Journal of Forensic Psychology Research and Practice

ISSN: 2473-2850 (Print) 2473-2842 (Online) Journal homepage: https://www.tandfonline.com/loi/wfpp21

Differences in Risk Factors for Violent, Nonviolent,

and Sexual Offending

Claudia E. van der Put, Mark Assink & Jeanne Gubbels

To cite this article: Claudia E. van der Put, Mark Assink & Jeanne Gubbels (2020): Differences
in Risk Factors for Violent, Nonviolent, and Sexual Offending, Journal of Forensic Psychology
Research and Practice, DOI: 10.1080/24732850.2020.1735248

To link to this article: https://doi.org/10.1080/24732850.2020.1735248

© 2020 The Author(s). Published with

license by Taylor & Francis Group, LLC.

Published online: 28 Feb 2020.

Submit your article to this journal

Article views: 100

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=wfpp21
JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE
https://doi.org/10.1080/24732850.2020.1735248

Differences in Risk Factors for Violent, Nonviolent, and

Sexual Offending
Claudia E. van der Put, Mark Assink, and Jeanne Gubbels
Research Institute of Child Development and Education, University of Amsterdam, Amsterdam,
Netherlands

ABSTRACT KEYWORDS
Whether risk factors for recidivism are equally predictive in Violent offending; sexual
different offender groups and across recidivism types is offending; nonviolent
a question of high clinical importance. Therefore, this study offending; risk factor;
aimed to examine (a) differences in impact of general delin- specialist offender;
generalist offender
quency risk factors for three different recidivism types, and (b)
differences in the presence of general delinquency risk factors
between five different offender groups. The studied sample
comprised 8,665 Dutch adult offenders, in which risk factors
for recidivism were assessed with the Recidivism Risk
Assessment Scales (RISc). Results showed that risk factors
were generally stronger related to nonviolent than to violent
recidivism, and only weakly or not at all related to sexual
recidivism. The total prevalence of risk factors was highest in
generalists (non-specialized offenders), followed by nonviolent,
violent, sexual, and one-time offenders. In violent offenders,
risk factors in the domains relationship with the partner/family,
emotional well-being, antisocial personality patterns, and anti-
social attitudes/cognitions were more prevalent than in non-
violent offenders, whereas in nonviolent offenders, risk factors
in the domains education/work, financial management/
income, antisocial friends/acquaintances, and drug abuse
were more prevalent. In conclusion, risk factors differ in pre-
valence and impact across offender groups and recidivism
types. Therefore, different treatment approaches are needed
for successfully reducing different recidivism types across
offender groups.

Research has shown that recidivism is generally high among offenders. For
example, Durose, Cooper, and Snyder (2014) showed that 67.8 percent of the
US state prisoners released in 2005 were arrested within three years of
release, and 76.6% were arrested within five years of release. High recidivism
rates are also found in other Western countries. In the Netherlands, 2-year
recidivism rates for adult offenders ranges from 34% to 47% depending on
whether the sample included arrested, convicted, or imprisoned persons
(Wartna, Blom, & Tollenaar, 2011). Given the detrimental effects of criminal
recidivism for both victims, offenders, and society as a whole, it is highly

CONTACT Claudia E. van der Put [email protected] Research Institute of Child Development and
Education, University of Amsterdam, P.O. Box 15780, Amsterdam, NG 1001, The Netherlands
© 2020 The Author(s). Published with license by Taylor & Francis Group, LLC.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives
License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction
in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
2 C. E. VAN DER PUT ET AL.

essential to break this cycle of offending by addressing predictors of recidi-

vism in interventions. In the past, many studies have been directed at
identifying risk factors for general recidivism and showed that variables
such as age, gender, number of committed offenses, mental health problems,
substance abuse, and antisocial associates are associated with reoffending
(see, for instance, Bonta, Law, & Hanson, 1998; Cottle, Lee, & Heilbrun,
2001; Gendreau, Little, & Goggin, 1996). Despite this body of research, far
less is known about differences in risk factors for specific types of recidivism,
such as nonviolent recidivism, violent recidivism, and sexual recidivism. It is
important to increase this knowledge, as for any specific type of recidivism,
some risk factors may be more important than other risk factors. Identifying
differences in risk factors across recidivism types is not only important for
a better understanding of why specific types of recidivism occur, but also for
advancing the practical tasks of risk and needs assessment. Therefore, the
aim of this study was to examine potential differences in risk factors across
specific types of recidivism.
A substantial number of meta-analysis on the effects of static and dynamic
risk factors for delinquent behavior has increased our knowledge of major,
moderate, and minor risk factors in different groups of offenders (see, for
instance, Assink et al., 2015; Bonta et al., 1998; Gendreau et al., 1996; Hanson
& Morton-Bourgon, 2005; Lipsey & Derzon, 1998). Based on the results of
these reviews, Andrews and Bonta (2010) formulated eight major risk factors
for criminal conduct, to which they referred to as the “Central Eight” risk
factors: (1) having a history of antisocial behavior, (2) antisocial associates,
(3) antisocial attitudes and cognitions, (4) having an antisocial personality,
(5) education/employment problems, (6) family/marital problems, (7) sub-
stance abuse, and (8) involvement in antisocial recreational activities. The
first four, referred to as the “Big Four” risk factors, have the strongest impact
on recidivism, and can be regarded as the four most important factors.
Interventions that successfully address the central eight risk factors have
been empirically associated with reduced recidivism (see, for instance,
Andrews, Bonta, & Hoge, 1990; Andrews & Bonta, 2010; Smith, Gendreau,
& Swartz, 2009). Further, because these eight risk factors substantially con-
tribute to a proper estimate of the risk of future criminal behavior, they are
often assessed in instruments for assessing the risk of recidivism, such as the
Level of Service/Case Management Inventory (LS/CMI; Andrews, Bonta, &
Wormith, 2004) and the Youth Assessment and Screening Instrument (YASI;
Orbis Partner, 2007).
Although the predictive value of risk factors has been well studied for
recidivism in general, much less attention has been paid to the predictive
value of risk factors for specific types of recidivism, such as violent, property,
and sexual recidivism. Bonta, Harman, Hann, and Cormier (1996) examined
differences in the predictive value between static risk factors for general
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 3

