Development of Attribution-Based Scales Associated With Proactive,Reactive,and Acquisitive Offending

Abstract

This study outlines the development of attribution scales to assess the antecedent processes of proactive, reactive, and acquisitive offending. Empirical keying, internal consistency, and item performance were the three test development strategies. Six samples (total N = 1,527) from parole, university, probation, and prison were used. The Proactive scale (10 items) reflected coldness and distance between offender and victims, the Reactive scale (10 items) reflected recklessness and excuses for violent behavior, and the Acquisitive scale (15 items) reflected less negative views of crime and choices. Convergent and discriminant validities were demonstrated by the Proactive scale stronger correlations with planning items, the Reactive scale stronger correlations with affective and person-based items, and the Acquisitive scale stronger correlations with other-based motives. Linking Proactive, Reactive, and Acquisitive scales to an offense provides psychological meaning that can benefit assessment and interventions.

Keywords

violence assessment proactive violence reactive violence acquisitive offending intervention

Developing taxonomies and understanding the nature of categories are basic tasks of science. These categories can inform appropriate groupings of clients (i.e., types of diagnosis, risk levels), provide guidelines for public policy (resources for high-risk groups, parameters of confinement) decisions, and suggest mutable content areas for intervention. The purpose of this article is to develop attribution-based measures reflecting the processes that are associated with proactive, reactive, and acquisitive offending. This is accomplished through three phases of scale development that incorporated five independent samples. The first phase used an empirical keying approach to classify items according to the reactive, proactive, and nonviolent acquisitive domains. Empirical keying includes items based upon an association with a targeted characteristic versus absence. The second phase used multiple methods (e.g., item-total, coefficient alpha, principal component analyses (PCAs), and Rasch item performance analyses) to delete items. Given the emphasis on the association with proactive, reactive, and acquisitive offending, the third phase reapplied the empirical keying approach.

Proactive, Reactive, and Acquisitive Constructs

A distinction can be made between proactive (i.e., planned, cold-blooded, instrumental) and reactive (i.e., affective, hot-blooded) driven actions among children (Poulin & Boivin, 2000), adolescents (Walters & DeLisi, 2015), and adults (Walters, 2009b). Among adult participants who are in prison, proactive aggression is associated with higher psychopathy scores (Cornell et al., 1996), yet they concluded that reactive actions are “the most basic form of aggression among criminal behavior” (p. 788). In addition, other research supports the prominent role of reactive measures in predicting future offending (Peterson, Skeem, Hart, Vidal, & Keith, 2010; Walters, 2011). Conversely, others have found high proactive scores to have more of a role in distinguishing between property and violent delinquency than reactive scores (Miller & Lynam, 2006).

Definitionally, acquisitive offending is a separate offense category. In contrast to proactive and reactive actions, acquisitive offending does not involve direct victim damage. Acquisitive offending has the goal of damaging another person’s property to secure goods or money (McGuire, 2004). The antecedents to acquisitive offending differ from violent offending. Focusing on the affective dimension, acquisitive offending is associated with positive emotional experiences, whereas violent crimes are associated with negative emotional experiences (Canter & Ioannou, 2004). Similarly, anger and hopelessness are experienced less by those who committed property offenses than those who committed assaultive and robbery offenses (Zamble & Quinsey, 1997). Antisocial peers may have more influence with acquisitive offending compared with violent offending (Antonaccio, Tittle, Botchkovar, & Kranidiotis, 2010). In a comparison of non, medium, and high levels of acquisitive offending groups, Wilson, Attrill, and Nugent (2003) found progressively higher scores on criminal thinking, impulsivity, and aggressive problem solving among these three groups. Thus, there is sufficient evidence to suggest that acquisitive offending is associated with unique processes, distinct from proactive and reactive processes.

Measurement and Content of Proactive, Reactive, and Acquisitive Domains

A standardized measure of proactive and reactive behaviors, the Reactive–Proactive Aggression Questionnaire (Raine et al., 2006) has shown construct validity and reliability across multiple samples (Baker, Raine, Liu, & Jacobson, 2008; White, Gordon, & Guerra, 2015). In a 10-year longitudinal study (ages 16 and 26), proactive aggression was associated with future antisocial behavior, including violence, whereas reactive aggression was associated with future negative affect, most notably anxiety (Fite, Raine, Stouthamer-Loeber, Loeber, & Pardini, 2010). When the Reactive–Proactive Aggression Questionnaire has been used as an outcome measure, antisocial personality and conduct disorder symptoms are more strongly related to proactive aggression (Steadham & Rogers, 2013).

The Psychological Inventory of Criminal Thinking Styles (PICTS; Walters, 1995) initially used three subscales (2 × Entitlement + (1.5 × Self-Assertion/Deception) + Historical) to measure the proactive domain and three subscales (2 × Cutoff + (1.5 × Problem Avoidance) + Current) to measure the reactive domain (Walters, 2006). With multiple controls and incremental validity tests, the PICTS proactive scale was associated with greater criminal history and substance abuse and greater nonaggressive incidents; the PICTS reactive scale was associated with greater aggressive incidents and more serious prison charges (Carr, Rosenfeld, Magyar, & Rotter, 2009; Walters, 2007; Walters, 2011).

One shortcoming of current measures is the use of static/historical items. A drawback to using past behaviors to predict future behaviors (i.e., Reactive–Proactive Aggression Questionnaire) is the potential circularity of the relationship. To address this, nonhistorical proactive and reactive measures have been used as predictors. Using a psychiatric high-risk violence sample, Skeem et al. (2004) found that proactive and reactive measures distinguished between two high-risk groups, with respect to past instrumental offenses. Moreover, Walters and Delisi (2015) observed that a proactive scale antisocial mediated the otherwise poor Psychopathy Checklist Factor 1 (core personality features) and crime relationship.

A second drawback to using static/historical is the lack of utility in measuring treatment change. Consequently, this may be the reason that the PICTS proactive domain has been less sensitive to treatment changes (Walters, 2006). With regard to the PICTS, a second shortcoming may be the scale development strategy. Both the proactive and reactive domains for the PICTS were created based on the composites of other scales, which may allow for a broad coverage but may limit conclusions on more narrowly defined aspects of the proactive and reactive measures. In the current study, scales were developed at the item level, using crime-related attributional items.

Currently, there is not a measure assessing the antecedent factors of acquisitive areas. Having such a measure has a benefit in risk assessments where the evaluator will have limited knowledge of the type of future offending (i.e., violent vs. nonviolent). Thus, in addition to measuring the processes that may lead to violent actions, antecedent factors for acquisitive offending are measured.

