, procedurally defaulted, or without merit. The
district court’s orders have not yet been appealed. Raglin is still on death row. Ohio department of corrections
webpage.
TEXAS
Nelson v. Dretke
Marlin Nelson's judgment for capital murder became final on October 4, 1993. The statute of limitations was
tolled by agreement of the Texas attorney general's office until the appointment of state habeas counsel, on
January 19, 1998. The statute was tolled again, 267 days later, when Nelson filed his state habeas petition.
After the state court denied relief, he had 98 days remaining, until December 19, 2002, to file his federal
petition. On September 17, 2002, Nelson moved in the SDTX for appointment of counsel. The district court
appointed counsel on March 13, 2003, three months after the expiration of the limitations period. On April
30, 2003, Nelson sought an extension and equitable tolling until June 13, 2003. While that motion was
pending, he filed his petition on August 22, 2003, 246 days after the limitations period had expired. The
petition contained 8 claims, including insufficient evidence for conviction and various errors in the sentencing
phase. The district court held that the petition was time-barred and denied equitable tolling for the period
during which the motion to appoint counsel was pending, noting precedent “upholding the expectation that a
death-row inmate, while entitled to counsel, should act independently to preserve his ability to seek redress in
federal court” by filing a skeletal petition. The court also denied tolling for the period after counsel was
appointed, noting that “if Nelson had not spent 258 days without actively seeking state relief, he would have
enjoyed more time to prepare a federal petition.” The Fifth Circuit in an unpublished opinion denied a
certificate of appealability, finding that jurists of reason would not find the district court's ruling to be
debatable. Nelson remains on death row.http://www.tdcj.state.tx.us/statistics/deathrow/drowlist/nelsonme.jpg
Prieto v. Dretke
Arnold Prieto was convicted of capital murder and sentenced to death. His conviction was affirmed on direct
appeal on December 16, 1998. He filed a petition for post-conviction relief, which was denied by the Texas
Court of Criminal Appeals on November 28, 2001. Prieto then filed a petition for a writ of habeas corpus in
the WDTX on August 2, 2002. Represented by appointed counsel, he filed an amended petition on
September 24, 2002, raising eight claims of error.
The district court found that Prieto’s conviction became final no later than March 17, 1999, and was tolled
while his state petition for post-conviction relief was pending, through November 28, 2001. Despite this
tolling, Prieto’s federal petition was filed 98 days after the AEDPA statute of limitations ended. The district
court refused to equitably toll the limitations period, holding that attorney neglect ordinarily does not justify
doing so. The court reasoned that there is no evidence that Prieto’s state habeas attorney was incapacitated by
her illness. Even more importantly, the alleged neglect of Prieto’s state habeas counsel did not last for the

Final Technical Report: Habeas Litigation in U.S. District Courts

113

entire duration of the limitations period. The district court also found that Prieto did not make any effort to
exercise diligence to pursue state or federal relief, could not show that he was misled by any
misrepresentation by a state agency, state court, or federal court, and did not allege that he was not aware of
the limitations period or otherwise precluded from filing a pro se petition on time.
Despite its finding that the petition was time-barred, the district court went on to address Prieto’s claims and
deny them all on the merits or for procedural reasons other than the statute of limitations. Prieto appealed and
the Fifth Circuit reversed the district court’s findings that the petition was untimely and that his jury
misconduct claim was procedurally barred. On remand, the district court held that the claim was in fact
procedurally barred and in the alternative was meritless and Teague-barred. The district court’s decision was
appealed and that appeal is still pending in the Fifth Circuit. Texas department of criminal justice webpage.
Rojas v. Cockrell
Leonard Uresti Rojas was convicted of capital murder and sentenced to death. His conviction became final
on February 9, 1999. On April 5, 2000, Rojas filed a motion for appointment of counsel in the NDTX. On
May 3, 2000, the court stayed his execution and on November 6, 2000, the court appointed him counsel.
Appointed counsel filed Rojas’s federal petition for a writ of habeas corpus on March 23, 2001, alleging as
claims of error 1) a due process violation when the court failed to define a necessary element of the offense of
capital murder in the jury charge, 2) insufficient evidence to sustain the “same criminal transaction” element
of capital murder, 3) insufficient evidence of future dangerousness, and 4) that the application of AEDPA’s
limitations period violates the Suspension Clause.
Rojas raised no arguments for equitable or statutory tolling and the district court dismissed the petition as
time-barred. Rojas appealed and the Fifth Circuit declined to issue a Certificate of Appealability. Rojas was
executed on December 4, 2002. http://www.tdcj.state.tx.us/statistics/deathrow/drowlist/rojas.jpg
Shannon v. Dretke
Willie Marcel Shannon was convicted of capital murder and sentenced to death on November 8, 1993. The
conviction and sentence were affirmed on December 11, 1996, and a motion for rehearing was denied on
January 29, 1997. Shannon did not petition the U.S. Supreme Court for a writ of certiorari. Shannon filed a
state post-conviction petition on April 13, 1998, which was denied on September 12, 2001. Shannon then
filed his federal petition for writ of habeas corpus in the SDTX (Houston) on March 1, 2002. The petition was
prepared by appointed counsel and raised 9 claims of error. The federal habeas proceedings were stayed from
August 27, 2002, until March 8, 2005, pending the disposition of Shannon’s motion for DNA testing in state
court. That motion was denied by the Texas state courts.
The court found that Shannon’s conviction became final on April 29, 1997, and was tolled from April 13,
1998, until September 12, 2001, while Shannon’s state post-conviction petition was being litigated. When the
state post-conviction petition was denied on September 12, 2001, Shannon had 17 more days to file a federal
petition. Shannon, however, did not file his petition until 151 days after the limitations period had expired.
Shannon argued that Pyles, which held that the statute of limitations for federal habeas petitions must be
tolled from the date a death row inmate requests state habeas counsel or the state court enters the statutorily
required findings for such appointment until the date counsel is actually appointed, should be applied in this
case. The district court found that even if that rule does apply and Shannon’s limitations period was tolled, his
petition was still untimely.
Shannon also argued that he is entitled to equitable tolling because his state habeas counsel delayed in seeking
appointment of federal habeas counsel and because time lapsed between his motion for appointment of
counsel and the date federal habeas counsel was appointed. The court rejected this argument, finding that