recidivism and static risk factors for violent recidivism, and found that risk
factors were more strongly related to general recidivism than to violent
recidivism. On the other hand, Hanson (2009) concluded in his review that
the predictive value of risk factors is similar for general, violent, and sexual
recidivism, with the exception that having deviant sexual interests is only
a risk factor for sexual recidivism. However, this conclusion was based on
a comparison of results of various meta-analyses, in which studies examining
different offender populations were included. In this review, risk factors for
general recidivism in adult offenders, risk factors for violent recidivism in
mentally disordered offenders, and risk factors for sexual recidivism in sexual
offenders were synthesized together, limiting the inferences that can be
drawn on possible differences in risk factors for general, violent, and sexual
recidivism. Schmidt, Campbell, and Houlding (2011) examined the predictive
validity of several risk assessment instruments in adolescents over a follow-
up period of on average 10 years. They found that each of the examined
instruments predicted general recidivism with moderate-to-large effect sizes.
However, variation in predictive validity across recidivism types was found.
For instance, the Youth Level of Service/Case Management Inventory
(YLS/CMI; Hoge & Andrews, 2002) only weakly predicted sexual recidivism,
whereas the youth version of the Psychopathy Checklist (PCL:YV; Forth,
Kosson, & Hare, 2003) was strongly predictive of sexual, violent, nonviolent,
and “technical” recidivism (e.g., dangerous driving, parole violations).
An important question that has not received much attention in research is how
risk factors for recidivism differ in nature and impact across offender groups, such
as violent offenders and property offenders. There is substantial research aimed at
identifying variables as risk factors for specifically sexual offense recidivism (see,
for instance, Hanson & Bussiere, 1998; Hanson & Morton-Bourgon, 2005; Mann,
Hanson, & Thronton, 2010; Prentky, Knight, & Lee, 1997), but only a small
number of studies is available on differences in risk factors across offender groups.
For instance, Craig, Browne, Beech, and Stringer (2006) examined characteristics
of sexual, violent, and general (non-sexual and nonviolent) offenders and found
that violent offenders have significantly more chaotic lifestyles, are more likely to
have a history of substance abuse, and display greater psychopathology than both
sexual offenders and general offenders. Seto and Lalumière (2010) conducted
a meta-analysis of studies comparing characteristics of adolescent sexual offenders
to characteristics of adolescent non-sexual offenders and found that male sexual
offenders had a much less extensive criminal history, fewer antisocial associates,
and fewer substance use problems than their non-sexual offending counterparts.
Lai, Zeng, and Chu (2016) focused on juvenile nonviolent versus juvenile violent
offenders and compared characteristics of nonviolent offenders, violent offenders,
and violent plus offenders, of whom the latter had committed both violent and
nonviolent offenses. They found that the violent plus offenders were younger, had
more static and dynamic risk factors, and were therefore more likely to reoffend
4 C. E. VAN DER PUT ET AL.

than nonviolent offenders. These studies reveal that it cannot be assumed in

general that the prevalence and importance of risk factors are equal across groups
of offenders.
An important task in forensic clinical practice is to assess the risk of
recidivism and to refer offenders to appropriate rehabilitative interventions
after assessing the dynamic risk factors, or criminogenic needs, that must be
targeted in these interventions (Andrews et al., 1990; Andrews & Bonta,
2010). In both types of assessment, it is essential that the appropriate risk
factors are measured and properly weighed to reach valid assessments. This
not only holds for preventing general recidivism, but also for preventing
specific recidivism, such as violent and nonviolent recidivism. As differences
in risk factors across recidivism types have not been studied extensively, the
primary aim of this study was to examine differences in the prevalence and
predictive value of risk factors for violent, nonviolent, and sexual recidivism
in a sample of Dutch offenders who belonged to one of five offender groups:
generalist offenders, specialized violent offenders, specialized sexual offen-
ders, and specialized nonviolent offenders. As research showed that recidi-
vism risks estimated by assessing the same set of risk factors may not be
equally valid across recidivism types (Schmidt et al., 2011), a second aim was
to examine to what extent a risk assessment instrument for general recidi-
vism can be used to estimate risks for violent recidivism, nonviolent recidi-
vism, and sexual recidivism. Pursuing these research aims is important for
strengthening risk and need assessment procedures in clinical practice, in
which not only generalist offenders, but also specialized offenders, need to be
properly referred to the most appropriate interventions.

Method
Sample
The sample consisted of 8,665 Dutch adult offenders, of whom 7,816 (90.2%)
were male and 849 (9.8%) were female. The mean age was 34.33 years
(SD = 11.91) and ranged from 18 to 86 years. Both men and women were
included in the sample, as no substantial gender differences were found in
the impact of dynamic risk factors for recidivism according to previous
research (Van der Knaap, 2012). All offenders had been referred to probation
services by a Dutch court. The offenders were referred to probation services
for violent offenses (51.0%), property offenses without violence (22.1%),
property offenses with violence (6.8%), drug offenses (9.0%), sex offenses
(6.2%), traffic offenses (1.4%), or other offenses (3.4%). The mean number of
previous committed offenses was 9.70.
Differences in characteristics and risk profiles were examined between the
following five offender groups: (1) generalists (offenders committing multiple
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 5

types of offenses): n = 6,258 (72.2%); (2) nonviolent specialists (offenders

committing only nonviolent offenses, including property offenses without
violence, public order offenses, drug offenses, traffic offenses, and other
nonviolent offenses): n = 1,030 (11.8%); (3) violent specialists (offenders
committing only violent offenses and/or property offenses with violence):
n = 169 (2.0%); (4) sexual specialists (offenders committing only sexual
offenses with or without violence): n = 50 (0.6%); and (5) one-time offenders
(offenders who have committed any offense just once): n = 1,158 (13.4%).

Measures
Risk and needs assessment
The Dutch instrument Recidive Inschattingsschalen (RISc; Van der Knaap,
Leenarts, & Nijssen, 2007 [Recidivism Risk Assessment Scales]), is used by
the Dutch probation services to assess an offender’s risk and needs for
planning treatment and rehabilitation, and to advise the prosecutor and the
court on appropriate measures. The RISc was specifically developed for
probation services in the Netherlands, but is based on the British developed
Offender Assessment System (OASys; Howard, Clark, & Garnham, 2003,
2006), which is used for diagnosis, needs assessment, and sanctions planning.
Because the OASys is conceptually based on the internationally widely used
Level of Service Inventory-Revised (LSI-R; Hollin & Palmer, 2006) and the
Assessment Case management and Evaluation System (ACE; Gibbs, 1999),
the RISc shows considerable similarities to the LSI-R and the ACE. The RISc
is designed for the following purposes: (a) to assess an offender’s likelihood
of recidivism; (b) to map out offending-related (care) needs; (c) to assess an
offender’s responsivity; and lastly (d) to obtain an indication of the need for
further specialized assessment. The instrument measures risk factors in the
following domains: (1) criminal history; (2) accommodation/living situation;
(3) education, work, and training; (4) financial management and income; (5)
relationship with partner/family; (6) antisocial friends/acquaintances; (7)
drug misuse; (8) alcohol misuse; (9) emotional well-being; (10) antisocial
personality patterns; and (11) antisocial attitudes/cognitions (see Table A1 in
Appendix A for sample items). In each domain, a varying number of factors
is measured, adding up to an assessment of 61 risk factors in total. Most
items are scored on a 3-point Likert scale with response categories 0 (no
problems), 1 (some problems), and 2 (significant problems). The scores on
the two items in the criminal history domain on recent and earlier com-
mitted offenses were combined into a single score for assessing an offender’s
criminal history. The likelihood to reoffend was expressed in a total RISc
score by adding up the scores on all individual scale scores.
The RISc is administered by trained probation officers, and assessment of
this instrument takes about four to five hours. In the administration
6 C. E. VAN DER PUT ET AL.

procedure, the probation officer first retrieves all available information on

the offender from official files and records. Next, the offender is interviewed
about all topics for which file information is unavailable, and finally, the RISc
is filled out. Previous research on the psychometric properties of the RISc
shows positive results on the inter-rater reliability, the internal consistency,
the construct validity, and the predictive validity (Van der Knaap & Alberda,
2009; Van der Knaap et al., 2007). The RISc data that were currently analyzed
were retrieved from the Dutch probation services between January 2010 and
December 2011.
To determine the internal consistency of the RISc domain scales in the sample
that was used in the current study, reliability analyses were performed for the
total sample as well as for the five different offender groups (i.e., generalists,
nonviolent specialists, violent specialists, sexual specialists, and one-time offen-
ders). Table 1 shows McDonalds total coefficients for all subscales in the total
sample and in the different offender groups. For the subscales “education, work,
and training”, “drug misuse”, “alcohol misuse”, and “antisocial personality
patterns” the value of McDonalds was >.80 in all subgroups. For the subscales
“financial management and income”, “emotional wellbeing”, and “antisocial
attitudes/cognitions”, the value of McDonalds was > .70 in all subgroups. For
the subscale “accommodation” the value of McDonalds was > .70 in all offender