The current effort proposes measuring proactive, reactive, and acquisitive processes within an attribution framework. Multiple lines of research have demonstrated the utility of attributions in understanding acquisitive offending (Gudjónsson & Singh, 1988), violence (Cima et al., 2007; Dutton, 1986; Gudjónsson & Singh, 1988; Weizmann-Henelius, Sailas, Viemero, & Eronen, 2002), and antisocial traits (Di Fazio, Kroner, & Forth, 1997; Fox, De Koning, & Leicht, 2003; Gudjónsson, 1984; Weizmann-Henelius et al., 2002). With an Icelandic criminal justice sample, external attributions were higher for violent than acquisitive offenses (Gudjónsson & Pétursson, 1991).

The items used to construct the three scales were written within an attributional framework that focused on the causes of crime (i.e., “Thinking that a victim can contribute to crime is wrong”). Thus, this focus on an attributional framework and the causes of crime restricts what is measured, thereby excluding direct assessment of broad psychological proactive, reactive, and acquisitive areas.

Scale Development Strategies

A class/prediction model, emphasizing prediction, and a dimensional model, emphasizing explanation, were used for developing the three scales. Loevinger (1957) argued that validation entails using the measurement model that bears a close relationship to the structure of the underlying processes to be measured. To reflect the class/prediction assessment model, an empirical keying approach was used (Reckase, 1996). The outcome measures used for item selection were defined by the criminal justice operationalized categories of proactive, reactive, and acquisitive offending. Some overlap of items for these three areas was allowed, as empirical studies have found some but not substantial differences among various offense groupings (Zamble & Quinsey, 1997). In addition to the class/prediction model, dimensional assessment model strategies of item-total correlations, coefficient alpha item deletion, PCA, and Rasch modeling were used.

The rationale for using both class/prediction and dimensional models was threefold. First, a goal of these three scales is to be value-added (both predictive and explanatory) to commonly used risk assessment instruments within forensic assessments. More recently, an argument has been made to use risk assessment instruments that are psychologically informed (Mann, Hanson, & Thornton, 2010). Second, the developed scales are intended to guide interventions according to broad, although well-defined theoretical constructs. Thus, there was a balance between construct strategies (Jackson, 1970) and measures having a relationship with an outcome (Borsboom, Mellenbergh, & van Heerden, 2004, p. 1062).

The third rationale for using both class/prediction and dimensional models was to develop appropriate measures to assess a client’s change, with the potential to indicate mechanisms of change (Bonta & Wormith, 2008). Skeem and Monahan (2011) argued that the assessment of risk should be integrated with the measurement of risk reduction. To accomplish this measurement of change within a risk assessment context, it is argued that both class/prediction and dimensional models can uniquely contribute. Five samples were used for scale development, and a final sample (Sample 6) assessed test–retest reliability and validity.

The goals for this article are twofold. The first goal is to develop a measure of proactive, reactive, and acquisitive scales that would reflect the antecedent factors for proactive, reactive, and acquisitive outcomes. Elevations on these scales would indicate whether intervention related to basic areas of offending is warranted. A second goal is to present reliability and validity data for the developed Proactive, Reactive, and Acquisitive scales.

Method

Participants

This study was constructed using six different samples to cover a wide spectrum of participants. Participants were from parole offices, university, prisons, and probation offices. Sample 1 consisted of 116 Canadian male participants who were on parole. The mean age was 34.1 (SD = 8.6). Racial composition was as follows: 85 (73.3%) White, nine (7.8%) Black, 14 (12.1%) North American Native/Metis/Inuit, five (4.3%) Asian, and three (2.6%) other. Marital status was as follows: 65 (56.1%) single, 35 (30.2%) common law, nine (7.6%) married, four (3.4%) divorced, and three (2.6%) separated. The most serious index offenses were as follows: 31 (26.7%) robbery, extortion; 25 (21.6%) drug offenses; 28 (24.1%) property, loitering; seven (6.0%) murder, attempted murder, manslaughter; seven (6.0%) assaults, threats; six (5.2%) sexual offenses; six (5.2%) fraud, forgery, false pretenses; three (2.6%) negligence, major driving offense; one (0.9%) possession of weapons or explosives; one (0.9%) kidnapping, confinement, hijacking; and one (0.9%) arson. The mean sentence length was 5.5 (SD = 4.4) years.

Sample 2 consisted of 50 U.S. Midwest university students selected from a sample of 238 criminal justice students who participated in a noncredit research project. The criteria for selection of these 50 students were a self-reported misdemeanor or felony behavior within the past 12 months. The mean age for the sample of 50 was 19.9 (SD = 1.4). Racial composition was as follows: 26 (52.0%) White, 14 (28.0%) Black, one (2.0%) North American Native, three (6.0%) Asian, and three (6.0%) Hispanic (two not recorded). One half (n = 25) were female. Employment status was as follows: four (8.0%) employed full-time, 24 (48.0%) part-time, and 22 (44.0%) unemployed.

Sample 3 consisted of 83 participants from a U.S. Midwest probation office (mean age = 27.0, SD = 9.7). Racial composition was as follows: 85 (73.3%) White, nine (7.8%) Black, 14 (12.1%) North American Native/Metis/Inuit, five (4.3%) Asian, and three (2.6%) other. The most serious index offenses were as follows: 31 (26.7%) robbery, extortion; 25 (21.6%) drug offenses; 28 (24.1%) property, loitering; seven (6.0%) murder, attempted murder, manslaughter; seven (6.0%) assaults, threats; six (5.2%) sexual offenses; six (5.2%) fraud, forgery, false pretenses; three (2.6%) negligence, major driving offense; one (0.9%) possession of weapons or explosives; one (0.9%) kidnapping, confinement, hijacking; and one (0.9%) arson. With regard to criminal history, 14 had a previous offense and 14 had a previous violent offense.

Sample 4 consisted of 216 Canadian male participants (mean age = 39.1, SD = 9.2) who resided in a minimum security institution. The Criminal Attribution Inventory (CRAI) was administered as part of a work placement screening battery. The majority of these participants were White (80%). No other demographic data were linked to these data.

Sample 5 consisted of 976 Canadian male participants confined to prison. The mean age was 35.5 (SD = 11.5). Racial composition was as follows: 726 (74.4%) White, 71 (7.3%) Black, 133 (13.6%) North American Native/Metis/Inuit, 23 (2.4%) Asian, and 23 (2.4%) other. Marital status was as follows: 471 (48.3%) single, 276 (28.3%) common law, 95 (9.7%) married, 55 (5.6%) divorced, and 66 (6.8%) separated. The most serious index offenses were as follows: 218 (22.3%) robbery, extortion; 82 (8.4%) drug offenses; 139 (14.2%) property, loitering; 140 (14.3%) murder, attempted murder, manslaughter; 129 (13.2%) assaults, threats; 120 (12.3%) sexual offenses; 12 (1.2%) fraud, forgery, false pretenses; 60 (6.1%) negligence, major driving offense; 17 (1.7%) possession of weapons or explosives; 25 (2.6%) kidnapping, confinement, hijacking; 12 (1.2%) arson; and 22 (2.3%) other. The mean sentence length was 5.5 (SD = 4.4) years.

