You are on page 1of 10

Journal of Criminal Justice 42 (2014) 203–212

Contents lists available at ScienceDirect

Journal of Criminal Justice

Employing mixed methods to explore motivational patterns of repeat


sex offenders
Joan A. Reid a,⁎, Eric Beauregard b, Karla M. Fedina c, Emily N. Frith c
a
School of Criminology and Justice Studies, University of Massachusetts Lowell, 113 Wilder Street, Suite 400, Lowell, MA 01854, United States
b
School of Criminology, Centre for Research on Sexual Violence, Simon Fraser University 8888 University Drive, Burnaby, British Columbia, V5A 1S6, Canada
c
Department of Rehabilitation and Mental Health Counseling, University of South Florida, 13301 Bruce B. Downs Blvd., Tampa, FL 33612-3807, United States

a r t i c l e i n f o a b s t r a c t

Available online 9 July 2013 Purpose: Understanding the motivation of sex offenders plays a key role in societal perception of victim cul-
pability and offender responsibility. This study identified patterns of offender motivation, assessed motiva-
tion stability across offenses, and estimated the influence of offense/victim specific factors on motivation in
comparison to offender factors.
Methods: Employing mixed-methods sequential explanatory research design, the study utilized confirmatory
factor analysis (CFA) to identify motivational factors underlying 346 crime events. Second, motivation stabil-
ity was assessed across offenses committed by 69 offenders. Finally, interview data were reviewed to assess
whether offender types emerged corresponding to CFA results.
Results: A motivational typology was identified comprised of five offender groupings built upon two underly-
ing constructs, one driven by sexual gratification and the other linked to anger/aggression. Minimal change in
motivation was observed across offenses committed by the same offender against different victims. Offense/
victim specific influence (22-23%) and offender related influence (77-78%) were similarly distributed with
both sexual and anger-driven motivation.
Conclusions: Motivation plays a key role in offending. Offense/victim specific factors have similar proportional
influence on crime motivation in comparison to offender factors irrespective of the type of motivation. Contrary
to popular public perception, victim blame is not warranted regardless of what motivates offenders.
© 2013 Elsevier Ltd. All rights reserved.

Introduction objective examination of evidence, and the skeptical evaluation of con-


clusions based on that evidence” (Palmer, 1988, p. 525). Relatedly, accu-
While researchers criticize the lack of empirical evidence regarding rate or inaccurate knowledge regarding criminal motivations of sex
the effectiveness of current strategies employed in preventing sexual offenders plays a key role in societal perceptions of victim culpability
violence (Casey & Lindhorst, 2009; Foubert, Godin, & Tatum, 2010; and offender responsibility. Recent studies have evaluated the impact
Tharp et al., 2011) and problems persist in determining what works of different types of offender motivation on public perception of victim
to reduce recidivism among sexual offenders (Olver, Stockdale, & blame for sex crimes (Angelone, Mitchell, & Lucente, 2012; Mitchell,
Wormith, 2011), conclusive evidence of the enduring imprint of Angelone, Kohlberger, & Hirschman, 2009; Sizemore, 2013). For exam-
sexual violence on victims (e.g., Campbell, Dworkin, & Cabral, 2009) ple, Mitchell et al. (2009) found that victims of rape by strangers were
demands the ardent pursuit of effective prevention and intervention perceived as more culpable in sexually-motivated offenses than in
efforts. In sum, sexual violence impacts the entire person – crushing violence-driven sex offenses. This disparity in victim blame is consid-
expectations of finding safety in the world, shattering affirmative be- ered a consequence of the notion that sexually-motivated offenses are
liefs about self-efficacy, and diminishing capacity for intimate rela- more likely to be prompted by victim characteristics and behaviors
tionships (Herman, 1992). than violence-driven offenses; thereby inferring that victims are more
Although researchers have expressed doubts concerning the utility culpable in cases of sexually-motivated crimes (Angelone et al., 2012;
of examining sex offender motivation as a means of refining prevention Mitchell et al., 2009). Providing even more compelling evidence of
efforts (Bryden & Grier, 2011); sexual violence, as any other type of the impact of any knowledge regarding offender motivation on victim
crime and societal problem, “is prevented by accurate knowledge blame, Angelone et al. (2012) found “simply providing knowledge
about its causes, and accurate knowledge can only be obtained by the of the perpetrator’s motivation [whether sexual or non-sexual] was
associated with decreased perceptions of the victim’s culpability and in-
⁎ Corresponding author. Tel.: +1 978 934 3956; fax: +1 978 934 3077. creased perceptions of the victim’s credibility in a case involving date
E-mail address: Joan_Reid@uml.edu (J.A. Reid). rape” (p. 2595).

0047-2352/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jcrimjus.2013.06.008
204 J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212

