DEPARTMENT OF PUBLIC SAFETY
RESPONSE TO
HOUSE CONCURRENT
RESOLUTION NO. 62
TWENTY-FIRST LEGISLATURE, 2002

December 2002

COMPARISON OF RECIDIVISM RATES AND RISK FACTORS BETWEEN
MAINLAND TRANSFERS AND NON-TRANSFERRED INMATES
A Study Requested by House Concurrent Resolution No. 62
House of Representatives, Twenty-First Legislature, 2002
State of Hawaii

Prepared by the
Department of Public Safety
December 2002

TABLE OF CONTENTS
Page
Introduction..............................................................................................................1
Background ..............................................................................................................2
Selection of Sampled Groups...................................................................................2
Data Sources ............................................................................................................3
Data Collection ........................................................................................................3
Measures of Recidivism...........................................................................................4
Measures of Risk Factors.........................................................................................5
Chi-Square Tests of Association..............................................................................6
Characteristics of Mainland Transfers and Non-transfers .......................................6
Recidivism Rates for Mainland Transfers and Non-transfers................................13
Risk Factors for Mainland Transfers and Non-transfers........................................14
Summary ................................................................................................................16
Recommendations..................................................................................................16
Attachment I: House Concurrent Resolution No. 62
Attachment II: Examples of Guidelines for Transfers at Various Dates
Attachment III: Risk and Needs Assessments

i

LIST OF TABLES
Page
Table 1:
Table 2:
Table 3:

Comparison of Mainland Transfers and Non-transfers by Gender................................7
Comparison of Mainland Transfers and Non-transfers by Ethnic Group or Race ........7
Comparison of Mainland Transfers and Non-transfers by Legal Status
during Incarceration ......................................................................................................8
Table 4: Comparison of Mainland Transfers and Non-transfers by Age at Release ...................9
Table 5: Comparison of Mainland Transfers and Non-transfers by Severity of Offense
during Incarceration .....................................................................................................10
Table 6: Comparison of Mainland Transfers and Non-transfers by Summary Offense
during Incarceration .....................................................................................................10
Table 7: Comparison of Mainland Transfers and Non-transfers by Length of Incarceration....11
Table 8: Comparison of Mainland Transfers and Non-transfers by Release Type....................12
Table 9: Comparison of Mainland Transfers and Non-transfers by Release Location..............13
Table 10: Comparison of Recidivism Rates between Mainland Transfers and Non-transfers ....14
Table 11: Comparison of Risk Factors between Mainland Transfers and Non-transfers ............15

ii

ACKNOWLEDGMENTS
The principal investigators for this study were Cheryl Rodrigues and Ken Hashi of the
Department of Public Safety, Research and Statistics Staff. Advice on data collection and
statistical analysis was provided by Janet Davidson of the Department of the Attorney General,
Criminal Prevention and Justice Assistance Division, Research and Statistics Branch. Electronic
records on arrests and convictions were provided by John Maruyama of the Department of the
Attorney General, Hawaii Criminal Justice Data Center. Access to parole case files and
interpretations of parole documents were provided by Tommy Johnson, Division Administrator;
Max Otani, Branch Administrator; David Chiang, Staff Services Supervisor; and the parole
officers and clerical staff of the Hawaii Paroling Authority.

iii

INTRODUCTION
In accordance to House Concurrent Resolution No. 62 passed by the State of Hawaii, House of
Representatives, Twenty-First Legislature, 2002 (Attachment I), this report presents the findings
of a study to compare the recidivism rates of inmates who were transferred to mainland
correctional facilities with similarly situated inmates who remained incarcerated in Hawaii. As
requested by the Resolution, the study focused on inmates who have been released or paroled for
at least three years.
(In this report, inmates who were transferred to Mainland facilities during their prison sentence
are referred to as ”mainland transfers”. Inmates who remained entirely incarcerated in Hawaii
are referred to as “non-transfers”.)
The criteria used to select inmates for transfer to the Mainland facilities made it difficult to create
a comparable group of inmates who were not selected. In general, the inmates selected for
transfer were required to be in good physical and mental health; evaluated to have no or low
institutional risk; had three or more years remaining on their minimum sentences; and no
pending charges.
Based on three levels of recidivism, the non-transfers had slightly higher recidivism rates than
the mainland transfers. These differences in rates for each level of recidivism between the two
groups, however, were not statistically significant, i.e. there is a large probability the magnitude
of the differences could have occurred largely by chance alone.
This study also compared the two groups on factors used to measure their propensity to
recidivate. Based on risk and needs assessments prior to their release, a larger share of mainland
transfers had “adequate” human capital than the non-transfers. This measure of human capital
was based on their total scores of reasoning and intellectual skills; educational and vocational
skills; attitude toward personal change; and emotional stability. The difference between the two
groups, moreover, was statistically significant, i.e. there is a small probability that the magnitude
of the difference could have occurred largely by chance alone.
Reasons for differences in recidivism and risk factors between the mainland transfers and nontransfers include the selection criteria used in transferring inmates; disparate characteristics and
loss of the sampled cases; the units of measures used for recidivism; and suitability of the risk
factors to the sampled cases.
Future studies should employ longitudinal research to more effectively collect data. In addition,
an inventory and evaluation of inmate programs should be conducted.

