Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Feb 1.
Published in final edited form as: Psychol Serv. 2016 May 30;13(3):272–282. doi: 10.1037/ser0000079

Effectiveness of a self-administered intervention for criminal thinking: Taking a chance on change

Johanna B Folk 1, David J Disabato 1, Jordan M Daylor 1, June P Tangney 1, Sharen Barboza 2, John S Wilson 2, Lynda Bonieskie 3, James Holwager 3
PMCID: PMC4980178  NIHMSID: NIHMS775535  PMID: 27243111

Abstract

The current study tested the effectiveness of a self-administered, cognitive-behavioral intervention targeting criminal thinking for inmates in segregated housing: Taking a Chance on Change (TCC). Participants included 273 inmates in segregated housing at state correctional institutions. Reductions in criminal thinking, as assessed by the Psychological Inventory of Criminal Styles-Simplified Version, were found in the general criminal thinking score as well as the proactive and reactive composite scores. Examination of demographic predictors of change (i.e., age, years of education, length of sentence) revealed older and more educated participants decreased in criminal thinking more than younger and less educated participants. For a subset of 48 inmates, completion of TCC was associated with significant reduction of disciplinary infractions. Reductions in reactive criminal thinking predicted reductions in disciplinary infractions. Although further research is needed to determine the effectiveness of TCC in reducing recidivism, the reductions in criminal thinking and disordered conduct suggest this is a promising intervention and mode of treatment delivery. By utilizing self-directed study at an accessible reading level, the intervention is uniquely suited to a correctional setting where staff and monetary resources are limited and security and operational issues limit the feasibility of traditional cognitive-behavioral group treatment.

Keywords: Criminal Thinking, Inmates, Psychological Inventory of Criminal Styles-Simplified Version, Restrictive Housing, Taking a Chance on Change

The U.S. Bureau of Justice Statistics census estimates that, on any given day, over 80,000 inmates are housed in restrictive housing units, including traditional disciplinary or administrative custody units, and 25,000 of these are in supermax facilities, where the sole purpose is to house large numbers of inmates in segregation (Steinbuch, 2014; Stephan, 2008). Inmates in restrictive housing are generally on lockdown in their cells for 23 hours per day, only coming out for showers, exercise in fenced areas, healthcare and mental health appointments, and institutional classification committee reviews (e.g., Sánchez, 2013). Inmates in long-term restrictive housing are more likely to reoffend after release from confinement than are general population inmates (Mears & Bales, 2009). Thus, it is imperative to identify interventions that can be delivered to these high-risk inmates.

Restrictive housing is used for diverse reasons and contains myriad challenges. Individuals can be placed in restrictive housing for rule violations (i.e., disciplinary segregation), posing a risk to security or safety in the facility (i.e., administrative segregation), or because they are deemed at risk in the general population (i.e., protective custody). Many of these individuals are severely mentally ill or cognitively disabled (Metzner & Fellner, 2010), are gang affiliated, or have accrued numerous and/or serious disciplinary infractions. These factors pose unique challenges to maintaining security within restrictive housing units, and facilities more broadly, including intensive demands on staffing resources. Containment, rather than rehabilitation, is the goal; treatment is often a low priority, if considered for these inmates at all.

One treatment target that is increasingly used to reduce recidivism risk in general population inmates, yet has remained relatively unexplored in inmates in restricted housing, is criminal thinking. Natural changes in criminal thinking over time suggest it is a dynamic aspect of personality (Tangney et al., 2015; Walters, 2003), and several correctional interventions have been designed to target these cognitive processes (e.g., Escaping the Cage, Batastini, Morgan, Kroner, & Mills, 2014; Thinking for a Change, Bush, Glick, & Taymans, 1997; Impact of Crime, Folk et al., 2015; Cognitive Interventions Program, National Institute of Corrections, 1996; Reasoning and Rehabilitation, Ross & Fabiano, 1985). The current study examines the effectiveness of a new, self-administered criminal thinking intervention deliverable to inmates in segregated housing: Taking a Chance on Change.