Table 1. Reliability analyses (McDonalds) for the different RISc subscales and in the different
subgroups of offenders.
Total sample Generalists Nonviolent Violent Sexual One-time
Scale (n = 8665) (n = 6258) (n = 1030) (n = 169) (n = 50) (n = 1158)
Criminal history (5 items) .750 .677 .610 .489 .620 .570
Accommodation/Living .832 .834 .839 .672 .837 .724
situation (4 items)
Education, work, and .872 .863 .854 .844 .847 .856
training (7 items)
Financial management .759 .750 .762 .751 .757 .769
and income (4 items)
Relationship with .659 .656 .696 .529 .710 .615
partner/family
(4 items)a
Antisocial friends/ .765 .765 .701 .727 .541 .605
acquaintances
(4 items)
Drug misuse (6 items) .880 .873 .868 .850 -b .848
Alcohol misuse (5 items) .886 .882 .857 .893 .862 .873
Emotional well-being .789 .775 .801 .819 .688 .830
(5 items)
Antisocial personality .860 .850 .829 .873 .816 .820
patterns (8 items)
Antisocial .847 .845 .824 .837 .860 .762
attitudes/cognitions
(5 items)
a
The scale “Relationship with partner/family” originally consisted of 5 items. However, because of the
relatively low value of McDonalds (.623) and item-rest correlation (.143) we removed the item “family
member has committed crimes in the past” from this scale.
b
The item variance in this offender group for this particular subscale was too low to calculate McDonalds.
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 7

groups with the exception of the violent offender group (is = .672). For the
subscale “antisocial friends/acquaintances”, the value of McDonalds was > .70 in
most offender groups with the exception of the sexual (= .541) and one-time
offender (is = .605) groups. For the subscale “relationship with partner/family”,
the value of McDonalds was > .60 in all offender groups with the exception of the
of violent offender group (= .529). Last, for the subscale “criminal history”
a McDonalds larger than .60 was found in most offender groups except in the
violent (= .489) and one-time offender (= .570) groups.

Recidivism
Recidivism was defined as any new conviction in a period of 3.5 years after
the assessment of the RISc, and was coded as 0 (no new conviction) or 1 (at
least one new conviction). The type of offenses committed in the follow up
period was also noted. Nonviolent recidivism comprised property crimes
without violence, public order offenses, drug offenses, traffic offenses, and
other nonviolent offenses. Violent recidivism comprised violent offenses and
property crimes with violence. Finally, sexual recidivism comprised sexual
offenses, such as sexual assault and rape.
Recidivism data were obtained from the national government database of
officially recorded information on convictions. The data used in this study
were retrieved in anonymous form, as the RISc scores and the recidivism rate
of each offender in the sample were merged using an anonymous identifier,
meaning that data could not be linked to identifiable or individual offenders.
Only the Dutch probation services could link the anonymous identifiers to
individual offenders.

Analyses
The statistical analyses were conducted stepwise. First, partial point-biserial
correlations (rpb) between the RISc domain scores and the various types of
recidivism were calculated to examine whether and how risk factors were
associated with the different types of recidivism. These correlations were
adjusted for gender and age, as preliminary results revealed significant
associations between these variables and each recidivism type. Second, the
correlation coefficients were transformed into Fisher z values, after which
a series of z tests were conducted to examine differences in the (strength of
the) association between a risk factor and the different types of recidivism.
Third, area under the receiver-operating-characteristic curve (AUC) values
were computed to examine the strength of the associations between the total
RISc score and the different recidivism types, as AUC values – in contrast to
correlations – are not sensitive to base rate differences. Fourth, bivariate
Pearson correlation coefficients were calculated to examine the strength of
the associations between the different RISc domain scores. Fifth, the relative
8 C. E. VAN DER PUT ET AL.

importance of the risk factors for each type of recidivism was determined by
building three separate logistic regression models. In these models, the
unique contribution of each RISc domain to the prediction of recidivism
was analyzed for nonviolent, violent, and sexual recidivism, respectively.
Fifth, chi-square tests were conducted to determine any differences in demo-
graphic characteristics between the offender groups. Cramer’s V values were
calculated to make inferences about the magnitude of differences. Last, to
find out if there are differences in scores on each of the (static or dynamic)
RISc domains between the five offender groups, a series of ANOVA’s was
performed in which was controlled for gender, origin, and age of offenders.

Ethical approval
Formal Institutional Review Board (IRB) approval for conducting this study
was not required, as it concerned analyses of secondary and de-identified
data, which does not pose harm to the participants, and therefore does not
require IRB approval. Accordingly, this study was ethically conducted given
the rules and guidelines of the Faculty Ethics Review Board (FMG-UvA) of
the University of Amsterdam, the Netherlands.

Results
Table 2 shows the partial point-biserial correlations between the RISc domain
scores and the different recidivism types, adjusted for gender and age. Significant
positive correlations were found between all domain scores and recidivism for
both nonviolent and violent recidivism. For sexual recidivism, the domain scores
criminal history, antisocial friends/acquaintances, emotional well-being, antiso-
cial personality patterns, and antisocial attitudes/cognitions were positively
correlated. According to Rice and Harris (2005), the criteria for a small, medium,
and large effect size at a 50% base rate are .10, .24, and .37, respectively. The
recidivism rates for nonviolent recidivism, violent recidivism, and sexual reci-
divism were 42.8%, 21.8%, and 1,9% respectively. Therefore, adjusted criteria for
small, medium, and large effect sizes were calculated using conversion formulae
(Rosenthal, 1991; Swets, 1986). For a 42.8% base rate (of nonviolent recidivism),
the criteria for small, medium, and large effect sizes are .10, .24, and .37,
respectively. For a 21.8% base rate (of violent recidivism), these criteria are.08,
.20, and .31, whereas for a 1.9% base rate (of sexual recidivism), these criteria are
.03, .07, and .11. For nonviolent recidivism, medium effect sizes were found for
the domains criminal history, education/work, antisocial friends, and drug
misuse, whereas small effect sizes were found for the domains accommodation,
financial management/income, alcohol misuse, emotional well-being, antisocial
personality patterns, and antisocial attitudes. For violent recidivism, medium
effect sizes were found for the domains criminal history, education/work, and
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 9