In terms of assisting in the healing process after sexual victimization, Piquero, 2004; Loewenstein, Nagin, & Paternoster, 1997). This research
exposure to victim blaming attitudes has consistently been found to pro- of behavioral economists and criminologists regarding the impact of
duce greater negative outcomes in survivors of sexual assault (Campbell, emotion and motivation on behavior suggests sex offending may be
Ahrens, Sefl, Wasco, & Barnes, 2001). Additionally, sexual assault survi- better understood by exploring these understudied forces (Loughran,
vors consider explanations of the causes of their victimization, particu- Paternoster, Piquero, & Pogarsky, 2011).
larly knowledge that may protect them from future harm, as beneficial More specifically, the role of motivation of sex offenders has long
and important for their healing (Ahrens, Cabral, & Abeling, 2009). been considered critical in developing targeted and effective treat-
Therefore, increasing the potential for avoidance of sexual victimiza- ment interventions for sex offenders. However, concerted efforts
tion and changing societal perceptions, even self-perceptions, regarding to understand motivation of sex offenders with application to crime
victims of sexual violence necessitates exploring the motivation of prevention have become sullied and stagnant due to somewhat in-
offenders. Accordingly, the purpose of this study was to identify motiva- flexible but well-intentioned perspectives surrounding the topic
tional constructs underlying 346 sex crimes committed by 69 repeat sex (Bryden & Grier, 2011). On one hand, the popular and entrenched
offenders incarcerated in a non-clinical setting using a mixed methods notion among victim advocates and rape crisis counselors espouses
approach. Based on previous research regarding motivations of sex that “the real cause of rape is gender inequality” (Campbell, Baker, &
offenders, two primary and independent motivational constructs were Mazurek, 1998, p. 473). Belief that the motivation of rapists is primarily
expected to be identified; one driven by a desire for sexual gratification generated by a need for male dominance rather than by a desire for sex-
and one linked to vindictiveness or aggression (Knight & Prentky, 1990; ual gratification has constructively altered the culture of victim-blaming
Knight, 1999). Utilizing a sample of sex offenses committed by repeat that typically surrounds sex crimes (Palmer, 1988). As shown by the
offenders facilitated the examination of stability in motivational pat- previously mentioned research, observers are more likely to assign re-
terns across offenses committed against different victims by the same sponsibility to victims of sexual assault if the offender is thought to be
offender and permitted the objective estimate of the proportional influ- motived by a desire for sexual gratification rather than by aggression
ence of offense or victim specific factors on crime motivation in compar- (Mitchell et al., 2009).
ison to offender factors. However, after a lengthy review of four decades of research and
commentary regarding motivation of rapists, Bryden and Grier
Redefining the role of motivation (2011) deduced, “No single line of research is conclusive, but we be-
lieve that the weight of the evidence indicates that sexual gratification
Enhancing opportunities for evading sexual victimization and re- is rapists’ most common (if not universal) goal” (p. 276). The re-
ducing recidivism rates among sexual offenders necessitates exploring viewers allow for the possibility of other additional motivations for
the common motivations underlying these crimes (Brantingham & rape but assert that “there is no reason to believe that they are often
Brantingham, 1993; Burgess, Commons, Safarik, Looper, & Ross, 2007; more influential than the sexual goal” (p. 276). So, while a reduction
Mustaine & Tewksbury, 2009; Robertiello & Terry, 2007). However, in victim blaming can be attributed to progressive affirmations
the study of motivation is considered by some to be relatively asserting that “the major motive for sexual assault is power … using
unproductive in terms of crime prevention. For example, Cornish and sex as the weapon” (Rape Advocacy Program, n.d., para. 3), scholarly
Clarke (2003) state that “situational crime prevention, with its selective research has consistently documented sexual arousal and excitement
attention to the process of crime commission itself, tends to treat the as a common, if not universal, motivating factor of sex offenders
offender's motivation as a ‘given’” (p. 50). As a result, research focused (for reviews, see Bryden & Grier, 2011; Robertiello & Terry, 2007).
on understanding offender motivation, a foundational component of In sum, incongruity exists between research findings linking the
several major criminological theories (Clarke & Cornish, 1985; Cohen motivation of sex offender to sexual gratification that may potentially
& Felson, 1979), lags behind criminological research focused on other el- facilitate societal tendencies toward victim blaming, and the popular-
ements of the crime process (Paulsen & Robinson, 2004; Sasse, 2005). ized script regarding the motive for sexual assault as power and control
Yet, motivation has been defined as “a dispositional variable that people that tends to result in less victim blaming. Therefore, a key purpose of
bring with them to every encounter” and also as a transactional phe- this study was to mitigate this incongruity by identifying motivational
nomenon “activated in any encounter by the demands, constraints, constructs underlying offenses by repeat sex offenders and assessing
and resources presented by the environment of action” (Lazarus, 1991, the relative stability of offender motivation across offenses involving
p. 819). In other words, motivation is not simply a trait, rather motiva- different victims. Perhaps in this way, a more accurate message may
tion emerges upon the fusion of “a particular motive trait with a suitable be found – one congruent with empirical findings that also minimizes
environment” (Lazarus, 1991, p. 819). Subsequently, motivation triggers victim blame.
and energizes behavior (Kleinginna & Kleinginna, 1981). Therefore,
without knowledge of offender motivation, both what constitutes a Previous research related to motivational typologies of sex offenders
criminogenic situation as well as the potentiality of crime within any
given situation remains unclear (Mustaine & Tewksbury, 2009). Based on the need for better treatment strategies for sex offenders,
Moreover, recent advances in research within the field of behav- research on sex offenders has frequently segregated samples of sex
ioral economics, as well as criminology, suggest that the influence of offenders by age of the victim (Beauregard, Leclerc, & Lussier, 2012;
motivation on choice and behavior needs to be reexamined (Ariely, Knight & Guay, 2006; Robertiello & Terry, 2007). In order to investigate
2008; Bouffard, 2011; Cullen, 2011; Piquero, Paternoster, Pogarsky, typologies of rapists, researchers selected offenders who assaulted only
& Loughran, 2011). Based on numerous studies, Ariely (2008) con- adults or adolescents over the age of 16, removing all other offenders
cluded that people tend to behave irrationally, yet in a predictable with reported offenses against children or younger adolescents as well
or systematic fashion, with their choices influenced by previously as those with offenses involving children, younger adolescents, and/or
understudied and misunderstood influences such as motivational adults from samples of sex offenders. Similarly, to study motivational
drives. Criminological researchers examining the links between internal typologies of child molesters, offenders who assaulted adults, older
constraints and externally imposed crime deterrents concluded that adolescents, or a mixture of victims across several age groupings were
motivation may overshadow any type of internal or external crime de- removed from the samples. Although separate typologies for rapists
terrent (Tittle & Botchkovar, 2005). Research studies, utilizing samples and child molesters may be useful in determining treatment modalities,
from the general population, have documented that people systemati- research has shown that sex offenders are not always specialized or
cally make irrational choices based on factors such as sexual arousal limited to one type of victim (i.e., assault only adults, adolescents
or anger (Ariely & Loewenstein, 2006; Bouffard, 2011; Carmichael & or children) but many sex offenders crossover between victim types
J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212 205

(Beauregard et al., 2012; Heil, Ahlmeyer, & Simons, 2003; Lussier, 2005; force and violence; however, their rage was not sexualized as with
Robertiello & Terry, 2007). sadistic offenders, nor was it solely directed at women as with vin-
As previously noted at the broadest level of classification, the dis- dictive offenders.
tinction in motivation of rapists (those who commit sex offenses
against adults or adolescents over the age of 16) has been commonly
Method
partitioned by whether the offense was sexually motivated or
non-sexually motivated (Bryden & Grier, 2011; Robertiello & Terry,
Based on the recommendation of Hicks and Sales (2006) and the
2007). Further specification of motivation of rapists hinged upon
need for analytically-derived motivational typologies of offenders,
these two basic classifications, with compensatory and sadistic types
the current study endeavored to identify motivational constructs
considered sexually-driven while power/control and opportunistic
using confirmatory factor analysis (CFA). Based on previous research,
types considered non-sexually motivated (for review, see Robertiello
two primary and independent motivational constructs were expected
& Terry, 2007). Motivation of those who commit sex crimes against
to be identified; one rooted in a desire for sexual gratification and
children has generally been categorized as fixated or regressed, with
one linked to vindictiveness or aggression (Knight & Prentky, 1990;
fixated offenders possessing a compulsive sexual attraction to chil-
Knight, 1999). To confirm the analytically-identified factors, explica-
dren and regressed offenders not being fixated on children but rather
tory interview data collected from the sample of repeat sex offenders
these offenders are thought to engage in offending due to stressors
regarding each offense were reviewed to assess whether offender
(Robertiello & Terry, 2007). However, in describing the motivational
types emerged comparable to the typological schematic of Knight
typology of child molesters, Knight and Guay (2006) noted that the
and Prentky (1990).
motivation typology of child molesters “could also be hypothetically
differentiated on the basis of dichotomous judgments of the presence
and absence of two motivations for rape – sex and aggression” (p. 35). Participants and procedures
Two types of child molesters labeled Compensatory and Impulse types
were expected to show lower levels of aggression, while the Impulse Data collected regarding 346 sexual offenses committed by a sam-
and Displaced-Aggression types were characterized as lower in sexual ple of 69 incarcerated repeat sex offenders were used in this study.
motivation (Knight & Guay, 2006). Repeat offenders were selected for the study sample, as they
The most relied upon research methodology for developing moti- assaulted more than one victim and it was possible to assess change
vation typologies of sex offenders, whether rapists or child molesters, in motivation across assaults. The sampling frame of this study
has involved the use of cluster analysis with investigators or clinical consisted of all male repeat sex offenders with at least one of their
experts used in the initial classification of offenders or to validate offenses involving a stranger and who had been sentenced to two or
the typologies (Knight & Guay, 2006). After reviewing the current more years of incarceration between 1994 and 2005 in the province
state of research on offender motivation, Hicks and Sales (2006) ad- of Quebec, Canada. Of the 92 offenders who met these criteria, 72
vocated that research advance beyond the use of cluster-based analy- agreed to participate and 69 provided qualitative data required by
ses, such as those based on clinical assessments, by first analytically the mixed-methods sequential explanatory design of the study. Of
identifying motivational factors and then validating these findings by the 20 offenders who met the study criteria yet did not participate,
assessing the degree to which groups emerge from the analytically- only 9 refused to participate. Eleven offenders were unavailable due
identified constructs. The advantage of analytic methods, such as confir- to their mental state, disciplinary issues, or transfer to different insti-
matory factor analysis (CFA), over group-based analysis that incorpo- tution. Depending on the number of offenses committed and partici-
rate clinically-derived taxonomies, is that such methods are more pant verbosity, the lengths of interviews ranged from 2 to 12 hours.
“agnostic” with respect to theory and less vulnerable to threats to inter- Information regarding the offenses provided during the interviews
nal validity stemming from investigator bias (Moffitt, 2006, p. 585). was compared with police reports. If discrepancies existed between
Although several motivational typologies have been proposed, offender accounts and police records, information in the police report
each with limited empirical validation (for reviews, see Robertiello & was used.
Terry, 2007; Knight & Guay, 2006); Knight and Prentky (1990) devel- Among the study participants, the majority were White (93%;
oped the motivational typology of sex offenders with the greatest n = 63) and mean age of the study sample at their first offense
level of empirical support and with potential for universal applicabil- was 30.7 years old (SD = 9.4). Rather than excluding certain sex
ity to all types of sex offenders (Knight & Guay, 2006). Knight and offenders based on victim age as is generally done when studying
Prentky proposed that two overarching and independent motivational motivation of sex offenders, this study used the complete sample of
constructs influence sex offenders: a) desire for sexual gratification, repeat sex offenders. The exclusion of offenders who show variation
and b) drive characterized by vindictiveness or aggression (see also in the ages of their victims was inconsistent with the purpose of the
Knight, 1999). Grounded upon these two drives, Knight and Prentky study that included investigating change or stability in motivation
(1990) hypothesized four types of sex offenders, which were further of sex offenders across offenses committed against different victims.
divided into subtypes. The first type, labeled opportunistic, was com- Within the sample were offenders who had exclusively sexually
prised of impulsive offenders who pursued available opportunities assaulted adult females (n = 16), female adolescents, with adoles-
in order to get immediate sexual gratification. These opportunistic cents defined as those 13 to 17 years old (n = 6), male adolescents
offenders did not exhibit excessive violence but rather only used (n = 1), girls with children defined as those 12 years old and under
aggression as a means of obtaining sexual gratification, with little re- (n = 6), boys (n = 4), or some combination (n = 36). This sample
gard for the victim. Second, Knight and Prentky described sexual of- of sex offenders, integrating offenders who victimized across different
fenders seeking sexual gratification who possessed extensive sexual age groups, parallels prior research findings indicating that the
fantasies and sexual preoccupation fused with aggression, dominance, majority of sex offenders do not specialize by victimizing one age
or coercion. These offenders were further categorized as sadistic group or victim type (e.g., Heil et al., 2003). The inclusion of offenses
and non-sadistic. A third type, labeled vindictive, was described as committed by offenders who exclusively victimized adults, only ado-
primarily angry toward women. As a result, sexual assaults perpe- lescents, or only children as well as those who victimized more indis-
trated by vindictive offenders included humiliation, degradation, criminately into one study sample is comparable to samples used
and physical harm of the victim. The fourth type, labeled pervasively in several prior studies regarding sex offender typologies (Knight &
angry, exhibited generalized anger and presented with a history of Prentky, 1990; Rosenberg & Knight, 1988). The number of offenses
antisocial and aggressive behavior. These offenders used excessive per offender ranged from 2 to 37 offenses. Thirty-two percent of
206 J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212