Final Technical Report: Habeas Litigation in U.S. District Courts

114

Shannon still had 40 days to file a petition once his federal habeas counsel was appointed (assuming that
Pyles applies). The district court dismissed Shannon’s petition as time-barred and denied a Certificate of
Appealability. The Fifth Circuit affirmed. Shannon was executed on November 8, 2006.
http://www.tdcj.state.tx.us/statistics/deathrow/drowlist/shannon.jpg
Pippin v. Dretke
Roy Lee Pippin was convicted of capital murder and sentenced to death. His conviction and sentence were
affirmed on direct appeal on May 21, 1997. On May 18, 1998, Pippin filed a state habeas corpus petition. He
then filed a second petition and a supplemental memorandum of law, raising new claims, which were treated
as successive petitions by the state court. On February 20, 2002, the Texas Court of Criminal Appeals denied
Pippin’s first habeas petition on the merits and dismissed the others as abuses of the writ. On October 7, 2002,
the U.S. Supreme Court denied certiorari. On June 20, 2002, Pippin filed a pro se federal petition for habeas
corpus in the SDTX (Houston). The next day, June 21, 2002, Pippin’s appointed counsel filed a federal
habeas petition on behalf of Pippin raising 18 claims of error.
On September 30, 2004, the district court held one of Pippin’s Brady claims (and a related ineffective
assistance of counsel claim) in abeyance pending further briefing, dismissed a prosecutorial misconduct claim
(presentation of perjured testimony) and several ineffective assistance claims as time-barred, and denied the
rest of the claims on the merits or found they were procedurally barred. The two claims that were dismissed as
time-barred were both raised for the first time in Pippin’s April 16, 2004, supplemental brief. The district
court did not discuss the relation back doctrine but instead stated that these two claims were clearly outside
the limitations period.
Pippin appealed the district court’s order but the Fifth Circuit held that it lacked jurisdiction to hear the appeal
because the district court’s order was not a final judgment. After additional briefing, the district court denied
Pippin’s remaining two claims on the merits. Pippin appealed and the Fifth Circuit denied a Certificate of
Appealability. The U.S. Supreme Court denied certiorari on October 16, 2006. Pippin was executed on
March 29, 2007. http://www.tdcj.state.tx.us/statistics/deathrow/drowlist/pippin.jpg

Final Technical Report: Habeas Litigation in U.S. District Courts

115

APPENDIX C. WRITS GRANTED
NON-CAPITAL CASES WITH AT LEAST ONE CLAIM GRANTED
Vivar v. Senkowski. Wilson Vivar was convicted of criminal possession of a weapon in the second degree
and assault in the second degree, then sentenced to concurrent sentences of 15 and five years. Vivar filed a
pro se petition for writ of habeas corpus in the EDNY (Brooklyn) challenging the sufficiency of the evidence
of both his convictions, as well as his 15-year sentence for the weapon possession. The court granted relief on
the insufficiency claim for the weapons conviction, finding that the prosecution failed to establish that the
firearm was a shotgun, which was a necessary element of the crime. The court also granted relief based on
the trial court’s failure to instruct the jury on the definition of a firearm, even though Vivar did not raise
this separate claim but merely discussed it somewhere in his petition. The insufficiency of evidence claim
related to the assault conviction was denied on the merits. The sentencing claim related to weapons conviction
was held to be moot since the court granted relief from the weapons conviction. The state did not appeal the
district court’s judgment. There is no other post-judgment information available, other than the fact that Vivar
was paroled on December 21, 2004. New York Department of Corrections Inmate Search.
Mack v. Folino et al. William Mack was convicted of third-degree murder, criminal conspiracy, and
possession of an instrument of crime. He was sentenced to concurrent terms of 24 to 40 years for murder, ten
to 20 years for conspiracy, and two years six months to five years for weapon possession. Through his
retained attorneys, he filed his petition in the EDPA (Philadelphia). The petition challenged Mack’s
conviction and raised three claims of error. The court denied two of the claims on the merits and granted relief
on the remaining claim, holding that there was a complete denial of cross-examination at a post-trial
hearing on Mack’s motion for extraordinary relief of a witness who claimed that someone other than
Mack is responsible for the murder in question. The district court held this error was not harmless and that
the only form of appropriate relief was a new hearing on the question of after-discovered evidence. The
district court ordered that the state grant Mack a new hearing, including a meaningful opportunity to crossexamine witnesses and present testimony. The district court’s decision was affirmed by the Third Circuit.
There is no further information regarding whether the state held a new hearing. Pennsylvania Department of
Corrections Inmate Search
Lockridge v. Adams. Bryan Lockridge was convicted of attempted murder of a police officer, possession of
cocaine for sale, and possession of a firearm by a felon. He was sentenced to life imprisonment. On appeal,
the California Court of Appeals affirmed his conviction but modified the sentence and stayed imposition of
the firearm conviction. Lockridge filed a pro se petition for a writ of habeas corpus in the CDCA (Los
Angeles) challenging his conviction on the ground that the trial court violated his Sixth Amendment right to
trial by jury and his Fourteenth Amendment due process rights when it refused his request to give a selfdefense instruction. The district judge agreed, and held that the trial court’s failure to instruct the jury was
not harmless. The state was ordered to provide Lockridge with a new trial or else release him. This decision
was affirmed by the Ninth Circuit.
Musgrave v. Spalding. Shelton Musgrave was convicted of first-degree murder with a deadly weapon
enhancement and was sentenced to 22 years imprisonment. On direct appeal the Washington State Court of
Appeals held that the trial court had erred in excluding the defense investigator’s testimony and photographs
but that the error did not require reversal. Musgrave filed, with retained counsel, a petition for a writ of habeas
corpus in the WDWA (Seattle) challenging his conviction. The petition raised three claims of error, two of
which were denied on the merits. The court granted the third claim which was that the exclusion of the
testimony and photographs of the investigator denied Musgrave the ability to effectively challenge the
credibility of the crucial witness to the crime. The witness’s testimony was the “lynchpin of both the State’s
case against the Petitioner and the Petitioner’s defense” and Musgrave’s inability to offer evidence related to
the credibility of that witness had a prejudicial effect on the jury’s verdict. The district court ordered that

