Who’s better at defending criminals? Does type of defense attorney matter in terms of
producing favorable case outcomes
By Thomas H. Cohen *
The role of defense counsel in criminal cases constitutes a topic of substantial importance for
judges, prosecutors, defense attorneys, scholars, and policymakers. What types of defense
counsel (e.g., public defenders, privately retained attorneys, or assigned counsel) represent
defendants in criminal cases and how do these defense counsel types perform in terms of
securing favorable outcomes for their clients? These and other issues are addressed in this article
analyzing felony case processing data from the Bureau of Justice Statistics (BJS). Specifically,
this paper examines whether there are differences between defense counsel type and the
adjudication and sentencing phases of criminal case processing. Results show that private
attorneys and public defenders secure similar adjudication and sentencing outcomes for their
clients. Defendants with assigned counsel, however, receive less favorable outcomes compared
to their counterparts with public defenders. This article concludes by discussing the policy
implications of these findings and possible avenues for future research.
Introduction
The Sixth Amendment to the U.S. Constitution establishes the right to counsel in Federal
criminal prosecutions. The U.S. Supreme Court expanded the defense counsel right for indigent
defendants in a series of cases decided in the 1960s and 1970s. One of the landmark decisions
that occurred during this period was Gideon v. Wainwright (372 U.S. 335 (1963)) where the U.S.
Supreme Court held that a defendant charged with a felony, including state crimes, had the right
to government provided counsel. The Supreme Court further extended the defense counsel right
to juvenile court proceedings in the In re Gault case (387 U.S. 1 (1967)) and to defendants facing
imprisonment for either felony or misdemeanor offenses in Argersinger v. Hamlin (407 U.S. 25
(1972)). More recently the Court held in Alabama v. Shelton (535 U.S. 654 (2002)) that indigent
defendants are entitled to court appointed counsel even when facing a suspended jail term.

                                                            
*
 Statistician, Bureau of Justice Statistics 810 7th Street, NW, Washington DC 20531 Email:
Thomas.H.Cohen@usdoj.gov. Special thanks to my colleagues at the Bureau of Justice Statistics including Duren
Banks and Michael Planty for reviewing and commenting on earlier drafts of this paper. The views and opinions
expressed in this paper are solely the author’s. They do not represent the views of the Bureau of Justice Statistics,
the Office of Justice Programs, or the U.S. Department of Justice. 

1
 
Electronic copy available at: http://ssrn.com/abstract=1876474

While the Supreme Court has recognized the right to defense counsel in most criminal
proceedings, it has not mandated how the provision of criminal defense should be provided by
the states. For this reason, states have adopted a variety of approaches to defending the
criminally accused including the utilization of some combination of public defender systems,
assigned counsel programs, or contract attorneys (Neubauer and Fradella, 2011). In addition,
defendants with the means can always hire their own attorneys for the purpose of criminal
defense. The ability of some defendants to hire their own attorneys, coupled with the fact that
many defendants do not have the means to pay for legal representation, raises questions of
attorney effectiveness. What types of defense counsel represent criminal defendants and how do
these attorneys perform in terms of securing favorable outcomes for their clients? Are private
attorneys more effective in keeping their clients from being convicted and protecting those
convicted clients from harsher punishments than public defenders or assigned counsel? How do
assigned counsel systems perform in comparison to public defenders? Are assigned counsel
securing dismissals and obtaining sentences at rates similar to that of public defenders?
This article will address these key questions about the role of defense counsel in criminal
cases. First, it will detail the types of defense counsel currently employed in state courts and
highlight prior research examining the effectiveness of public defenders, assigned counsel and
private attorneys. The data utilized in this article will be described and then a bivariate analysis
examining how frequently the various types of defense counsel including public defenders,
assigned counsel and private attorneys are employed and whether key case outcomes including
the likelihood of conviction, incarceration, and sentence length vary by these defense attorneys
will be provided. This paper will then build on the bivariate findings by attempting to model the
outcomes of conviction, incarceration, and sentence length for the purpose of discerning whether

2
 
Electronic copy available at: http://ssrn.com/abstract=1876474

associations at the bivariate level hold when other factors related to these outcomes are
statistically controlled. Model results will be explicated and the article will conclude by
discussing the possible repercussions and implications inherent in the current research.
Types of defense counsel in state courts
Although the U.S. Supreme Court requires states to provide representation to indigent
defendants, the method of defense counsel provision is not specified. Consequently, states have
adopted different approaches to providing counsel for poor defendants. The major types of
publicly financed defense counsel representation provided by the states include some
combination of public defender systems, assigned counsel programs, or contract attorneys
(Neubauer and Fradella, 2011; Spangenberg and Beeman, 1995). These systems of indigent
defense are applied in a blended format throughout the states. Some states, for example, employ
statewide public defender systems but still utilize contract or assigned counsel in conflict cases
or as a means of alleviating heavy caseloads. Other states have no centralized mechanism of
public defense and employ differing methods of indigent representation at the local level with
some counties using public defenders and others employing contract attorneys or assigned
counsel in the same state (Farole and Langton, 2010; Spangenberg and Beeman, 1995). 1
Of all the methods of indigent representation, the system of assigned counsel is perhaps
the oldest (Neubauer and Fradella, 2011). Assigned counsel systems involve the appointment by
the courts of private attorneys as needed from a list of available attorneys. Assigned counsel
systems consists of either an ad hoc structure where private attorneys are appointed by judges on
a case by case basis or coordinated systems in which an administrator oversees the appointment
                                                            
1

While there are no nationwide statistics on the prevalence of these three forms of indigent defense, a survey of
indigent defense systems in the nation’s 100 most populous counties conducted in 2000 showed public defenders
handling 82%, assigned counsel 15%, and contract attorneys 3% of the 4.2 million cases disposed of in these
counties (DeFrances and Litras, 2000).

3
 
Electronic copy available at: http://ssrn.com/abstract=1876474

of counsel (Neubauer and Fradella, 2011; Spangenberg and Beeman, 1995). Assigned counsel
systems have been criticized for appointing attorneys with inadequate skills, experience, and
qualifications to represent indigent defendants. This problem is especially acute in counties with
ad-hoc assignment systems where recent law school graduates or attorneys of marginal
capabilities will sometimes take clients as a means of gaining trial experience or supplementing
income (Neubauer and Fradella, 2011; Iyengar, 2007; Beck and Shumsky, 1997; Spangenberg
and Beeman, 1995; Gist, 1989-1990). Scholars specializing in indigent defense, however, argue
that these weaknesses can be overcome by the establishment of administrative oversight
organizations whose purpose is to ensure that appointed counsel have the requisite skills and
qualifications to provide adequate defense. These oversight boards can also provide supervision,
training, and support for attorneys selected to participate in assigned counsel systems
(Spangenberg and Beeman, 1995).
Compared to assigned counsel systems, contract attorneys are a more recent approach for
providing indigent representation through the private market (Neubauer and Fradella, 2011).
Contract attorneys involve governmental units reaching agreements with private attorneys, bar
associations, or law firms to provide indigent defense services for a specific dollar amount and
time period (Spangenberg and Beeman, 1995; Worden, 1993, 1991). Although contract systems
can limit the costs governments pay for indigent defense, critics argue that these systems could
reduce the quality of representation as law firms underbid each other in an effort to secure
competitive contracts. In some markets, however, contract systems have failed to reduce the
costs of indigent defense and have actually resulted in higher defense costs as a result of not
enough attorneys being available to generate competitive markets (Spangenberg and Beeman,
1995; Worden, 1990-1991).

4
 

Of all forms of indigent defense, the most popular and widely used are public defender
programs. Under a public defender system, salaried staff attorneys render criminal indigent
defense services through a public or private nonprofit organization or as direct government
employees. The first public defender program started in Los Angeles County in 1913 and spread
gradually until the Supreme Court decisions of Gideon and Argersinger resulted in a more rapid
expansion (Neubauer and Fradella, 2011; Spangenberg and Beeman, 1995). The administration
and funding of public defender programs occurs at either the state or county levels. In 2007, 22
states administered and provided funding to public defender offices at the state level, while in the
remaining 27 states and the District of Columbia, public defender offices were funded and
administered at the local level (Farole and Langton, 2010; Langton and Farole, 2010).
Public defender programs have a variety of strengths which have been discussed
extensively in the literature. The principle benefits of the public defender system are that it
provides indigent defendants with access to professional legal staff with the training, experience,
and skills to provide adequate legal defense. Public defenders offices can also employ
investigative and other support services that might not be available through assigned counsel or
contract programs (Neubauer and Fradella, 2011; Spangenberg and Beeman, 1995). Lastly,
intensive interactions with prosecutors and judges enable public defenders to forge close
relationships with key members of the courtroom workgroup ensuring that these attorneys are
well positioned to strike favorable bargains for their clients (Hartley, Miller, and Spohn, 2010;
Fleming, Nardulli, and Eisenstein, 1992; Champion, 1989; Heumann, 1978).
Criticisms of public defender programs center on issues related to funding and cooptation. In many jurisdictions, public defender programs are not allocated enough resources to
keep up with expanding caseloads which could prevent them from adequately representing their

5
 

clients (Spangenberg and Beeman, 1995). Another key criticism concerns the phenomena of
public defenders being pressured by members of the courtroom workgroup to emphasize rapid
case processing over vigorous criminal defense. Public defenders often work in an environment
in which judges and prosecutors stress the need to process large numbers of defendants who are
either factually or legally guilty. Some argue that the rapid processing of these defendants, as
opposed to an adversarial forum where prosecutors and public defenders forcefully represent
their positions before neutral judges, defines the true nature of the courtroom workgroup and
criminal case processing (Heumann, 1978; Blumberg, 1966-1967). By placing a premium on the
expeditious disposition of factually or legally guilty defendants, public defenders face enormous
pressures to cooperate with judges and prosecutors by encouraging their clients to plead guilty
rather than mount strong adversarial defenses (Hartley, Miller, and Spohn, 2010; McCoy, 1993;
Fleming, Nardulli, and Eisenstein, 1992; Feeley, 1992; Heumann, 1978).
Defendants who don’t wish to avail themselves of indigent representation, and who have
the means of doing so, can hire a private attorney. Since private attorneys are not part of the
courtroom workgroup, they cannot be as easily pressured into emphasizing expeditious case
resolution over vigorous advocacy. In theory, private attorneys should be able to forcefully
represent their clients without taking into consideration their relationships with the local judges
or prosecutors. Private attorneys might also have the financial resources to mount a stronger
defense than their indigent counterparts (Neubauer and Fradella, 2011; Hartley, Miller, and
Spohn, 2010).
In spite of these potential strengths, several factors call into question whether private
attorneys are truly superior to indigent counsel. Private attorneys rarely have the opportunity to
specialize solely in the practice of criminal defense. Unlike their public defender counterparts,

6
 

most private practices have a multitude of different case types which prevents these attorneys
from developing the levels of professionalism and expertise approaching that of public defenders
in the area of criminal law and procedure (Hanson, Ostrom, Hewitt, and Lomvardias, 1992).
Also, the weaker relationships between private attorneys and other courtroom actors might
hinder these attorneys from striking deals with prosecutors that public defenders, who are more
integrally ensconced with the court community, might be able to reach (Hartley, Miller, and
Spohn, 2010; Fleming, Nardulli, and Eisenstein, 1992; Feeley, 1992; Champion, 1989).
Prior research on the effect of attorney type on case outcomes
There is an extensive literature comparing different attorney types in terms of their
effectiveness in securing favorable outcomes for their clients (Feeney and Jackson, 1990-1991).
Most of this literature examines whether public defenders represent their clients as effectively as
private counsel in terms of securing acquittals or dismissals, keeping their clients from being
incarcerated, or ensuring that the shortest possible sentences are imposed on their clients
(Hanson, Ostrom, Hewitt, and Lomvardias, 1992). 2 Evaluations comparing public defenders to
assigned counsel or contract attorneys regarding the abovementioned outcome characteristics are
less frequent. A brief overview of the literature examining whether public defenders do better at
criminal defense than private attorneys or assigned/contract counsel is provided below.
The majority of studies show defendants with public defenders receiving adjudication,
incarceration, and sentencing outcomes that are not appreciably different compared to those with
private attorneys (Feeney and Jackson, 1990-1991). Some notable studies comparing public
defenders to private attorneys include Hartley, Miller and Spohn (2011) examination of counsel
                                                            
2

Another type of literature examines the relationship between attorneys and their clients through a variety of “input”
measures including number of meetings, time of first meeting, and level of legal work for a particular case (Harlow,
2000; Hanson, Ostrom, Hewitt, and Lomvardias, 1992). Since this article focuses on outcomes and not attorney
client relationships, this line of research is omitted from further discussion.