Sample 6 consisted of 87 U.S. Midwest male participants who were confined to prison for sexual offenses. The mean age was 48.2 (SD = 13.0). Racial composition was as follows: 68 (78.2%) White, 13 (14.9%) Black, two (2.3%) Hispanic, and four (4.6%) other. Marital status was as follows: 40 (46.0%) single, nine (10.3%) married, 34 (39.1%) divorced, and four (4.6%) separated/widowed. The most serious index offenses were as follows: 25 (27.7%) aggravated criminal sexual abuse, 19 (21.8%) predatory criminal sexual assault, 15 (17.2%) aggravated criminal sexual assault, and 12 (13.8%) other. The number of years served was 13.0 (SD = 9.1) years.

Measures

CRAI

The CRAI is a self-report measure that assesses the endorsement of various causes of crime (Kroner & Mills, 2003). The CRAI has six scales, with each scale composed of 10 items.¹ The participant is asked to agree or disagree with each statement (dichotomous responding). The CRAI’s instructions define crime for the respondent as, “what YOU know the average type of crime to be.” Thus, the instructions ask for a general perception of the cause of crime, not referencing the participant’s specific crimes. The six original scales are Psychopathology (e.g., “Criminal behavior is often caused by mental illness”), Personal (e.g., “Crimes occur because of lifelong traits inside the person”), Victim (e.g., “The victim has a part in the beginning of many crimes”), Alcohol (e.g., “Alcohol makes people commit crime”), Society (e.g., “When crime occurs, society should be partially blamed”), and Randomness (e.g., “Unexpected events can result in crime”). Past research has demonstrated the dynamic nature of the scales and concurrent validity (Kroner, Hemmati, & Mills, 2006; Kroner & Mills, 2004).

Measures of Criminal Attitudes and Associates (MCAA)

The MCAA (Part B) is a self-report measure (46 items) that assesses antisocial attitudes covering areas of entitlement, antisocial intent, violence, and antisocial associates (Mills & Kroner, 2001). The instructions ask the respondent to agree or disagree with each stated antisocial attitude. The MCAA has demonstrated concurrent (Mills, Kroner, & Forth, 2002), criterion (Bäckström & Björklund, 2008), construct (Mandracchia & Morgan, 2012), and predictive validities across a range of samples (Boduszek, Hyland, Pedziszczak, & Kielkiewicz, 2012; Van Hiel, Hautman, Cornelis, & De Clercq, 2007).

Analyses

Empirical keying

The empirically keying phase used three samples to categorize items into reactive, proactive, and acquisitive groups. The most serious offense (assault, attempted murder, manslaughter, robbery) of each participant in Sample 1 (parole office) did not have sufficient detail to make a distinction between proactive and reactive domains. Thus, items from the MCAA were used to categorize the violent crimes into proactive and reactive categories. The MCAA five proactive items (“There is nothing wrong with beating up a child molester,” “Sometimes you have to fight to keep your self-respect”) and the five reactive items (“It’s understandable to hit someone who insulted you,” “Someone who makes you very angry deserves to be hit”) were respectively totaled. An endorsement of one of the five items resulted in the offense being categorized as proactive (0 = not present, 1 = present) or reactive (0 = not present, 1 = present). At this stage, placement of a CRAI item in both proactive and reactive categories (~15%) was allowed. For the empirically keying phase, a recorded crime of theft, fraud, forgery, false pretenses, and possession of a weapon met the criteria for an acquisitive offense.

Sample 2 (college students) was used to examine the acquisitive keying of the CRAI items. The criterion measure was a sum of two items reflecting the wrongful acquisition of grades (“Times cheated on exams,” “Times plagiarized on assignment or essay”). Each item had a six-level response (0 = never, 1 = once, 2 = twice, 3 = thrice, 4 = 4 times, and 5 = 5 times or more, but for the empirically keying 0 = never, 1 = present). In this sample, there was no information regarding violence, which precluded the CRAI items being categorized according to reactive or proactive categories. Sample 3 (probation office) used the following offense criteria for reactive (assault, any aggravated offense), proactive (robbery), and acquisitive (burglary, larceny, auto theft, fraud, stolen/damaged property) categories. Thus, the acquisitive criteria were predetermined for each sample. Although not exactly similar, the operationalization did meet the basic acquisition definition.

To determine whether a CRAI item could be keyed as a reactive, proactive, or acquisitive item, a point system, based on correlations, was used. If a CRAI item had a zero-ordered correlation of .15 or greater with reactive, proactive, or acquisitive criteria, then 1 point was assigned. These points were tallied across the three samples (parole office, university, and probation office).

Statistical procedures

The within-scale item performance phase used item-total correlations, coefficient alpha item deletion, and PCA to reduce the number of items keyed into the reactive, proactive, and acquisitive categories (Sample 5). A correlation matrix of dichotomous data may have problems of misspecification and normal distribution, which can result in biased standard errors, overestimation of the number of factors, and underestimation of the factor loadings (Woods, 2002). A tetrachoric correlation matrix was created for the CRAI data. This procedure infers a latent Pearson correlation with the assumption of bivariate normality and eliminates the effect of response frequency, correcting for the reduction in dichotomous correlations (see Table 1).

Table 1.

Sample Descriptions.

Sample	Place	Sample Size	Empirical Keying	Statistical Procedure
Sample 1	Parole Office (CND)	116	Yes
Sample 2	University (USA)	50	Yes
Sample 3	Probation Office (USA)	83	Yes	Rasch
Sample 4	Prison (CND)	216	No	Rasch
Sample 5	Prison (CND)	976	No	item-total, alpha, PCA
Sample 6	Prison (USA)	86	No	alpha, correlation (validity)

Note. CND = Canada; PCA = principal component analysis.