Final Technical Report: Habeas Litigation in U.S. District Courts

116

Musgrave be released or retried. The district court’s decision was appealed but the parties agreed to jointly
dismiss the appeal.
Kang v. Green et al. Catherine C. Kang was charged with vehicular manslaughter. At her jury trial a defense
witness testified that the decedent in this case had not been wearing a seatbelt, evidence which is not
admissible under Ohio state law. The prosecution objected and the judge declared a mistrial. The case was reassigned and a new trial date was set. The defense filed a motion to dismiss based on double jeopardy
grounds, which was denied. A continuance was granted and the case was stayed pending the outcome of
Kang’s filing of a petition for a writ of habeas corpus in the SDOH (Eastern Division). Kang’s petition, filed
with the assistance of retained counsel, raised two claims: 1) that the trial judge abused his discretion in
declaring a mistrial in violation of double jeopardy, and 2) the exclusion of the evidence violated Kang’s
constitutional rights. The court dismissed Kang’s evidentiary challenge as not cognizable in habeas corpus
because the claim raised a state law evidentiary issue. The district court then granted Kang’s double jeopardy
claim, finding there was no manifest necessity for declaring a mistrial. The court reasoned that the jury had
already heard evidence from which they could have concluded that the decedent was not wearing a seatbelt,
and a curative instruction could have solved the problem. Also at issue in this case was whether Kang was “in
custody.” The court held that Kang was in custody, even though she was released on her own recognizance,
because she was required to appear at a future trial date and a failure to appear would result in her being held
in contempt of court and in a warrant being issued for her arrest. The court ordered that respondent be barred
from re-trying Kang on the vehicular manslaughter charge. The decision was not appealed.
Deltessandro v. Knowles. Steven Deltessandro was convicted of robbery, assault with force likely to produce
great bodily injury, and aggravated mayhem. He was sentenced to five years’ imprisonment for the robbery
and life without parole for the aggravated mayhem. After his state appeal failed, Deltessandro filed a pro se
petition for a writ of habeas corpus in the NDCA (San Francisco) raising seven claims of error. The district
court granted Deltressandro’s claim that his confrontation clause rights were denied when the trial court
excluded impeaching evidence of witness Easley’s prior criminal activity. The court found that the error
was harmless with regard to the assault and robbery convictions but not with regard to the aggravated
mayhem conviction since Easley’s testimony was the only evidence that established the specific intent
necessary for an aggravated mayhem conviction. Of the six remaining claims, one was procedurally barred
and the rest were denied on the merits. The court vacated Deltessandro’s conviction for aggravated mayhem
and ordered that the state retry him or re-sentence him within 90 days. The court’s decision was not appealed.
It is unknown whether Deltessandro was retried or re-sentenced.
NON-CAPITAL GRANTS REVERSED
Brewster v. Alameda. Thomas Brewster pleaded guilty to possession of a controlled substance and received
six years’ imprisonment. He filed a pro se petition for writ of habeas corpus in the EDCA (Sacramento)
challenging his conviction. The district judge dismissed the search and seizure claim based on Stone v.
Powell, but granted Brewster’s claim that counsel was ineffective for failing to raise the meritorious
Fourth Amendment claim on appeal. The district court found that a retrial would be pointless since the
illegally seized evidence would be suppressed and there would be no evidence justifying a conviction. The
district court ordered that Brewster be released. The district court’s decision was reversed by the Ninth
Circuit. Its opinion was not available to determine the reason for reversal.