7
 

type effects in criminal cases processed in Cook County; Williams (2002) study of the efficacy
of private vs. public defenders in a northern Florida jurisdiction; Hanson, Ostrom, Hewitt, and
Lomvardias (1992) evaluation of defense counsel effectiveness in nine jurisdictions; and
Nardulli, Eisenstein, and Flemming (1988) exploration of defense counsel as part of a general
examination of criminal case processing in nine counties located in several mid-western states. 3
For the most part, these studies found that defendants represented by public defenders do not
receive disadvantageous outcomes compared to their counterparts with private attorneys.
Although most studies find public defenders and private attorneys securing similar
outcomes for their clients, some research shows private attorneys doing better at criminal
defense. A study conducted by Hoffman, Rubin, and Shepherd (2005) showed defendants with
private attorneys receiving shorter prison sentences than those with public defenders. Other
studies have found that attorney type can influence pretrial release decisions, charge reductions,
and sentencing outcomes in certain contexts (Hartley, Miller and Spohn, 2011). Lastly, several
studies show private attorneys obtaining favorable results for some outcomes such as
incarceration decisions but not for the remainder of outcomes including likelihood of conviction
and length of imposed sentence (Hanson, Ostrom, Hewitt, and Lomvardias, 1992).
Empirical studies comparing public defenders to assigned attorneys or contract counsel
are less frequent. Several earlier studies including Hermann, Single, and Boston (1977)
evaluation of defense counsel types among 3,000 felony cases processed in Los Angeles, New
York, and Washington, DC and others conducted by Radtke, Semple, and Cohen (1982) and
Clarke and Koch (1980) found that defendants represented by public defenders and assigned
counsel received similar outcomes (Feeney and Jackson, 1990-1991). The National Center for
                                                            
3

For an excellent summary of earlier studies on this issue see Feeney and Jackson (1990 – 1991). For a more recent
overview of this topic, see Hartley, Miller and Spohn (2011).

8
 

State Courts evaluation of several indigent defense types including public defenders, assigned
counsel, and contract attorneys also showed little discernible differences among these various
types of defense counsel (Hanson, Ostrom, Hewitt, and Lomvardias, 1992).
Several recent studies comparing public defenders to assigned counsel, however, have
found assigned counsel producing less favorable outcomes for their clients. In a study comparing
defendants represented by public defenders to those with court appointed counsel in federal
district courts, Iyengar (2007) found that defendants with assigned counsel were more likely to
be convicted and receive longer sentences than defendants with public defenders. Another study
conducted by Roach (2010) using a sub-sample of felony cases from the State Court Processing
Statistics (SCPS) project discerned assigned counsel obtaining noticeably less favorable
outcomes for their clients compared to public defenders. In addition to these studies, research has
also shown assigned counsel performing poorly when compared to private attorneys (Beck and
Shumsky, 1997; Champion, 1989).
In summary, these studies have advanced our understanding of how counsel type may
influence case processing outcomes. Overall, these studies show private attorneys and public
defenders obtaining similar results for their clients; however, they also provide some evidence
suggesting that assigned counsel are less effective advocates than other types of criminal
attorneys. While this research has illuminated how the different types of defense counsel might
affect case processing outcomes, there are some limitations that could be addressed by the
current study. First, most studies involve comparisons of either public defenders to private
attorneys or public defenders to assigned counsel. Few research efforts attempt to examine the
impact of private attorneys, public defenders, and assigned counsel on case outcomes
simultaneously. In addition, many of these studies have at most a limited number of jurisdictions

9
 

or relied on relatively dated data. Moreover, the few multijurisdictional studies on this topic did
not take advantage of the statistical techniques currently available to account for the hierarchical
nature of their data. Nor did many of the abovementioned studies address the problem of possible
sample selection biases that could occur as case outcomes are followed from charging through
adjudication and sentencing (Berk, 1983). Lastly, there have been few recent studies comparing
public defenders to assigned counsel in state court systems. Of the two research efforts
mentioned, one focused on comparing assigned counsel to public attorneys in the federal courts
and the other used a sub-sample of felony case processing data from the State Court Processing
Statistics (SCPS) project (Roach, 2010; Iyengar, 2007). 4
This article will attempt to address these limitations by using more updated state level
felony data, by examining data covering a larger and more varied number of jurisdictions, and by
applying statistical techniques that can handle the sampling framework and sample selection
biases of the data being analyzed. This article will also attempt a more in-depth comparison of
assigned counsel to public defenders than has been attempted in prior studies. 5 Specifically, this
article will attempt to address the following research questions:
•

How do public defenders, private attorneys, and assigned counsel compare in
terms of securing favorable outcomes for their clients? Are defendants
represented by private attorneys or assigned counsel more or less likely to be
convicted, incarcerated, or sentenced to prison than their counterparts represented
by public defenders? Are defendants with private attorneys or assigned counsel
sentenced to shorter or longer periods of confinement than those with public
defenders?

                                                            
4

Roach (2010) used data only in instances where random assignment between public defenders and assigned
counsel could be supported with the SCPS data. For these reasons, counties relying solely on public defenders or
assigned counsel were excluded from that analysis.
5
It should be noted that assigned counsel refers to both attorneys working under an assigned counsel system and
contract attorneys. For reasons that are further described in the methodology section, the data examined do not
distinguish assigned from contract attorneys.

10
 

•

Are defendants represented by private attorneys or assigned counsel receiving
more or less advantageous outcomes for certain offense categories? For example,
are defendants charged with violent crimes in a better or worse position if they are
represented by private attorneys or assigned counsel compared to their
counterparts with public defenders?

•

What are the characteristics of defendants represented by the different types of
defense attorneys? Are defendants retaining private counsel different in terms of
their offense charge seriousness, criminal history, and demographic characteristics
compared to defendants represented by public defenders or assigned counsel? Are
indigent defendants represented by assigned counsel or public defenders
comparable concerning their most serious criminal charges, criminal histories,
and demographic characteristics?

Data used to examine the effect of defense counsel on criminal case outcomes
This article analyzed data from the State Court Processing Statistics (SCPS) series,
covering felony cases filed in May of even-numbered years in 2004 and 2006. SCPS is a biennial
data collection series sponsored by the Bureau of Justice Statistics that examines felony cases
processed in a sample of 40 of the nation’s 75 most populous counties. The SCPS sample is a 2stage stratified sample, with 40 of the nation’s 75 most populous counties selected at stage one
and a systematic sample of state court felony filings (defendants) within each county selected at
stage two. Counties selected to participate in SCPS provide a list of defendants charged with a
felony on certain randomly selected business days of May of an even number year and these
cases are followed until case disposition or May 31st of the following year. Weights are applied
so that these data represent felony case processing for the entire month of May in the nation’s 75
most populous counties. 6
Each SCPS data collection tracks approximately 15,000-16,000 felony defendants for up
to one year. A variety of information is collected on these defendants including the types of
arrest charges filed against felony defendants, conditions of pretrial release, and pretrial
                                                            
6

For more information about the SCPS methodological framework, see Cohen and Kyckelhahn (2010) report titled
Felony Defendants in Large Urban Counties, 2006 at http://bjs.ojp.usdoj.gov/.

11
 

misconduct. The adjudication outcomes encompassing the dismissal, diversion, guilty plea, and
trial conviction rates for felony defendants are also recorded. For those defendants convicted,
information on the sentencing of felony defendants including the imposition of prison, jail, and
probation sentences are provided. The SCPS project, moreover, obtains data on the defendant’s
demographic characteristics, criminal justice status at the time of arrest, and prior arrests and
convictions.
In addition to this case and defendant level data, the SCPS series also obtains information
on the types of defense counsel representing felony defendants in state courts located in the
nation’s 75 most populous counties. Specifically, SCPS collects information on whether the
defendant was represented by a public defender, assigned counsel, or private attorney at the time
of case adjudication. SCPS also identifies those defendants who decided to proceed without an
attorney (pro-se) at the time of case adjudication.
There are many challenges associated with collecting data on defense counsel type for
felony defendants in state courts. First, many jurisdictions do not capture information on defense
counsel in a way that is readily accessible for the SCPS project. In prevision iterations of SCPS,
the difficulties associated with collecting defense counsel information resulted in a high
proportion of felony cases with missing data for this particular element. For example, defense
counsel data were missing for approximately two-fifths of defendants in the 1992 and 1994
SCPS collections and were not available for about a third of defendants tracked in the 1996
through 2002 SCPS series. The 2004 and 2006 SCPS projects witnessed a decline in missing
defense counsel data to about 25% of all SCPS cases. Since the two most recent SCPS iterations

12
 

had fewer defendants with missing defense counsel data, this article focuses on these newer
datasets and excludes earlier years. 7
The other challenge inherent in collecting information on defense counsel type is that
there can be substantial churn in who represents defendants in court. Defendants, at times, will
ask for or demand a change in defense counsel and attorneys have been known to withdraw from
cases in which they are in conflict with clients over legal strategy or compensation.
Unfortunately, by focusing on attorney type at the date of case disposition, SCPS is unable to
measure or ascertain possible movements of attorneys on and off specific cases. Lastly, SCPS
does not distinguish assigned counsel from contract attorneys. Both types of defense counsel
representation are treated as the same and are labeled as “assigned counsel” for the purposes of
this article. Although it cannot be stated with certainty what proportion of assigned counsel are
contract attorneys, an examination of indigent defense systems in the 100 most populous
counties showed contract attorneys accounting for a relatively small percent of indigent defense
systems in highly populated jurisdictions (DeFrances and Litras, 2000).
Before delving into the statistical models examining the impact of defense counsel on
case outcomes, it’s important to use descriptive techniques to see how frequently public
defenders, assigned counsel and private attorneys are employed in felony cases and examine
whether felony defendants are similar or different in terms of their offense seriousness, criminal
                                                            
7

Although the defense counsel field has improved, the fact that this information was missing for 25% of cases raises
the possibility that the results in this article could change if defense counsel information were available for all cases
tracked in the SCPS sample. While it’s impossible to determine with certainty what would happen if defense counsel
data were obtainable for all SCPS cases, the influence of missing data can checked through an imputation procedure.
Imputation techniques were used to estimate defense counsel type for cases that did not have this information. These
techniques assessed the effect of missing data by comparing results in which defense counsel type were estimated
against results in which the categories of defense representation were not available. Results were not appreciably
different between the imputed and non-imputed data at the bivariate and multivariate levels. Since only minor
changes occur with using imputed data and since it’s generally not suggestible to impute when data are missing for
over 10% of cases, this article did not rely on the imputed data for further analysis.