Next, Rasch modeling was used to assess the scales’ item discriminating power across different groups (New rule 4, Embretson, 1996). To accomplish this, the partitioning of the sample (probation office vs. prison sample) and partitioning at score levels (low score vs. high score in the present study) were calculated as a mathematical function of person and item parameters (Hendriks, Fyfe, Styles, Skinner, & Merriman, 2012). This approach allows scales to be modified (i.e., removal of items) without attempting to change the model of the construct (cf. classical test theory). Consistent item parameter estimates (invariance) are assessed among subgroups (using partition criteria) and provide evidence that the Rasch model is maintained. Bias occurs when cases at the same overall level do not have the same probability of reporting item content. Only one aspect of the Rasch modeling is used for the current analyses, that of an item diagnostic to determine how well a item is functioning to measure the scale construct. After combining Samples 3 and 4, two partition criteria of sample (probation office vs. prison sample) and the total score (low score vs. high score) were used. Individual items were evaluated according to the Wald test, which allows for model testing of single items. The model was estimated in R, version 3.4.1 (R Core Team, 2016) with the package eRm (Mair, Hatzinger, Maier, & Rusch, 2015). The poorest items (greatest bias) for either partition (place and score level) on each scale were removed.

Finally, in keeping with the empirical keying emphasis, the two items with the most points (item keying phase) that were deleted in the item performance phase were placed back onto their respective scales.

Results

Empirically Keying Phase

This phase used Samples 1, 2, and 3 to empirically key the CRAI items into reactive, proactive, and acquisitive categories. Zero-ordered correlations were calculated for each of the 60 CRAI item responses and the reactive, proactive, and acquisitive criteria. Using the correlation cutoff of .15, 36 items were categorized as reactive, 33 items as proactive, and 46 items as acquisitive.

Within-Scale Item Performance Phase

Using Sample 5, each item from one of the three keyed scales was correlated with the other two total scale scores. The four items with the strongest correlations with the noncorresponding (n = 2) total scale scores were deleted, as they were too strongly related to another dimension. Then, the two weakest Cronbach’s alpha contributors to each keyed scale were deleted. Using the tetrachoric correlation matrix, each scale was subjected to a PCA. The expected components were set to two for each scale, which the results demonstrated was acceptable. Items were retained if they had loading above .45 and were a part of the strongest component. Items associated with the weaker component were deleted. This resulted in scale item totals of 11 (Reactive), 11 (Proactive), and 15 (Acquisitive).

Next, each item was evaluated with the Wald test to determine the Rasch model fit in a combined Sample 3 and 4 data set. The poorest fitting item (p < .001) for either the place (probation office vs. prison sample) or mean (low score vs. high score) partitioning for each scale was dropped (two items were dropped from the acquisitive scale because one item failed to enter the Rasch model). Finally, the two items per scale that were deleted through the item performance phase (five of the six through PCA) with the highest number of points from the empirical keying phase were placed back onto the scales: Proactive (“Society supports behaviors which are related to crime,” “Being mentally sick has nothing to do with crime”; total = 12 items), Reactive (“A lot of crime happens when people are in the wrong place at the wrong time,” “Crime cannot be blamed on the victim”; total = 12 items), and Acquisitive (“Alcohol has very little to do with crime,” “Drinking a lot of alcohol can result in crime”; total = 15 items) (see the appendix for scale items).

Reliability and Validity

Table 2 contains the reliability analyses. Coefficient alphas ranged from .56 to .74 across the four samples (Sample 5 was excluded as this sample used internal consistency methods for scale development). Sample 6 had a 3-month test–retest for 28 cases. Test–retest of the three scales ranged from .41 to .48. The correlations among the three scales were .51 (Proactive vs. Reactive), .83 (Proactive vs. Acquisitive), and .52 (Reactive vs. Acquisitive).

Table 2.

Cronbach’s Alpha and Test-Retest Coefficients for Proactive, Reactive, and Acquisitive Scales.

Scale	No. of Items	Cronbach’s Alpha			Test–Retest
Scale	No. of Items	Sample 1	Sample 3	Sample 4	Sample 6	Sample 6 (n = 28)
Proactive	10	.71	.58	.70	.74	.41
Reactive	10	.55	.64	.68	.65	.45
Acquisitive	15	.71	.56	.68	.58	.48

Evidence for convergent and discriminant validities is presented in Table 3. The targeted items were not used in the previous analyses. The items emphasizing planning had stronger correlations with the Proactive scale than the other two scales. The items emphasizing affective and person-based processes had stronger correlations with the Reactive scale than the other two scales. The items emphasizing other-based motives had stronger correlations with the Acquisitive scale than the other two scales.

Table 3.

Correlations Among Targeted Items and Proactive, Reactive, and Acquisitive Scales.

Targeted Items	Proactive	Reactive	Acquisitive
Planning emphasis
Making new friends will be hard for me	.18	.00	.08
It will be tough to find a good place to live	.15	−.09	.12
Unexpected events can result in crime	.35	.20	.26
Crime is not likely to happen at random^a	.35	.25	.30
When at home, there will be times I won’t know what to do with my time	.48	.34	.43
Affective and person-based emphasis
I expect to do fun things on the spur of the moment	.09	.52	.10
People who do crime do so because of their personality traits.	.17	.30	.15
Alcohol can be blamed for a lot of crime	.37	.46	.36
Being mentally sick has nothing to do with crime^a	.22	.31	.19
Only I can decide what is right and wrong	.13	.22	.09
Other-based motives
Victims of crime are usually innocent bystanders^a	.14	−.03	.25
Chance is a big part of crime	.28	.26	.32
I would be open to cheating certain people	−.03	−.11	.15

Note. N = 87. Expected stronger correlations are in bold.

Reverse-keyed items.

Discussion

Planning to do damage to someone (proactive), reacting violently (reactive), and taking things from others (acquisitive) can be construed as separate categories of offending. Dynamic measurement and the distinctively attributional nature of the items may allow for unique aspects of three basic types of offending to be captured. The five samples used for development of the Proactive, Reactive, and Acquisitive scales included participants from parole offices, university, probation offices, and prison. Multiple strategies were incorporated in the development of these scales, which is consistent with other test developers (Nader et al., 2013; Stein et al., 2013; Ward, Gibson, Boman, & Leite, 2010). The application of these scales can be for treatment recommendation and evaluation, adjunct to risk assessment, and the explanation of causality of crime and criminal behavior.

In examining the item content of the Proactive scale, there is an emphasis on placing culpability on victims, purposefully creating distance between perpetrators and victims, which is congruent with higher rates of strangers in proactive crimes (Cornell et al., 1996). This focus on victims was absent in the Reactive scale. The Proactive scale had a more abstract focus (i.e., societal blame) emphasizing personal disengagement from crime events, which may reflect the “coldness” of proactive violence. Past research has found a lack of empathy and less conscientiousness with proactive crimes (Cornell et al., 1996; Miller & Lynam, 2006).