Final Technical Report: Habeas Litigation in U.S. District Courts

117

CAPITAL CASES WITH AT LEAST ONE CLAIM GRANTED , summarized, by state 165
ALABAMA
Lawhorn v. Haley. James Lawhorn was convicted of capital murder and sentenced to death. He filed his
petition in the NDAL (Eastern) raising 44 claims and sub-claims of error. The court granted relief on
Lawhorn’s claim that he was subjected to unconstitutional delay in securing a judicial determination of
probable cause for his warrantless arrest in violation of the Fourth Amendment. The district court found
that the proper remedy for this error is to suppress the confession that Lawhorn gave as a result of that delay.
The district court ordered that Lawhorn’s conviction be vacated. The district court also found that Lawhorn’s
trial counsel was ineffective for failing to make a closing argument at the penalty phase and the proper
remedy for this was to vacate Lawhorn’s death sentence. Respondent appealed the district court’s judgment,
the result in the Eleventh Circuit is unknown. Lawhorn remains on death row. Offender Information for James
Lawhorn
ARIZONA
Amaya-Ruiz v. Cluff. Jose Amaya-Ruiz was convicted of first-degree murder, manslaughter, theft, and firstdegree burglary. He was sentenced to death. His manslaughter conviction was reversed by the Supreme Court
of Arizona. Amaya-Ruiz then filed a petition for writ of habeas corpus in the DAZ (Tucson), which was
denied. The Ninth Circuit affirmed. The state of Arizona then filed a motion for a competency examination
because there was reason to believe that Amaya-Ruiz was incompetent for execution. The superior court
found that Amaya-Ruiz was incompetent. After several treatment sessions his competency was deemed
restored and a warrant of execution was issued. Amaya-Ruiz then filed another petition for a writ of habeas
corpus in the same district court challenging the state’s finding that competency was restored. After an
evidentiary hearing, the district judge granted Amaya-Ruiz’s claim of incompetency for execution and
ordered he could not be executed until his competency was restored and the determination of that restoration
was made using constitutionally adequate procedures. Amaya-Ruiz appealed to the Ninth Circuit but later
voluntarily dismissed his appeal. He is currently serving a life sentence. Offender Information for Jose
Amaya-Ruiz
Laird v. Schriro. Kenneth Laird was convicted of capital murder and sentenced to death for his involvement
in a robbery, kidnapping, and murder. He filed a petition raising 24 claims in the DAZ (Phoenix). Several
claims were determined to be procedurally barred and several others were denied on the merits. The court
granted relief on Laird’s claim that his death sentence violated the Eighth Amendment, as determined in
Roper v. Simmons, because he was under the age of 18 at the time the offense was committed. Laird
appealed the rejection of his other claims and the Ninth Circuit affirmed the district court’s decision.
Offender Information for Kenneth Laird
OHIO
Mardrigal v. Bagley. Jamie Madrigal was convicted of aggravated murder for his alleged involvement in the
robbery of a Kentucky Fried Chicken restaurant. His petition filed in NDOH (Cleveland) challenged his
conviction and sentence. The district court granted Madrigal’s claim that the trial court violated the
confrontation clause by admitting the out-of-court statements of a co-defendant. The Ohio Supreme
Court had earlier found that this error was harmless. The district court disagreed and held that the error was
not harmless since the main issue at trial was the identity of the killer and the co-defendant’s statement denied
culpability and implicated Madrigal. The rest of Madrigal’s claims were denied. The district court ordered
that Madrigal be released from custody or retried. Both parties appealed the district court’s judgment and the
165

Court of appeals status reported as of March 2007 unless otherwise noted.

Final Technical Report: Habeas Litigation in U.S. District Courts

118

Sixth Circuit affirmed. He is now serving a sentence of 120 years to life imprisonment. Offender Information
for Jamie Madrigal

Taylor v. Bagley. Michael Taylor was convicted of aggravated murder and sentenced to death. His petition
in the NDOH (Cleveland) challenging his conviction and sentence included a claim that the evidence was
insufficient to support his conviction for aggravated murder. The district court granted the writ based on
this claim, and found that there was not enough evidence to prove that there was prior calculation and design,
which is a necessary element of aggravated murder. The rest of Taylor’s claims were denied. Both parties
appealed. Taylor died on January 27, 2004 and the appeal was dismissed.
PENNSYLVANIA
Holloway v. Horn. Arnold Holloway was convicted of first-degree murder and sentenced to death. He filed a
petition for a writ of habeas corpus in the EDPA (Philadelphia) challenging his conviction and sentence. The
court denied relief on all of Holloway’s guilt-phase claims, including a Batson claim, which was found to be
procedurally barred and meritless. The district court then granted relief on Holloway’s claim that his penaltyphase counsel was ineffective for failing to investigate mental health issues and present a mental health
expert. The district court ordered that a writ of habeas corpus be issued unless Holloway was re-sentenced.
Both parties appealed and the Third Circuit reversed, finding that Holloway’s Batson claim was not
procedurally barred and should be granted. On remand the district court ordered that Holloway be retried
within 120 days. There is no information about Holloway on the Pennsylvania Department of Corrections
website.
Rivers v. Horn. Delores Rivers was convicted of first-degree murder and related charges and was sentenced
to death. She filed a petition for writ of habeas corpus in the EDPA (Philadelphia) raising 58 claims of error
challenging conviction and sentence. The parties entered into a stipulation by which habeas proceedings
would be terminated in exchange for Rivers being sentenced to life imprisonment. As part of that
stipulation, the district court granted Rivers’s claim that trial counsel was ineffective for failing to
investigate and present mitigating evidence during the penalty phase and denied the rest of the claims.
Rivers agreed to waive her right to any further appeals as part of the stipulation. Pennsylvania Department of
Corrections Offender Search
TEXAS
Arroyo v. Dretke. Randy Arroyo was convicted of capital murder and sentenced to death. He filed a petition
for a writ of habeas corpus in the WDTX (San Antonio) challenging his conviction and sentence, raising 12
claims of error, including a few sub-claims. Relief was granted on Arroyo’s claim that his death sentence
violated the Eighth Amendment, as determined in Roper v. Simmons, because he was under the age of 18 at
the time of the murder. The state commuted Arroyo’s sentence to life imprisonment. Arroyo appealed the
district court’s decision and the appeal is still pending in the Fifth Circuit. Texas Offender Search
Brewer v. Dretke. Brent Brewer was convicted of capital murder and sentenced to death. He filed a petition
NDTX (Amarillo) challenging his sentence. The court granted relief on Brewer’s claim that the special issues
given to the jury at Brewer’s penalty phase did not allow the jury to give effect to relevant mitigating
evidence. The rest of Brewer’s claims were denied. The district court ordered that Brewer’s sentence either be
commuted to life or he be given a new sentencing hearing. Respondent appealed and the Fifth Circuit
reversed. The U.S. Supreme Court granted certiorari and reversed on April 25, 2007 (Brewer v.
Quarterman). Texas Offender Search