1

BACKGROUND
Number and Location of Transfers in Out-of-State Contracted Facilities
From December 1995, the Department of Public Safety has housed inmates from Hawaii in
contracted facilities in various states. Initially, about 300 male inmates or 14.2 percent of the
total male prison inmates were transferred to facilities in Texas. By June 30, 1999, 1,095 Hawaii
male inmates (38.1 percent of the total male prison inmates); and 83 female inmates (32.2
percent of the total female prison inmates) were housed in out-of-state contracted facilities in
Minnesota, Oklahoma, and Tennessee. As of November 11, 2002, 1,238 Hawaii male inmates
(36.1 percent of the total male prison inmates) were housed in contracted facilities located in
Oklahoma and Arizona; and 64 Hawaii female inmates (14.5 percent of the total female prison
inmates) were housed in a contracted facility located in Oklahoma.
Selection of Inmates for Transfers to Out-of-State Contracted Facilities
In general, the inmates selected for transfers to the out-of-state facilities were required to be in
good physical and mental health; evaluated to have no or low institutional risk; had three or more
years remaining on their minimum sentences; and no pending hearings or charges. The transfers
also needed approval by the Department of Public Safety, Health Care Office and the
Department of Public Safety, Sex Offender Treatment Program. Inmates were further screened
to determine the status of any parole hearing or trial. Notifications were sent to the Hawaii
Paroling Authority; Office of the Prosecuting Attorney; Office of the Public Defender; and the
Circuit Courts of the Hawaii Judiciary.
The selection criteria, however, were not strictly applied and were changed at various times. For
example, inmates with two years remaining on their minimum sentence; or inmates who could
not be moved to minimum facilities in Hawaii were later included. Some examples of the
criteria used for various dates are listed in Attachment II.
SELECTION OF SAMPLED GROUPS
Based on the criteria in House Concurrent Resolution No. 62, only inmates released until June
30, 1999 were included in this study.
A total of 185 mainland transfers (146 males and 39 females) were released between January 1,
1996 and June 30, 1999 . Of this total number, one inmate was released between January 1,
1996 and June 30, 1996. Three inmates were released between July 1, 1996 and June 30, 1997.
During Fiscal Years 1998 and Fiscal Years 1999, 37 and 144 mainland transfers were released,
respectively. All of these inmates were included as the initial sample for the study.
2

For the sample of non-transfers, a total of 185 prison inmates (sentenced felons and parole
violators; 146 males and 39 females) were selected. Because the largest shares of the mainland
transfers were released during Fiscal Year 1998 and Fiscal Year 1999, the sample was selected
entirely from these two fiscal years. Of the 890 non-transfers who were released during Fiscal
Year 1998, the sample selected for the non-transfers corresponded to the sample selected for the
mainland transfers from Fiscal Years 1996 to 1998, i.e. 28 males prison inmates and 13 female
prison inmates. Of the 1,039 non-transfers who were released during Fiscal Year 1999, the
sample selected for 1999 were 118 male prison inmates and 26 female prison inmates.
More exact matching of the non-transfers to the mainland transfers was not conducted due to
limitations on time and staff. Matching inmates based on the selection criteria used for transfers
such as their offenses at sentencing, prior criminal history, security classification, history of
institutional misconducts, physical and mental health would have required extensive review of
inmate case files.
DATA SOURCES
Three sources were used in this study to obtain demographic, incarceration, parole, and rearrest
data for each sampled case:
1. Historical electronic inmate records from the Corrections Information System (CIS) of
the Department of Public Safety.
2. Active and inactive parole case files from the Hawaii Paroling Authority; and parole
records from the electronic data base of the Hawaii Paroling Authority.
3. Printouts in electronic form on arrests and corresponding dispositions from the Offender
Based Transaction Statistics \ Comprehensive Criminal History (OBTS\CCH) provided
by the Hawaii Criminal Justice Data Center of the Department of the Attorney General.
DATA COLLECTION
Demographic information, prison admission and release dates, legal status (i.e. sentenced felon
or parole violator), release type and locations; and charge and sentencing information were
initially obtained from historical electronic inmate records of the CIS.