13
 

histories, and demographic characteristics across these various forms of defense counsel. A
descriptive analysis can also illuminate whether key case outcomes including the likelihood of
conviction, incarceration, and sentence length vary by these defense attorneys without the
complexities inherent in statistical model building. The descriptive results follow below.
Descriptive analysis of characteristics of felony defendants by attorney type and
relationship between attorney type and case outcomes
An analysis of defense counsel among felony defendants shows the vast majority
employing some form of indigent representation. In 2004 and 2006, about 80% of defendants
charged with a felony in the nation’s 75 most populous counties reported having public
defenders or assigned counsel, while 20% hired an attorney (table 1). Among the estimated
69,000 felony defendants using publicly financed defense services, approximately three-fourths
were represented by public defenders. Defendants charged with property or drug crimes were
slightly more likely to have been represented by public defenders or assigned counsel (80%) than
those charged with public-order (74%) or violent (76%) offenses. Interestingly, about 2% of
felony defendants proceeded pro-se, meaning that they represented themselves in court. Since
this research is interested in examining the relationship between defense counsel type and
criminal case processing outcomes, these pro-se defendants are excluded from the remainder of
this analysis. 8

                                                            
8

See Hashimoto (2007) for an analysis of the implications of proceeding without an attorney in criminal cases.

14
 

Table 1. Types of defense counsel, by most serious arrest charge category,
in the nation's 75 most populous counties, 2004 & 2006

Most serious
Number of
arrest charges defendants
All defendants
87,661
Violent
19,059
Property
26,844
Drug
32,387
Public-order
9,371

Percent of felony defendants represented by Public
Assigned Private
defender
counsel attorney
Pro - se
60 %
19 %
20 %
2%
57
19
22
2
59
20
18
2
63
17
18
2
55
20
25
1

Analyzing the distribution of offense charges across the defense counsel categories
provides another method for examining the variation of legal counsel for felony defendants. For
the general and specific SCPS offense categories, defendants with public defenders and assigned
counsel were charged with relatively similar offenses 9 ; however, defendants with private
attorneys had a different distribution of offense charges compared to defendants with public
defenders or assigned counsel (table 2). 10 In terms of specific offense charges, a greater
percentage of rape/sexual assault, drug trafficking, and public-order defendants clustered around
private attorneys compared to their indigent counterparts. For example, 7% of defendants
represented by private attorneys were charged with rape or other violent 11 crimes, while about
4% of defendants with public defenders or assigned counsel were charged with these offenses.
Private attorneys also had a higher proportion of their clients charged with drug trafficking
(17%) compared to their assigned counsel (13%) or public defender (14%) counterparts. In
                                                            
9

Chi-square statistic of 2.5, p > .05 demonstrates no statistically significant difference between the offense
distributions for defendants represented by public defenders and assigned counsel. For the remainder of the
descriptive section, chi-square and other tests of significance are reported only in instances where the categories of
defense counsel significantly differ from each other.
10
Chi-square tests of offense distributions between private attorneys and public defenders (23.4, p < .001) and
private attorneys and assigned counsel (4.1, p < .05).
11
A large portion of “other” violent offenses includes non-rape sexual offenses.

15
 

addition, private attorneys represented more defendants charged with public-order offenses
(14%) than public defenders (10%) or assigned counsel (11%).
Although the offense distributions between defendants represented by public defenders
and assigned counsel were not significantly different, there were some disparities worth noting.
In particular, a slightly higher proportion of defendants with public defenders were charged with
other drug (e.g., drug possession) offenses (25%) than their equivalents with assigned counsel
(21%). Also, defendants with assigned attorneys witnessed a somewhat higher number of
charges involving very violent crimes such as murder, rape, or robbery (8%) than defendants
with public defenders (6%).

16
 

Table 2. Comparing types of defense counsel, by most serious arrest charge category,
in the nation's 75 most populous counties, 2004 & 2006
Most serious
arrest charges
General categories
Violent
Property
Drug
Public-order

Felony defendants represented by Public defender
Assigned counsel
Private attorney

Specific categories
Murder
Rape
Robbery
Assault
Other violent
Burglary
Larceny-theft
Motor vehicle theft
Forgery
Fraud
Other property
Drug sales
Other drug
Weapons
Driving-related
Other public order
Number of defendants

21 %
30
39
10

22 %
33
34
11

25 %
29
33
14

0.4 %
0.5
5.5
11.1
3.3
8.8
8.6
4.0
2.7
2.4
4.1
14.2
24.8
3.4
2.9
3.6

0.7 %
0.9
6.7
11.2
2.6
8.9
8.3
3.9
3.3
4.7
3.6
13.3
20.8
2.5
3.2
5.4

0.6 %
1.6
5.1
11.8
5.4
7.3
8.0
1.7
3.0
5.0
3.8
17.3
15.9
3.9
5.0
4.7

52,337

16,613

17,101

Comparing demographic characteristics shows virtually no differences in terms of age or
gender across these various forms of defense counsel but reveals private attorneys representing a
greater percentage of whites than their indigent counterparts (table 3). 12 For example, 36% of
defendants retaining private attorneys were white compared to 29% with assigned counsel and
26% with public defenders. Conversely, a higher proportion of defendants represented by public
defenders (44%) or assigned counsel (47%) were black than defendants with the means to hire
their own attorneys (34%).
                                                            
12

Defendants represented by private attorneys had significantly different racial characteristics compared to
defendants with public defenders (chi-square = 20.2, p < .001) or assigned counsel (chi-square = 16.3, p < .001).

17
 

Table 3. Comparing types of defense counsel, by demographic characteristics, in the nation's
75 most populous counties, 2004 & 2006
Demographic
Felony defendants represented by characteristics Public defender Assigned counsel Private attorney
Gender
Male
82 %
82 %
83 %
Female
18
18
17
Race
White
Black
Hispanic
Other

26 %
44
27
2

29 %
47
23
1

36 %
34
27
3

Mean Age

32 yrs.

32 yrs.

31 yrs.

A defendant’s criminal background constitutes another area of potential differences
between defendants represented by indigent counsel and private attorneys. In general, criminal
backgrounds were less common among defendants who retained private counsel. Nearly a third
(31%) of defendants with private attorneys had no previous arrest history, while only about a
fifth of defendants with public defenders (17%) or assigned counsel (19%) had never been
arrested (table 4). 13 Convictions mirrored arrest history with approximately half of defendants
with public defenders or assigned counsel having at least one prior felony conviction compared
to 36% of defendants with private attorneys. 14 Unlike those with private counsel, the criminal
histories of defendants represented by public defenders and assigned counsel were nearly
identical. Only in the area of criminal justice status (e.g., on probation, parole, etc) were

                                                            
13

Defendants with private attorneys were significantly less likely to have prior arrest records compared to
defendants with public defenders (chi-square = 87.1, p < .001) or assigned counsel (chi-square = 60.0, p < .001).
14
Prior felony convictions significantly less common among defendants with private attorneys than defendants with
public defenders (chi-square = 88.9, p < .001) or assigned counsel (chi-square = 34.5, p < .001).

18
 

defendants with public defenders manifesting a more serious criminal history than defendants
with assigned counsel.

Table 4. Comparing types of defense counsel, by defendant criminal history, in the nation's
75 most populous counties, 2004 & 2006
Percent of felony defendants represented by Criminal history Public defender Assigned counsel Private attorney
Active criminal
41 %
33 %
28 %
justice status
Most serious
prior arrest
Felony
Misdemeanor
None

72 %
11
17

69 %
12
19

54 %
14
31

Most serious
prior conviction
Felony
Misdemeanor
None

51 %
17
32

50 %
17
33

36 %
17
47

This analysis has examined whether felony defendants represented by public defenders,
assigned counsel, or private attorneys had similar or different characteristics by several criteria
including offense charges, demographics, or criminal history. The next part focuses on
adjudication and sentencing outcomes and examines whether they vary by defense counsel type.
In the nation’s 75 most populous counties, the overall conviction rates were about the
same for felony defendants represented by public defenders (73%) or hired attorneys (72%)
(table 5). Defendants with assigned counsel, in comparison, faced a higher likelihood (78%) of

19
 

conviction. 15 Among those defendants convicted, approximately 90% with assigned counsel or
public defender representation, and about 85% with private attorneys, were convicted of a
felony. The remaining 10-15% of defendants were convicted of a misdemeanor across these
defense counsel categories. The vast majority of convicted defendants plead guilty irrespective
of who represented them in court with only a slightly higher percentage of defendants with
private attorneys using the trial option (4%) compared to defendants with public defenders or
assigned counsel (3%).
Convicted defendants represented by public defenders or assigned counsel were more
likely than those hired by private attorneys to be sentenced to incarceration. About two-thirds of
convicted defendants with private attorneys were sentenced to either prison or jail; in
comparison, 78% of convicted defendants represented by assigned counsel and 74% represented
by public defenders received an incarceration sentence. 16 The percent of defendants sentenced to
incarceration did not differ significantly between defendants with public defenders or assigned
counsel. When examining the type of incarceration sentence imposed, convicted defendants
represented by assigned counsel were significantly more likely to receive prison sentences
compared to those represented by either public defenders or private attorneys. Nearly half (46%)
of convicted defendants with an assigned counsel received a prison sentence, while
approximately a third of convicted defendants with retained counsel (29%) or public defender
(32%) representation were sentenced to prison. 17

                                                            
15

Defendants with assigned counsel were significantly more likely to be convicted than defendants with public
defenders (chi-square = 7.9, p < .01) or private attorneys (chi-square = 9.1, p < .001).
16
Higher percentage of defendants with assigned counsel (chi-square = 11,.6, p < .01) and public defenders (chisquare = 19.9, p < .001) received incarceration compared to defendants with private attorneys.
17
Significant differences in prison incarceration for defendants represented by assigned counsel compared to
defendants with public defenders (chi-square = 5.9, p < .01) or private attorneys (chi-square = 16.1, p < .001)

20
 

Among convicted defendants sentenced to serve time either in prison or jail, those using
public defenders received shorter average sentences than those with private attorneys or assigned
counsel. Defendants with public defenders were sentenced to an average of 23 months of
confinement, while those with hired attorneys or assigned counsel were sentenced to
incarceration terms averaging 31 and 35 months, respectively. 18 The average incarceration
periods between defendants represented by assigned and private counsel were not significantly
different.