Table 2. Point-Biserial correlations between RISc scales and nonviolent, violent, and sexual
recidivism, and differences in correlations between recidivism types (Expressed in Fisher’s z).
Nonviolent Violent Sexual Sexual
recidivism recidivism recidivism Violent vs. Sexual vs. vs.
(n = 3708) (n = 1888) (n = 164) Nonviolent Nonviolent Violent
Criminal history .335** .243** .032** 3.55*** 3.93*** 2.63**
Accommodation/Living .208** .135** .011 2.66** 2.49* 1.52
situation
Education, work, and .257** .203** .021 2.02* 3.00** 2.25*
training
Financial management .224** .151** −.004 2.68** 2.88** 1.90+
and income
Relationship with .077** .139** −.003 −2.22* 1.00 1.74+
partner/family
Antisocial friends/ .268** .170** .029** 3.64*** 3.05** 1.74+
acquaintances
Drug misuse .243** .181** −.006 2.30* 3.15** 2.30*
Alcohol misuse .187** .157** −.001 1.09 2.36* 1.94+
Emotional well-being .104** .126** .023* −0.79 1.01 1.26
Antisocial personality .222** .230** .041** −0.30 2.29* 2.35*
patterns
Antisocial .207** .179** .025* 1.03 2.30* 1.90+
attitudes/cognitions
Total RISc score .356** .274** .026* 3.22** 4.30*** 3.11**
All partial correlations were adjusted for gender and age.
+
p <.10; *p <.05; **p <.01; *** p <.001

antisocial personality patterns, whereas small effect sizes were found for all other
domain scores. For sexual recidivism, small effect sizes were found for criminal
history and antisocial personality patterns. The AUC value of the total RISc
score was 0.723 (95% CI [0.712, 0.733]) when predicting nonviolent recidivism,
0.704 (95% CI [0.692, 0.717]) when predicting violent recidivism, and 0.585
(95% CI [0.546, 0.625]) when predicting sexual recidivism.
To examine differences in the strength of the correlations between RISc
domains and the three recidivism types, Fisher’s z tests were performed.
Table 2 reveals that most significant differences in correlations were found
between sexual and nonviolent recidivism. The domains criminal history,
accommodation/living situation, education/work/training, financial manage-
ment/income, antisocial friends/acquaintances, drug misuse, alcohol misuse,
antisocial personality patterns, and antisocial attitudes/cognitions were all
more strongly related to nonviolent recidivism than to sexual recidivism.
Further, the domains criminal history, accommodation/living situation, edu-
cation/work/training, financial management/income, antisocial friends/
acquaintances, and drug misuse were more strongly related to nonviolent
recidivism than to violent recidivism. Finally, the domains criminal history,
education/work/training, drug misuse, and antisocial personality patterns
were more strongly related to violent recidivism than to sexual recidivism.
To examine the strength of the associations between the different domain
scores, Pearson correlation coefficients were calculated (see Table 3). The
10 C. E. VAN DER PUT ET AL.

Table 3. Pearson correlation coefficients between the RISc domain scores.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
1. Criminal history 1.00
2. Accommodation/Living situation .34 1.00
3. Education, work, and training .47 .41 100
4. Financial management/income .37 .50 .47 1.00
5. Relationship with partner/family .26 .28 .33 .23 1.00
6. Relationship with friends/ .48 .38 .48 .43 .19 1.00
acquaintances
7. Drug misuse .44 .39 .46 .46 .24 .48 1.00
8. Alcohol misuse .28 .21 .21 .20 .21 .20 .26 1.00
9. Emotional well-being .21 .28 .39 .23 .40 .25 .32 .24 1.00
10. Antisocial personality patterns .50 .38 .56 .36 .43 .49 .40 .31 .49 1.00
11. Antisocial attitudes/cognitions .47 .33 .44 .31 .25 .44 .31 .21 .24 .70 1.00
All correlation coefficients are significant (p <.001)

strength of the association between most domain scores was medium to

large. In addition, multivariate logistic regression analyses were performed,
separately for the three recidivism types (see Table 4). The regression ana-
lyses showed only very small to small odds ratios ranging between 0.92 and
1.13 for most domain scores, meaning that the unique predictive value of the
domain scores (above all other domain scores) was (very) small.
As for the demographic characteristics of offenders, Table 5 reveals that
significant differences in multiple characteristics were found between offen-
der groups. In specific, the percentage of males was significantly higher in
sexual offenders (100.0) and generalists (94.2) than in violent offenders
(89.9), and the percentage of males was significantly higher in violent offen-
ders (89.9) than in nonviolent (78.3%) and one-time offenders (78.9%).
Further, sexual offenders consisted of significantly more native Dutch offen-
ders (90.0%) than generalists (72.4%), nonviolent offenders (73.5%), violent
offenders (60.4%), and one-time offenders (72.4%). The highest percentage of
non-Western offenders was found in violent offenders (35.5). As for the
mean age of offenders, sexual offenders were the oldest (Mage = 44.32)
compared to the other groups of offenders. Except for the gender differences,
all differences were only small in magnitude according to the criteria of
Cohen (1988).
Last, Table 6 reveals differences in RISc domain scores between the five
groups of offenders. The results revealed that the total RISc score was highest
for generalist offenders (M = 62.45), followed by nonviolent offenders
(M = 43.80), violent offenders (M = 36.92), sexual offenders (M = 34.60),
and one-time offenders (M = 23.63). In general, the results showed that
generalists scored highest on all domains followed by nonviolent offenders,
violent offenders, sexual offenders, and one-time offenders. Nonviolent
offenders scored on average significantly higher than violent offenders on
the domains criminal history, education/work/training, financial manage-
ment/income, relationship with friends/acquaintances, and drug misuse,
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 11

Table 4. Multiple logistic regression analyses predicting three recidivism types from the different
RISc domains with age and gender as control variables.
Nonviolent recidivism Violent recidivism Sexual recidivism
(n = 3708) (n = 1888) (n = 164)
B SE Wald χ2 OR B SE Wald χ2 OR B SE Wald χ2 OR
Criminal history .13 .01 252.69*** 1.13 .09 .01 109.09*** 1.10 .05 .03 4.45* 1.05
Accommodation/ .05 .01 14.89*** 1.05 −.01 .01 .31 .99 .03 .04 .43 1.03
Living situation
Education, work, .05 .01 29.60*** 1.05 .04 .01 13.75** 1.04 .00 .03 .00 1.00
and training
Financial .03 .02 2.76+ 1.03 .01 .02 .16 1.01 −.09 .05 3.01 .92
management
and income
Relationship with −.04 .01 11.30** .96 .04 .01 9.08** 1.04 −.08 .04 3.39 .96
partner/family
Relationship with .09 .02 30.32*** 1.09 −.01 .02 .23 .99 .05 .05 1.10 1.05
friends and
acquaintances
Drug misuse .04 .01 20.96*** 1.04 .03 .01 11.47** 1.03 −.06 .03 5.28* .94
Alcohol misuse .07 .01 65.48*** 1.07 .06 .01 35.65*** 1.06 −.03 .03 1.04 .97
Emotional well- −.04 .01 11.54*** .96 −.02 .01 3.30 .98 .04 .04 1.04 1.04
being
Antisocial .00 .01 .02 1.00 .08 .01 35.32*** 1.08 .11 .04 8.67** 1.11
personality
patterns
Antisocial .01 .01 .11 1.01 −.01 .02 .85 .99 −.05 .04 1.17 .95
attitudes/
cognitions
Gender .17 .09 3.46 1.18 .49 .12 15.61*** 1.63 1.97 .72 7.49** 7.15
Age −.03 .00 134.19*** .97 −.03 .00 110.05*** .97 −.03 .01 14.26*** .97
Constant −.92 .13 52.47*** .40 −2.45 .17 216.69*** .09 −5.54 .78 50.05*** .00
χ2(13) 1702.74*** 997.49*** 77.54***
B = Unstandardized regression weight; SE = Standard error; OR = Odds ratio; * p <.05; ** p <.01; *** p <.001.