3

Information for each of the sampled cases were printed on a data collection form. Data on these
forms were verified using information from the active and inactive case files at the Hawaii
Paroling Authority. In addition, information was collected on the risk and needs assessments and
on parole violations and revocations. At the Hawaii Paroling Authority, case files for nine
mainland transfers and four non-transfers were not located, even after two to three additional
attempts to find them. Consequently, the resulting sample was 176 for mainland transfers and
181 for non-transfers.
Information on rearrests for each sampled case were obtained using printouts of arrests and
corresponding dispositions generated from the electronic OBTS\CCH records. For each arrest
date, the most serious offense (excluding parole violations) and its corresponding disposition was
collected.
MEASURES OF RECIDIVISM
Recidivism was based on all arrests, convictions, parole violations, and revocations which
occurred between the inmate’s date of release and June 30, 2002. Moreover, three measures or
levels of recidivism were used in this study. The first level was based on reports of 1) any arrest
for a new offense or 2) parole violation. The second level was based on reports of 1) any
conviction for any new offense or 2) parole revocation. The third level was based on reports of
1) any conviction for any felony offense or 2) parole revocation.
A rigorous and scientifically valid measure of recidivism, however, was difficult to construct
since the following questions would be difficult to answer:
-

“At what stage or involvement with the criminal justice system by an ex-offender
does his situation constitute a return to criminal behavior: undetected crime, arrest,
interrogation as a suspect, booking, arraignment, court case filing, prosecution,
conviction, parole violation involving a new offense, parole violation involving a
technical violation, etc.?

-

For what period of time after his release should an ex-offender be monitored to
determine if he returns to a mode of criminal behavior, lifetime, one day, one year,
two years…?

-

How do we incorporate the number of ex-offenders who commit new offenses, but
who are not apprehended, into a “recidivism” rate?”1

1

“The Irrelevance of Recidivism to Criminal Justice Evaluation” by Robert Roesch and Lonnie Fouty, Florida
Department of Corrections, March 30, 1978.

4

MEASURES OF RISK FACTORS
Five “risk factors” were used to compare differences between the mainland transfers and nontransfers on their propensity to recidivate prior to release. These factors were based on the same
measures in the report entitled Parole Decision Making in Hawaii (August 2001) 2 used to predict
success and failure on parole. Moreover, the five risk factors were derived from determining the
inter-correlations of the items in the risk and needs assessments by the Hawaii Paroling
Authority (Attachment III). Inter-correlations (factor analysis) of the total 22 items in the
assessments resulted in the five factors (groups or clusters) shown in the following table3:

Criminal History

Human Capital

Conventional Life
Style
Suitable
companions

Community and
Personal Stability
Suitability of
residence

Alcohol Problem
and Self Control
Alcohol use
problem

Prior felony
convictions

Reasoning and
intellectual skills

Prior conviction on
selected property
offenses

Educational and
vocational skills

Drug use and
problems or
dependence

Number of address
changes in year
prior to sentenced

Sexual conduct

Prior probation
and parole
revocations

Attitude toward
change

Percent of time
employed in year
prior to sentence

Marital and family
relationships

Age at first
conviction or
adjudication

Emotional stability

Quality of
employment

Personal financial
management skill

As explained in Parole Decision Making in Hawaii, “for each parolee, a score was computed for
each of the five factors by summing the scores on the component items and dividing by the
number of items. The list of all the factor scores was then divided at the midpoint, with
individual scores falling in either the “low risk” or “high risk” half.”
For this study, the median values of the factor scores were calculated and used to determine high
and low levels for “Criminal History” and for “Alcohol Problem and Self Control”; and adequate
or inadequate levels for “Human Capital”, “Conventional Life Style”, and “Community and
Personal Stability”. Furthermore, only the initial risk and needs assessments usually completed
within one year of the date of release were used. Consequently, risk factors were not calculated
for inmates with assessments more than one year prior to their release or for inmates with
assessments after their release.