Table 5. Comparing types of defense counsel, by felony case processing outcomes, in the nation's
75 most populous counties, 2004 & 2006
Felony case
processing outcomes
Adjudication outcomes
Convicted
Not convicted
Other outcome

Felony defendants represented by Public defender Assigned counsel Private attorney
73 %
22
6

78 %
14
8

72 %
23
6

Conviction level
Felony conviction
Misdemeanor conviction

88 %
12

91 %
9

85 %
15

Type of conviction
Guilty plea
Bench trial
Jury trial

97 %
1
2

98 %
1
2

97 %
2
2

Most serious sentence
Incarceration
Prison
Jail

74 %
32
42

78 %
46
32

65 %
29
36

Non - incarceration
Probation
Other

26 %
23
3

22 %
20
2

35 %
30
5

23 mths.

35 mths.

31 mths.

Mean sentence length
(in months)

                                                            
18

Confidence intervals show public defenders garnering significantly shorter sentences for their clients than those
with private attorneys or assigned counsel.

21
 

So far, this article has provided some interesting findings concerning whether defendants
differ in several key characteristics across the various types of defense counsel and the possible
relationships between defense counsel types and case outcomes. In terms of defendant
characteristics, these findings show that defendants represented by assigned counsel and public
defenders have remarkably similar characteristics. In general, defendants receiving legal
representation through these two forms of indigent counsel are charged with relatively
comparable crimes and have similar criminal histories and demographic characteristics. In
comparison, defendants with the means to hire their own attorneys are exemplified by different
attributes compared to their indigent counterparts. These defendants tend to have less serious
criminal backgrounds and are charged with an array of offenses both more and less serious
compared to their contemporaries with indigent counsel. For example, private attorneys
represented a greater proportion of defendants charged with sexual and drug trafficking crimes
than those with assigned counsel or public defenders. However, these attorneys also provided
legal advocacy to more defendants charged with less serious public-order offenses compared to
defendants who could not afford to hire their own attorneys. Lastly, private attorneys represented
minorities less frequently than public defenders or assigned counsel.
While these findings offer some intriguing insights into who gets what types of defense
counsel, their overall repercussions are limited by the nature of the SCPS data. There are a
variety of socio-economic factors that could influence the distribution of defense counsel types
over felony defendants including income levels, employment status, community ties, and
residential stability that for reasons related to cost and accessibility are currently not collected in
SCPS. Nevertheless, these findings suggest that defendants with public defenders and assigned

22
 

counsel have relatively similar attributes and that the major differences occur when comparing
defendants with indigent counsel to those with private attorneys.
The similarities between assigned counsel and public defenders, however, do not carry
over when examining case outcomes. Here, the descriptive analysis shows defendants with
assigned counsel receiving outcomes that, on the whole, are less favorable compared to
defendants with public defenders or private attorneys. In general, defendants with assigned
counsel are more likely to get convicted and sentenced to prison than their equivalents who are
represented by public defenders or who have the means to hire their own attorneys. Moreover,
defendants with assigned counsel were sentenced to longer periods of confinement than those
with public defenders. Another finding concerned the underwhelming evidence in support of the
proposition that private attorneys secure better outcomes for their clients. Overall, the descriptive
section showed that defendants who hired their own attorneys were just as likely to get convicted
and actually received longer sentences compared to defendants represented by public defenders.
The one area in which private attorneys seemed to be doing better involved the decision by
courts to incarcerate defendants. The descriptive analysis found defendants with private attorneys
being incarcerated less frequently compared to their counterparts with indigent counsel.
In conclusion, these findings suggest that indigent defendants who are represented by
assigned counsel are receiving less favorable outcomes compared those with public defenders or
private attorneys. They also imply that hiring an attorney does not automatically guarantee
superior results; although, there is some evidence that private attorneys are keeping their clients
out of prison or jail to a greater extent than indigent counsel. Although illuminating, these results
do not attempt to control for other factors that could influence these key outcomes such as
offense charge severity or prior criminal history. Therefore, the next part of this article applies

23
 

multivariate techniques to examine the relationship between defense counsel type and
adjudication and sentencing outcomes.
Modeling conviction, incarceration, and sentence length outcomes by defense counsel type
In this section, multivariate statistical techniques are used to further refine our
understanding of the relationship between defense counsel type and case outcomes. Multivariate
analysis can help us disentangle the effects of defense counsel type from other factors such as
criminal history or offense charge severity that could influence adjudication and sentencing
outcomes. For example, if the multivariate analysis shows that defendants with assigned counsel
are still more likely to be convicted or sentence to prison than defendants with public defenders,
net of other controls, that would provide more confidence of the extant findings and conclusions.
The multivariate analysis proceeds in several parts. First, the dependent and independent
variables used in the statistical models are detailed. Next, the discussion highlights the technical
issues associated with modeling adjudication and sentencing outcomes. Model results are then
explicated and conclusions offered regarding the model’s overall findings.
Dependent variables
There are four major dependent variables analyzed in the multivariate section including
conviction, incarceration, prison, and sentence length. The conviction variable examines whether
a defendant charged with a felony was eventually convicted of either a felony or misdemeanor,
and the incarceration and prison variables measure whether a convicted defendant was
incarcerated in a county jail or state prison facility. The incarceration variable has a broader
scope because it tracks convicted defendants receiving prison or jail sentences, while the prison
variable identifies only those convicted defendants sentenced to state prison. Unlike the previous
three dependent variables that have dichotomous outcomes, sentence length is a continuous

24
 

variable measuring the length in months of both prison or jail terms combined. The dependent
variables of conviction, incarceration, prison, and sentence length are typically used in research
examining case processing outcomes among different types of defense counsel (Roach, 2010;
Iyengar, 2007; Hanson, Ostrom, Hewitt, and Lomvardias, 1992; Feeney and Jackson, 19901991).
Independent variables
The primary independent variable of interest is type of defense counsel which includes
defendants represented by public defenders, private attorneys, or assigned counsel. In the
regression models, public defenders represent the reference category meaning that the case
processing outcomes of defendants with private attorneys or assigned counsel are compared to
their counterparts with public defender representation. Other independent variables include
extra-legal and legal covariates measuring defendant demographics, most serious arrest or
conviction charges, criminal justice status and history, monetary bond amounts, case processing
time, and type of conviction. Along with attorney type, these extra-legal and legal covariates
have been found to be significantly associated with various court outcomes related to conviction
and sentencing. Moreover, several researchers have concluded that these variables constitute key
factors in court adjudication and sentencing decisions in the context of the SCPS data file (Piehl
and Bushway, 2007; Steffensmeier and Demuth, 2006; Demuth and Steffensmeier, 2004;
Weidner, Frase, and Pardoe, 2004).
The extra-legal factors in the SCPS data file measure a defendant’s gender, race/ethnicity,
and age. Gender is a single dummy variable. Race/ethnicity is categorized into three dummy
variables including White non-Hispanic, which serves as the reference category, Black nonHispanic, and Hispanic. The defendant’s age is continuous variable; however, an age-squared

25
 

term has been added to the model because prior research has shown age to have a non-linear
relationship with court sentencing decisions (Steffensmeier and Demuth, 2006; Demuth and
Steffensmeier, 2004; Steffensmeier, Krammer, and Ulmer, 1995). 19
Among the legal factors, prior research has shown offense severity and criminal history
accounting for the most important predictors in regression models examining adjudication and
sentencing decisions (Steffensmeier and Demuth, 2006; Demuth and Steffensmeier, 2004).
Offense severity is measured with 13 dummy variables representing either the most serious arrest
charges, for models examining the likelihood of conviction, or the most serious conviction
charges, for models analyzing factors related to incarceration, prison, and sentence length
outcomes. Specific offense types measured include drug possession, which serves as the
reference category, murder, rape, robbery, assault, burglary, larceny, motor vehicle theft, other
property, drug trafficking, weapons, driving/public-order, and misdemeanor 20 offenses. In terms
of measuring criminal history, two covariates were employed in the regression models. The first
is a dummy variable representing defendants with a criminal justice status (e.g., on pretrial
release, probation, parole, or other criminal justice status) at the time of arrest. The second
criminal history factor contains three dummy variables quantifying a defendant’s criminal
conviction history. These include no conviction history, which serves as the reference category,
prior felony conviction, and prior misdemeanor conviction. 21
The remaining independent variables measures bail amounts set by the court, time from
arrest to adjudication, and type of conviction. Monetary bond serves as an important determinant
                                                            
19

The age-squared term was also mean centered in order to avoid multicollinearity issues associated with using an
age and age-squared term (Steffensmeier and Demuth, 2006).
20
Misdemeanor offenses are included only in the incarceration, prison, and sentence length models and not the
initial models examining likelihood of conviction.
21
It should be noted that SCPS contains other potential criminal history covariates including prior arrest and
incarceration history. Preliminary models revealed high correlation levels between prior arrest, conviction, and
incarceration history and that the most parsimonious models were ones which excluded some of these criminal
history factors.

26
 

of pretrial release decisions with higher bond amounts associated with increased likelihoods of
pretrial detention. Monetary bond also provides an indirect measure of case seriousness because
more serious cases tend to garner higher bond amounts (Cohen and Reaves, 2007). Monetary
bond is measured through five dummy variables with no bond serving as the reference category,
and bond amounts of $1-$9,999; $10,000-$24,999; $25,000-$49,999; and $50,000 or more 22
providing measures of escalating monetary bail. The case processing time variable measures the
number of days from arrest to case adjudication for each felony defendant. This variable has
been logged transformed so that the regression assumptions of linearity and constant variance are
not violated in the models. The final covariates include three dummy variables detailing the
mode of conviction which includes the reference category of guilty plea, bench trial conviction,
and jury trial conviction. Numerous adjudication and sentencing studies have demonstrated that
the defendants convicted through trial receive harsher sentences compared to their counterparts
who plead guilty (Piehl and Bushway, 2007; Steffensmeier and Demuth, 2006; Demuth and
Steffensmeier, 2004). The various legal and extra-legal independent variables described above
are also detailed in table 6. 23

                                                            
22

The $50,000 or more variable includes cases in which the court did not set bond or refused to the release the
defendant under any circumstances.
23
Several covariates described above measure relatively similar attributes implicating multicollinearity issues.
Variance inflation factors (VIF) were calculated to test for potential multicollinearity in the models. The VIF
calculations revealed that multicollinearity was not an issue for these analyzes.