The Reactive scale contains items with substantial alcohol content, which is consistent with other research (Peterson et al., 2010). Fite et al. (2010) found alcohol misuse and binge drinking to be uniquely associated with reactive aggression (compared with proactive). An examination of the bully-victimization and delinquency relationship compared reactive criminal thinking and anger as mediators (Walters & Espelage, 2017). Only reactive criminal thinking mediated the bully-victimization and delinquency relationship, suggesting that recklessness, which frequently is a consequence of drinking, is a key factor. From an attributional perspective, this reactive emphasis may be viewed as an excuse for the violent behavior (Bushman & Cooper, 1990). There is a strong mental health focus, which reflects person-based, actor explanations for crime.

The Acquisitive scale has a mixture of abstract and concrete items, but includes items with conditional language that refers to degree or frequency variability. These items include terms of “often,” “some choice,” “can contribute,” “it is difficult,” “feel some,” and “contributes.” This language is consistent with other research that describes antecedent factors for acquisitive behavior as less negative or finite (Canter & Ioannou, 2004; Zamble & Quinsey, 1997). Several items refer to choice, which is consistent with investigations into acquisitive offending. Shover (2010) examined those who had committed grand larceny, burglary, robbery, and auto theft and used the idea of “criminal calculus” to frame the results.

In the development of the three scales, item overlap was allowed. It was assumed that these three negative outcomes (proactive, reactive, and acquisitive offending) are different according to basic categories (i.e., kind), which would favor no overlap, but measuring severity within each category was of interest. Thus, to capture as much potential variance as possible, some overlap was allowed. One consequence of an overlapping strategy is relatively lower alpha levels (see Table 2). These levels are slightly lower than other studies using criminal justice–involved samples (.85-.89, Ruiz, Douglas, Edens, Nikolova, & Lilienfeld, 2012), but within the range of other measures of similar length (.50-.86, MacKenzie, Goodstein, & Blouin, 1987; .36-.82, Palmer & Hollin, 2003; .48-.80, Walters, 2005). Similarly, the test–retest coefficients were at the lower end of other studies using criminal justice–involved samples (.68-.84, Cima et al., 2007; .18-.92, Fisher, Reynolds, Wood, & Johnson, 2004; .34-.75, Walters & Willoughby, 2000), but within the range of other measures (.31-.48, Kroner, Reddon, & Beckett, 1991; .40-.48, Walters, 2003). Contributing to the lower test–retest may be the dynamic nature of the items. This lower scale test-retest will faciliate the measure treatment change and change over time.

Assessing three basic areas of proactive, reactive, and acquisitive may have application to classification, assessment, and intervention. Relating the current scales to these criminal justice activities is suggestive of external validity. With regard to classification, the criminal justice system uses basic types of crime for placement and levels of supervision, often using a violent versus nonviolent category. Including the proactive/reactive distinction has implications for management strategies. For example, emotional control may not be sufficient to warrant a reduction in supervision levels for those whose violence is generated by proactive, calculating motivations. With regard to assessment, a general risk estimate may not facilitate an explanation as to why violence has occurred. A proactive/reactive explanation may offer insight into the role of high-risk contexts. Thus, for elevated reactive scores, affectively charged environments may result in high-risk situations, which may guide assessment recommendations.

Proactive, reactive, and acquisitive distinctions may inform appropriate intervention efforts for participants involved in the criminal justice system. Empirical support for such applications comes from recent research examining treatment outcome according to the types of offending. Differential treatment results based on other proactive and reactive scale scores occur for criminal justice anger-management interventions (Walters, 2009a). This study’s proactive measure yielded no pre–post differences, whereas significant pre–post reductions were demonstrated for the reactive measure. Proactive types of offenses have purposeful and external components, which are reflected in the CRAI Proactive scale’s emphasis on societal values. Consequently, interventions that use alternative cognitions (i.e., Socratic questioning, through records) and skill-based methods to challenge offending rationales and values may be of greater benefit (Fite et al., 2010; Walters, 2006). This may include behavioral experiments that make alternative cognitions believable (Bennett-Levy, 2003; Gannon, 2016).

The CRAI Reactive and Acquisitive scales have strong potential for evaluating many current treatment interventions. The Control of Violence for Angry Impulsive Drinkers program was completed by 56 randomly assigned participants who had a history of alcohol-related violence (Bowes et al., 2014). With a 17-month follow-up period, this treatment group had 13% less violent recidivism than the treatment as usual controls. A key element of the intervention was reference to angry aggression affected by drinking. Success in reducing reactive violence may be indicated by scores from the Reactive scale, which has references to alcohol.

The Enhanced Thinking Skills (ETS) program is a widely used intervention intended to reduce recidivism. However, for a sample of 21,000 participants, reduction in recidivism was not realized for acquisitive offending (Travers, Mann, & Hollin, 2014). A focus of the ETS program is to address impulsive behavior and poor problem solving, dynamics that are not exclusively central to acquisitive offending. The researchers argued that appropriate treatment for acquisitive offending will likely involve addressing rational choices. The Acquisitive scale, with an emphasis on victims and their role in crime, may capture rational choice components that are related to acquisitive offending.

Limitations and Future Directions

This article has made an assumption that there are different kinds of offending and has sought to examine the antecedents to these basic offense categories. A different perspective is that differences in crime are best captured by solely degrees of severity and not by different categories (i.e., kinds, typologies). Using two Philadelphia birth cohort data sets, support for explaining crime based on degree of severity has been stronger than for explaining crime according to types of crime (violent vs. other; Brame, Mulvey, & Piquero, 2001). The CRAI Proactive, Reactive, and Acquisitive scales were developed to account for severity (within each scale), but this was done within basic categories.

Patterning the development of scales after offense behavior has its problems. First, the recorded offense often does not accurately reflect the actual offending behavior for a variety of reasons (e.g., plea bargaining, prosecutorial discretion). It may be a leap to assume that offenses are behavioral indicators of this proposed attribution framework. Second, the use of an individual charge may not account for criminally versatile individuals who are likely to be well represented in the criminal justice samples included in this study.

The current Proactive, Reactive and Acquisitive scales are purportedly dynamic. This, though, does not necessarily mean that these scales will function to measure a mechanism of change. Future research will require a pre–post-assessment with a follow-up assessment. Within a risk assessment context, research into the incremental validity to a standardized risk assessment instrument will provide an indication of the predictive and explanatory utility of the scales.

Summary

Prediction in the criminal justice system, whether forensic or correctional, typically does not delineate the psychological processes associated with basic categories of offenses. The present study used multiple samples with divergent scale creation techniques for the development of the Proactive, Reactive, and Acquisitive scales. These scales provide the foundation to link causal mechanisms to basic categories of offenses. This simplistic link can assist in the explanation of past offenses, provide direction for intervention and preventive strategies, and assist in the evaluation of intervention and preventive strategies.