Table 5. Demographic characteristics of offenders in different offender groups.

Variable Generalists Nonviolent Violent Sexual One-time
Categorical N % n % n % n % n % χ2 p V
Gender 450.95 <.01 .23
Male 5894 94.2a 806 78.3b 152 89.9 c 50 100.0a 913 78.9b
Female 364 5.8a 224 21.7b 17 10.1 c 0 0.0a 244 21.1b
Origin 44.97 <.01 .05
Native 4526 72.4a 755 73.5a 102 60.4b 45 90.0 c 835 72.4a
Dutch
Non- 1502 24.0a 225 21.9a,b 60 35.5 c 4 8.0d 245 21.2b
Western
Western 227 3.6a 47 4.6a,b 7 4.1a,b 1 2.0a,b 74 6.4b
Continuous N M n M n M n M n M F p η2
Age 6258 33.06a 1030 34.64b 169 35.24abd 50 44.32 c 1158 37.01d 39.14 <.01 .02
Generalists = Offenders committing multiple types of offenses; Nonviolent = Offenders committing only nonviolent
offenses including property offenses without violence, public order offenses, drug offenses, traffic offenses, and
other nonviolent offenses; Violent = Offenders committing only violent offenses and/or property offenses with
violence; Sexual = Offenders committing only sexual offenses; One-time = Offenders who have committed any
offense just once.
Percentages with the same subscript letter do not differ significantly from each other at the.05 level of
significance.
 12

Table 6. Differences in mean RISc domain scores between the five offender groups.
Generalists (n = 6258) Nonviolent (n = 1030) Violent (n = 169) Sexual (n = 50) One-time (n = 1158)
RISc Domain M (SD) M (SD) M (SD) M (SD) M (SD) F†
Criminal history 7.05 (3.62)a 4.69 (3.13)b 3.60 (2.77)c 4.14 (2.70)b,c 1.21 (1.75)d 701.06***
C. E. VAN DER PUT ET AL.

Accommodation/Living situation 1.79 (2.29)a 1.37 (2.10)b 0.93 (1.45)b,c 0.84 (1.69)b,c 0.67 (1.33)c 73.09***
Education, work, and training 5.69 (3.90)a 4.39 (3.70)b 3.46 (3.32)c 2.72 (3.43)c,d 2.49 (3.15)d 192.86***
Financial management and income 2.65 (2.28)a 2.52 (2.26)a 1.43 (1.74)b 0.88 (1.60)b 1.23 (1.74)b 111.17***
Relationship with partner/family 3.37 (2.26)a 2.06 (2.01)b 3.81 (2.06)a 3.10 (2.07)a,b 2.07 (2.64)b 96.45***
Antisocial friends/acquaintances 2.42 (2.07)a 2.13 (1.84)b 0.92 (1.47)c 1.16 (1.49)c 0.89 (1.32)c 141.77***
Drug misuse 3.44 (3.57)a 2.31 (3.13)b 1.25 (2.27)c 0.24 (1.02)c 0.77 (1.90)c 157.53***
Alcohol misuse 2.82 (3.12)a 1.59 (2.47)b 1.46 (2.42)b,c 1.12 (2.16)b,c 0.87 (1.97)c 137.82***
Emotional well-being 3.72 (2.66)a 3.12 (2.73)b 3.39 (2.85)a,b 4.38 (2.27)a 2.86 (2.77)b 57.99***
Antisocial personality patterns 8.05 (3.72)a 5.42 (3.41)b 7.33 (3.92)a 6.92 (3.57)a 4.86 (3.43)c 237.42***
Antisocial attitudes/cognitions 3.81 (2.61)a 2.64 (2.37)b 3.09 (2.44)b 3.34 (2.32)a,b 1.89 (2.00)c 149.49***
Total RISc score 62.45 (31.46)a 43.80 (26.09)b 36.92 (20.58)c 34.60 (21.17)b,c,d 23.63 (17.36)d 445.35***
Generalists = Offenders committing multiple types of offenses; Nonviolent = Offenders committing only nonviolent offenses including property offenses without violence, public
order offenses, drug offenses, traffic offenses, and other nonviolent offenses; Violent = Offenders committing only violent offenses and/or property offenses with violence;
Sexual = Offenders committing only sexual offenses; One-time = Offenders who have committed any offense just once.
Values with the same subscript do not differ significantly from each other.
† The ANOVA’s were controlled for gender, origin, and age of offenders.
* p <.05; ** p <.01; *** p <.001.
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 13

whereas violent offenders scored on average significantly higher than non-

violent offenders on the RISc domains relationship with partner/family and
antisocial personality patterns. The results further showed that nonviolent
offenders scored on average significantly higher than sexual offenders on the
domains education/work/training, financial management/income, antisocial
friends/acquaintances, and drug misuse, whereas sexual offenders scored on
average higher than nonviolent offenders on the domains emotional well-
being and antisocial personality patterns. No significant differences were
found between violent and sexual offenders in the average domain scores
and RISc total score.

Discussion
The aims of the present study were: (1) to examine differences in the
predictive value of general delinquency risk factors in predicting nonviolent,
violent, and sexual recidivism, and (2) to examine differences in the pre-
valence of risk factors between generalist, violent, sexual, nonviolent, and
one-time offenders. In general, the results revealed that risk factors measured
with the RISc were more strongly related to nonviolent than to violent
recidivism, and only weakly, or not at all, related to sexual recidivism. In
addition, differences were found in the unique contribution of risk factors to
the prediction of the different recidivism types. The total prevalence of risk
factors was highest in generalists (non-specialized offenders), followed by
nonviolent, violent, sexual offenders, and one-time offenders. In violent
offenders, risk factors in the domains relationship with the partner/family,
emotional well-being, antisocial personality patterns, and antisocial attitudes/
cognitions were more prevalent than in nonviolent offenders, whereas in
nonviolent offenders, risk factors in the domains education/work, financial
management/income, antisocial friends/acquaintances, and drug abuse were
more prevalent. Between violent and sexual offenders no differences were
found in the prevalence of risk factors.
Previous research showed that risk factors are more strongly related to
general recidivism than to violent and sexual recidivism (e.g., Hanson,
2009; Olver, Stockdale, & Wormith, 2014), but differences in risk factors
for violent and nonviolent recidivism have been far less studied. Our
results indicate that different risk factors are more strongly associated
with nonviolent than with violent recidivism, and that risk factors were
least associated with sexual recidivism. This implies that nonviolent reci-
divism can be better predicted by general delinquency risk factors than
violent recidivism, and that sexual recidivism cannot be predicted ade-
quately by general delinquency factors. Given these results, it may be
reasoned that the threshold for committing a new violent offense may be
higher than committing a new nonviolent offense, because the former
14 C. E. VAN DER PUT ET AL.

often have more serious consequences for victims and offenders.