Final Technical Report: Habeas Litigation in U.S. District Courts

119

Colella v. Cockrell. Paul Colella was convicted of capital murder for two killings and sentenced to death. He
filed a petition in the SDTX (Brownsville) challenging his conviction and sentence. The court granted
respondent’s Motion for Summary Judgment as to several of Colella’s claims. An evidentiary hearing was
held and the parties subsequently agreed to settle the case. Respondent conceded that Colella’s trial
counsel was ineffective for failing to investigate and prepare the case for trial. The parties agreed that
Colella would plead guilty and as the result of a plea bargain, he would receive a 20-year sentence with
credit for time served. Texas Offender Search
Draughon v. Dretke. Martin Draughon was convicted of capital murder and sentenced to death. He filed a
petition for a writ of habeas corpus in the SDTX (Houston) challenging his conviction and sentence, raising
31 claims. After an evidentiary hearing, the district judge granted relief on Draughon’s claim that his trial
counsel was ineffective for failing to retain a ballistics expert at both the guilt and punishment phases of
his trial. The district court found that a ballistics expert would have testified that the fatal bullet ricocheted
off something else before striking the victim, which supports Draughon’s assertion that the killing was
unintentional and he therefore lacked the requisite intent for capital murder. The rest of Draughon’s claims
were denied. The district court ordered that Draughon be released from custody unless he was granted a new
trial or his capital murder conviction vacated and a new sentencing hearing granted. Both parties appealed to
the Fifth Circuit, which affirmed the district court’s judgment. 427 F. 3d 286. Draughon is listed on the Texas
Department of Corrections Offender Search website as Draughton, his conviction reduced to murder, he is
serving to 40 years.
Goynes v. Dretke. Theodore Goynes was convicted of capital murder and sentenced to death. He filed a
petition for a writ of habeas corpus in the SDTX (Houston) challenging his conviction and sentence and
raising nine claims of error, five of which were denied on the merits and three of which were procedurally
defaulted. The court granted relief on the remaining claim: that during the punishment phase the special
issues instruction did not allow the jury to consider and give effect to Goynes’s mitigating evidence
regarding mental illness and his low IQ, thus violating due process. The district court’s decision was
appealed. Goynes moved for permission to file a successive petition, but the Fifth Circuit denied his motion
as moot because of his grant of relief from his sentence. The Fifth Circuit then denied a Certificate of
Appealability on the claims that Goynes appealed. The re-sentencing was supposed to occur sometime in the
fall of 2006. According to the Texas Department of Corrections website, Goynes is still on death row. Texas
Offender Search
Guidry v. Dretke. Howard Guidry was sentenced to death for a murder-for-hire. Guidry filed a petition for a
writ of habeas corpus in the SDTX (Houston) challenging his conviction and sentence, raising four claims of
error, two of which were granted. The district court first found that Guidry’s Fifth Amendment rights were
violated when he was questioned by police, and subsequently confessed, without having an attorney present.
The district court found that Guidry requested to have counsel present and the police tricked him by
saying that they had spoken to his attorney and his attorney had given them permission to question
him. The court found that this error was not harmless. Relief was also granted on Guidry’s claim that the
admission of hearsay testimony into evidence violated the Confrontation Clause. The Texas Court of
Criminal Appeals found this error was harmless, but the district court disagreed. Guidry’s other two claims
were denied on the merits. The district court ordered that Guidry be released from custody unless a new trial
was conducted within 180 days. Respondent appealed and the Fifth Circuit affirmed. According to the Texas
Department of Corrections website, Guidry is still on death row. Texas Offender Search
Johnson v. Dretke. Eddie Johnson filed a petition for a writ of habeas corpus in the NDTX (Fort Worth)
challenging his capital murder conviction and death sentence. The court granted relief on Johnson’s claim that
he was under the age of 18 at the time of his offense, making his death sentence unconstitutional under
Roper v. Simmons. The rest of Johnson’s claims were denied on the merits. The district court ordered that
Johnson’s sentence be vacated and he be either re-sentenced within 180 days or released from custody. The

Final Technical Report: Habeas Litigation in U.S. District Courts

120

district court’s judgment was not appealed. Johnson was re-sentenced to life imprisonment. Texas Offender
Search
Little v. Dretke. Leo Little was convicted of capital murder and sentenced to death for a killing in the course
of committing and attempting to commit robbery and kidnapping. Little filed a petition for a writ of habeas
corpus in the WDTX (San Antonio), challenging his death sentence. Little’s petition raised three claims of
error. A fourth claim of error was added after the U.S. Supreme Court decided Roper v. Simmons. The
respondent conceded that Little was under the age of 18 at the time of the offense and that his death
sentence therefore violated the Eighth Amendment under Roper v. Simmons. Respondent also waived
the exhaustion defense with regard to this claim. The district judge ordered that Little be granted a new
punishment trial, or that his death sentence be vacated and a new sentence less than death be imposed. The
court did not address the merits of the rest of Little’s claims. The district court’s decision was not appealed.
According to the Texas Department of Corrections website, Little is now serving a life sentence. Texas
Offender Search
Saldano v. Texas Department of Criminal Justice-Institutional Division. Victor Saldano was convicted
of capital murder for a kidnapping and murder, then sentenced to death. On direct appeal Saldano complained
that his Eighth Amendment rights were violated at the punishment phase when the jury considered his race as
evidence to support its determination of future dangerousness. The Solicitor General of the State of Texas
conceded this error and the United States Supreme Court vacated Saldano’s sentence and remanded the case
to the Court of Criminal Appeals of Texas. On remand, the Texas Court of Criminal Appeals reinstated
Saldano’s conviction and death sentence. Saldano filed a petition for writ of habeas corpus in the EDTX
(Beaumont) challenging his sentence based on claims that included an ineffective assistance of counsel claim.
Respondent conceded this error and agreed that Saldano should receive a new sentencing hearing, but the
claim was dropped when Saldano filed an amended petition. The amended petition contained a single claim,
challenging the trial court’s admission of expert testimony stating that Saldano’s Hispanic ethnicity was a
positive factor for predicting future dangerousness. The respondent conceded this error but the amici
curiae, the District Attorney and Prosecuting Attorney, argued that the claim was procedurally defaulted, and
in the alternative, harmless error. The District Attorney also moved to intervene in this case. The district court
denied the motion for intervention and held that the admission of the testimony was constitutional error and
the procedural default and harmless error defenses were waived by the respondent. The district court ordered
that Saldano be released from custody unless a new punishment hearing was held within 180 days or his death
sentence was reduced to life imprisonment. The District Attorney appealed the district court’s decision. The
Fifth Circuit affirmed the district court’s denial of the motion to intervene and then dismissed the appeal
since the District Attorney is not a party to the case. According to the Texas Department of Corrections
website, Saldano is still on death row. Texas Offender Search
Williams v. Dretke. Bruce Williams was convicted of capital murder and sentenced to death. He filed a
petition for a writ of habeas corpus in the NDTX (Dallas) challenging his sentence. The petition raised four
claims of error. Relief was granted on Williams’s claim that his death sentence violated the Eighth
Amendment because he was under the age of 18 at the time of the offense. The district court found that
the rest of Williams’s claims were moot because they challenged his sentence and not his conviction. The
district court ordered that Williams be released unless his sentence was commuted to life imprisonment within
sixty days. The district court’s decision was not appealed and the State of Texas commuted Williams’s
sentence to life imprisonment. Texas Offender Search
Willis v. Dretke. Ernest Willis was convicted of capital murder for causing death in the course of
committing arson on a habitation and sentenced to death. After direct appeal, Willis filed a state petition for a
writ of habeas corpus. The trial judge vacated the conviction and sentence but was reversed by the Texas
Court of Criminal Appeals. Willis filed a petition for a writ of habeas corpus in the WDTX (Pecos)
challenging his conviction and sentence. The court granted Willis’s claims that his due process rights were