Footnotes

Appendix

Proactive, Reactive, and Acquisitive Scales.

Proactive items (“–” = reverse-keyed items)

11. Society’s rigid rules have very little to do with criminal behavior. (–)

21. It is unfair to blame victims for crime. (–)

23. Society and its rules has little to do with crime occurring. (–)

29. General society contributes to much of the violence on the street. (+)

33. Thinking that a victim can contribute to crime is wrong. (–)

35. It is difficult to see how society can be blamed for crime. (–)

41. Society supports behaviors which are related to crime. (+)

45. A victim’s behavior is not related to crime. (–)

52. Being mentally sick has nothing to do with crime. (–)

53. Society’s mess contributes to crime. (+)

59. Society cannot cause crime. (–)

60. Sometimes crime just happens. (+)

Reactive items

2. One cannot blame alcohol for crime. (–)

10. Criminal behavior is often caused by mental illness. (+)

14. Alcohol does not cause criminal behavior. (–)

16. Doing crime and having a mental illness are totally separate. (–)

22. People who have mental problems are more likely to do crime. (+)

32. High crime rates are related to drinking. (+)

34. Crime can be blamed on being somewhat messed-up psychologically. (+)

38. Blaming alcohol for the majority of crime does NOT make sense. (–)

48. A lot of crime happens when people are in the wrong place at the wrong time. (+)

50. Alcohol has very little to do with crime. (–)

57. Crime cannot be blamed on the victim. (–)

58. Most crimes are related to mental difficulties. (+)

Acquisitive items

15. Victims of crime often exaggerate what happened to them. (+)

19. Crime is not caused by one’s personality. (–)

24. For the most part, people get involved in crime by chance. (+)

27. When a crime occurs, victims have some choice as to their involvement. (+)

29. General society contributes to much of the violence on the street. (+)

33. Thinking that a victim can contribute to crime is wrong. (–)

35. It is difficult to see how society can be blamed for crime. (–)

39. Victims frequently add to their stories. (+)

47. Authority in society is not related to doing crime. (–)

48. A lot of crime happens when people are in the wrong place at the wrong time.(+)

50. Alcohol has very little to do with crime. (–)

51. Victims should feel some responsibility. (+)

53. Society’s mess contributes to crime. (+)

56. Drinking a lot of alcohol can result in crime. (+)

57. Crime cannot be blamed on the victim. (–)

Acknowledgements

Gratefully acknowledged are Toni Hemmati, Jennifer Welsh, and Kimberly Staley for their efforts with data collection.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Daryl G. Kroner, PhD, is a professor in the Department of Criminology and Criminal Justice at Southern Illinois University. His current research interests include risk assessment, measurement of intervention outcomes, interventions among offenders with mentally illness, and criminal desistance.

Brett A. Lacey, MS, is a PhD student in the Department of Criminology and Criminal Justice at Southern Illinois University. His research focuses on multidisciplinary development and analysis of offender and neighborhood classification systems and tools. He is interested in the implementation of classification tools to influence policy and social changes to improve the effectiveness and accuracy of proper assessment of corrective measures for criminal offenders and collective efficacy improvements in high crime areas to achieve optimal results and efficiency.

Mary Louise Cashel, PhD, is an associate professor in the Department of Psychology at Southern Illinois University. Her research focuses on violence prevention, enhancing school climate and safety, and community collaboration. Much of her work has addressed assessment and intervention with juveniles in the justice system.

References

Antonaccio

Tittle

C. R.

Botchkovar

Kranidiotis

(2010). The correlates of crime and deviance: Additional evidence. Journal of Research in Crime & Delinquency, 47, 297-328. doi:10.1177/0022427810365678

Bäckström

Björklund

(2008). The Measures of Criminal Attitudes and Associates (MCAA): Further testing of structural and criterion-related validity. Criminal Justice and Behavior, 35, 1398-1410. doi:10.1177/0093854808322239

Baker

L. A.

Raine

Liu

Jacobson

K. C.

(2008). Differential genetic and environmental influences on reactive and proactive aggression in children. Journal of Abnormal Child Psychology, 36, 1265-1278. doi:10.1007/s10802-008-9249-1

Bennett-Levy

(2003). Mechanisms of change in cognitive therapy: The case of automatic thought records and behavioural experiments. Behavioural and Cognitive Psychotherapy, 31, 261-277. doi:10.1017/S1352465803003035

Boduszek

Hyland

Pedziszczak

Kielkiewicz

(2012). Criminal attitudes, recidivistic behaviour, and the mediating role of associations with criminal friends: An empirical investigation within a prison sample of violent offenders. Europe’s Journal of Psychology, 8(1), 18-31. doi:10.5964/ejop.v8i1.296

Bonta

Wormith

J. S.

(2008). Risk and need assessment. In Raynor

McIvor

(Eds.), Developments in social work with offenders (pp. 131-152). London, England: Jessica Kingsley Publishers.

Borsboom

Mellenbergh

G. J.

van Heerden

(2004). The concept of validity. Psychological Review, 111, 1061-1071. doi:10.1037/0033-295X.111.4.1061

Bowes

McMurran

Evans

Oatley

Williams

David

(2014). Treating alcohol-related violence: A feasibility study of a randomized controlled trial in prisons. The Journal of Forensic Psychiatry & Psychology, 25, 152-163. doi:10.1080/14789949.2014.895025

Brame

Mulvey

E. P.

Piquero

A. R.

(2001). On the development of different kinds of criminal activity. Sociological Methods & Research, 29, 319-341. doi:10.1177/0049124101029003003

10.

Bushman

B. J.

Cooper

H. M.

(1990). Effects of alcohol on human aggression: An integrative research review. Psychological Bulletin, 107, 341-354. doi:10.1037/0033-2909.107.3.341

11.

Canter

D. V.

Ioannou

(2004). Criminals emotional experiences during crimes. International Journal of Forensic Psychology, 1, 71-81.

12.

Carr

W. A.

Rosenfeld

Magyar

Rotter

(2009). An exploration of criminal thinking styles among civil psychiatric patients. Criminal Behaviour and Mental Health, 19, 334-346. doi:10.1002/cbm.749

13.

Cima

Merckelbach

Butt

Kremer

Knauer

Schellbach-Matties

(2007). It was not me: Attribution of blame for criminal acts in psychiatric offenders. Forensic Science International, 168, 143-147. doi:10.1016/j.forsciint.2006.07.004

14.

Cornell

D. G.

Warren

Hawk

Stafford

Oram

Pine

(1996). Psychopathy in instrumental and reactive violent offenders. Journal of Consulting and Clinical Psychology, 64, 783-790. doi:10.1037/0022-006X.64.4.783

15.