Therefore, there may be other specific risk factors that substantially con-
tribute to the risk of violent recidivism, which were not measured in the
instrument (RISc) that was used in the present study for assessing general
delinquency risk factors. This may be in particular the case for sexual
recidivism, as it is more difficult to determine risk factors for sexual
recidivism than for other types of criminal recidivism given the low base-
rate (e.g., Spice, 2012).
The bivariate associations between the domain scores and nonviolent or
violent recidivism were small to medium in magnitude for most domain
scores. However, the regression analyses produced only very small odds
ratios for most domain scores. This means that the unique predictive value
of the domain scores (above all other domain scores) was (very) small, which
may be due to the medium to high correlation coefficients that were found
for the associations between the different domain scores. Further, we found
that the predictive value of the total RISc score corresponded with a high
effect size for predicting nonviolent recidivism (AUC = .723), with a medium
effect size for predicting nonviolent recidivism (AUC = .704), and with
a small effect size for predicting sexual recidivism (.585) (given the criteria
of Rice & Harris, 2005). Previous research also showed that risk assessment
instruments for general recidivism are only weak predictors for sexual reci-
divism (e.g., Schmidt et al., 2011; Seto & Lalumière, 2010; Viljoen, 2009).
A specific risk assessment instrument seems to be needed for predicting
sexual reoffending to an acceptable degree, in which specific risk factors,
such as deviant sexual interests, prior sexual offenses, and deviant victim
choices are assessed (Hanson & Bussiere, 1998; Hanson & Morton-Bourgon,
2005).
The results revealed that the total prevalence of risk factors was higher in
nonviolent offenders than in violent offenders. In violent offenders, we found
that risk factors in the domains relationship with the partner/family, emo-
tional well-being, antisocial personality patterns, and antisocial attitudes/
cognitions were more prevalent than in nonviolent offenders, whereas in
nonviolent offenders, risk factors in the domains education/work, financial
management/income, antisocial friends/acquaintances, and drug abuse were
more prevalent. To a more general extent, it seems that individual risk factors
(emotional well-being, antisocial personality patterns, and antisocial atti-
tudes/cognitions) are more prevalent in violent offenders, whereas environ-
mental risk factors (education/work, financial management/income,
antisocial friends/acquaintances) are more prevalent in nonviolent offenders.
Lai et al. (2016) who studied differences in risk factors between nonviolent
and violent youth offenders found no significant difference between non-
violent and violent offenders in the total prevalence of risk factors.
Nonviolent offenders had higher risk scores than violent offenders in the
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 15

criminal history, family circumstances/parenting, substance abuse, and atti-

tudes/orientation domains, whereas violent offenders had higher risk scores
in the peer relation and personality/behavior domains. The results of our
study are therefore only partly in line with the results of Lai and colleagues,
which may be due to differences in the populations studied (youth versus
adult offenders, and Asian versus European offenders).
We did not find any significant differences in risk domain scores between
violent and sexual offenders, but this may be due to insufficient statistical
power in detecting differences given the relatively low number of violent
(n = 169) and sexual offenders (n = 50). Seto and Lalumière (2010) con-
ducted a meta-analysis on differences between adolescent sexual and non-
sexual offenders and found that sexual offenders had much less extensive
criminal careers, fewer affiliations with antisocial peers, and fewer substance
use problems than non-sexual offenders. In addition, Craig et al. (2006)
found that violent offenders were more problematic than sexual offenders
in terms of having a chaotic lifestyle, suffering from psychopathology, and
abusing substances. In the present study, we did find differences in levels of
risk factors in sexual than in violent offenders, but these differences were
non-significant.
Several limitations need to be discussed. First, we analyzed a sample of
Dutch adult offenders, which raises the question to what degree the current
findings can be generalized to non-Dutch populations. However, previous
studies found that the impact of risk factors for delinquency was equivalent
in American and Dutch samples (Esbensen & Weerman, 2005; Van der Put,
2012), so we expect that our results can at least to some extent be generalized
to populations in other Western countries. Second, we only focused on the
predictive value of risk factors and not protective factors. Previous studies
showed that protective factors may buffer or mitigate the impact of risk
factors, or decrease the likelihood of violent recidivism and the development
of youth violence (Lodewijks, de Ruiter, & Doreleijers, 2010; Lösel &
Farrington, 2012; Ullrich & Coid, 2011). Furthermore, some evidence is
found for the predictive value of a number of protective factors for sexual
reoffending (De Vries Robbé, 2015). Therefore, it is interesting to direct
future research on examining differences in the impact of protective factors
between recidivism types as well as offender groups, and on unraveling how
protective factors interact with risk factors. Third, recidivism was defined as
any new conviction registered in official court records in a 3.5 year follow up
period. Using official records may imply an underestimation of the true
recidivism prevalence, as, for instance, minor offenses have not (always)
been recorded. On the other hand, both officially registered offenses and self-
reports on committed offenses have their limitations in analyzing recidivism
data (Breuk, 2007), and this should be taken into account when interpreting
results of studies on recidivism. Fourth, the majority of the total sample of
16 C. E. VAN DER PUT ET AL.

offenders comprised generalist offenders, whereas smaller numbers of

sampled offenders comprised the three specialized offender groups.
Especially the sexual offender group was small (n = 50), which has an effect
on the statistical power to detect significant results. Other studies also found
that most offenders are generalists rather than specialists (Seto & Lalumière,
2010; Simon, 1997). However, it may be possible that offenders, who were
initially identified as specialists, turned out to be generalists, which may in
particularly be the case for relatively young offenders who just started their
offending career. A final limitation is that the data were retrieved from the
Dutch probation services. These data were collected for clinical purposes and
not specifically for scientific research, which may have affected the accuracy
of the data collection.

Conclusion
The present results contribute to our knowledge of differences in risk factors
for various types of recidivism between offender groups. Several implica-
tions for clinical practice can be derived from our results. First, we found
differences in levels of risk factors between different offender groups as well
as differences in the impact of risk factors across recidivism types. For
violent offending, risk factors in the domains relationship with the part-
ner/family, emotional well-being, antisocial personality patterns, and anti-
social attitudes/cognitions seem relatively important, whereas the risk
domains education/work/training, financial management/income, antisocial
friends/acquaintances, and drug misuse seem relatively important for non-
violent offending. This indicates that in clinical practice, different
approaches are needed to treat offenders so that specific types of recidivism
can be prevented successfully. According to the Risk Need Responsivity
(RNR) model for effective offender rehabilitation, offender therapy should
focus on those criminogenic needs that are most strongly related to criminal
behavior (e.g., Andrews & Bonta, 2010). Our findings provide more insight
into the variety of these criminogenic needs across offender groups, which is
important to take into account in both assessment procedures and referral
of offenders to the most appropriate interventions. These interventions
should not only target an offender’s criminogenic needs – which may differ
across offender types -, but should also be delivered with the right intensity
as prescribed by the risk principle of the RNR model (Andrews & Bonta,
2010).
Consistent with findings of previous studies (Schmidt et al., 2011; Viljoen
et al., 2009), we found that sexual recidivism could not be predicted by the
general risk assessment instrument that we used in the present study. This
indicates that different and more specialized risk assessment instrument are
needed to successfully predict sexual reoffending. However, sexual offenders
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 17

are approximately three times more likely to reoffend by committing a non-

sexual offense than a sexual offense (e.g., Langström & Grann, 2000; McCann
& Lussier, 2008; Rasmussen, 1999; Vandiver, 2006). Therefore, in sexual
offenders, it is not only important to assess the risk of sexual recidivism
with a specialized instrument, but also to assess the risk of recidivism in
other offense types with a more general risk assessment instrument.

Authors’ contributions
CvdP designed the study, performed the statistical analyses, helped drafting the manuscript
and critically reviewed the manuscript. MA helped drafting the manuscript and critically
reviewed the manuscript. JG searched and reviewed previous literature and drafted the
manuscript. All three authors contributed to and approved the final version of the
manuscript.