2

Parole Decision Making in Hawaii (August 2001) by Social Science Research Institute, University of Hawaii at
Manoa and the Research and Statistics Branch, Crime Prevention and Justice Assistance Division, Hawaii State
Department of the Attorney General.
3
Ibid., page 29.

5

Risk factors could not be calculated when initial assessments were not completed because presentence investigation reports were waived; for inmates who were released on the expiration date
of their minimum sentence; or for inmates who were incarcerated out-of-state and a face-to-face
assessment could not be conducted prior to their parole hearing.
Inter-correlations (factor analysis) of the risk and needs assessments for the sampled inmates in
this study were not conducted due to limited time available to complete the study. Consequently,
the findings on the risk factors presented in this study should be interpreted with caution.
CHI-SQUARE TESTS OF ASSOCIATION
Comparison of recidivism rates and risk factors between the “mainland transfers” and “nontransfers” were based on frequencies depicted in a two-dimensional statistical table and
evaluated by using chi-square tests of association. This test of association is used to assess the
statistical significance of differences based on the actual frequencies and calculated (expected)
frequencies. The larger the difference of the actual frequencies relative to the expected
frequencies, the larger the chi-square statistic and the more likely the difference is statistically
significant, i.e. only a small the probability the difference would have been produced by chance
alone. Generally, the .05 or lower level of significance (i.e. probability) is used to accept the
differences as not having occurred largely by chance alone.
CHARACTERISTICS OF MAINLAND TRANSFERS AND NON-TRANSFERS
Comparisons of the sampled inmates of mainland transfers and non-transfers are depicted on the
following demographic and institutional characteristics:
•
•
•
•
•
•
•
•
•

Gender
Ethnicity or race
Legal status at release
Age at release
Severity of offense during incarceration
Summary offense during incarceration
Length of incarceration
Release type
Release location

Characteristics of the two groups were similar by gender; ethnicity or race; and age at release.
However, compared to the non-transfers, a larger share of the mainland transfers were sentenced
felons; convicted for felony A offenses and for violent offenses; incarcerated for two or more
years; released to parole supervision; and released to a federal agency or moved out of state.
6

Males comprised about eight out of ten of the sampled mainland transfers and non-transfers.
(Table 1).
TABLE 1
Comparison of Mainland Transfers and Non-transfers
by Gender
Mainland
Transfers
Gender

Number

Percent

Non-transfers
Number

Percent

Total Number Released

176

100.0%

181

100.0%

Male
Female

142
34

80.7%
19.3%

145
36

80.1%
19.9%

By ethnicity and race, a slightly larger share of the mainland transfers were “Other” (21.0
percent) and Caucasian (19.3 percent) than the non-transfers (Table 2). For the non-transfers, a
slightly higher proportion were Hawaiian or Part Hawaiian (42.0 percent) than the mainland
transfers (39.8 percent).
TABLE 2
Comparison of Mainland Transfers and Non-transfers
by Ethnic Group or Race
Mainland
Transfers
Ethnic Group or Race

Number

Percent

Non-transfers
Number

Percent

Total Number Released

176

100.0%

181

100.0%

Caucasian
Hawaiian/Part Hawaiian
Asian
Other
Unknown

34
70
21
37
14

19.3%
39.8%
11.9%
21.0%
8.0%

31
76
23
30
21

17.1%
42.0%
12.7%
16.6%
11.6%

7

Most of the mainland transfers were incarcerated as sentenced felons (77.8 percent) compared to
the non-transfers (62.4 percent) (Table 3). Conversely, a larger share of the non-transfers were
incarcerated as parole violators (37.6 percent) than the mainland transfers (16.5 percent).

TABLE 3
Comparison of Mainland Transfers and Non-transfers
by Legal Status during Incarceration

Mainland
Transfers
Legal Status

Number

Percent

Non-transfers
Number

Percent

Total Number Released

176

100.0%

181

100.0%

Sentenced Felon
Parole Violator
Other

137
29
10

77.8%
16.5%
5.7%

113
68
0

62.4%
37.6%
0.0%

8

At the time of their release, mainland transfers were generally the same age as the non-transfers
(Table 4). The median and mean ages (34.7 years and 33.7 years) for the mainland transfers
were slightly lower than the non-transfers (35.1 years and 34.8 years). Slightly larger
proportions of the mainland transfers (27.3 percent) were 30 to 34 years old compared to the
non-tranfers (22.1 percent). Conversely, a slighty higher share of the non-transfers (25.4
percent) were 35 to 39 years old compared to the mainland transfers (20.5 percent).
TABLE 4
Comparison of Mainland Transfers and Non-transfers
by Age at Release
Mainland
Transfers
Age Group
Total Number Released
20 years or less
20 to 24 years
25 to 29 years
30 to 34 years
35 to 39 years
40 to 44 years
45 years or older
Unknown