Di Fazio

Kroner

D. G.

Forth

A. E

. (1997). The attribution of blame scale with an incarcerated sample: Factor structure, reliability and validity. Criminal Behaviour and Mental Health, 7, 153-164. doi:10.1002/cbm.160

16.

Dutton

D. G.

(1986). Wife assaulter’s explanations for assault: The neutralization of self-punishment. Canadian Journal of Behavioural Science, 18, 381-390. doi:10.1037/h0079964

17.

Embretson

S. E.

(1996). The new rules of measurement. Psychological Assessment, 8, 341-349. doi:10.1037/1040-3590.8.4.341

18.

Fisher

D. G.

Reynolds

G. L.

Wood

M. M.

Johnson

M. E.

(2004). Reliability of arrest and incarceration questions on the risk behavior assessment. Crime & Delinquency, 50(1), 24-31. doi:10.1177/0011128703258872

19.

Fite

P. J.

Raine

Stouthamer-Loeber

Loeber

Pardini

D. A.

(2010). Reactive and proactive aggression in adolescent males: Examining differential outcomes 10 years later in early adulthood. Criminal Justice and Behavior, 37, 141-157. doi:10.1177/0093854809353051

20.

Fox

De Koning

Leicht

(2003). The relationship between attribution of blame for a violent act and EPQ-R sub-scales in male offenders with mental disorder. Personality and Individual Differences, 34, 1467-1475. doi:10.1016/S0191-8869(02)00128-9

21.

Gannon

T. A.

(2016). Forensic psychologists should use the behavioral experiment to facilitate cognitive change in clients who have offended. Aggression and Violent Behavior, 27, 130-141. doi:10.1016/j.avb.2016.02.008

22.

Gudjónsson

G. H.

(1984). Attribution of blame for criminal acts and its relationship with personality. Personality and Individual Differences, 5(1), 53-58. doi:10.1016/0191-8869(84)90137-5

23.

Gudjónsson

G. H.

Pétursson

(1991). The attribution of blame and type of crime committed: Transcultural validation. Journal of the Forensic Science Society, 31, 349-352. doi:10.1016/S0015-7368(91)73167-8

24.

Gudjónsson

G. H.

Singh

K. K.

(1988). Attribution of blame for criminal acts and its relationship with type of offence. Medicine, Science and the Law, 28, 301-303. doi:10.1177/002580248802800407

25.

Hendriks

Fyfe

Styles

Skinner

S. R.

Merriman

(2012). Scale construction utilising the Rasch unidimensional measurement model: A measurement of adolescent attitudes towards abortion. Australasian Medical Journal, 5, 251-261. doi:10.4066/AMJ.2012.952

26.

Jackson

D. N.

(1970). A sequential system for personality scale development. Current Topics in Clinical and Community Psychology, 2, 61-96.

27.

Kroner

D. G.

Mills

J. F.

(2003). The CRiminal Attribution Inventory: User guide. Kingston, Ontario: Authors.

28.

Kroner

D. G.

Hemmati

Mills

J. F.

(2006). Measuring criminal attributions with a normative instructional set: Is there a difference? Legal and Criminological Psychology, 11, 219-227. doi:10.1348/135532505X67585

29.

Kroner

D. G.

Mills

J. F.

(2004). The criminal attribution inventory: A measure of offender perceptions. Journal of Offender Rehabilitation, 39, 15-29. doi:10.1300/J076v39n04_02

30.

Kroner

D. G.

Reddon

J. R.

Beckett

(1991). Basic Personality Inventory clinical and validity scales: Stability and internal consistency. Journal of Psychopathology and Behavioral Assessment, 13, 147-154. doi:10.1007/BF00961428

31.

Loevinger

(1957). Objective tests as instruments of psychological theory: Monograph supplement 9. Psychological Reports, 3, 635-694. doi:10.2466/pr0.1957.3.3.635

32.

MacKenzie

D. L.

Goodstein

L. I.

Blouin

D. C.

(1987). Personal control and prisoner adjustment: An empirical test of a proposed model. Journal of Research in Crime and Delinquency, 24(1), 49–68. doi:10.1177/0022427887024001004

33.

Mair

Hatzinger

Maier

M. J.

Rusch

(2015). eRm: Extended Rasch Modeling (Version 0.15-5). Retrieved from http://cran.r-project.org/web/packages/eRm/index.html

34.

Mandracchia

J. T.

Morgan

R. D.

(2012). Predicting offenders’ criminogenic cognitions with status variables. Criminal Justice and Behavior, 39(1), 5-25. doi:10.1177/0093854811425453

35.

Mann

R. E.

Hanson

R. K.

Thornton

(2010). Assessing risk for sexual recidivism: Some proposals on the nature of psychologically meaningful risk factors. Sexual Abuse: A Journal of Research and Treatment, 22, 191-217. doi:10.1177/1079063210366039

36.

McGuire

(2004). Property offences. In Hollin

C. R.

(Ed.), The essential handbook of offender assessment and treatment (pp. 253-268). West Sussex, UK: John Wiley.

37.

Miller

J. D.

Lynam

D. R.

(2006). Reactive and proactive aggression: Similarities and differences. Personality and Individual Differences, 41, 1469-1480. doi:10.1016/j.paid.2006.06.004

38.

Mills

J. F.

Kroner

D. G.

(2001). Measures of Criminal Attitudes and Associates (MCAA). Selby, Ontario, Canada: Authors.

39.

Mills

J. F.

Kroner

D. G.

Forth

A. E.

(2002). Measures of Criminal Attitudes and Associates (MCAA): Development, factor structure, reliability, and validity. Assessment, 9, 240-253. doi:10.1177/1073191102009003003

40.

Nader

I. W.

Niederkrotenthaler

Schild

A. H. E.

Koller

Tran

U. S.

Kapusta

N. D.

. . . Voracek

(2013). Development of a scale to assess knowledge about suicide postvention using item response theory. Suicide and Life-threatening Behavior, 43, 174-184. doi:10.1111/sltb.12006

41.

Palmer

E. J.

Hollin

C. R.

(2003). Using the Psychological inventory of criminal thinking styles with English prisoners. Legal & Criminological Psychology, 8(2), 175–187.

42.

Peterson

Skeem

J. L.

Hart

Vidal

Keith

(2010). Analyzing offense patterns as a function of mental illness to test the criminalization hypothesis. Psychiatric Services, 61, 1217-1222. doi:10.1176/ps.2010.61.12.1217

43.