Disclosure Statement
No potential conflict of interest was reported by the authors.

References
Andrews, D. A., & Bonta, J. (2010). The psychology of criminal conduct (5th ed.). New
Providence, NJ: LexisNexis Matthew Bender.
Andrews, D. A., Bonta, J., & Hoge, R. D. (1990). Classification for effective rehabilitation:
Rediscovering psychology. Criminal Justice and Behavior, 17, 19–52. doi:10.1177/
0093854890017001004
Andrews, D. A., Bonta, J., & Wormith, S. J. (2004). The Level of Service/Case Management
Inventory (LS/CMI). Toronto, Canada: Multi-Health Systems.
Assink, M., Van der Put, C. E., Hoeve, M., De Vries, L. A., Stams, G. J. J. M., & Oort, F. J.
(2015). Risk factors for persistent delinquent behavior among juveniles: A meta-analytic
review. Clinical Psychology Review, 42, 47–61. doi:10.1016/j.cpr.2015.08.002
Bonta, J., Harman, W. G., Hann, R. G., & Cormier, R. B. (1996). The prediction of recidivism
among federally sentenced offenders: A re-validation of the SIR scale. Canadian Journal of
Criminology, 38, 61–97. doi:10.3138/cjcrim.38.1.61
Bonta, J., Law, M., & Hanson, K. (1998). The prediction of criminal and violent recidivism
among mentally disordered offenders: A meta-analysis. Psychological Bulletin, 123,
123–142. doi:10.1037/0033-2909.123.2.123
Breuk, R., Clauser, C., Stams, G. J. J. M., Slot, W., & Doreleijers, T. H. (2007). The validity of
questionnaire self-report of psychopathology and parent-child relationship quality in
juvenile delinquents with psychiatric disorders. Journal of Adolescence, 30, 761–771.
doi:10.1016/j.adolescence.2006.10.003
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. New York, NY:
Routledge Academic.
Cottle, C. C., Lee, R. J., & Heilbrun, K. (2001). The prediction of criminal recidivism in
juveniles: A meta-analysis. Criminal Justice and Behavior, 28, 367–394. doi:10.1177/
0093854801028003005
18 C. E. VAN DER PUT ET AL.

Craig, L. A., Browne, K. D., Beech, A., & Stringer, I. (2006). Differences in personality and
risk characteristics in sex, violent and general offenders. Criminal Behaviour and Mental
Health, 16, 183–194. doi:10.1002/cbm.618
De Vries Robbé, M., Mann, R. E., Maruna, S., & Thornton, D. (2015). An exploration of
protective factors supporting desistance from sexual offending. Sexual Abuse: A Journal of
Research and Treatment, 27, 16–33. doi:10.1177/1079063214547582
Durose, M. R., Cooper, A. D., & Snyder, H. N. (2014). Recidivism of prisoners released in 30
states in 2005: Patterns from 2005 to 2010. Washington, DC: US Department of Justice,
Office of Justice Programs, Bureau of Justice Statistics.
Esbensen, F. A., & Weerman, F. M. (2005). Youth gangs and troublesome youth groups in the
United States and the Netherlands: A cross-national comparison. European Society of
Criminology, 2, 5–37. doi:10.1177/1477370805048626
Forth, A. E., Kosson, D. S., & Hare, R. D. (2003). The psychopathy checklist: Youth version.
Toronto, ON: Multi-Health Systems.
Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of the predictors of adult
offender recidivism. What Works! Criminology, 34, 575–607. doi:10.1111/j.1745-9125.1996.
tb01220.x
Gibbs, A. (1999). The assessment, case management and evaluation system. Probation
Journal, 46(3), 182–186.
Hanson, R. K. (2009). The psychological assessment of risk for crime and violence. Canadian
Psychology, 50, 172–182. doi:10.1037/a0015726
Hanson, R. K., & Bussiere, M. T. (1998). Predicting relapse: A meta-analysis of sexual
offender recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348–362.
doi:10.1037/0022-006X.66.2.348
Hanson, R. K., & Morton-Bourgon, K. E. (2005). The characteristics of persistent sexual
offenders: A meta-analysis of recidivism studies. Journal of Consulting and Clinical
Psychology, 73, 1154–1163. doi:10.1037/0022-006X.73.6.1154
Hoge, R. D., & Andrews, D. A. (2002). The youth Level of Service/Case Management
Inventory. Toronto, ON: Multi-Health Systems.
Hollin, C. R., & Palmer, E. J. (2006). Criminogenic need and women offenders: A critique of
the literature. Legal and Criminological Psychology, 11, 179–195. doi:10.1348/
135532505X57991
Howard, P., Clark, D., & Garnham, N. (2003). Evaluation and validation of the Offender
Assessment System (OASys). London, U.K.: OASys Central Research Unit.
Howard, P., Clark, D., & Garnham, N. (2006). An evaluation of the Offender Assessment
System (OASys) in three pilots 7999–2001. London, U.K.: Prison and Probation Services.
Lai, V., Zeng, G., & Chu, C. M. (2016). Violent and nonviolent youth offenders: Preliminary
evidence on group subtype. Youth Violence and Juvenile Justice, 14, 313–329. doi:10.1177/
1541204015615193
Langström, N., & Grann, M. (2000). Risk for criminal recidivism among young sex offenders.
Journal of Interpersonal Violence, 15, 855–871. doi:10.1177/088626000015008005
Lipsey, M. W., & Derzon, J. H. (1998). Predictors of violent or serious delinquency in
adolescence and early adulthood: A synthesis of longitudinal research. In R. Loeber &
D. P. Farrington (Eds.), Serious and violent juvenile offenders: Risk factors and successful
interventions (pp. 86–105). Thousand Oaks, CA: Sage.
Lodewijks, H. P., de Ruiter, C., & Doreleijers, T. A. (2010). The impact of protective factors in
desistance from violent reoffending: A study in three samples of adolescent offenders.
Journal of Interpersonal Violence, 25, 568–587. doi:10.1177/0886260509334403
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 19