offenders reported information on two offenses, 30% on three of- Table 1


fenses, 12% on four offenses, and 9% reported on five or more offenses. Descriptive statistics and variance explained for all indicators in CFA (N = 346)

Indicators
Indicators of motivation EMOTIONS Positive Negative Multinomial Variance
Responses Explained
Motivation may be self-reported or detected by inference from R2
other types of indicators. Assessing motivation through inference Primary emotion 61% 58% sexually aroused 62%
has advantages over explicitly self-reported motivation in that it is before offense 3% happy/calm
less likely to be filtered by offender interpretation or biased by offender 39% 25% angry/frustrated
inadequacies in self-awareness (Hicks & Sales, 2006). Motivation is as- 7% anxious/agitated
7% sad/empty/lonely
sumed to activate behavior, and to be integrated with and influenced Primary emotion 83% 82% sexually aroused 84%
by cognitions and emotions (Kleinginna & Kleinginna, 1981; Lazarus, during offense 1% happy/calm
1991). Therefore, motivation can be inferred by behaviors, cognitions, 17% 12% angry/frustrated
and emotions. With specific application to motivations of sex offenders, 2% anxious/agitated
3% sad/empty/lonely
Prentky and Knight (1991) enumerated indicators useful for differenti-
Primary emotion 28% 19% happy/calm 25%
ating motivation among sex offenders including offender emotion, after offense 9% sexually aroused
offender thoughts and attitudes toward victims, and offender behavior 72% 41% guilty/regret
linked to victim treatment. 21% anxious/agitated
Accordingly, analogous variables available in the data were utilized 7% sad/empty/lonely
3% angry/frustrated
as manifest indicators to identify motivational factors underlying the
346 sex offenses assessed in this study. Displayed in Table 1 are the COGNITIONS YES NO
descriptive statistics of the 14 indicators of offender emotions, cogni-
Victim provoked offense 17% 83% 27%
tions, and behaviors that were included in the analysis. Indicators of Victim consented 18% 82% 92%
offender emotion available in the data were offender self-reported Victim deserved offense 4% 96% 36%
primary emotion before, during, and after the offense. These indica- Victim highly attractive 54% 46% 32%
tors were originally collected as multinomial variables. To facilitate
Positive Negative Multinomial
inclusion in factor analysis the indicators were dichotomized, with Responses
“0” indicating positive emotions and “1” indicating negative emotions.
Victim emotion 20% 12% sexually aroused 79%
The breakdown of the original multinomial variables and the descrip-
during offense 8% happy/calm
tive statistics for the modified binominal variables are shown in 80% 67% anxious/afraid
Table 1. 12% angry/frustrated
Also displayed in Table 1 are dichotomous indicators of offender 1% sad/empty/lonely
cognitions incorporated in the factor analysis including the belief
BEHAVIORS YES NO
that the victim consented, that the victim provoked the offense, and
that the victim deserved the offense. Also, offender rating of victim Force used 41% 59% 91%
Physical harm 20% 80% 46%
attractiveness was included as an indicator of offender cognition. In
Verbal hostility 56% 44% 94%
addition, offender report of victim emotion during the offense was Humiliated victim 10% 90% 57%
included as a cognitive indicator. As with the indicators of offender Selection of victim based 63% 37% 42%
emotion, to facilitate inclusion in the factor analysis this indicator on vulnerability
Selection of victim 36% 64% 48%
was dichotomized into a binominal variable. For all dichotomous
based on physical
yes/no variables, no was coded “0” and yes was coded “1”. Indicators characteristic
of offender behavior associated with victim treatment incorporated in
the factor analysis are also noted in Table 1 including force used,
physical harm of the victim, verbal hostility toward the victim, and standard error estimates when using data that are coarsely catego-
purposed humiliation of the victim. Other behavioral indicators in- rized and/or non-normally distributed (Brown, 2006; Flora & Curran,
cluded in the analysis were victim selection based on vulnerability 2004; Muthén & Muthén, 1998-2012). In addition, as the study data
and victim selection based on physical features.1 were collected regarding offenses committed by repeat offenders,
the analysis was specified as complex. This analytic specification
Analytic plan is appropriate for analyzing data obtained by stratification with
non-independence of observations due to cluster sampling (Muthén
A sequential quantitative-qualitative mixed method approach was & Muthén, 1998-2012). More specifically, this type of analysis pro-
used in this study (Maruna, 2010). Specifically, the statistical analysis vides standards errors and a chi-square test of model fit that takes
of quantitative data related to the 346 sexual offenses was followed into account non-independence of observations (Asparouhov, 2005).
by the examination of data collected from open-ended interview To begin assessment of the hypothesized model, the 14 factor
questions regarding the sexual offenses. The qualitative analysis of indicators were entered into the model based on face validity in
the detailed and explicatory interview data was used to validate and congruence with Knight and Prentky’s (1990) motivational factors
clarify the findings from the quantitative analysis. reflective of a desire for sexual gratification and anger or aggression
(see Fig. 1).2 Next, the CFA model was assessed based on results
Quantitative analysis: confirmatory factor analysis from an array of tests designed to assess different aspects of global
CFA was utilized in this study to confirm whether the selected model fit. A non-significant chi-square suggests that that the hy-
manifest indicators included in analysis were indicative of two under- pothesized CFA model fits the data adequately. Other indicators of
lying motivational factors in accordance with the motivational cate- adequate model fit include: Comparative Fit Index (CFI) of .95 or
gories proposed by Knight and Prentky (1990). Due to the inclusion higher; Tucker-Lewis Index (TLI) of .95 or higher; Root Mean Square
of dichotomous indicators, the weighted least squares mean and of Approximation (RMSEA) of .06 or less; and Weighted Root Mean
variance adjusted (WLSMV) estimator available with Mplus 7 was Square Residual (WRMR) of 1.0 or less (Hu & Bentler, 1999; Kline,
utilized as it is designed to provide more accurate parameter and 2005; Yu, 2002).
J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212 207

scores on both motivational factors were grouped together. The de-


gree of stability and change in offender motivation across offenses
was also assessed to validate the usefulness of the CFA findings re-
garding underlying motivational factors.