Criminal Thinking: A Meaningful Target for Intervention

Criminal thinking is defined as “attitudes, beliefs, and rationalizations that offenders use to justify and support their criminal behavior” (Walters, 2012, p. 272). It supports criminal behavior before, during, and after the offense (Maruna & Mann, 2006, p. 155). Criminal thinking is ripe for intervention due to its predictive validity. The construct is a “Big Four” risk factor associated with recidivism (Andrews & Bonta, 2010) and significantly predicts recidivism in state and federal inmates (Walters, 2012) as well as jail inmates (Caudy et al., 2015). Criminal thinking is uniquely related to violent recidivism beyond conventional risk assessment (Mills, Kroner, & Hemmati, 2004), recidivism among sex offenders (Walters, Deming, & Casbon, 2015), and to disciplinary infractions during incarceration (Gendreau, Goggin, & Law, 1997; Simourd, 1997; Tangney et al., 2012; Walters 1996; Walters & Elliott, 1999; Walters & Geyer, 2005; Walters & Schlauch, 2008). When controlling for other variables, criminal thinking as assessed by the Psychological Inventory of Criminal Thinking Styles (PICTS; Walters, 1995) remains a significant predictor of disciplinary infractions (Walters & Schlauch, 2008).

Many researchers accept the proactive versus reactive classification of criminal thinking. Both predict general recidivism equally (Walters, 2012). Preliminary findings suggest proactive criminal thinking drives calculated criminal behavior and is associated with an offender's positive outcome expectancies (Walters, 2007). Reactive criminal thinking predicts disciplinary infractions during incarceration (Walters & Geyer, 2005) and is associated with poor impulse control and hostile attribution biases (Walters, 2007).

Proactive and reactive criminal thinking are correlated dimensions (with r values ranging from .60 to .76) that motivate antisocial behavior (Cuadra, Jaffee, Thomas, & DiLillo, 2014; Disabato et al., 2015; Ragatz, Anderson, Fremouw, & Schwartz, 2011; Walters, 2005; Walters et al., 2015).

Interventions Targeting Criminal Thinking

Criminal thinking can be reduced through intervention. The National Institute of Corrections (NIC) promotes Thinking for a Change (Bush et al., 1997) as a group intervention focusing on cognitive restructuring, social skills training, and concrete problem solving. It has reduced criminal thinking and recidivism among probationers (Golden, Gatchel, & Cahill, 2006; Lowenkamp, Hubbard, & Makarios, 2009) and increased social problem-solving skills and decreased agreement with criminal thinking errors among state inmates (Bickle, 2015).

Reasoning and Rehabilitation (Robinson & Porporino, 2001) and Enhanced Thinking Skills (Sadlier, 2010) reduced self-reported criminal thinking with moderate effect sizes (Blud, Travers, Nugent, & Thornton, 2003). Reductions assessed by the PICTS (Walters, 1995) were most evident on the Cognitive Indolence subscale, a reactive criminal thinking style that assesses poor critical thinking skills and an overreliance on cognitive shortcuts for dealing with problems. Simourd's Criminal Attitude Program (CAP; Simourd, 2007) led to statistically significant reductions in criminal attitudes (Cohen's d = 0.58), as assessed by the Criminal Sentiments Scale-Modified (Shields & Simourd, 1991) among male offenders (Simourd, Olver, & Brandenburg, 2015).

Lifestyle Issues is a psychoeducational program focusing on factors believed to be instrumental in initiating, transforming, and maintaining a criminal lifestyle (Walters, 1990); criminal thinking is a maintenance factor in the program. In a sample of 85 male federal inmates, current criminal thinking as assessed by the PICTS (Walters, 1995), decreased, whereas historical criminal thinking did not (Walters, Trgovac, Rychlec, DiFazio, & Olson, 2002). These studies highlight some of the existing evidence that criminal thinking is amenable to intervention among a wide range of offenders.