Number

Percent

Non-transfers
Number

Percent

176

100.0%

181

100.0%

0
11
37
48
36
27
16
1

0.0%
6.3%
21.0%
27.3%
20.5%
15.3%
9.1%
0.6%

0
17
35
40
46
28
15
0

0.0%
9.4%
19.3%
22.1%
25.4%
15.5%
8.3%
0.0%

Average

Mainland
Transfers

Mean age in years
Median age in years

34.7
33.7

9

Non-transfers
35.1
34.8

Compared to the non-transfers, larger percentages of the mainland transfers were incarcerated for
Felony A offenses (16.5 percent) and Felony B offenses (44.9 percent) (Table 5). In contrast,
about one-half of the non-transfers were incarcerated for Felony C offenses compared to 35.2
percent for the mainland transfers.
TABLE 5
Comparison of Mainland Transfers and Non-transfers
by Severity of Offense During Incarceration

Mainland
Transfers
Severity of Offense
Total Number Released
Felony A
Felony B
Felony C
Misdemeanor
Unknown

Number

Percent

Non-transfers
Number

Percent

176

100.0%

181

100.0%

29
79
62
4
2

16.5%
44.9%
35.2%
2.3%
1.1%

18
67
93
2
1

9.9%
37.0%
51.4%
1.1%
0.6%

By summary offense, a greater share of the mainland inmates (31.8 percent) than non-transfers
(23.2 percent) were incarcerated for violent offenses (Table 6). Non-transfers, on the other hand,
had a larger percentage who were incarcerated for drug offenses (34.3 percent) than the mainland
transfers (27.8 percent).
TABLE 6
Comparison of Mainland Transfers and Non-transfers
by Summary Offense During Incarceration

Mainland
Transfers
Summary Offense
Total Number Released
Violent
Sex
Property
Drugs
Other

Number

Percent

Non-transfers
Number

Percent

176

100.0%

181

100.0%

56
8
53
49
10

31.8%
4.5%
30.1%
27.8%
5.7%

42
6
60
62
11

23.2%
3.3%
33.1%
34.3%
6.1%

10

The length of incarceration for mainland transfers were longer than the non-transfers. For the
mainland transfers, larger shares of them were incarcerated from 24 months to less than 36
months (28.4 percent); and for 36 months or more (39.8 percent) (Table 7). Non-transfers had a
larger percentage incarcerated for less than 12 months (30.4 percent) and for 12 months to less
than 24 months (35.4 percent). Similarly, the mean and median length of incarceration for the
mainland transfers (36.3 months and 30.5 months) were higher than for the non-transfers (23.6
months and 19.1 months).

TABLE 7
Comparison of Mainland Transfers and Non-transfers
by Length of Incarceration

Mainland
Transfers
Length of incarceration
Total Number Released
Less than 12 months
12 mo.s to less than 24 mo.s
24 mo.s to less than 36 mo.s
36 months or more

Number

Percent

Non-transfers
Number

Percent

176

100.0%

181

100.0%

9
47
50
70

5.1%
26.7%
28.4%
39.8%

55
64
29
33

30.4%
35.4%
16.0%
18.2%

Length of Incarceration

Mainland
Transfers

Mean length in months
Median length in months

36.3
30.5

Non-transfers

23.6
19.1

The differences in the length of stay reflect the differences in the legal status between the
mainland transfers and non-transfers. A greater share of the mainland transfers were sentenced
felons compared to the proportion for the non-transfers. These inmates were incarcerated for
more than one year. In contrast, the non-transfers had a larger percentage of parole violators
who were generally incarcerated for a period ranging from several months to over one year.