Qualitative analysis: comparative template analysis of interview data


As recommended by Hicks and Sales (2006), additional clarifying
and explicatory interview data collected from the offenders regarding
each sexual offense were reviewed to determine if a motivational ty-
pology of sexual offenders emerged based on the CFA-derived factors
scores. Specifically, the interview data were analyzed to explore
whether the groupings based on the analytically-derived factor scores
were comparable to the typological schematic provided by the prior
research of Knight and Prentky (1990). To facilitate the analysis of
the interview data, this study utilized template analysis (TA), a sys-
tematic method of analysis designed for use with interview data that
involves several ordered steps cumulating in a data-derived template
used to interpret and frame the study findings (King, 1998) supported
by ATLAS.ti 5.0 software (Muhr & Friese, 2004).
For this study, a priori themes were created based on the research
of Knight and Prentky (1990) including themes contrasting general-
ized anger versus specific anger toward women, impulsive actions
verses planned scheming, sexualized aggression verses instrumental
aggression, and purposed humiliation of victim based in sexualized
versus retaliatory fantasies. During the first step of data analysis, the
explicatory interview data were coded based on the a priori themes.
If these data were encompassed by the a priori themes, they were so
coded. If not, new themes were created or existing themes modified
to accommodate the data. To diminish the likelihood of intercoder
variation that could result from systematic differences between the-
matic framework of each coder, structured discussion with coders
followed each step in the data analysis.

Findings

Confirmatory factor analysis

The two-factor model was entered based on expected patterns of


Fig. 1. Hypothesized two-factor model.
emotions, behaviors, and cognitions in relation to two different sex
crime motivations (see Fig. 1). 3 This hypothesized model evidenced
To further assist in the process of evaluating model fit, Mplus pro- good fit to the observed patterns in the data, χ2(69) = 70.03, p =
vides modification indices (Muthén & Muthén, 1998-2012). Different .44. Supplementary fit indices also demonstrated good fit (CFI = .99;
from the previously mentioned model fit indictors that provide TLI = .99; RMSEA = .01; WRMR = .70). No modification indicators
an evaluation of global model fit, modification indices are useful in were greater than 4.00, suggesting that no substantial model changes
evaluating models for specific points of ill fit (Brown, 2006). A modi- were needed due to improve particular areas of the model (Brown,
fication index for a specific parameter notes the expected decrease in 2006). However, three non-significant paths with weak effect sizes
the overall model Chi-square test statistic (i.e., how the model overall were removed, slightly improving model fit. The model modifications
fit would improve) if the particular parameter was adopted, fixed, or aligned with the general theoretical assumptions of the study. The cog-
freed in the specified model (Brown, 2006). Models that adequately nition indicator “victim provoked the offense” loaded significantly on
represent the observed patterns in the data, or have good fit, should Factor 2 but not Factor 1, so the model was modified to include this var-
produce modification indices that are small in magnitude (Brown, iable as an indicator for Factor 2 only. The cognition indicators, “victim
2006). Modification indices with values greater than 4.00 were eval- deserved the offense” and “victim emotion during the offense”, loaded
uated to determine if changes in the model were needed due to par- significantly on Factor 1 but not on Factor 2; so these were modified
ticular areas of poor fit (Brown, 2006). as indicators for Factor 2 only. Displayed in Table 2, the final model
Subsequent to the evaluation of model fit, the factor loadings of the evidenced adequate fit to the data. The chi-square was not significant,
indicators were assessed to determine whether they provided support χ2(72) = 72.51, p = .46. Supplementary fit indices also demonstrated
for the study hypotheses. The factor indicators were all dichotomous good fit (CFI = .99; TLI = .99; RMSEA = .01, with 90% confidence in-
variables and consequently the estimate for these variables took the terval .00 -.03; WRMR = .73).
form of a conditional probability based on the value of predictor vari- Next, the factors loadings of the indicators were examined.
ables with probit coefficients reported (Liao, 1994). The findings of Sixteen of 18 factor loadings were statistically significant with most
good model fit, along with strong and significant loadings of the factor exhibiting strong effect sizes. The emotion indicators, primary emo-
indicators on the latent factors, were used to determine the quality of tion before, during, and after the offense loaded positively on to
the model (Kline, 2005). Once the CFA was completed, standardized Factor 1 with moderate effect sizes, signifying that experiencing neg-
factor scores associated with each of the identified factors for each ative emotions (i.e., anger, agitation) before and during the offense,
sex offense in the dataset were computed. To facilitate the next step and negative emotions (i.e., guilt, anxiety) after the offense were as-
of the analytic strategy, the offenses with similar standardized factor sociated with Factor 1 (see Table 1). Two cognitive indicators, belief
208 J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212