Lösel, F., & Farrington, D. P. (2012). Direct protective and buffering protective factors in the
development of youth violence. American Journal of Preventive Medicine, 43, 8–23.
doi:10.1016/j.amepre.2012.04.029
Mann, R. E., Hanson, K., & Thronton, D. (2010). Assessing risk for sexual recidivism: Some
proposals on the nature of psychologically meaningful risk factors. Sexual Abuse, 22(2),
191–217. doi:10.1177/1079063210366039
McCann, K., & Lussier, P. (2008). Antisociality, sexual deviance, and sexual reoffending in
juvenile sex offenders: A meta-analytical investigation. Youth Violence and Juvenile Justice,
6, 363–385. doi:10.1177/1541204008320260
Olver, M. E., Stockdale, K. C., & Wormith, J. S. (2014). Thirty years of research on the Level
of Service Scales: A meta-analytic examination of predictive accuracy and sources of
variability. Psychological Assessment, 26(1), 156. doi:10.1037/a0035080
Orbis Partners. (2007). Long-term validation of the Youth Assessment and Screening
Instrument (YASI) in New York State Juvenile probation. Retrieved from http://www.
criminaljustice.ny.gov/opca/pdfs/YASI-Long-Term-Validation-Report.pdf
Prentky, R. A., Knight, R. A., & Lee, A. F. S. (1997). Risk factors associated with recidivism
among extrafamilial child molesters. Journal of Consulting and Clinical Psychology, 65, 141
−149. doi:10.1037/0022-006X.65.1.141
Rasmussen, L. (1999). Factors related to recidivism among juvenile sexual offenders. Sexual
Abuse, 11, 69–85. doi:10.1023/A:1021332929575
Rice, M. E., & Harris, G. T. (2005). Comparing effect sizes in follow-up studies: ROC Area,
Cohen’s d, and r. Law and Human Behavior, 29, 615–620. doi:10.1007/s10979-005-6832-
7
Rosenthal, R. (1991). Meta-analytic procedures for social research. Newbury Park, CA: Sage.
Schmidt, F., Campbell, M. A., & Houlding, C. (2011). Comparative analyses of the YLS/CMI,
SAVRY, and PCL:YV in adolescent offenders: A 10-year follow-up into adulthood. Youth
Violence and Juvenile Justice, 9, 23–42. doi:10.1177/1541204010371793
Seto, M. C., & Lalumière, M. L. (2010). What is so special about male adolescent sexual
offending? A review and test of explanations through meta-analysis. Psychological Bulletin,
136, 526–575. doi:10.1037/a0019700
Simon, L. M. J. (1997). Do criminal offenders specialize in crime types? Applied and
Preventive Psychology, 6, 35–53. doi:10.1016/S0962-1849(05)80064-2
Smith, P., Gendreau, P., & Swartz, K. (2009). Validating the principles of effective interven-
tion: A systematic review of the contributions of meta-analysis in the field of corrections.
Victims & Offenders, 4, 148–169. doi:10.1080/15564880802612581
Spice, A., Viljoen, J. L., Latzman, N. E., Scalora, M. J., & Ullman, D. (2012). Risk and
protective factors for recidivism among juveniles who have offended sexually. Sexual
Abuse, 25(4), 347–369. doi:10.1177/1079063212459086
Swets, J. A. (1986). Indices of discrimination or diagnostic accuracy: Their ROCs and implied
models. Psychological Bulletin, 99, 100–117. doi:10.1037/0033-2909.99.1.100
Ullrich, S., & Coid, J. (2011). Protective factors for violence among released prisoners—
Effects over time and interactions with static risk. Journal of Consulting and Clinical
Psychology, 79, 381–390. doi:10.1037/a0023613
Van der Knaap, L. M., & Alberda, D. L. (2009). De predictieve validiteit van de Recidive
Inschattingsschalen (RISc) [Predictive validity of the Recidivism Risk Assessment Scales
(RISc)]. Den Haag, the Netherlands: Ministerie van Justitie, WODC.
Van der Knaap, L. M., Alberda, D. L., Oosterveld, P., & Born, M. P. (2012). The predictive
validity of criminogenic needs for male and female offenders: Comparing the relative
impact of needs in predicting recidivism. Law and Human Behavior, 36, 413–422.
doi:10.1037/h0093932
20 C. E. VAN DER PUT ET AL.

Van der Knaap, L. M., Leenarts, L. E. W., & Nijssen, L. T. J. (2007). Psychometrische
kwaliteiten van de Recidive Inschattingsschalen (RISc). Interbeoordelaarsbetrouwbaarheid,
interne consistentie en congruente validiteit [Psychometric qualities of the Recidivism Risk
Assessment Scales (RISc). Inter-rater reliability, internal consistency, and convergent
validity]. Den Haag, the Netherlands: Ministerie van Justitie, WODC.
Van der Put, C. E., Deković, M., Stams, G. J. J. M., Hoeve, M., & Van der Laan, P. H. (2012).
Het belang van vroegtijdig ingrijpen bij jeugdcriminaliteit [The importance of early inter-
vention in juvenile crime]. Kind & Adolescent, 33(1), 2–20. doi:10.1007/s12453-012-0001-9
Vandiver, D. (2006). A prospective analysis of juvenile male sex offenders: Characteristics and
recidivism rates as adults. Journal of Interpersonal Violence, 21, 673–688. doi:10.1177/
0886260506287113
Viljoen, J. L., Elkovitch, N., Scalora, M. J., & Ullman, D. (2009). Assessment of reoffense risk
in adolescents who have committed sexual offenses: Predictive validity of the ERASOR,
PCL: YV, YLS/ CMI,and Static-99. Criminal Justice and Behavior, 36, 981–1000.
doi:10.1177/0093854809340991
Wartna, B. S. J., Blom, M., & Tollenaar, N. (2011). De WODC-Recidivemonitor. Retrieved
from https://www.wodc.nl/binaries/de-wodc-recidivemonitor-2011-20110803_tcm28-
78142.pdf
 JOURNAL OF FORENSIC PSYCHOLOGY RESEARCH AND PRACTICE 21

Appendix A

Table A1. RISc domains, examples of domain items, and response options.
Domain Example of item Response options
Criminal history
Number of previously committed 0 = no committed offenses, 1 = 1 to 3 committed
offenses offenses, 2 = more than 3 committed offenses
Accommodation/ Accommodation track record (in 0 = no, 1 = some experience with homelessness,
Living terms of periods of homelessness) 2 = more than 6 months of homelessness
situation
Quality of the living environment 0 = living environment does not contribute to
criminal behavior, 1 = living environment
contributes somewhat to criminal behavior,
2 = lives in criminal neighborhood/in close
proximity to victims
Education, work, Work experience and employment 0 = employed and does not quit job before a new
and training track record job is found, 1 = is in general employed, but quits
before a new job is found, 2 = not employed, or
unclear employment track record
Attitude toward education, work, 0 = motivated, 1 = somewhat motivated, 2 = not
or training motivated
Financial Current financial situation 0 = stable and appropriate, 1 = not always stable
management and appropriate, 2 = no insight in financial
and income situation
Gambling addiction or other 0 = no, 2 = yes
addiction
Relationship Quality of current relationship 0 = mutual relationships, 1 = problems with
with partner/ with partner, family, and other relationships, 2 = destructive, harmful
family relatives relationships
Family member has police record 0 = no, 2 = yes
Antisocial Friends and acquaintances 0 = rejects criminal behavior, 1 = is partly involved
friends/ in other’s criminal behavior, 2 = mostly criminal
acquaintances friends
Negative influence of friends 0 = no, 1 = is being used by friends, 2 = totally
dependent on friends
Drug misuse The offender’s criminal behavior is 0 = no, 1 = connections with criminal behavior,
linked to his or her drug use 2 = connections with criminal behavior and
violence
Drug use 0 = no, 1 = yes
Alcohol misuse Current alcohol use is problematic 0 = does not drink, 1 = drinking has some
influence, 2 = problematic drinker
The offender’s criminal behavior is 0 = no, 1 = connections with criminal behavior,
linked to his or her alcohol use 2 = connections with criminal behavior and
violence
Emotional well- Self-destructive behavior 0 = no, 2 = current or past self-destructive
being behavior
Mental problems 0 = no, 1 = no link with criminal behavior,
2 = long term mental health problems
Antisocial Impulsivity 0 = not impulsive, 1 = somewhat impulsive,
personality 2 = very impulsive
patterns
Dominant behavior 0 = not dominant, 1 = somewhat dominant,
2 = very dominant
Antisocial Pro-criminal attitudes 0 = accepts guilt, 1 = ambivalent feelings,
attitudes/ 2 = feels that crime pays off
cognitions
Motivation to change 0 = motivated, 1 = somewhat motivated, 2 = not
motivated