11

A slightly higher share of the mainland transfers (90.9 percent) than non-transfers (86.2 percent)
were released and placed under the supervision of the Hawaii Paroling Authority (Table 8). In
contrast, the percentage of non-transfers (12.2 percent) who were discharged from parole was
greater than the mainland transfers (2.8 percent). In general, inmates are discharged from parole
when the date of their maximum sentences have expired; or because of good behavior, illness, or
other exceptions.
TABLE 8
Comparison of Mainland Transfers and Non-transfers
by Release Type

Mainland
Transfers
Release Type

Number

Percent

Non-transfers
Number

Percent

Total Number Released

176

100.0%

181

100.0%

Released to Parole Superv.
Discharged from Parole
Other

160
5
11

90.9%
2.8%
6.3%

156
22
3

86.2%
12.2%
1.7%

For the mainland transfers, larger shares were released to a federal agency (13.6 percent) or
moved out of state (10.2 percent) than the non-transfers (Table 9). For non-transfers, higher
percentages were released to the community or family (72.4 percent); and to treatment programs
(19.9 percent.).

12

TABLE 9
Comparison of Mainland Transfers and Non-transfers
by Release Location

Mainland
Transfers
Release Location

Number

Percent

Non-transfers
Number

Percent

Total Number Released

176

100.0%

181

100.0%

Community or Family
Federal Agency
Moved Out of State
Treatment Programs
Other
Not Reported

112
24
18
21
1
0

63.6%
13.6%
10.2%
11.9%
0.6%
0.0%

131
6
2
36
1
5

72.4%
3.3%
1.1%
19.9%
0.6%
2.8%

Note: Community or Family includes releases to parole, time served, and moved to neighbor islands.
Other includes bail or bond releases; and changes to pretrial or sentenced status.

RECIDIVISM RATES FOR MAINLAND TRANSFERS AND NON-TRANSFERS
The recidivism rates were based only on sampled cases who resided in Hawaii and were under
parole supervision. Because losses of samples occurred from inmates who died, moved out of
state, or were under the custody of a Federal agency, the sample of mainland transfers dropped to
134 and non-transfers fell to 173.
For each level of recidivism, the mainland transfers had lower proportions than the non-transfers
(Table 10). For the first level, i.e. any arrest or parole violation, 77.6 percent of the mainland
transfers recidivated compared to 80.9 percent of the Hawaii transfers. For the second level, i.e.
any conviction or parole revocation, 63.4 percent of the mainland transfers recidivated compared
to 67.6 percent of the Hawaii transfers. For the third level, i.e. any felony conviction or parole
revocation, 50.0 percent of the mainland transfers recidivated compared to 51.5 percent of the
non-transfers. The differences between the two groups for each level, however, were not
statistically significant.

13

TABLE 10
Comparison of Recidivism Rates
Between Mainland Transfers and Non-transfers
MAINLAND
TRANSFERS
First Level
Yes
No
Chi-square value:
Level of significance :

Second Level
Yes
No
Chi-square value:
Level of significance :

Third Level
Yes
No
Chi-square value:
Level of significance :

134 100.0%
104 77.6%
30 22.4%

HAWAII
INMATES
173 100.0%
140 80.9%
33 19.1%

0.508
0.456

134 100.0%
85 63.4%
49 36.6%

173 100.0%
117 67.6%
56 32.4%

0.591
0.442

134 100.0%
67 50.0%
67 50.0%

173 100.0%
89 51.4%
84 48.6%

0.063
0.802

RISK FACTORS FOR MAINLAND TRANSFERS AND NON-TRANSFERS
The total sample used for comparing the propensity to recidivate was based on 85 mainland
transfers and 109 non-transfers. Of the five risk factors, only the difference in “human capital”
was statistically significant (Table 11). A larger share of the mainland transfers with “adequate”
human capital was 72.9 percent compared to the non-transfers with 51.4 ,percent. The likelihood
of such a large difference between the two groups occurring by chance would be less than 3 out
of 1000.

14

TABLE 11
Comparison of Risk Factors
Between Mainland Transfers and Non-transfers

MAINLAND
TRANSFERS
Criminal History
High
Low
Chi-square value:
Level of significance :

Human Capital
Inadequate
Adequate
Chi-square value:
Level of significance :

Conventional Lifestyle
Inadequate
Adequate
Chi-square value:
Level of significance :

Community and
Personal Stability
Inadequate
Adequate
Chi-square value:
Level of significance :

Alcohol Problem and
Self Control
High
Low
Chi-square value:
Level of significance :