27
 

Table 6. Profile of variables utilized in conviction, incarceration, and sentence length models
Variables in model
Type of attorney
Public defender
Private attorney
Assigned counsel
Gender
Female defendant
Race/Hispanic origin
White
Black, non-Hispanic
Hispanic, any race
Age at arrest
Age
Age-Squared
Most serious arrest/
conviction offense
Drug possession
Murder
Rape
Robbery
Assault
Burglary
Larceny
Motor vehicle theft
Other property
Drug trafficking
Weapons
Driving or public order
Misdemeanor
On probation, parole,
or other status at arrest
Most serious
prior conviction
No prior conviction
Felony
Misdemeanor
Monetary bail amounts
No bail set
$1 - $9,999
$10,000 - $24,999
$25,000 - $49,999
$50,000 or more
Time from arrest to
adjudication
Type of conviction
Guilty plea
Bench trial
Jury trial

Coding
0 = Reference category
1 = Private attorney
1 = Assigned counsel
1 = Female defendant
0 = Reference category
1 = Black defendant
1 = Hispanic defendant
Age at arrest
Age at arrest (mean centered and squared)

0 = Reference category
1 = Murder
1 = Rape
1 = Robbery
1 = Assault
1 = Burglary
1 = Larceny
1 = Motor vehicle theft
1 = Other property offenses
1 = Drug trafficking
1 = Weapons
1 = Driving or other public-order
1 = Misdemeanor
1 = On probation, parole, or
criminal justice status at arrest

0 = Reference category
1 = Prior felony conviction
1 = Prior misdemeanor conviction
0 = Reference category
1 = Bail amounts $1-$9,999
1 = Bail amounts $10,000-$24,999
1 = Bail amounts $25,000-$49,999
1 = Bail amounts $50,000 or more or no bond set
Number of months from
arrest to adjudication (logged transformed)
0 = Reference category
1 = Bench trial
1 = Jury trial

Modeling conviction, incarceration, prison, and sentence length outcomes
Binary probit regression models were used to estimate the probability of being convicted
and sentenced to incarceration or prison and ordinary least squares (OLS) regression models
were employed to analyze sentence length. Binary probit regression is one widely accepted
28
 

method for analyzing the effects of multiple independent factors on dichotomous or binomial
outcomes such as whether defendants charged with felony offenses are convicted, sentenced to
incarceration, or sent to state prison, while OLS regression is a commonly employed technique
for examining the effects of multiple factors on a continuous dependent variable such as sentence
length (Long and Freese, 2006; Pardoe, 2006).
There were several technical issues associated with modeling court adjudication and
sentencing outcomes. One primary issue concerned the sampling structure of the data being
analyzed. To reiterate, the SCPS data are drawn from a random sample of days in 40 counties
and are weighted to represent cases processed in the 75 most populous counties during the month
of May. When the regressions utilize these weighted data, the large number of weighted cases
might result in statistical significance for nearly all the variables in the model. Finite population
correction adjustments were used to adjust for the weighted structure of the SCPS data by
accounting for the probability of a county and case appearing in the sample. In addition, the
county based nature of SCPS potentially could violate the independence of observation
assumptions inherent in regression modeling. In other words, the fact that the SCPS data hail
from different counties means that important differences in local legal culture and sentencing
frameworks need to be accounted for in these models. Hence, the standard errors in these models
were further adjusted in order to take into account the clustering of cases by their primary
sampling units (e.g., counties). Clustering standard errors by their primary sampling units allows
for the models to take into account the effects of unmeasured county characteristics such as
differences in sentencing systems, criminal laws, and localized culture of criminal case
processing.

29
 

Another issue involved sample selection bias. By focusing on incarceration and sentence
length outcomes, in addition to examining the conviction decision, this article analyzes subsamples of the SCPS cohort. The first sub-sample is created for the incarceration analysis which
excludes defendants who were not convicted, while the second sub-sample is generated for the
sentence length decision, which excludes convicted defendants who were not incarcerated. Two
statistical techniques were employed to control for this narrowing funnel of defendants moving
from conviction to incarceration and sentencing. 24 The initial statistical approach involved
employing a probit model with Heckman selection to model incarceration and prison decisions.
A probit model with Heckman selection is one commonly used technique to model outcomes
such as incarceration and prison where a segment of defendants (e.g., those who are not
convicted) are excluded (Cameron and Trivedi, 2009). The second approach utilized an OLS
Heckman selection model to examine sentence length outcomes. The OLS Heckman selection
technique is another accepted approach for modeling a continuous outcome such as sentence
length where a segment of defendants (e.g., convicted defendants who were not incarcerated)
have been excluded (Cameron and Trivedi, 2009; Heckman, 1976). The utilization of these
selection models helped correct for biases associated with selecting convicted defendants for the
purposes of modeling incarceration decisions and selecting incarcerated defendants for the
purposes of modeling sentence length outcomes. 25
For the OLS Heckman regression, there were some additional adjustments that should be
noted. First, the dependent variable sentence length was transformed using a natural log because
                                                            
24

It should be noted that a selection effect also occurs from arrest to the initial charging decision. Unfortunately,
SCPS is unable to determine how many arrested defendants were charged with a felony. The starting point for SCPS
is the moment a prosecutor files a felony charge; hence, these data cannot be used to model prosecutor charging
decisions from the point of arrest.
25
 The Stata commands “heckprob” and “heckman” were utilized to model sentencing and incarceration decisions.
Since Heckman relies on a two-stage modeling approach, the selection equation should have at least one variable
that is not in the primary equation. In this article, the bail amounts have been placed in the primary equations but
excluded from the selection equations.

30
 

of the non-normal distribution of the regression residuals. Because interpreting a log transformed
sentence can be difficult, a proportional change in the log transformed sentence was calculated
by exponentiating it and then subtracting one (Pardoe, 2006). This calculation provided an
expected proportional change in the log transformed sentence length for defendants represented
by private attorneys and assigned counsel compared to public defenders net of other controls in
the SCPS data file. Also, before the variable sentence length was log transformed, life sentences
were recoded to 720 months (60 years) and the influence of outliers on the dependent variable
sentence length was checked by examining the studentized residuals for each of the regression
models. For the most part, these models showed studentized residuals within appropriate
parameters, however, there were 392 cases - 1% of convicted defendants – with sentences of
less than a week that were excluded because their studentized residuals scores were below -3
indicating that these cases were outliers.
For each dependent variable, regression models were run for all felony defendants and
then separately for felony defendants charged with or convicted of violent, property, drug, or
public-order offenses. Running separate regression models allows for an examination of whether
defendants represented by private attorneys or assigned counsel are receiving more or less
advantageous outcomes for certain offense categories. For example, private attorneys may be
doing a better job representing felony defendants charged with violent offenses compared to their
counterparts charged with property or drug crimes.
Model findings
Table 7 contains the results of the probit and Heckman models examining the likelihood
of conviction, incarceration, and prison by defense attorney type for the combined populations of
SCPS defendants. All three multivariate models show defendants with private attorneys

31
 

receiving outcomes that were not significantly different compared to their counterparts with
public defenders. In other words, defendants who hired a private attorney were just as likely to
get convicted, incarcerated, or sentenced to prison compared to indigent defendants with public
defenders. These results are somewhat at odds with the descriptive section which showed
defendants with private attorneys being sentenced to incarceration less frequently than those with
public defenders. While the probit model does show the Z-score for incarceration being reduced
by .118 for defendants represented by private attorneys, that reduction was not statistically
significant when other factors in the model were taken into account.
A more interesting finding involves the outcomes for defendants with assigned counsel
representation. According to the probit models, defendants with assigned counsel were
significantly more likely to be convicted and sentenced to prison, net of controls, compared to
defendants represented by public defenders. For example, the Z-scores measuring the likelihood
of conviction and state imprisonment for defendants with assigned counsel were .232 and .305
higher, respectively, than for defendants with public defenders. The only outcome in which
assigned counsel garnered similar results to public defenders was for incarceration. Defendants
represented by assigned counsel were just as likely to receive some form of incarceration as
defendants with public defenders. Despite similarities in incarceration outcomes, the combined
models provide evidence that defendants represented by assigned counsel received significantly
worse outcomes in terms of being convicted and sentenced to prison compared to their
counterparts who were represented by public defenders.
Before delving into the offense specific models, it is worth noting the impact of other
covariates on the likelihood of conviction, incarceration, and prison. Concerning demographics,
females were less likely to be incarcerated and sentenced to prison than males and for

32
 

race/ethnicity, blacks had a greater likelihood of being sentenced to incarceration or prison than
whites. Hispanics were just as likely to get convicted and sentenced to prison as whites;
however, these defendants had a higher likelihood of receiving an incarceration sentence. Age
did not have an impact on conviction; yet, the models show age having a curvilinear influence on
the likelihood of receiving a prison sentence.
The influence of the most serious offense or conviction charges depended upon the
outcome being examined. For the conviction model, several of the more serious offense charges
such as rape or assault were associated with a lower likelihood of conviction compared to drug
possession defendants. In comparison, the incarceration model witnessed all defendants having a
greater likelihood of incarceration than defendants convicted of drug possession. Lastly, the
probability of being sentenced to prison was higher for the more serious offense categories such
as murder, rape, robbery, assault, or burglary than for defendants convicted of drug possession.
Among the remaining variables, criminal history and monetary bonds were associated
with higher likelihoods of defendants being convicted, incarcerated, and sentenced to prison,
while case processing time and conviction types had mixed effects on these outcomes.
Specifically, defendants with a prior felony conviction and higher bond amounts were
significantly more likely to be convicted, incarcerated, and sent to state prison net of controls.
Longer case processing times were associated with increased likelihoods of conviction and
prison but not incarceration. Lastly, defendants convicted by bench or jury trial were
significantly more likely to receive a state prison sentence compared to their counterparts who
plead guilty. A trial conviction, however, was not associated with increased probabilities of
receiving any form of incarceration.

33
 

Table 7. Probit regression models comparing types of defense counsel, by conviction, incarceration (in/out), and prison
(prison vs. jail or other non-incarceration sentence) outcomes, in the nation's 75 most populous counties, 2004 & 2006

Variables in model
Type of attorney
Private attorney
Assigned counsel
Gender
Female defendant
Race/Hispanic origin
Black, non-Hispanic
Hispanic, any race
Age at arrest
Age
Age-Squared
Most serious arrest/
conviction offense
Murder
Rape
Robbery
Assault
Burglary
Larceny
Motor vehicle theft
Other property
Drug trafficking
Weapons
Driving or public order
Misdemeanor

Probit model of
conviction outcome
Coefficient
& standard errors

Probit model with
sample selection of
incarceration outcome
Coefficient
& standard errors

-0.063
(0.034)
0.232 ***
(0.068)

-0.118
(0.068)
0.074
(0.118)

0.034
(0.060)
0.305 **
(0.101)

0.029
(0.034)

-0.137 ***
(0.029)

-0.220 ***
(0.024)

-0.144 ***
(0.043)
0.013
(0.037)

0.109 **
(0.037)
0.202 ***
(0.040)

0.153 ***
(0.045)
0.013
(0.034)

0.000
(0.001)
0.000
(0.000)

0.004 *
(0.002)
0.000 **
(0.000)

0.006 ***
(0.001)
-0.001 ***
(0.000)

-0.253
(0.170)
-0.260
(0.132)
0.019
(0.125)
-0.304
(0.111)
0.243
(0.110)
0.241
(0.125)
0.242
(0.118)
0.088
(0.098)
0.258
(0.102)
0.129
(0.113)
0.279
(0.098)
---

1.311
(0.411)
0.926
(0.198)
0.594
(0.114)
0.526
(0.100)
0.552
(0.118)
0.342
(0.114)
0.740
(0.137)
0.311
(0.085)
0.368
(0.113)
0.474
(0.149)
0.293
(0.109)
0.178
(0.067)

1.501
(0.347)
0.685
(0.134)
0.697
(0.086)
0.266
(0.068)
0.204
(0.076)
0.065
(0.082)
0.185
(0.082)
0.020
(0.080)
0.067
(0.072)
0.306
(0.075)
0.059
(0.085)
-1.020
(0.094)