Final Technical Report: Habeas Litigation in U.S. District Courts

121

violated by the State’s administration of medically inappropriate antipsychotic drugs without his
consent, that the State suppressed evidence favorable and material to the sentencing determination,
and that Willis received ineffective assistance of counsel at the guilt-innocence phase and the sentencing
phase. All other claims raised in the petition were denied (Willis maintained he was innocent and argued the
state did not present a theory of alleged motive at trial). The district court ordered that the state grant Willis a
new trial by November 18, 2004 or else release him from custody. The State chose not to appeal the decision
or to retry Willis. He was released from custody on October 6, 2004. Texas Offender Search.
Capital Grants, Appeals Still Pending 166
OHIO
Johnson v. Bagley. Rayshawn Johnson was convicted of felony murder and sentenced to death in 1998.
After state appeal and post-conviction proceedings, Johnson filed his petition in 2002 in the SDOH
(Cincinnati), raising 26 claims. The judge ordered an evidentiary hearing on Johnson's claim that he was
denied effective assistance of counsel during the mitigation phase of his trial, and granted relief on this
claim. The state appealed the federal court's order to commute Johnson's death sentence or grant him a new
hearing, and the appeal is pending in the Sixth Circuit. Offender Information for Rayshawn Johnson
Bies v. Bagley. Michael Bies was convicted of capital murder and sentenced to death. The state court found
that Bies was mildly mentally retarded but could still be executed. Bies filed a petition for a writ of habeas
corpus in the SDOH (Cincinnati) challenging his conviction and sentence. Bies’s petition raised 135 claims
and sub-claims of error. Atkins v. Virginia was subsequently decided and respondent tried to argue it should
be able to re-try Bies on the issue of whether he is mentally retarded. Bies’s nineteenth claim for relief, which
argued that the state’s insistence on re-trying his mental retardation claim violates double jeopardy, was
bifurcated from the rest of his claims, some of which were found to be unexhausted. The court then granted
that claim and ordered that Bies’s capital sentence be vacated and that he be re-sentenced. Respondent
appealed and that appeal is still pending in the Sixth Circuit. Offender Information for Michael Bies
D’Ambrosio v. Bagley. Joe D’Ambrosio was convicted of aggravated murder, aggravated felony murder,
kidnapping, and aggravated burglary. He was sentenced to death. He filed a petition for a writ of habeas
corpus in the NDOH (Cleveland) challenging his conviction and sentence. The court granted D’Ambrosio’s
claim that the prosecution failed to disclose exculpatory evidence prior to trial in violation of Brady v.
Maryland. The rest of D’Ambrosio’s claims were denied on the merits or procedurally barred. The district
court ordered that D’Ambrosio’s convictions and sentence be set aside or a new trial be granted. Both parties
appealed and that appeal is still pending in the Sixth Circuit. Offender Information for Joe D'Ambrosio

166

Court of appeals status reported as of March 2007 unless otherwise noted.