Barriers to Delivering Interventions in Restrictive Housing Units

Current interventions are effective at reducing criminal thinking, and some reduce behavioral outcomes such as recidivism, but their methods of delivery present several barriers to implementation, particularly in restrictive housing units. The interventions outlined above are designed to be delivered to groups of inmates, led by a trained facilitator. Although an often-effective mode of treatment delivery, group treatment can be problematic in correctional settings due to scarce staffing and budgetary resources (Pattavina & Taxman, 2013; Taxman & Belenko, 2012), as well as limited treatment space. Group treatment also excludes inmates confined to restrictive or segregated housing, who are unable to participate due to security concerns. To reach inmates in restrictive/segregated housing with mental health issues, some correctional facilities will use therapeutic cubicles: phone-booth sized barred cells arrayed to permit a number of inmates to participate in a group intervention while remaining safe from each other. Although these cubicles can be used for criminal thinking interventions, they are expensive and require extensive time from security officers.

We were able to locate only one intervention specifically designed for or used with inmates in restrictive/segregated housing that targets criminal thinking: Escaping the Cage (Batastini et al., 2014). Escaping the Cage is a self-directed intervention aimed at maximizing adaptive behaviors so individuals with serious mental illness can return to the general population with decreased risk to themselves and others. The intervention consists of nine treatment modules with 46 total handouts that inmates complete at their own pace. Mental health clinicians provide feedback on completed tasks and assignments and discuss specific challenges or personal areas of growth with participants. Criminal thinking is a target, but not the primary focus, of the intervention. Although efforts are underway to evaluate Escaping the Cage (e.g., Miller & Seirup, 2015), there is currently no published research demonstrating its effectiveness.

The Current Study

The current study presents an evaluation of a program that attempts to overcome some of the implementation barriers of existing criminal thinking interventions as outlined above, focusing primarily on reducing criminal thinking. The Taking a Chance on Change (TCC) program is a psychoeducational, in-cell, self-administered intervention that can be administered to inmates in restrictive or segregated housing. The low resources required to deliver TCC make it a feasible alternative for inmates in restrictive housing where provision of one-on-one or group-based criminal thinking interventions may be challenging to deliver. The current study assesses the effectiveness of TCC in reducing criminal thinking and disciplinary infractions among inmates held in restrictive housing. We expect to see decreases in criminal thinking from pre- to post-intervention overall, as well as in the specific proactive and reactive dimensions.

We hypothesize reductions in reactive criminal thinking will be greater in magnitude than those for proactive criminal thinking. Reactive aggression is largely a product of intense negative emotions and poor emotion regulation (Davidson, Putnam, & Larson, 2000; Wilkowski & Robinson, 2008). The TCC curriculum includes topics such as anger and stress management, which may be more directly relevant to reactive criminal thinking. Inmates may learn how to cognitively reappraise hostile attribution biases and regulate intense feelings of anger or impulsivity through TCC modules. Although some research suggests proactive aggression is also a product of emotion dysregulation (Orobio de Castro, Merk, Koops, Veerman, & Bosch, 2005), other research suggests it stems from callous and unemotional traits that are difficult to change (e.g., Marsee & Frick, 2007). Therefore, we hypothesize only small decreases in proactive criminal thinking.

For a subset of participants, we also examine changes in disciplinary infraction rates from pre- to post-intervention. We hypothesize the intervention will lead to reductions in disciplinary infractions when comparing the six months prior to and following completion of the intervention. Although we cannot test mediation within the current study design, we expect decreased criminal thinking from pre- to post-intervention is the mechanism by which disciplinary infractions will be reduced because criminal thinking is a known predictor of disciplinary infractions during incarceration.

Predicting changes in criminal thinking and disciplinary infractions

In addition to considering whether TCC is effective in reducing criminal thinking and disciplinary infractions, we examine whether some individuals experience greater reductions than others. Three potentially relevant individual difference factors we consider are age, education level, and length of prison sentence. These hypotheses are exploratory in nature and the analyses are intended to test the generalizability of effects.

Age

On the one hand, the identities of younger individuals are more malleable and may be more apt to change based on participation in treatment. But it is also possible that younger inmates are more open to the influences of deviant peers during incarceration and invested in exploring their criminal identity, resulting in resistance to changing their patterns of criminal thinking. As such, they would be less likely to benefit from the intervention, relative to their older peers, who have greater experience, including experience with the potential negative consequences of criminal thinking and behavior.