*

**
*

*

**

**

Continued on next page

34
 

Probit model with
sample selection of
prison outcome
Coefficient
& standard errors

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

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

*

***

***

Continued from previous page
Table 7. Probit regression models comparing types of defense counsel, by conviction, incarceration (in/out), and prison
(prison vs. jail or other non-incarceration sentence) outcomes, in the nation's 75 most populous counties, 2004 & 2006

Variables in model
On probation, parole,
or other status at arrest
Most serious
prior conviction
Felony
Misdemeanor
Monetary bail amounts
$1 - $9,999
$10,000 - $24,999
$25,000 - $49,999
$50,000 or more
Time from arrest to
adjudication - natural log
Type of conviction
Bench trial
Jury trial
Constant

Probit model of
conviction outcome
Coefficient
& s tandard errors

Probit model with
sample selection of
incarceration outcom e
Coefficient
& standard errors

Probit model with
sample selection of
pris on outcome
Coefficient
& s tandard errors

0.116 **
(0.040)

0.017
(0.053)

0.048
(0.041)

0.286 ***
(0.052)
0.191 ***
(0.043)

0.375 ***
(0.038)
0.143 **
(0.044)

0.401 ***
(0.055)
-0.137 **
(0.043)

0.144
(0.046)
0.371
(0.050)
0.563
(0.099)
0.608
(0.079)

0.115
(0.058)
0.279
(0.076)
0.418
(0.071)
0.621
(0.074)

0.107
(0.048)
0.225
(0.063)
0.355
(0.068)
0.621
(0.059)

**
***
***
***

0.100 **
(0.032)
---

*
***
***
***

*
***
***
***

-0.049
(0.026)

0.063 *
(0.027)

-0.055
(0.082)
0.100
(0.140)

0.278 **
(0.090)
0.352 **
(0.118)

-0.300

0.161
-1.047
(0.155)
(0.175)
Rho
-0.777 ***
-0.807 ***
-(0.053)
(0.027)
Number of observations
21,048
19,341
19,342
Population size
77,538
71,013
71,016
Notes: Model (1) includes probit regress ion of conviction outcome; model (2) includes Heckman model
of incarceration outcome, and model (3) includes Heckman model of pris on outcome.
The Heckman model includes equation accounting for the selection effect of conviction on
the incarceration and prison outcome. As teris ks indicate category difference
from the following significance levels : *>=.05, **>=.01, ***>=.001.
Models adjusted to account for clustering of s tandard errors at the county level
and s tratified s ampling pattern. Finite population correction adjustments als o
used to account for probability of county and cas e appearing in sample.
Model standard errors in parentheses. Overall model fit s tatis tics not provided through
s urvey command in Stata. Reference group for defens e counsel type is public defender.

Table 8 provides separate model results of defense counsel’s impact for defendants
charged with or convicted of violent, property, drug, or public-order offenses. Overall, the
models show defendants with private attorneys receiving similar outcomes compared to their

35
 

counterparts with public defenders. For example, although the probit models indicate reduced Zscores among violent felony defendants represented by private attorneys in terms of their
likelihood of being convicted, incarcerated, and sentenced to prison, the reduction of these Zscores was not statistically significant when other factors in the models were controlled. A
similar pattern holds when comparing the outcomes for drug offenders with private attorneys and
public defenders.
The only offense categories where having a private attorney made some difference was
for defendants charged with or convicted of property or public-order offenses. In the property
category, defendants with retained counsel were less likely to be convicted than defendants with
public defenders. Among public-order defendants, having a private attorney reduced the Z-score
probability of receiving an incarceration sentence by .329. These effects, however, were not
incredibly strong or consistent. For example, property defendants with private attorneys were just
as likely to receive incarceration for state prison as their equivalents with public defenders.
Among public-order defendants, the likelihood of conviction or state prison was essentially the
same for defendants represented by public defenders and private attorneys.
While the likelihood of conviction, incarceration, and state prison was not significantly
different for defendants represented by private attorneys and public defenders across the major
SCPS offense categories, the same pattern does not hold for assigned counsel. Indigent
defendants represented by assigned counsel received worse case outcomes, particularly for
property or drug crimes, than their public defender counterparts. For example, property and drug
defendants with assigned counsel had statistically higher Z-scores in terms of their likelihood of
being convicted and sentenced to state prison than property or drug defendants with public
defenders. The association of assigned counsel with inferior results, however, is not has

36
 

consistent for violent or public-order defendants. Violent defendants represented by assigned
counsel were more likely to be convicted but had similar Z-scores in terms of being incarcerated
or sentenced to state prison compared to those with public defenders. Public-order defendants
with assigned counsel did not manifest any statistically appreciable differences compared to
public defenders regarding the likelihood of conviction, incarceration, or state imprisonment.
Table 8. Probit regression models comparing types of defense counsel for specific offenses, by conviction, incarceration (in/out) and prison (prison
vs. jail or other non-incarceration sentence) outcomes, in the nation's 75 most populous counties, 2004 & 2006

Most serious offense
by type of counsel
Violent
Private attorney
Assigned counsel
Property
Private attorney
Assigned counsel
Drug
Private attorney
Assigned counsel

Probit model of
conviction outcome
Coefficient
& standard errors

Probit model with
sample selection of
incarceration outcome
Coefficient
& standard errors

Probit model with
sample selection of
prison outcome
Coefficient
& standard errors

-0.061
(0.056)
0.186 **
(0.061)

-0.121
(0.075)
0.024
(0.096)

-0.122
(0.066)
0.079
(0.064)

-0.123 *
(0.057)
0.215 **
(0.077)

-0.074
(0.079)
-0.057
(0.108)

-0.006
(0.090)
0.242 **
(0.091)

-0.049
(0.062)
0.299 **
(0.107)

-0.068
(0.084)
0.270
(0.152)

0.130
(0.084)
0.510 ***
(0.133)

Public-order
Private attorney

0.023
-0.329 *
-0.106
(0.071)
(0.137)
(0.078)
Assigned counsel
0.070
-0.176
0.079
(0.142)
(0.211)
(0.153)
Notes: Model (1) includes probit regression of conviction outcome; model (2) includes Heckman model
of incarceration outcome, and model (3) includes Heckman model of prison outcome.
The Heckman model includes equation accounting for the selection effect of conviction on
the incarceration and prison outcome. Asterisks indicate category difference
from the following significance levels: *>=.05, **>=.01, ***>=.001.
Models adjusted to account for clustering of standard errors at the county level
and stratified sampling pattern. Finite population correction adjustments also
used to account for probability of county and case appearing in sample.
All variables in table 7 are included in the regression models.
Model standard errors in parentheses.
Overall model fit statistics not provided through survey command in Stata.
Reference group for defense counsel type is public defender.

37
 

Table 9 highlights results of the OLS Heckman regression model examining sentence
length by defense counsel type. Overall, the analysis yielded patterns of association between
defense counsel type and sentence length that were somewhat similar to that of the bivariate
results. Although defendants represented by private attorneys received log transformed sentences
that were 11% proportionally higher compared to defendants using public defenders, the
differences in sentence length between defendants represented by these forms of defense counsel
were not statistically different. In a finding mirroring the bivariate results, defendants with
assigned counsel received log transformed sentences that were 26% proportionately longer than
defendants with public defenders. The differences in sentence length between defendants with
assigned counsel and public defenders were statistically significant.
Before examining the relationship between sentence length and defense counsel type for
the specific SCPS offenses, the influence of other covariates on sentence length should be noted.
In regards to demographics, females received shorter sentences than males, and for
race/ethnicity, blacks were sentenced to incarceration terms that were no different than whites.
Hispanics, however, had log transformed sentences that were 15% shorter than whites. The
model also shows age having a curvilinear effect with increases in age associated with longer log
transformed sentences but with older defendants receiving shorter sentences.
The influence of the conviction charges on sentence length depended upon their severity.
In general, higher sentences were associated with the more serious offense categories such as
murder, rape, robbery, assault, burglary, or drug trafficking, while shorter sentences were
correlated with less serious offenses. The remaining variables including criminal history,
monetary bonds, case processing time, and conviction type were associated with longer
sentences. Specifically, defendants with a prior felony conviction and higher bond amounts were

38
 

significantly more likely to receive longer log transformed sentences net of controls. Longer case
processing times were associated with increases in sentence length. Lastly, defendants convicted
by bench or jury trial were significantly more likely to receive a longer incarceration sentence
compared to their counterparts who plead guilty.

Table 9. Heckman regression model comparing types of defense counsel, by sentence length outcomes,
in the nation's 75 most populous counties, 2004 & 2006

Variables in model
Type of attorney
Private attorney

Coefficient &
standard error

Assigned counsel
Gender
Female defendant
Race/Hispanic origin
Black, non-Hispanic
Hispanic, any race
Age at arrest
Age
Age-Squared
Most serious
conviction offense
Murder
Rape
Robbery
Assault
Burglary
Larceny
Motor vehicle theft
Other property
Drug trafficking
Weapons
Driving or public order
Misdemeanor

0.100
(0.057)
0.230
(0.070)

Proportional
change in
log sentence/a
10.6 %
25.9 **

-0.215
(0.037)

-19.4 %***

0.055
(0.042)
-0.157
(0.037)

5.7 %

0.005
(0.002)
-0.000
(0.000)

2.142
(0.198)
1.130
(0.158)
1.042
(0.114)
0.237
(0.100)
0.275
(0.112)
0.034
(0.088)
0.089
(0.090)
0.073
(0.088)
0.197
(0.087)
0.171
(0.176)
0.007
(0.082)
-0.690
(0.078)

-14.5 ***

0.5 %**
0.0 ***

751.3 %***
209.6 ***
183.4 ***
26.7 *
31.7 *
3.4
9.3
7.5
21.8 *
18.7
0.7
-49.8 ***

Continued on next page

39
 

Continued from previous page
Table 9. Heckman regression model comparing types of defense counsel, by sentence length outcomes,
in the nation's 75 most populous counties, 2004 & 2006

Variables in model
On probation, parole,
or other status at arrest
Most serious
prior conviction
Felony
Misdemeanor
Monetary bail amounts
$1 - $9,999
$10,000 - $24,999
$25,000 - $49,999
$50,000 or more
Time from arrest to
adjudication
Type of conviction
Bench trial
Jury trial
Constant
Rho

Coefficient &
standard error

Proportional
change in
log sentence/a

0.214
(0.037)

23.9 %***

0.289
(0.078)
-0.165
(0.046)

33.5 %***

-0.078
(0.086)
0.110
(0.088)
0.335
(0.108)
0.583
(0.111)

-15.2 ***

-7.5 %
11.6
39.8 **
79.1 ***

0.270
(0.015)

31.0 %***

0.484
(0.091)
0.834
(0.104)

62.3 %***
130.3 ***

0.865
-0.556 ***

Number of observations

13,690

Population size
49,846
Notes: Model includes Heckman model of sentence length outcome.
The Heckman model includes equation accounting for the selection effect of
incarceration on the sentence length outcome. Asterisks indicate category difference
from the following significance levels: *>=.05, **>=.01, ***>=.001.
Models adjusted to account for clustering of standard errors at the county level
and stratified sampling pattern. Finite population correction adjustments also
used to account for probability of county and case appearing in sample.
Model standard errors in parentheses.
Overall model fit statistics not provided through Heckman survey command in Stata.
Reference group for defense counsel type is public defender.
a/Dependent variable sentence length was transformed using a natural log because of non-normal distribution
of regression residuals. Interpreting log transformed sentences involves exponentiating it and
and subtracting one. This calculation provides an expected proportional change in sentence length
for defendants represented by private attorneys and assigned counsel compared to public defenders.