Final Technical Report: Habeas Litigation in U.S. District Courts

122

OKLAHOMA
Powell v. Mullins. Paris Powell was convicted of first-degree malice-aforethought murder of one victim and
shooting with intent to kill another. Powell filed a petition for a writ of habeas corpus in the WDOK
(Oklahoma City) challenging his conviction and sentence. The court granted relief, after an evidentiary
hearing, on Powell’s first and second claims alleging that his convictions were obtained using evidence
known by the prosecutor to be false and that exculpatory evidence was suppressed in violation of Brady
v. Maryland. The rest of the claims were found to be moot. Both parties appealed the district court’s judgment
and that appeal is still pending in the Tenth Circuit. Offender Information for Paris Powell
PENNSYLVANIA
Bond v. Beard. Jesse Bond was convicted of first-degree murder, robbery, possessing an instrument of
crime, and conspiracy, and then sentenced to death. Bond filed a petition for a writ of habeas corpus in the
EDPA (Philadelphia), raising 18 claims and sub-claims of error in his petition. All of Bond’s claims
regarding his conviction were denied on the merits, but the court granted relief from Bond’s death sentence
because of his counsel’s failure to investigate and present mitigating evidence at the sentencing phase.
The district court also found that the prosecutor made improper comments during sentencing. The district
court ordered that Bond be re-sentenced within 120 days or else be sentenced to life imprisonment. Both
parties appealed and that appeal is still pending in the Third Circuit. Pennsylvania Department of
Corrections Offender Search
Lewis v. Horn. Reginald Lewis was convicted of capital murder and sentenced to death for a stabbing. He
filed his petition in the EDPA (Philadelphia) challenging his conviction and sentence. Relief was granted on
Lewis’s claim that his trial counsel was ineffective for failing to present mitigating evidence during the
penalty phase regarding his mental illness and his traumatic upbringing. The rest of Lewis’s sentencing
claims were not addressed since relief was granted on this ground. His conviction-related claims were all
denied. The district court ordered that Lewis be given a new sentencing hearing or be sentenced to life
imprisonment. Both parties appealed the district court’s judgment and that appeal is still pending in the Third
Circuit. Pennsylvania Department of Corrections Offender Search
Rollins v. Horn. Saharris Rollins was convicted of first-degree murder, robbery, and possession of an
instrument of crime, then sentenced to death. His petition, filed in the EDPA (Philadelphia), was granted
based on Rollins’s claims that he was denied effective assistance of counsel when his trial attorney failed
to adequately prepare for the penalty phase and failed to investigate mitigating evidence regarding
Rollins’s childhood, as well as his claim that the judge incorrectly instructed the jury that it must
unanimously agree on mitigating circumstances before giving effect to those mitigating factors at
sentencing. The district court ordered that Rollins be sentenced to life imprisonment if the state does not
conduct a new sentencing hearing within 180 days. Respondent appealed and that appeal is still pending.
Pennsylvania Department of Corrections Offender Search
Thomas v. Horn. Brian Thomas was convicted of first-degree murder, burglary, involuntary deviate sexual
intercourse, and rape. He received a death sentence. Thomas filed a petition for a writ of habeas corpus in the
EDPA (Philadelphia) challenging his conviction and sentence. The court denied Thomas’s claims
challenging his conviction but granted relief from his death sentence due to ineffective assistance of counsel
at sentencing. The district court ordered that Thomas be granted a new sentencing hearing or have his
sentence commuted to life without parole. Both parties appealed and that appeal is still pending in the Third
Circuit. Pennsylvania Department of Corrections Offender Search

Final Technical Report: Habeas Litigation in U.S. District Courts

123

TEXAS
Adams v. Dretke. John Adams was convicted of capital murder and sentenced to death for killing a woman
during the robbery of her home. His petition filed in the NDTX (Dallas) challenged his conviction and
sentence, raising six claims of error. The court granted Adams’s claim that his counsel was ineffective in the
investigation and presentation of mitigating evidence at the penalty phase of trial. The rest of Adams’s
claims were denied on the merits. The district court ordered that a new punishment trial be granted.
Respondent appealed and that appeal is still pending in the Fifth Circuit. Adams remains on death row.
Texas Offender Search
Williams v. Dretke. Nanon Williams was convicted of capital murder and sentenced to death. He filed a
petition for a writ of habeas corpus in the SDTX (Houston) challenging his conviction and sentence. The
court granted Williams’s claim that his death sentence violates the Eighth Amendment, as determined in
Roper v. Simmons, because he was under the age of 18 at the time of the offense. The rest of Williams’s
claims were denied. The district court ordered that Williams be released from custody unless his sentence is
commuted to life imprisonment within 180 days. Both parties appealed and that appeal is still pending in the
Fifth Circuit. Williams is currently serving a life sentence. Texas Offender Search
Capital Grants Reversed 167
ALABAMA
Callahan v. Haley. James Callahan was convicted of capital murder and sentenced to death. His conviction
was reversed on direct appeal and a new trial was conducted. He was convicted again and sentenced to death
a second time. That conviction was affirmed on appeal. He filed a petition for a writ of habeas corpus in the
NDAL (Eastern) challenging his conviction and sentence. The petition raised 24 claims of error. The district
judge disagreed with the magistrate judge's recommendation that the petition be denied, and granted relief on
two of Callahan’s claims. The district court held that Callahan’s Sixth Amendment right to a fair and
impartial tribunal was violated when the state judge was present during custodial interrogation, where
Callahan made a confession without the presence of an attorney. The district court also held that
Callahan’s counsel was ineffective at the sentencing stage for failing to present various types of
mitigating evidence. The district court did not address the rest of Callahan’s claims. The district court
ordered that Callahan be released within 90 days unless a new trial is scheduled. A Certificate of
Appealability was granted with regards to eight of Callahan’s claims despite the fact that the district court had
addressed two of those claims. The Eleventh Circuit vacated the district court’s judgment without prejudice
and remanded the case to the district court for it to address the six appealed claims it had failed to address
earlier. On remand, those six claims were denied. The Eleventh Circuit reversed the grant on appeal and
denied relief. Callahan remains on death row. http://www.doc.alabama.gov/deathrow.asp.
OHIO
Biros v. Bagley. Kenneth Biros was convicted of aggravated robbery, attempted rape, and aggravated
murder. He was sentenced to death. The conviction for aggravated robbery was dismissed by the Ohio Court
of Appeals. The Supreme Court of Ohio then upheld the death sentence and reinstated the aggravated robbery
conviction. Biros filed a petition challenging his conviction and sentence in the NDOH (Cleveland). The
court granted relief on Biros’s claim that the indictment was defective because it did not include the
language from O.R.C. § 2929.04(A)(7) that “either the offender was the principal offender in the
commission of the aggravated murder or, if not the principal offender, committed the aggravated
murder with prior calculation and design.” The district court held that this was structural error and Biros’s
167

Court of appeals status reported as of March 2007 unless otherwise noted.