Education

Given the self-directed nature of the intervention, it is possible education level will be relevant to how much individuals benefit. Although the reading level of the program is designed for individuals with limited academic skills, individuals with higher levels of education may be better able to engage with the material since it is somewhat akin to completing homework assignments, with which they may have more experience.

Sentence Length

With regard to sentence length, one possibility is that individuals with longer pending sentence lengths may be less likely to decrease criminal thinking and disciplinary infractions following participation in TCC than those with shorter sentences. Individuals with longer sentences may be more willing to enhance connections to the criminal community in order to survive within the correctional environment, and criminal thinking and misconduct may enhance these connections and minimize being victimized. Individuals with shorter sentences, in contrast, may be preparing for release into the community in the near future and thus might capitalize on motivation to make changes to aid in their reentry. On the other hand, some research indicates that inmates serving life without the possibility of parole show similar rates of disciplinary infractions to inmates serving comparatively shorter sentences (Cunningham & Sorensen, 2006). Thus, sentence length may not be significantly related to inmates' criminal thinking, institutional misconduct, or ability to benefit from TCC.

In addition to providing evidence for the effectiveness of the TCC intervention, results of these analyses will help determine whether the intervention is more effective for certain types of individuals, providing empirically-derived guidance for more targeted treatment delivery.

Method

Taking a Chance on Change (TCC)

Taking a Chance on Change (TCC) is a structured in-cell treatment program designed to provide inmates housed in long-term restrictive/segregated units with the opportunity to participate in psychoeducational programming. The purpose of TCC is to address cognitive and behavioral deficits, as well as challenges common among inmates in long-term restrictive housing such as impulse control, anger, emotion regulation, effective communication, goal-setting, and long-term planning.

TCC is divided into eight units, each of which includes between four and eight modules (50 total modules). The eight units are: Preparing for Change, Self-Awareness/Goal Setting, Identifying and Changing Mistaken Beliefs, Effective Problem-Solving, Effective Communication, Anger Management, Stress Management, and Relapse Prevention. Each module contains a handout and worksheet that inmates review and complete. At the conclusion of a specific unit, a module reviewing the concepts discussed and an in-cell/open-book assessment is conducted to reinforce the inmate's understanding of the unit's concepts.

Completion of the entire TCC program can require between nine and twelve months; however, inmates may participate in individual TCC units without completing the entire program. TCC is typically provided to inmates on long-term restrictive housing units during mental health cell-to-cell rounds. Each week, mental health staff conducting rounds distribute a handout and worksheet for the next module and collect the worksheet completed by the inmate for the prior week's homework. Participants have the opportunity to briefly discuss the module's content with mental health staff during the rounds.

Although the reading level of TCC was designed for individuals with limited academic skills, basic literacy skills are a necessary requisite for participation. Lexile analyses indicate the eight TCC units require an 8th grade reading level on average (SD = 1.4)1. TCC is available in both English and Spanish and has been successfully implemented within six different state correctional systems.

Participants

The sample consisted of 273 inmates in segregated housing at six state correctional institutions operated by the Maryland Department of Public Safety and Correctional Services. Participation in TCC is voluntary. Individuals assigned to segregated housing are seen routinely by a mental health professional during cell-to-cell rounds. All inmates who will be spending more than 90 days in segregated housing are made aware of the TCC program upon admission to segregated housing. Individual inmates who are determined to have the need for a cognitive intervention based on their pattern of rule-violating behavior or mental health needs, as well as inmates who are referred for the program by security staff, are prioritized for inclusion. Inmates can also refer themselves to the program by making a request. Participation within a segregated housing unit may be limited (e.g., 12 to 20 inmates at a time) due to availability of mental health staff. Individuals who do not complete assignments may be removed from the program so other individuals can access the materials. There are no negative security consequences for inmates who choose not to participate or who choose to stop participation once they have begun. There is also no formal security incentive for participation in the program.