Table 10 provides separate model results of defense counsel’s impact on the log
transformed sentence for defendants convicted of violent, property, drug, or public-order crimes.
With the exception of the drug offense category, defendants convicted of violent, property, or
public-order offenses received similar sentences regardless of whether they were represented by
a private attorney or public defender. Only the drug offense category manifested sentences that

40
 

were appreciably different between private attorneys and public defenders; however, for this
offense category, convicted defendants with private counsel were sentenced to periods of
confinement 37% longer than their counterparts with public defenders.
In a finding paralleling the bivariate analysis, longer sentences were associated with
defendants represented by assigned counsel compared to their public defender equivalents. For
example, defendants convicted of violent crimes received log transformed sentences that were
33% higher if they had assigned counsel than if they were represented by public defenders.
Among convicted property defendants, the log transformed sentences were 22% higher for
defendants with assigned as opposed to public defender representation. Defendants convicted of
public-order offenses received log transformed sentences that were 40% higher if they received
legal assistance through an assigned counsel rather than public defender program. Only in the
drug offense category were sentences between defendants with assigned counsel and public
defender representation not significantly different.

41
 

Table 10. Heckman regression models comparing types of defense counsel for specific offenses,
by sentence length outcomes, in the nation's 75 most populous counties, 2004 & 2006

Most serious offense
by type of counsel
Violent
Private attorney
Assigned counsel
Property
Private attorney
Assigned counsel
Drug
Private attorney
Assigned counsel

Coefficient &
standard error
0.004
(0.085)
0.283
(0.089)
-0.080
(0.066)
0.203
(0.068)
0.318
(0.132)
0.081
(0.129)

Proportional
change in
log sentence
0.4 %
32.7 ***

-7.7 %
22.5 ***

37.4 %***
8.4

Public-order
Private attorney

0.014
1.4 %
(0.149)
Assigned counsel
0.337
40.0 **
(0.129)
Notes: Model includes Heckman models of sentence length outcome.
The Heckman model includes equation accounting for the selection effect of
incarceration on the sentence length outcome. Asterisks indicate category difference
from the following significance levels: *>=.05, **>=.01, ***>=.001.
Models adjusted to account for clustering of standard errors at the county level
and stratified sampling pattern. Finite population correction adjustments also
used to account for probability of county and case appearing in sample.
Model standard errors in parentheses.
All variables in table 9 are included in the regression models.
Overall model fit statistics not provided through Heckman survey command in Stata.
Reference group for defense counsel type is public defender.
a/Dependent variable sentence length was transformed using a natural log because of non-normal distribution
of regression residuals. Interpreting log transformed sentences involves exponentiating it and
and subtracting one. This calculation provides an expected proportional change in sentence length
for defendants represented by private attorneys and assigned counsel compared to public defenders.

Discussion of model results and primary findings
This article sought to examine the role of defense counsel in felony case processing and
investigate whether key case outcomes are similar or different by defense counsel type. For the
most part, the extant analysis does not support the contention that utilizing a private attorney is
equated with superior outcomes. Overall, the models show that defendants who hire private
attorneys were just as likely to be convicted, incarcerated, and sentenced to prison as their
counterparts with public defenders. For those defendants sentenced to incarceration, the
sentences imposed were not statistically different between defendants represented by public
42
 

defenders or retained counsel. Even though the offense specific models produced some minor
exceptions to these findings, they were not consistent or strong enough to negate the general
pattern of negligible differences between defendants represented by public defenders and private
attorneys.
That the findings showed defendants represented by public defenders and private
attorneys receiving similar outcomes should not be too surprising. As discussed in the literature,
the majority of studies show defendants with public defenders receiving adjudication,
incarceration, and sentencing outcomes that were relatively similar compared to those with
private attorneys (Hartley, Miller, Sophn, 2010; Hanson, Ostrom, Hewitt, and Lomvardias, 1992;
Feeney and Jackson, 1990-1991). Even studies that found defendants with private attorneys
receiving favorable outcomes in certain contexts are not entirely at odds with the current
research. Similar to the Hanson, Ostrom, Hewitt, and Lomvardias (1992) report, this article also
found defendants represented by private attorneys receiving incarceration less frequently at the
bivariate level. However, when other factors in the SCPS data file are statistically controlled, the
differences in incarceration rates between defendants represented by private attorneys and public
defenders becomes insignificant.
Rather than contradicting the prior research, this study produces findings that are in line
with others evidencing little to no differences in key case outcomes between defendants
represented by private attorneys and public defenders. That this is so can probably be explained
by the levels of professionalism inherent in many public defender offices. In many states and
locales, the establishment of the public defender office has given rise to a group of attorneys with
significant experience, expertise, and specialization in criminal law and procedure. Moreover,
since the defendants being examined are those charged with felony offenses, the public defenders

43
 

assigned to represent these defendants are often required to have significant courtroom training
before moving into the practice of felony defense (Hanson, Ostrom, Hewitt, and Lomvardias,
1992). Lastly, the close working relationships public defenders develop with other courtroom
workgroup actors including prosecutors and judges allows them at least the opportunity to strike
favorable deals for their clients (Hartley, Miller, and Spohn, 2010; Fleming, Nardulli, and
Eisenstein, 1992; Heumann, 1978).
In summary, the competence public defenders show “in their knowledge of court
procedures and practices, their abilities to negotiate the most favorable outcomes for their clients,
and their success in knowing how to achieve favorable outcomes expeditiously” (Hanson,
Ostrom, Hewitt, and Lomvardias, 1992: 105) places them in a position to rival their counterparts
in the private defense bar. Many private attorneys, who engage in criminal defense, have other
practice areas as well and hence, do not have the resources, training, or expertise to practice
criminal defense at levels equaling the public defender’s office (Neubauer and Fradella, 2011).
The fact that very few private attorneys can engage solely in the practice of criminal defense
places them in a position where they can at best provide legal advocacy that equals the public
defender system.
Although this article supports the contention that defendants with private attorneys and
public defenders receive similar outcomes, some caution should be used before concluding that
hiring an attorney makes no difference under any circumstances. As revealed in the bivariate
analysis, defendants retaining private attorneys are appreciably different compared to indigent
defendants. On the whole, they tend to be whiter, have less substantial criminal histories, and are
charged with offenses that are both more and less serious compared to defendants who lack the
resources to hire their own attorneys. While the regression models have attempted to account for

44
 

these factors, there are other unmeasured covariates that, if controlled, could perhaps reveal a
different story. For example, it is possible that attorney experience has a paramount effect on
case outcomes, and that private attorneys who are able to develop significant expertise and
practice areas in criminal defense, obtain outcomes that exceed those of public defenders.
Unfortunately, SCPS does not measure attorney experience and hence is unable to discern how
the degree of attorney expertise could influence case outcomes. Thus, although this research can
assess attorney effectiveness in overall felony case processing, it cannot tease out how private
attorneys with significant expertise in criminal practice compare to public defenders. Future
research analyzing the role of the private defense bar might want to focus on the issue of attorney
expertise and examine how that factor impacts on case outcomes.
A more interesting, and in some ways troubling, finding concerns the role of assigned
counsel in felony case processing. In general, defendants represented by assigned counsel
received the least favorable outcomes in that they were convicted and sentenced to state prison at
higher rates compared to defendants with public defenders. These defendants also received
longer sentences than those who had public defender representation. Although the offense
specific analyzes did not always find significant associations between assigned counsel and the
case processing outcomes being modeled, for several of these models the likelihood of
conviction and state imprisonment, as well as the length of sentence, were found to be
significantly higher for defendants with assigned counsel representation. The patterns of assigned
defense counsel representation and unfavorable case outcomes held even when the various
factors in the SCPS data file were statistically controlled.
The findings reported in this article diverge somewhat from the prior research on this
topic. While earlier studies found negligible differences in case processing outcomes between

45
 

assigned counsel and public defenders, several more recent studies including Roach (2010) and
Iyengar (2007) noted significant differences between defendants with these differing forms of
public defense. To reiterate, both papers found defendants with assigned counsel being subjected
to appreciably worse outcomes in terms of likelihood of conviction and severity of punishment
compared to defendants with public defenders. That the current research aligns with these studies
provides further empirical evidence calling into question the assigned counsel system as a form
of adequate indigent defense.
While this paper has highlighted correlations between being represented by an assigned
attorney and receiving less favorable case processing outcomes, it is important to note that one
should be careful before using these results to infer causation or conclude that the assigned
counsel system is an inferior form of indigent defense. There are many factors that could
potentially be associated with the case outcomes of conviction, incarceration, and sentence
length that are not measured directly in SCPS including strength of the evidence, levels of
victimization, aggravating factors such as weapon use, or mitigating circumstances including the
defendant’s employment status, family relations, or community ties. Even though this article has
attempted to control for a variety of key covariates such as offense severity, criminal history, bail
status, demographics, and the SCPS sample design, the data cannot be modeled in a way that
controls for every variable associated with these criminal case outcomes.
The inability to account for all factors associated with felony case processing outcomes
opens the possibility of selection biases explaining some of the differences between assigned
counsel and public defenders. For example, many SCPS jurisdictions employ both public
defenders and assigned counsel to represent indigent defendants and in these counties, it is
possible that the public defenders are able to select cases where they are more likely to prevail or

46
 

produce outcomes entailing less severe punishments for their clients. In these jurisdictions, a
selection process where assigned counsel are provided the worst cases in terms of conviction and
punishment probability, rather than inferior legal advocacy, could explain why public defenders
produce better outcomes for their clients. Selection bias could be especially problematic in
jurisdictions where heavy caseloads necessitate the utilization of assigned counsel as a means of
alleviating public defender workloads. Unless random assignment is applied, public defenders
could offload their less favorable cases into the assigned counsel system.
While the possibility of selection bias cannot be fully discounted, an alternative
explanation of the assigned counsel system providing an inadequate form of indigent defense
should also be considered. As noted by several scholars, the assigned counsel system has been
disparaged for appointing attorneys with inadequate skills and qualifications to represent
indigent defendants (Neubauer and Fradella, 2011; Iyengar, 2007; Spangenberg and Beeman,
1995). This line of reasoning has become more prevalent in some recent studies including
Iyengar’s (2007) analysis of assigned counsel in federal courts and Roach’s (2010) examination
of assigned counsel in several SCPS jurisdictions. In addition to comparing case outcomes, both
authors attempted to discern qualification levels of these two forms of indigent counsel. Iyengar
examined the law school attended and bar passage date of public defenders and assigned counsel
in three federal districts, while Roach collected similar information for a group of assigned
attorneys in one SCPS jurisdiction (Franklin, OH). Both authors found evidence supporting the
proposition that assigned counsel have less experience and hail from lower quality law schools
than their public defender counterparts.
Although this article does not examine the issue of attorney qualifications, the fact that
some studies have found differences in attorney quality, as measured by law school attended and