Final Technical Report: Habeas Litigation in U.S. District Courts

124

death sentence must be set aside. The rest of Biros’s claims were denied. The Sixth Circuit disagreed and held
that the error was subject to harmless error analysis and in this case it was harmless. The district court’s
judgment was reversed and relief was denied. Biros remains on death row. Offender Information for Kenneth
Biros
TEXAS
Chester v. Johnson. Elroy Chester was convicted of capital murder for a killing in the course of burglarizing
a house. He was sentenced to death. He filed his petition in the EDTX (Texarkana), challenging his sentence
only. The petition raised three claims of error, two of which were denied as moot. The court granted Chester’s
claim that his death sentence violated the Eighth Amendment, as determined in Atkins v. Virginia,
because Chester is mentally retarded. The district court ordered that Chester be released from custody
unless new proceedings were held to determine whether he is mentally retarded, or in the alternative, that his
sentence be commuted to life imprisonment. Respondent appealed and the Fifth Circuit vacated the district
court’s judgment and remanded, finding that the Atkins claim was unexhausted. On remand, the district
court dismissed the petition without prejudice. It is unclear whether Chester has since exhausted his state
court remedies. Chester remains on death row. Texas Offender Search
Conner v. Dretke. Johnny Conner was convicted of capital murder and sentenced to death. He filed a
petition in the SDTX (Houston) challenging his conviction and sentence. The petition raised three claims of
error, two of which were denied on the merits. The court granted relief on Conner’s claim that his trial
counsel was ineffective for failing to investigate and present evidence of his medical history. Conner’s
medical history was important because he had a leg injury that interfered with his ability to run and the
eyewitnesses in this case all agreed that the perpetrator ran from the scene of the crime. Conner’s other two
claims were denied on the merits. The district court ordered that Conner be released from custody unless he is
granted a new trial. Respondent appealed and the Fifth Circuit reversed. Conner remains on death row. Texas
Offender Search
Foster v. Dretke. Kenneth Foster was convicted of capital murder, which resulted from his involvement in a
string of armed robberies. He was sentenced to death. Foster filed a petition for a writ of habeas corpus in the
WDTX (San Antonio) challenging his conviction and sentence, raising 14 claims for relief, including a few
sub-claims. Eight of those claims were denied on the merits and one was barred by Teague v. Lane. Two
claims related to newly discovered evidence of actual innocence were barred as not cognizable on federal
habeas review. The district court granted relief on Foster’s fourth, fifth, and seventh claims. The district court
found that there was no evidence that Foster actually killed the victim or intended to kill the victim or another
person. Foster’s sentencing jury was never given the opportunity to decide whether he was a major
participant in the armed robbery conspiracy that resulted in murder. Because this finding was never made by
the jury, the district court held that Foster’s death sentence violated the Eighth Amendment because it is
not supported by the factual findings mandated by Emmund and Tison and it violated Ring v. Arizona.
Both parties appealed the decision and the Fifth Circuit denied a Certificate of Appealability. Foster then
asserted an actual innocence claim. The Fifth Circuit denied a Certificate of Appealability on the actual
innocence claim but went on to discuss the Eighth Amendment claim for which the district court had granted
relief. The Fifth Circuit reversed, finding that Ring v. Arizona does not apply retroactively and that the state
court did in fact make the findings required by Emmund and Tison. The Supreme Court denied certiorari.
Foster remains on death row. Texas Offender Search.

Final Technical Report: Habeas Litigation in U.S. District Courts

Appendix D. List of District names and numbers
Capital case sample
AO number District
13
26
3A
39
40
41
42
47
48
70
73
78
87

PA-E
AL-N
FL-M
TX-N
TX-E
TX-S
TX-W
OH-N
OH-S
AZ
CA-C
NV
OK-W

125

Final Technical Report: Habeas Litigation in U.S. District Courts

126

Appendix E. List of variables collected 168
Per case variables
Coder Name
Date coded
Sample
Where coded
Documentation
available for coding
Petitioner sex
Petitioner race
Not a state hab case
Proceeding challenged
Conviction date
Number of counts of conviction
Conviction type
Most serious count of conviction
1st Offense Attempt/Conspiracy
Second most serious count of conviction
2d off Attempt/Conspiracy
Third most serious count of conviction
3d off Attempt/Conspiracy
List additional convictions
Sentence date
Judgment date
Sentence type
Length of sentence
Prior direct appeal or no
Prior state collateral review or no
Representation on state PCR
Date state judgment final for SOL (cap cases only)
Date of first docket entry in case
Date federal petition first filed
Date first counseled petition filed
Amended petition filed
IFP ruling
Representation at time of first filing in case
Attorney added
Atty substituted
Date atty starts
Attorney appointed for what
Answer filed
pleading by st requesting dismissal or sj
Reply by petitioner
State record filed
Discovery
Evidentiary Hearing
Evidentiary hearing by who
Magistrate R&R on disposition of any claim
DCt stayed death sentence in state
Prior petition dismissed w/o prej referred to in record
168

Additional variables derived from the data collected or from other sources (e.g., AO data sets, FJC documentation)
were added to the data set for analysis. A complete list with codebook will be included with the data sets submitted for
archiving.

Final Technical Report: Habeas Litigation in U.S. District Courts

127

Record reflects fed case put on hold for exhaustion
Date of stay for exhaustion, consideration by state ct entered
Date stay lifted
Date case first terminated in DCt
Docket number of 1st termination order
Appellate information*
Notice of Appeal filed
CoA considered by DJ
Date DJ COA ruling docketed
Total no of docket entries

Per claim variables:
Type of claim raised
Disposition by DJ, type
Defense of SOL ruling by DJ
Reason DJ rejected SOL
Defense of PD ruling by DJ
Reason DJ rejected PD
Def of exhaustion ruling by DJ
Defense of Teague ruling by DJ
Successive petition
waiver of review ruling
Reason for DJ denial/dismissal
Alternative reason for denial/dismissal
Std of review applied if reached merits
Application of Harmless error

*appellate information was not available for cases that had recently terminated or were still pending.