85 100.0%
37 43.5%
48 56.5%

HAWAII
INMATES
109 100.0%
40 36.7%
69 63.3%

0.931
0.335

85 100.0%
23 27.1%
62 72.9%

109 100.0%
53 48.6%
56 51.4%

9.320
0.002

85 100.0%
46 54.1%
39 45.9%

109 100.0%
51 46.8%
58 53.2%

1.026
0.311

85 100.0%
42 49.4%
43 50.6%

109 100.0%
49 45.0%
60 55.0%

0.381
0.537

85 100.0%
32 37.6%
53 62.4%
0.648
0.421

15

109 100.0%
35 32.1%
74 67.9%

SUMMARY
Some differences in the recidivism rates between the mainland transfers and non-transfers were
found. Proportionally smaller shares of mainland transfers recidivated at each of the three levels.
These differences, however, were not statistically significant. By risk factors, a larger share of
the mainland transfers had more ‘adequate” human capital than the non-transfers. The
difference, moreover, was statistically significant.
Several possible reasons for differences in recidivism and risk factors between the mainland
transfers and non-transfers include:
•
•
•
•
•

Selection criteria used in transferring inmates to contracted out of state facilities
A larger share of the non-transfers being released as parole violators (37.6 percent) than
the mainland transfers (16.5 percent). Prior parole experience was found to be a
statistically significant predictor of recidivism4.
Disparities in the loss of the sampled cases due to case files not being located; no
assessments; releases to federal agencies; and residency in another state.
Unit of measures used for recidivism
Appropriateness of the item scores used to construct the risk factors

RECOMMENDATIONS
Future studies should employ longitudinal research; and include an inventory and evaluation of
inmate programs.
A longitudinal study would allow closer monitoring of when, why, and to whom changes in the
selection process for transfers occurred. At the same time, more comparable samples can be
selected as the selection process occurs. A longitudinal study would distribute the time and
effort for collecting demographic, institutional, and release data over a longer period of time;
have case files be more accessible; and increase the scope of the information to be collected.
Also, this method would track the comparable samples over many points in time and identify
changes and trends which can help to better “explain” differences in recidivism and risk factors.
An inventory and (process and outcome) evaluation of inmate programs in mainland facilities
and Hawaii facilities should be conducted to determine the quantitative and qualitative
differences between the facilities. The information may help to explain some of the differences
in recidivism between mainland transfers and non-transfers.
Finally, a study should also be conducted to determine the relationship between the elements that
comprise “human capital” and the recidivism rates for non-transfers. Findings that confirm a
strong relationship may be used to develop or change programs during incarceration and parole.
4

“Survival on Parole”, by Social Science Research Institute, University of Hawaii at Manoa; and the Department of
the Attorney General, State of Hawaii

16

ATTACHMENT II
EXAMPLES OF GUIDELINES FOR TRANSFER AT VARIOUS DATES
For December 1995 transfers
1.
2.
3.
4.
4.
5.

Tentative parole date 1999 or later
Classified as Close, Medium, or Minimum custody
Have no pending litigation requiring return to Hawaii in 3 years
Have no extreme medical problems
Have no debilitating mental health problems or use of psychotropic medications.
Exclude inmates classified as Maximum custody

Inmates should be screened through the Department of Public Safety, Health Care Office;
Hawaii Paroling Authority; Hawaii State Judiciary; Office of the Prosecuting Attorney,
Office of the Public Defender, and Department of Public Safety, Sex Offender Treatment
Program prior to transfer.
For list of inmates to be considered for transfer (needed by February, 12, 1997)
1.
2.
3.
4.

Must be classified as medium or close custody.
Must have four years or more remaining on their sentences.
Inmates with long-term minimum sentences.
Inmates with out-of-state detainers. However, do not include inmate with Immigration
and Naturalization detainers.
5. Inmates with mental health problems who are stabilized by medication.
6. Inmates whose home-base (family) is on the mainland.
7. Inmates who can be released to the mainland.
For list of inmates to be considered for transfer (requested by May 1, 1997)
1. More than two years before their longest tentative parole date or discharge date.
2. No for pending charges and detainers.
3. No histories of:
A. Misconduct’s and/or criminal charges involving assaultive behavior.
B. Escapes or attempted escapes.
C. Sex offenses
4. Medical, mental health and dental clearance.
Note: Mental health clearance to include mental stability and suicidal behavior.
Inmates on psycho tropic medication must be stabilized.
5. Identify balance of maximum sentence.
6. Include any relevant comments regarding the above information. Also include gang
information and protective custody needs if you have this information.