Table 2 regarding two independent motivational constructs (Knight &


CFA unstandardized, standardized, and critical ratio values CFA (standard errors and Prentky, 1990; Knight, 1999). In addition, the R2 corresponding to the
confidence intervals in parentheses; N = 346)
manifest variables indicate that the identified factors explain a consid-
Parameter Estimate Unstd. Coeff. Standarized Critical erable portion of the variance (between 19% and 94%) (see Column 5
(Std. Error) Coeff. [95% CI] Ratio (Z) of Table 1). For further reference, the unstandardized coefficients, stan-
Factor 1 Estimates dardized coefficients with 95% confidence intervals, and standard errors
Factor 1 by Primary emotion before .47 (.17) .45 [.13, .77] 2.80 for all model estimates are reported in Table 2.
offense
The degree of stability or change in offender motivation across of-
Factor 1 by Primary emotion during .37 (.14) .35 [.08, .63] 2.58
offense fenses was evaluated in two ways. First, the intraclass correlation
Factor 1 by Primary emotion after offense .38 (.18) .36 [.03, .70] 2.15 (ICC) for each set of factor scores was calculated to determine the
Factor 1 by Victim consented -.94 (.06) -.90 [-1.03, -.76] -16.35 consistency of the factor scores across offenses by the same offender.
Factor 1 by Victim deserved offense .63 (.11) .60 [.40, .81] 5.53 ICC represents the proportion of overall variance at the between
Factor 1 by Victim emotion during .93 (.07) .89 [.76, 1.02] 12.52
offender level. The ICC for the factor labeled anger motivation was
offense
Factor 1 by Force used .95 (.04) .95 [.88, 1.03] 25.81 .79, indicating that 79% of the variance in the anger motivation factor
Factor 1 by Physical harm .71 (.10) .68 [.49, .88] 6.85 scores was at the offender or between offender level, while 21% of the
Factor 1 by Verbal hostility 1.02 (.07) .97 [.90, 1.04] 15.35 anger motivation variance was within offender or at the level of the
Factor 1 by Humiliated victim .79 (.10) .75 [.58, .93] 8.13
offenses. The ICC for the factor labeled sexual motivation was .78, indi-
Factor 2 Estimates
Factor 2 by Primary emotion before -.94 (.24) -.65 [-.92, -.39] -3.92 cating that 78% of the variance in the sexual motivation factor scores
offense was at the offender or between offender level, while 22% of the sexual
Factor 2 by Primary emotion during -1.22 (.36) -.85 [-1.11, -.59] -3.36 motivation variance was within offender or at the level of the offenses.
offense Importantly, the ICC of factor scores for anger motivation and for
Factor 2 by Primary emotion after offense -.51 (.31) -.35 [-.71, .01] -1.63
sexual motivation are almost identical, indicating that the variance
Factor 2 by Victim provoked offense .75 (.32) .52 [.06, .98] 2.37
Factor 2 by Victim consented .51 (.28) .36 [.05, .67] 1.85 in crime motivation is similarly distributed when comparing variation
Factor 2 by Victim highly attractive .81 (.33) .56 [.15, .98] 2.43 at the between offender and within offender levels.
Factor 2 by Victim selection by .94 (.26) .65 [.40, .92] 3.67 Second, offenses with similar standardized factor scores on both
vulnerability
motivational factors were grouped together (see Fig. 2). Offenses by
Factor 2 by Victim selection by physical .70 (.15) .70 [.40, .99] 4.65
characteristic
the same offender were examined to assess stability or change in
Factor 1 with Factor 2 .01 (.11) .01 [-.30, .32] 0.07 motivation across offenses. Of the 69 offenders in the sample, five
2 offenders evidenced qualitative change in type of motivation across
Note: χ (72) = 72.51, p = .46; CFI = .99; TLI = .99; RMSEA = .01; WRMR = .73.
offenses (i.e., possessed high levels of sexual motivation with low
anger for one offense and high levels of anger with low sex motivation
that the victim deserved the assault and that the victim experienced for another offense) (see Fig. 3). Review of the five offenders with
negative emotions during the offense (i.e., fear, anger) were positively a qualitative change in motivation revealed that the factor scores of
related to Factor 1 with strong effect sizes. One cognitive indicator, be- the offenses committed these five offenders were moderate and not
lief that the victim consented, was negatively related to Factor 1 with extreme. In other words, the offenses committed by these offenders
a strong effect size. Four behavioral indicators, force used, physical did not score extremely high or extremely low on either anger motiva-
harm of the victim, verbal hostility toward the victim, and purposed tion or sexual motivation. Detailed examination of the individual
humiliation of the victim were all significantly associated to Factor 1 offenses of these offenders revealed exceptionally varied offenses
with strong effect sizes. Based on the factor loadings, anger motivation based on victim age (e.g., 9-year-old child victim of one offense vs.
was selected as the best fitting label for Factor 1, corroborating Knight 55-year-old adult victim of another offense) or assault characteristics
and Prentky’s (1990) prior research regarding anger or aggression as a (e.g., exhibition without any physical contact with the victim vs.
primary motivation of certain sex offenders. sexual assault with penetration). Of the remaining 64 offenders, 39 ex-
Two indicators of emotion, primary emotion before and during perienced no change in type or level of motivation across offenses and
the offense, were negatively associated with Factor 2 with strong 25 offenders had offenses with either slightly higher or lower levels of
effect sizes, signifying that experiencing positive emotion (i.e., sexual sexual or anger motivation but did not qualitatively change motiva-
arousal) before and during the offense were associated with Factor 2 tion. These findings of stability in the motivational groupings across
(see Table 1). Four cognitive indicators – the victim provoked the offenses committed by the majority of offenders provided evidence
offense, the victim regarded as highly attractive, victim selected by that the analytically-identified latent factors are tapping into underlying
vulnerability, and victim selected by physical characteristics – were and relatively stable motivational constructs.
positively associated with Factor 2 with strong effect sizes. One cogni-
tive indicator, offender belief that the victim consented, was related to Comparative template analysis of interview data
Factor 2 with a moderate effect size but failed to reach statistical sig-
nificance at the predetermined level (z = 1.85, p = .07). The offender Available elucidatory interview data were assessed by comparing
indicator of emotion after the offense was negatively associated with and contrasting themes within the groupings of offenses segmented
Factor 2 indicating that a positive emotional experience (i.e., happi- by factors scores (see Fig. 2) in order to determine whether offender
ness, sexual arousal) after the assault. However, this loading also did types emerged from the data corresponding to the analytically-
not reach statistical significance (z = –1.63, p = .10). Based on the identified motivation factors. Based on analysis of offender explana-
factor loadings of these indicators, sexual motivation was selected as tions or rationale for behaviors and treatment of victims, five major
the best fitting label for Factor 2, corroborating Knight and Prentky’s types of offenders emerged. Four types paralleled the prototypical
(1990) prior research regarding a desire for sexual gratification as a offenders within the motivational typology developed by Knight
key motivation for certain types of sex offenders. and Prentky (1990). One additional type of offender emerged that re-
As displayed in Table 2, Mplus provided the correlation coefficient sembled the opportunistic offender, yet was differentiated from other
between the two identified factors, showing that the two identified types of offenders by extensive and sustained drug or alcohol use.
motivational factors were independent from one another. The conclu- The explanations and rationale of offenses within the low-sex/
sion from this analysis is that two nearly orthogonal motivations under- low-anger quadrant of Fig. 2 revealed an offender type that closely
lie the sexual offenses within this sample, supporting prior research resembled the opportunistic offender within the Knight and Prentky
J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212 209

High-Sex/Low-Anger High-Sex/High-Anger
“Sexually Fixated” “Sadistic”
Factor Scores on Sex Motivation (M=0; SD=.56)

36
36

“Vindictive”

Low-Sex/Low-Anger
“Substance High-Anger/Low-Sex
“Opportunistic”
User”

Factor Scores on Anger Motivation (M=0; SD=.77)

Fig. 2. Emergent motivational patterns of sex offenders displayed on scatterplot of anger motivation factor score by sexual motivational factor score for offenses labeled by offender
(N = 346).

(1990) typology. Twelve offenders (17% of the sample) within this Characteristic statements regarding scheming or what may be la-
grouping committed offenses involving 50 victims (12% of sample). beled premeditated opportunism were given by Offender 65, stating
Two major themes emerged from the elucidatory data: a) easy access “he thought about the assaults in his fantasies in an unstructured way,
to potential victims, and b) some scheming or anticipating the “right but when the occasion presented itself he made his move with the victims
moment” to commit the assaults. Offenders in this group emphasized who happened to be there;” and by Offender 48, stating “he had not
accessibility to victims when providing rationale for their behavior, really premeditated the assaults but when he is in the presence of victims
with the majority (7 out of 12) explaining that contact with victims and the timing is perfect, then he starts to put his act together to bring
occurred in their home or place of employment. These opportunistic the victims where he wants…he manipulates young adolescent girls
offenders described how their employment facilitated prolonged or in the context of tattooing.” Three offenders mentioned occasional
repeated contact with potential victims or they revealed easy access marijuana use, and Offender 34 stated that he gave marijuana to victims
to victims because they occupied positions of authority or trust with “to make them uninhibited and ease them into sexual behaviors.” Of-
the victims. Prototypical explanations regarding ease of access were fenders 16 and 64 represented unusual cases within this set of of-
given by Offender 33 who explained that “victims were coming to fenders. Offender 16 detailed 11 assaults involving touching victims’
play with his son or biking on the street or were playing and swimming breasts. This offender admitted to intravenous drug use that he de-
at his house” and by Offender 34 who stated that he “selected victims scribed as his way of “minimizing feelings of inferiority he has had since
from his young employees or young customers of his store.” his accident left him mentally and physically handicapped.” Offender
63 reported heavy drug and alcohol use, stating that “the assaults
were unplanned and totally random” and “when under the influence of
substances, he would feel horny and try to have sex no matter what.” The
Low Sex/ varied ages of victims assaulted by Offender 63 – females aged 33, 11,
Low Anger and 6 years old – suggests randomness. The similarity of these two
9 7 offenders with the substance user offender described later indicates
that they most likely best fit in that category.
5 The clarifying explanations of the offenses within the high-sex/
No Change low-anger quadrant of Fig. 2 revealed an offender type that closely re-
High-Anger/ High Sex/
Low Sex Low-Anger 39 sembled the sexually fixated offender within the Knight and Prentky
(1990) typology. Twenty offenders (29% of the sample) within this
grouping committed offenses involving 96 victims (28% of the sample).
4 5 Themes emerged including: a) strong sexual fantasies and sexual im-
pulses, b) planned manipulation and luring of victims, and c) formation
High Sex/ of a contrived moral code. Offenders in this group emphasized sexual
High Anger
arousal, sexual fantasies, and overwhelming sexual impulses. For exam-
ple, Offender 10 stated that “he did not want to give children ‘that kind of
Fig. 3. Motivational change and stability across offenses by repeat offenders (N = 69). love’ but his sexual impulses were too overwhelming.” Similarly, Offender
210 J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212