Poulin

Boivin

(2000). Reactive and proactive aggression: Evidence of a two-factor model. Psychological Assessment, 12, 115-122. doi:10.1037/1040-3590.12.2.115

44.

Raine

Dodge

Loeber

Gatzke-Kopp

Lynam

Reynolds

. . . Liu

(2006). The reactive–proactive aggression questionnaire: Differential correlates of reactive and proactive aggression in adolescent boys. Aggressive Behavior, 32, 159-171. doi:10.1002/ab.20115

45.

R Core Team. (2016). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Available from http://www.R-project.org/

46.

Reckase

M. D.

(1996). Test construction in the 1990s: Recent approaches every psychologist should know. Psychological Assessment, 8, 354-359. doi:10.1037/1040-3590.8.4.354

47.

Ruiz

M. A.

Douglas

K. S.

Edens

J. F.

Nikolova

N. L.

Lilienfeld

S. O.

(2012). Co-occurring mental health and substance use problems in offenders: Implications for risk assessment. Psychological Assessment, 24(1), 77-87. doi:10.1037/a0024623

48.

Shover

(2010). Aging criminals: Changes in the criminal calculus. In Cromwell

P. F.

(Ed.), In their own words: Criminals on crime: An anthology (5th ed., pp. 409-419). New York, NY: Oxford University Press.

49.

Skeem

J. L.

Monahan

(2011). Current directions in violence risk assessment. Current Directions in Psychological Science, 20(1), 38–42. doi:10.1177/0963721410397271

50.

Skeem

J. L.

Mulvey

E. P.

Appelbaum

Banks

Grisso

Silver

Robbins

P. C.

(2004). Identifying subtypes of civil psychiatric patients at high risk for violence. Criminal Justice and Behavior, 31, 392-437. doi:10.1177/0093854803262585

51.

Steadham

J. A.

Rogers

(2013). Predictors of reactive and instrumental aggression in jail detainees: An initial examination. Journal of Forensic Psychology Practice, 13, 411-428. doi:10.1080/15228932.2013.847350

52.

Stein

N. R.

Schorr

Litz

B. T.

King

L. A.

King

D. W.

Solomon

Horesh

(2013). Development and validation of the coping with terror scale. Assessment, 20, 597-609. doi:10.1177/1073191111411668

53.

Travers

Mann

R. E.

Hollin

C. R.

(2014). Who benefits from cognitive skills programs? Differential impact by risk and offense type. Criminal Justice and Behavior, 41, 1103-1129. doi:10.1177/0093854814543826

54.

Walters

G. D.

(1995). The Psychological Inventory of Criminal Thinking Styles: Part I. Reliability and preliminary validity. Criminal Justice and Behavior, 22, 307-325. doi:10.1177/0093854895022003008

55.

Walters

G. D.

(2003). Development of a self-report measure of outcome expectancies for crime. Journal of Offender Rehabilitation, 37(1), 1-10. doi:10.1300/J076v37n01_01

56.

Walters

G. D.

(2005). Incremental validity of the Psychological Inventory of Criminal Thinking Styles as a predictor of continuous and dichotomous measures of recidivism. Assessment, 12(1), 19–27. https://doi.org/10.1177/1073191104270662

57.

Walters

G. D.

(2006). Proactive and reactive composite scales for the Psychological Inventory of Criminal Thinking Styles (PICTS). Journal of Offender Rehabilitation, 42, 23-36. doi:10.1300/J076v42n04_02

58.

Walters

G. D.

(2007). Predicting institutional adjustment with the Psychological Inventory of Criminal Thinking Styles composite scales: Replication and extension. Legal and Criminological Psychology, 12(1), 69-81. doi:10.1348/135532505X83587

59.

Walters

G. D.

(2009a). Anger management training in incarcerated male offenders: Differential impact on proactive and reactive criminal thinking. International Journal of Forensic Mental Health, 8, 214-217. doi:10.1080/14999010903358995

60.

Walters

G. D.

(2009b). Latent structure of a two-dimensional model of antisocial personality disorder: Construct validation and taxometric analysis. Journal of Personality Disorders, 23, 647-660. doi:10.1521/pedi.2009.23.6.647

61.

Walters

G. D.

(2011). Predicting recidivism with the Psychological Inventory of Criminal Thinking Styles and Level of Service Inventory–Revised: Screening version. Law and Human Behavior, 35, 211-220. doi:10.1007/s10979-010-9231-7

62.

Walters

G. D.

DeLisi

(2015). Psychopathy and violence: Does antisocial cognition mediate the relationship between the PCL: YV factor scores and violent offending? Law and Human Behavior, 39, 350-359. doi:10.1037/lhb0000123

63.

Walters

G. D.

Espelage

D. L.

(2017). Mediating the bullying victimization–delinquency relationship with anger and cognitive impulsivity: A test of general strain and criminal lifestyle theories. Journal of Criminal Justice, 53, 66–73. doi:10.1016/j.jcrimjus.2017.09.007

64.

Walters

G. D.

Willoughby

F. W.

(2000). The Psychological Inventory of Drug-Based Thinking Styles (PIDTS). Alcoholism Treatment Quarterly, 18, 51-66. doi:10.1300/J020v18n02_04

65.

Ward

J. T.

Gibson

C. L.

Boman

Leite

W. L.

(2010). Assessing the validity of the retrospective behavioral self-control scale: Is the general theory of crime stronger than the evidence suggests? Criminal Justice and Behavior, 37, 336-357. doi:10.1177/0093854809359673

66.

Weizmann-Henelius

Sailas

Viemero

Eronen

(2002). Violent women, blame attribution, crime, and personality. Psychopathology, 35, 355-361. doi:10.1159/000068590

67.

White

B. A.

Gordon

Guerra

R. C.

(2015). Callous–unemotional traits and empathy in proactive and reactive relational aggression in young women. Personality and Individual Differences, 75, 185-189. doi:10.1016/j.paid.2014.11.031

68.

Wilson

Attrill

Nugent

(2003). Effective interventions for acquisitive offenders: An investigation of cognitive skills programmes. Legal and Criminological Psychology, 8(1), 83-101.

69.

Woods

C. M.

(2002). Factor analysis of scales composed of binary items: Illustration with the Maudsley Obsessional Compulsive Inventory. Journal of Psychopathology and Behavioral Assessment, 24, 215-223. doi:10.1023/A:1020770831134

70.

Van Hiel

Hautman

Cornelis

De Clercq

. (2007). Football hooliganism: comparing self-awareness and social identity theory explanations. Journal of Community & Applied Social Psychology, 17(3), 169–186. doi:10.1002/casp.902

71.

Zamble

Quinsey

V. L.

(1997). The criminal recidivism process. Cambridge, UK: Cambridge University Press.