47
 

number of years in practice, provides some evidence supporting the contention that the structure
of the assigned counsel system explains the findings highlighted in this study. Other evidence in
support of this argument hails from the fact that, as measured by SCPS, defendants represented
by assigned counsel and public defenders are markedly similar. Indigent defendants with both
forms of defense counsel tend to be charged with similar offenses and have comparable criminal
histories. Moreover, the multivariate models show that at least in regards to the covariates related
to offense severity, criminal history, demographics, bail amounts, conviction type, and sample
design, the differences in case processing outcomes between assigned counsel and public
defenders are still statistically significant. Whether these findings would change if additional
covariates were brought into models is unfortunately a question that cannot be addressed in the
current research.
Conclusion
This article sought to examine the role of defense counsel in criminal cases and analyze
differences in case outcomes between defendants represented by public defenders, assigned
counsel, and private attorneys. BJS data tracking defendants charged with a felony offense in the
nation’s 75 most populous counties in 2004 and 2006 were employed to address issues including
the types of defense counsel representing felony defendants and the performance of these
attorneys in terms of securing favorable outcomes for their clients. Results from this research
show private attorneys and public defenders securing similar adjudication and sentencing
outcomes for their clients, while assigned counsel generated less favorable outcomes in terms of
likelihood of conviction, state imprisonment, and sentence length.
The negligible differences in case outcomes between defendants with public defenders
and private attorneys should not be too surprising. A plethora of research dating back to the

48
 

1970s shows that hiring a private attorney does not raise the odds of escaping a conviction or
receiving a less harsh sentence. In a sense, this article has produced findings in line with prior
studies on the topic comparing case outcomes among defendants with private attorneys and
public defenders. What this article has not done is examine issues related to how the experience
and qualifications of private attorneys impact on case processing. It is quite possible that private
attorneys who have significant expertise and practice areas in criminal defense are able to secure
favorable outcomes for their clients at rates exceeding the public defense bar. Future research
may want to test the efficacy of private attorneys who are experienced in criminal defense and
compare how these attorneys stack up to public defenders with similar backgrounds.
A more problematic finding concerns the role of assigned counsel in felony criminal
defense. In theory, defendants represented by assigned counsel and public defenders should
receive similar outcomes because they are comparable in terms of the offenses they are charged
with, their criminal backgrounds, and their socio-economic characteristics since both sets of
defendants could not afford a private attorney. The fact that defendants with assigned counsel
receive less favorable outcomes raises the possibility that these attorneys are being assigned
cases that are more likely to result in a conviction and longer sentence compared to their public
defender counterparts. An alternative explanation would be that the assigned counsel system may
be seriously impaired by funding and other organizational issues in its ability to utilize
competent attorneys with sufficient expertise in criminal defense.
Although this article cannot provide enough measures to truly ascertain whether selection
bias or an inadequate assigned counsel system explains these findings, the extant study raises the
possibility of serious problems with the practice of assigning private attorneys to represent
indigent defendants. Many jurisdictions rely on assign counsel to represent indigent defendants

49
 

and this is especially the case for less populous counties or states without centralized state
funded public defender systems. Defendants who are defended by assigned counsel could be
seriously disadvantaged compared to those who are represented by public defenders. The
discrepancy in attorney quality and competence raises issues of fairness and justice in the
nation’s state criminal courts.
It is imperative to stress that more research is needed before concluding that the assigned
counsel system is seriously flawed as a means of providing legal assistance to indigent
defendants. Future research into this area might want to employ more rigorous techniques such
as random assignment to evaluate the effectiveness of assigned counsel. 26 Subsequent research
might also examine the efficacy of different types of assigned counsel systems. For example,
perhaps assigned counsel systems with strict administrative oversight are more effective
compared to instances with little oversight or monitoring. Irrespective of the research techniques
employed, more effort is needed to begin providing nuance and texture to the findings
highlighted in this paper.

                                                            
26

See Huang, Chen, and Lin (2010) for an interesting example of how random assignment was used to compare
public defenders and assigned counsel in the Taiwanese courts.

50
 

References
Beck, J. and R. Shumsky (1997). “A Comparison of Retained and Appointed Counsel in Cases
of Capital Murder,” Law and Human Behavior 21(5): 525-538.
Berk, R. (1983). “An Introduction to Sample Selection Bias in Sociological Data,” American
Sociological Review 48: 386-398.
Blumberg, A. (1966-1967). “The Practice of Law as Confidence Game: Organization Cooptation
of a Profession,” Law and Society Review 1: 15-39.
Cameron, A. and P. Trivedi (2009). Microeconomics Using Stata. College Station, TX: Stata
Press
Champion, D. (1989). “Private Counsels and Public Defenders: A Look at Weak Cases, Prior
Records, and Leniency in Plea Bargaining,” Journal of Criminal Justice 17: 253-263.
Clarke and Koch (1980). “Juvenile Court: Therapy or Crime Control, and Do Lawyers Make A
Difference?” Law and Society Review 14: 263.
Cohen, T. and B. Reaves (2007). Pretrial Release of Felony Defendants in State Courts.
Washington, D.C.: Bureau of Justice Statistics
Cohen, T. and T. Kyckelhahn (2010). Felony Defendants in Large Urban Counties, 2006.
Washington, D.C.: Bureau of Justice Statistics
DeFrances, C. and M. Litras (2000). Indigent Defense Services in Large Counties,
1999. Washington, D.C.: Bureau of Justice Statistics
Demuth, S. and Steffensmeier, D. (2004) “Ethnicity Effects on Sentencing Outcomes in Large
Urban Courts: Comparisons Among White, Black, and Hispanic Defendants,” Social
Science Quarterly 85: 994-1011.
Farole, D. and L. Langton (2010). County-Based and Local Public Defender Offices, 2007.
Washington, D.C.: Bureau of Justice Statistics
Feeley, M. (1992). The Process is the Punishment: Handling Cases in a Lower Criminal Court.
New York, New York: Russell Sage Foundation.
Feeney, F. and P. Jackson (1990-1991). “Public Defenders, Assigned Counsel, Retained
Counsel: Does Type of Criminal Defense Counsel Matter?” Rutgers Law Journal 22:
361-456.
Flemming, R., P. Nardulli, and J. Eisenstein. (1992). The Craft of Justice: Politics and Work in
Criminal Court Communities. Philadelphia, PA: Pennsylvania University Press.

51
 

Gist, N. (1989-1990). “Assigned Counsel: Is the Representation Effective?” Criminal Justice 4:
16-48.
Hanson, R., B. Ostrom, W. Hewitt, and C Lomvardias (1992). Indigent Defenders Get the Job
Done and Done Well. Williamsburg, VA: National Center for State Courts.
Harlow, C. (2000). Defense Counsel in Criminal Cases. Washington, D.C.: Bureau of Justice
Statistics
Hartley, R., H. Miller, and C. Spohn (2010). “Do You Get What You Pay For?: Type of Counsel
and its Effect on Criminal Court Outcomes,” Journal of Criminal Justice 38:10631070.
Hashimoto, E. (2007). “Defending the Right of Self-Representation: An Empirical Look at the
Pro Se Felony Defendant,” North Carolina Law Review 85: 423.
Heckman, J. (1979). “Sample Selection Bias as a Specification Error,” Econometrica 47:153-161
Hermann, R., E. Single, and J. Boston (1977). Counsel for the Poor: Criminal Defense in Urban
America. Lexington, MA: Lexington Books
Heumann, M. (1978). Plea Bargaining: The Experiences of Prosecutors, Judges, and Defense
Attorneys. Chicago, IL: The University of Chicago Press.
Hoffman, M., P. Rubin, and J. Shepherd (2005-2006). “An Empirical Study of Public Defender
Effectiveness: Self Selection by the Marginally Indigent,” Ohio State Journal of
Criminal Law 3: 223-255
Huang, KC., KP. Chen, and CC. Lin (2010). “Does the Type of Criminal Defense Counsel
Affect Case Outcomes? A Natural Experiment in Taiwan,” International Review of
Law and Economics 30: 113-127.
Iyengar, R. (2007). An Analysis of the Performance of Federal Indigent Defense Counsel.
Cambridge, MA: National Bureau of Economic Research – Harvard University
Langton, L. and D. Farole (2010). State Public Defender Programs, 2007. Washington, D.C.:
Bureau of Justice Statistics
Long, J. and J. Freese (2006). Regression Models for Categorical Dependent Variables Using
Stata, 2nd Edition. College Station, TX: Stata Press
Nardulli, P., J. Eisenstein, and R. Flemming (1988). The Tenor of Justice: Criminal Courts and
the Guilty Plea Process. Chicago, IL: University of Illinois Press
McCoy, C. (1993). Politics and Plea Bargaining: Victims Rights in California. Philadelphia, PA:
University of Pennsylvania Press.
52
 

Neubauer, D. and H. Fradella (2011). America’s Courts and the Criminal Justice System.
Belmont, CA: Wadsworth.
Pardoe, Iain. (2006). Applied Regression Modeling: A Business Approach. New York, NY:
Wiley.
Piehl, A, and S. Bushway (2007). “Measuring and Explaining Charge Bargaining,” Journal of
Quantitative Criminology 23: 105-125.
Radtke, P., P. Semple, and L. Cohen (1982). Alternative Models for Defending the Indigent in
Virginia. Washington, D.C.: American Institutes for Research.
Roach, M. (2010). “Explaining the Outcome Gap between Different Types of Indigent Defense
Counsel: Adverse Selection and Moral Hazard Effects,” available at Social Science
Research Network: http://ssrn.com/abstract=1839651
Spangenberg, R. and M. Beeman (1995). “Indigent Defense Systems in the United States,” Law
And Contemporary Problems 58: 31-50.
Steffensmeier, D., J. Kramer, and J. Ulmer (1995). “Age Differences in Sentencing,” Justice
Quarterly 12: 583-602.
Steffensmeier, D. and S. Demuth (2006). “Does Gender Modify the Effects of Race-ethnicity on
Criminal Sanctioning? Sentences for Male and Female White, Black, and Hispanic
Defendants,” Journal of Quantitative Criminology 22: 241-261.
Weidner, R., R. Frase, and I. Pardoe (2004). “Explaining Sentence Severity in Large Urban
Counties: A Multilevel Analysis of Contextual and Case Level Factors,” The Prison
Journal 84: 184-207.
Williams, M. (2002). “A Comparison of Sentencing Outcomes for Defendants with Public
Defenders versus Retained Counsel in a Florida Circuit Court,” The Justice Systems
Journal 23: 249-257.
Worden, A. (1990-1991). “Privatizing Due Process: Issues in the Comparison of Assigned
Counsel, Public Defender, and Contracted Indigent Defense Systems,” The Justice
Systems Journal 14 (3): 390-418.
Worden, A. (1993). “Counsel for the Poor: An Evaluation of Contracting for Indigent Defense,”
Justice Quarterly 10(4) 613-638.
Cases cited
Alabama v. Shelton (535 U.S. 654 (2002))
Argersinger v. Hamlin (407 U.S. 25 (1972))
53
 

Gideon v. Wainwright (372 U.S. 335 (1963))
In re Gault (387 U.S. 1 (1967))

54