41 detailed his typical behaviors prior to offending, stating he would they killed victims because they were “fascinated with death and
“drink to reduce his inhibitions and watch porn for stimulation and to wanted to know how it felt to kill,” with another offender asserting
maintain sexual fantasies … he then would hunt for someone to fulfill his that “killing makes him feel good.” Other offenders who killed victims
fantasy scenarios.” Offender 11 stated that “before the assaults he felt sex- were intoxicated and used excessive force resulting in the death of
ually aroused … planned the assaults as an elaboration of his fantasies.” the victims. A third subset of offenders stated that they killed their
Advanced planning and manipulation were described by Offender victims to avoid being caught and returning to prison.
50 who assaulted 18 males from 12 to 15 years of age and stated that Explanations of the offenses within the high-sex/high-anger quad-
“he used drugs to attract youth to his home and porn to lower victims’ rant of Fig. 2 revealed an offender type paralleling the sadistic offender
inhibitions … he introduced victims to games that would lead to sex.” within Knight and Prentky’s (1990) typology. Eight offenders (12%
Numerous offenders within this group (12 out of 20) noted limits of the sample) within this group committed offenses involving 84
to what they would or would not do based on what they considered victims (24% of the sample). Themes emerged including: a) excessive
harmful or wrong. For example, Offender 39 reported that “he did porn use, b) highly planned assaults, c) acting out deviant sexual fan-
not go further than oral sex and sexual contact because anything more tasies, and d) humiliation and force being experienced as sexually
would be rape, stating ‘I still have my principles.’” Five of the 20 of- arousing. Describing eroticized aggression, Offender 36 stated that
fenders mentioned using drugs or alcohol occasionally or prior to “he would rent up to eight S&M movies per day … he planned the sexual
the assault to reduce their inhibitions, but none described their sub- assaults in his fantasies… transforming his place into a sex slave prison.
stance use as habitual or excessive. Offender 32 stood out from the His accomplices brought him girls … very vulnerable girls with no place
other offenders in this group as he was the only offender with of- to go … he tortures them and he finds it sexually exciting to see them
fenses that scored below the mean on anger motivation who he killed suffer.” Offender 6 admitted “regular use of porn … stalking potential
a victim. Based on his explanation of his behavior, he did not kill the victims … being sexually aroused from what he is doing to the victim
victim out of anger or frustration but rather explained the murder as and from acting upon his deviant sexual fantasies.” Offender 21 was an
“part of his plan.” extreme case within the sample, assaulting 37 females, including 11
Twenty-five offenders (36% of the sample) committed 87 offenses women, 8 adolescents, and 18 girls. He reported that “he planned the
(25% of the sample) plotted within the high-anger/low-sex quadrant sexual assaults in his fantasies … in each case he urinated in victims’
of Fig. 2. Two offender types emerged from the analysis of the mouth during oral sex because he felt angry … and he was sexual aroused
interview data linked to the high-anger/low-sex quadrant. One type by acting out his deviant sexual fantasies.”
resembled the opportunistic offenders described above, yet the ma-
jority of offenders within this set emphasized heavy and continual Discussion
drug or alcohol use when explaining or rationalizing their behavior.
The second type strongly resembled the vindictive offender depicted The findings of this study support and extend prior research of
by Knight and Prentky (1990). Providing an analytically-based differ- Knight and Prentky (1990) by identifying an analytically-derived mo-
entiation of the two types, the anger motivation factor scores of the tivational typology of sex offenders built upon two underlying motiva-
offenses of vindictive offenders were primarily plotted one standard tions driving sexual offenses. One underlying motivational factor was
deviation above the mean while the anger motivation factor scores rooted in a desire for sexual gratification and the other was linked to
of the offenses linked to the substance user offender scored lower anger, vindictiveness, or aggression. Offenders with the lowest levels
(see Fig. 2). of these two motivational drives emphasized the influential effect
Themes emerged from the interview data of the substance user of- of opportunity on their criminal behavior. Sex offenders primarily
fender concerning: a) continual intoxication from drug or alcohol use, driven by sexual impulses revealed advanced planning of their crimes
b) facing hardships and failed relationships, c) impulsive offending corresponding to strong and persisting sexual fantasies. Sex offenders
behavior with minimal planning, and d) being determined to get sex primarily energized by anger rationalized sex offending as getting
however necessary. For example, Offender 55 reported that he “used payback or taking revenge for past insults by women, triggering
cocaine and pot every day to escape his pain and suffering … he feels offenses incorporating excessive humiliation and degradation of
trapped.” Offender 25 described himself as “always intoxicated and victims. Certain sex offenders who exhibited aggressive and impulsive
under the influence … he wants to have consensual sex but uses force tendencies defined their daily existence by excessive and continual
to get what he wants when the victims refuse.” Offender 22 admitted intoxication. While Knight and Prentky (1990) did not specify a dis-
“he was addicted to cocaine and alcohol” and stated “the assaults were tinct offender type primarily characterized by substance use, some
not planned when he met the victims, but when leaving the bar with studies have segregated substance users “whose clinical pictures
them he knew he would get sex whether they wanted it or not.” were dominated by substance abuse” into a distinct class, warning
Themes emerged from the interview data of the vindictive of- that “the consequential inclusion of these alcoholic offenders in
fenders including: a) rage and revenge fantasies involving women other groups may both confuse our understanding of the alcoholic
from the past or present, b) purposed humiliation, and c) advanced sexual offender and obscure the salient characteristics of the other
planning and stalking. Offender 8 detailed his treatment of the victim groups” (Rosenberg & Knight, 1988, p. 408). Lastly, sex offenders
by stating that he intentionally used “physical humiliation to treat her driven by both sex and anger merged these two motivations triggering
like a whore.” Offender 31 described rage toward women, stating that offenses characterized by eroticized aggression, with the offenders
“he was angry and frustrated because of problems with his wife … was reporting sexual arousal while causing and watching the suffering
looking for a victim the same age as his spouse ….feels that women are and humiliation of victims.
all whores that laugh at him and take advantage of him….and he gets Additionally, stability rather than change in crime motivation
revenge through sex and humiliation.” Offender 38 described stalking was observed across offenses committed by the same offender.
victims, highlighting that “he was prowling near bars and nightclubs While approximately 40% of the offenders showed minimal variation
and would sometimes follow a victim for several days to learn her in the level of motivation across offenses, only five offenders showed
route … he wanted to humiliate women because of the humiliation he change in type of motivation. Moreover, approximately 22% of varia-
felt from women who rejected him because of his unattractive appearance.” tion in crime motivation across all offenses, whether sexually- or
Importantly, 12 of the 14 offenses resulting in victim deaths linked to anger-driven motivation, was observed within offenders, indicating
this sample were plotted within the high-anger/low-sex quadrant of that 78% of variation in crime motivation for sex offenses was ob-
Fig. 2. Of those offenders who provided explanatory detail regarding served between offenders rather than being offense specific or victim
their rationale for killing victims, several offenders explained that specific. Comparable apportioning of offense/victim specific influence
J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212 211

(22-23%) and offender related influence (77-78%) on sexual and Notes


anger-driven motivation indicates that the impact of offense/victim
factors does not vary depending of the type of crime motivation 1. Although a number of these items selected as indicators of offender motivation
appear to be victim characteristics or behaviors, the items describing victims are based
of the offender. Contrary to popular public perception, victim blame
on self-report from offenders and reflect offenders’ perceptions. For example, when
is not reasonable or warranted regardless of what motivated the asked if the victim dressed in a provoking fashion, numerous offenders responded
offender. yes, although the victims were children as young as 5, 6, 7, 8, and 10 years old. In
While the study findings provide additional empirical evidence describing such victims, one offender explained why he selected two 8-year-old girls
as victims, stating that “they looked like easy girls. He had already heard them talk
of a motivational typology of sex offenders, several limitations of in a coarse way about sex.” Another offender who assault 18 young adolescent boys
the study should be mentioned. The sample utilized for the study (12-15 years old) stated that he choose victims who were exhibitionist, effeminates,
was comprised of serious, incarcerated male repeat sex offenders, extraverted, lightly dressed and who were talking openly about sex.” Offender judg-
ments regarding victims who were all strangers to them were most likely contrived
preventing generalization of the findings to female sex offenders victim characteristics based on offender perceptions and desires rather than actual vic-
and restricting application to sex offenders who are not involved in tim characteristics.
the criminal justice system. In addition, the crime events examined 2. In order to confirm that a two-factor model provided the best fit to the data,
exploratory factor analysis (EFA), which makes no assumptions about the particular
were those known to the criminal justice system and may differ configuration of the indicators, was used to compare model fit of a one-factor model,
in some way from those which were not disclosed. The information two-factor model, and three-factor model.
regarding the offenses was based primarily on retrospective data 3. As previously noted, EFA was used to compare model fit of a one-factor model,
two-factor model, and three-factor model to access which of these models provide the
collection. Such retrospectively collected data are not considered as best fit to the data. First, a difference in chi-square statistics test was conducted for the
accurate or reliable as data collected and documented at the time one-factor model in comparison to the two-factor model using Mplus (Muthén & Muthén,
of its occurrence (Butz, 1981). Recall and self-report of past life 1998-2012). The difference between the one-factor and two-factor models was deter-
mined to be significant (χ2(13) = 32.09, p = .002), indicating that the two-factor model
events, particularly criminal activities, can be incomplete or inaccu-
provided superior fit in comparison to the one-factor model. Next, a difference in chi-
rate. To minimize this limitation, the self-report data from the of- square statistics test was conducted for two-factor model in comparison to a three-
fenders was compared to information from official records whenever factor model. The difference between the two-factor and three-factor models was not
possible. significant (χ2(12) = 16.85, p = .155), indicating that the three-factor model did not
provide superior fit in comparison to the two-factor model. These results provided some
In addition, the study data were collected regarding offenses com- empirical support for the two-factor model prior to conducting the two-factor CFA.
mitted by repeat offenders. To offset this limitation, the WLSMV estima-
tor available with Mplus 7 was utilized as it is designed to provide
References
more accurate parameter and standard error estimates when analyzing
data obtained by stratification with non-independence of observations Ahrens, C. E., Cabral, G., & Abeling, S. (2009). Healing or hurtful: Sexual assault survivors’
interpretations of social reactions from support providers. Psychology of Women
due to cluster sampling (Muthén & Muthén, 1998-2012). Also, when
Quarterly, 33(1), 81–94. http://dx.doi.org/10.1111/j.1471-6402.2008.01476.x.
identifying latent constructs, the hazards of reification or the naming Angelone, D. J., Mitchell, D., & Lucente, L. (2012). Predicting perceptions of date rape: An ex-
fallacy may occur, resulting in the treatment of unobservable constructs amination of perpetrator motivation, relationship length, and gender role beliefs. Journal
as explicitly observed measures (Kline, 2005). For example, the latent of Interpersonal Violence, 27, 2582–2602. http://dx.doi.org/10.1177/0886260512436385.
Ariely, D. (2008). Predictably irrational: The hidden forces that shape our decisions.
construct labeled “anger motivation” could have been labeled aggres- New York: Harper Collins.
sion or another related term. As noted, anger was not the only negative Ariely, D., & Loewenstein, G. (2006). The heat of the moment: The effect of sexual
emotion reported before or during offenses but “anger/frustration/ arousal on sexual decision making. Journal of Behavioral Decision Making, 19, 87–98.
http://dx.doi.org/10.1002/bdm.501.
aggression” it was the most commonly reported negative emotion Asparouhov, T. (2005). Sampling weights in latent variable modeling. Structural Equation
(64% prior to offense, 68% during the offense). So, in choosing a single Modeling, 12, 411–434. http://dx.doi.org/10.1207/s15328007sem1203_4.
descriptive as a label for this latent construct, anger was selected as Beauregard, E., Leclerc, B., & Lussier, P. (2012). Decision Making in the Crime Commission
Process Comparing Rapists, Child Molesters, and Victim-Crossover Sex Offenders.
the best option. Criminal Justice and Behavior, 39(10), 1275–1295. http://dx.doi.org/10.1177/
As evidenced by the study findings, both the type and strength of 0093854812453120.
motivation play a key role in the process of offending. Contemplating Bouffard, J. (2011). “In the heat of the moment”: Mediating versus moderating relation-
ships between sexual arousal and perceived sanctions. Journal of Crime and Justice,
offender motivation beyond just “a given” may inform prevention
34, 24–44. http://dx.doi.org/10.1080/0735648X.2011.554745.
efforts by indicating which environments or situations potentiate Brantingham, P. L., & Brantingham, P. J. (1993). Environment, routine, and situation:
the occurrence of a criminal event. As a result, risk prevention and Toward a pattern theory of crime. In R. V. Clarke & M. Felson (Eds.), Routine Activity
and Rational Choice. Advances in Criminological Theory, Vol. 5. (pp. 259–294). New
victim recovery strategies built upon empirically-grounded knowl-
Brunswick, NJ: Transaction Publishers.
edge regarding offender motivation could enhance social understand- Brown, T. (2006). Confirmatory factor analysis for applied research. New York: Guilford
ing, reduce victim blaming, and strengthen individual self-efficacy Press.
(Angelone et al., 2012). Knowledge regarding variation in offender Bryden, D. P., & Grier, M. M. (2011). Search for Rapists’ “Real Motives”. The Journal of Criminal
Law and Criminology, 101, 171–277 (http://dx.doi.org/0091-4169/11/10101-0171).
motivations for sexual assault could also enhance prevention efforts Bryne, J., & Roberts, A. (2007). New directions in offender typology design, development,
targeting potential offenders. A more comprehensive and inclusive and implementation: Can we balance risk, treatment and control? Aggression and
approach, incorporating information regarding both sexual and Violent Behavior, 12, 483–492. http://dx.doi.org/10.1016/j.avb.2007.02.009.
Burgess, A. W., Commons, M. L., Safarik, M. E., Looper, R. R., & Ross, S. N. (2007). Sex
anger/aggression motivations for sexual assault may facilitate greater offenders of the elderly: Classification by motive, typology, and predictors of sever-
self-awareness and better choices when encountering opportunities ity of crime. Aggression and Violent Behavior, 12(5), 582–597. http://dx.doi.org/
for offending. Additionally, enhanced understanding of offender mo- 10.1016/j.avb.2007.02.006.
Butz, W. P. (1981). Why collect retrospective data? Santa Monica, CA: Rand Corporation.
tivation may improve treatment options for offenders. Studies show Campbell, R., Ahrens, C. E., Sefl, T., Wasco, S. M., & Barnes, H. E. (2001). Social reactions
that most past and current interventions for offenders focus on risk to rape victims: Healing and hurtful effects on psychological and physical health
level and behavioral avoidance with a low rate of success (Bryne & outcomes. Violence and Victims, 16(3), 287–302.
Campbell, R., Baker, C. K., & Mazurek, T. L. (1998). Remaining radical? Organizational
Roberts, 2007). Perhaps, inadequate treatment effects are due to a
predictors of rape crisis centers' social change initiatives. American Journal of Com-
mismatch of treatment with offender, and greater comprehension of munity Psychology, 26(3), 457–483. http://dx.doi.org/10.1023/A:1022115322289.
the underlying motivations driving offenders could facilitate more Campbell, R., Dworkin, E., & Cabral, G. (2009). An ecological model of the impact of
sexual assault on women's mental health. Trauma, Violence & Abuse, 10(3), 225–246.
effective individualized treatment strategies.
http://dx.doi.org/10.1177/1524838009334456.
Carmichael, S., & Piquero, A. R. (2004). Sanctions, perceived anger, and criminal
Acknowledgement offending. Journal of Quantitative Criminology, 20, 371–393. http://dx.doi.org/10.1007/
s10940-004-5869-y.
Casey, E. A., & Lindhorst, T. P. (2009). Toward a multi-level, ecological approach to the pri-
The authors thank Melanie Gingras for translation of interview mary prevention of sexual assault prevention in peer and community contexts. Trau-
data used in this study. ma, Violence & Abuse, 10(2), 91–114. http://dx.doi.org/10.1177/1524838009334129.
212 J.A. Reid et al. / Journal of Criminal Justice 42 (2014) 203–212

Clarke, R. V., & Cornish, D. B. (1985). Modeling offenders' decisions: A framework for Maruna, S. (2010). Mixed method research in criminology: Why not go both ways? In
research and policy. Crime and Justice, 6, 147–185 (JSTOR URL: http://www.jstor. A. R. Piquero & D. Weisburd (Eds.), Handbook of quantitative criminology. London:
org/stable/1147498). Springer. http://dx.doi.org/10.1007/978-0-387-77650-7.
Cohen, L. E., & Felson, M. (1979). Social change and crime rate trends: A routine activity Mitchell, D., Angelone, D. J., Kohlberger, B., & Hirschman, R. (2009). Effects of offender
approach. American Sociological Review, 44, 588–608 (JSTOR URL: http://www. motivation, victim gender, and participant gender on perceptions of rape victims and
jstor.org/stable/2094589). offenders. Journal of Interpersonal Violence, 24(9), 1564–1578. http://dx.doi.org/
Cornish, D., & Clarke, R. (2003). Opportunities, precipitators and criminal decisions: 10.1177/0886260508323662.
A reply to Wortley’s critique of situational crime prevention. In M. Smith & D. Moffitt, T. E. (2006). Pathways in the life course to crime. In F. T. Cullen & R. Agnew
Cornish (Eds.), Theory for Practice in Situational Crime Prevention. Crime Prevention (Eds.), Criminological theory: Past to present (pp. 502–521). Los Angeles: Roxbury
Studies, Vol. 16. (pp. 151–196). Monsey, NY: Criminal Justice Press. Publishing.
Cullen, F. T. (2011). Beyond adolescent-limited criminology: Choosing our future-The Muhr, T., & Friese, S. (2004). User’s manual for ATLAS.ti 5.0 (2nd ed.). Berlin: Scientific
ASC 2010 Sutherland Address. Criminology, 49, 287–330. http://dx.doi.org/10.1111/ Software Development.
j.1745-9125.2011.00224.x. Mustaine, E. E., & Tewksbury, R. (2009). Transforming potential offenders into motivated
Flora, D. B., & Curran, P. J. (2004). An empirical evaluation of alternative models of es- ones: Are sex offenders tempted by alcohol and pornography? Deviant Behavior,
timation for confirmatory factor analysis with ordinal data. Psychological Methods, 30(7), 561–588. http://dx.doi.org/10.1080/01639620802467821.
9, 466–491. http://dx.doi.org/10.1037/1082-989X.9.4.466. Muthén, L. K., & Muthén, B. O. (1998). Mplus user’s guide (7th ed.). Los Angeles:
Foubert, J. D., Godin, E., & Tatum, J. (2010). In their own words: Sophomore college men Author.
describe attitude and behavior changes resulting from a rape prevention program Olver, M. E., Stockdale, K. C., & Wormith, J. S. (2011). A meta-analysis of predictors of
two years after their participation. Journal of Interpersonal Violence, 25, 2237–2257. offender treatment attrition and its relationship to recidivism. Journal of Consulting
http://dx.doi.org/10.1177/0886260509354881. and Clinical Psychology, 79(1), 6–21. http://dx.doi.org/10.1037/a0022200.
Heil, P., Ahlmeyer, S., & Simons, D. (2003). Crossover sexual offenses. Sexual Abuse: A Journal of Palmer, C. T. (1988). Twelve reasons why rape is not sexually motivated: A skeptical
Research and Treatment, 15, 221–236. http://dx.doi.org/10.1023/A:1025031325230. examination. Journal of Sex Research, 25(4), 512–530. http://dx.doi.org/10.1080/
Herman, J. L. (1992). Trauma and recovery. New York: Basic Books. 00224498809551479.
Hicks, S. J., & Sales, B. D. (2006). Criminal profiling : Developing an effective science and Paulsen, D. J., & Robinson, M. B. (2004). Spatial aspects of crime: Theory and practice.
practice (1st ed.). Washington, DC: American Psychological Association. Boston: Allyn & Bacon.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure Piquero, A., Paternoster, R., Pogarsky, G., & Loughran, T. (2011). Elaborating the individual
analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, difference component in deterrence theory. Annual Review of Law and Social Science,
6, 1–55. http://dx.doi.org/10.1080/10705519909540118. 7, 335–360. http://dx.doi.org/10.1146/annurev-lawsocsci-102510-105404.
King, N. (1998). Template analysis. In G. Symon & C. Cassell (Eds.), Qualitative methods Prentky, R. A., & Knight, R. A. (1991). Identifying critical dimensions for discrimi-
and analysis in organizational research (pp. 118–134). London: Sage Publications. nating among rapists. Journal of Consulting and Clinical Psychology, 59, 643–661.
Kleinginna, P. R., & Kleinginna, A. M. (1981). A categorized list of emotion definitions, http://dx.doi.org/10.1037/0022-006X.59.5.643.
with suggestions for a consensual definition. Motivation and Emotion, 5, 345–379. Rape Advocacy Program. (n. d.). Defining sexual violence: Myths and facts. Retrieved on
http://dx.doi.org/10.1007/BF00993889. February 25, 2013 from http://www.rvap.org/pages/myths__facts/#crime_of_passion
Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.). Robertiello, G., & Terry, K. J. (2007). Can we profile sex offenders? A review of sex offender
New York: Guildford Press. typologies. Aggression and Violent Behavior, 12(5), 508–518. http://dx.doi.org/
Knight, R. A. (1999). Validation of a typology for rapists. Journal of Interpersonal Violence, 10.1016/j.avb.2007.02.010.
14, 297–323. http://dx.doi.org/10.1177/088626099014003006. Rosenberg, R., & Knight, R. A. (1988). Determining male sexual offender subtypes using
Knight, R. A., & Guay, J. P. (2006). The role of psychopathy in sexual offenders against women. cluster analysis. Journal of Quantitative Criminology, 4(4), 383–410. http://dx.doi.org/
In C. J. Patrick (Ed.), Handbook of psychopathy (pp. 512–532). New York: John Wiley. 10.1007/BF01065346.
Knight, R. A., & Prentky, R. A. (1990). Classifying sexual offenders: The development of Sasse, S. (2005). "Motivation" and routine activities theory. Deviant Behavior, 26, 547–570.
corroboration of taxonomic models. In W. L. Marshall, D. R. Laws, & H. E. Barbaree http://dx.doi.org/10.1080/01639620500218260.
(Eds.), Handbook of sexual assault (pp. 23–52). New York: Plenum. Sizemore, O. J. (2013). The role of perpetrator motivation in two crime scenarios. Journal of
Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press. Interpersonal Violence, 28(1), 80–91. http://dx.doi.org/10.1177/0886260512448843.
Liao, T. F. (1994). Interpreting probability models: Logit, probit, and other generalized linear Tharp, A. T., DeGue, S., Lang, K., Valle, L. A., Massetti, G., Holt, M., et al. (2011). Commen-
models. Thousand Oaks, CA: Sage Publications. tary on Foubert, Godin, & Tatum (2010): The evolution of sexual violence prevention
Loewenstein, G., Nagin, D. S., & Paternoster, R. (1997). The effect of sexual arousal on and the urgency for effectiveness. Journal of Interpersonal Violence, 26, 3383–3392.
expectations of sexual forcefulness. Journal of Research in Crime and Delinquency, http://dx.doi.org/10.1177/0886260510393010.
34, 443–473. http://dx.doi.org/10.1177/0022427897034004003. Tittle, C. R., & Botchkovar, E. V. (2005). Self‐control, criminal motivation and deter-
Loughran, T. A., Paternoster, R., Piquero, A. R., & Pogarsky, G. (2011). On ambiguity in rence: An investigation using Russian respondents. Criminology, 43(2), 307–354.
perceptions of risk: Implications for criminal decision making and deterrence. http://dx.doi.org/10.1111/j.0011-1348.2005.00010.x.
Criminology, 49, 1029–1061. http://dx.doi.org/10.1111/j.1745-9125.2011.00251.x. Yu, C. Y. (2002). Evaluating cutoff criteria of model fit indices for latent variable models
Lussier, P. (2005). The criminal activity of sexual offenders in adulthood: Revisiting with binary and continuous outcomes. (Doctoral dissertation). University of
the specialization debate. Sexual Abuse: A Journal of Research and Treatment, 17, California (Retrieved on February 25, 2013 from http://www.statmodel.com/
269–292. http://dx.doi.org/10.1007/s11194-005-5057-0. download/Yudissertation.pdf).

You might also like