Of those who reported demographic information2, 99% were male (n = 225). Mean age was 30.4 years (SD = 8.16) and ranged from 18 to 62 (n = 195). Mean years of education was 11.3 (SD = 1.32) and ranged from 8 to 16 (n = 177). Mean length of sentences was 24.6 years (SD = 27.2) and ranged from 1 to 130 years (n = 192). In regards to self-reported offense, 75% of inmates indicated their most severe index offense was a violent crime, 8% a property crime, 9% a drug related crime, and 7% reported their index offense as “other” (n = 190). Since participants did not have the option to classify their offense into multiple categories or report multiple crimes, the statistics reported here may not adequately characterize the sample's criminal behavior.

Procedure and Pre/Post Measures

Prior to the start of TCC, participants were given a survey packet to complete containing a demographic questionnaire, the pre/post measure of criminal thinking, and an informed consent that described the TCC program. As noted, participation in the TCC program was voluntary. Participants were allowed to complete the eight TCC units at their own pace, which ranged from 52 to 402 days (M = 192.9; SD = 106.4). Three participants did not complete all eight units and instead completed only 2, 4, and 5 units. After the intervention, all participants, including the three who completed only a portion of the TCC, were re-administered the measure of criminal thinking.

The Psychological Inventory of Criminal Styles – Simplified Version (PICTS-SV) was adapted from the original PICTS (Version 4.0) by the senior authors (Barboza and Wilson) after permission to modify the original PICTS was provided by Dr. Walters, the developer of the original PICTS. The simplified version lowered the reading level required for the instrument to the sixth grade but otherwise sought to maintain the content and intent of the original PICTS items. Disabato et al. (2015) replicated the factor structure and high reliability of the original PICTS in the PICTS-SV. Intercorrelations for the PICTS-SV (Disabato et al., 2015) and the original PICTS (Walters et al., 2015) in sexual offender samples suggest the two versions are commensurate.

The PICTS-SV includes 80 self-report items, to which participants respond on an asymmetrical 4-point Likert scale (1 = disagree; 2 = uncertain; 3 = agree; 4 = strongly agree) without a specified time interval. Excluding the 16 validity items and 8 items loading on the original Sentimentality subscale,3 56 PICTS-SV items measure criminal thinking. Identical to the original PICTS, proactive and reactive criminal thinking composite scores for the PICTS-SV are calculated by summing items from their respective a priori criminal thinking subscales (Walters et al., 2011). The proactive criminal thinking subscale contains 32 items; a sample item is “I thought the victims of my crimes deserved what they got or should have known better.” The reactive criminal thinking subscale contains 24 items; a sample item is “I tend to act without thinking when I'm under stress.” A general criminal thinking total score is computed by summing all 56 criminal thinking items.

Numbers of disciplinary infractions were recorded to provide a behavioral outcome of the TCC intervention. Disciplinary infractions were obtained from prison records and reflect a range of rule violations including but not limited to assaults, threats, weapons, disruptive behavior, interfering with operations, and disobeying orders. Participants' number of disciplinary infractions from six months prior to the start of the TCC served as the pre-treatment behavioral indicator. Number of disciplinary infractions from the end of the TCC intervention to 6 months after served as the post-treatment behavioral indicator. The number of disciplinary infractions ranged from 0 to 5 both before and after the TCC intervention.

Unplanned Missing Data

A total of 273 participants completed the pre-intervention criminal thinking assessment, but only 197 completed the post-intervention criminal thinking assessment, resulting in 72.2% study retention. Although no data were recorded on the sources of study attrition, there are two hypothesized reasons for the missing data post-intervention. First, some participants left segregated housing, transferred prisons, or were released to the community during the study and were no longer able to participate. No post-intervention data were obtained for these participants. Second, some participants chose to discontinue the intervention and were dropped from the study. Reasons for discontinuation are unavailable to the researchers.

To further understand the potential reasons for missing criminal thinking data, we conducted Welch's t-tests for unequal variances comparing those with and without post-intervention criminal thinking data. The Welch's t-test computes the standard errors separately for each group and then uses the Satterthwaite approximation of degrees of freedom (Ruxton, 2006). Missing post-intervention criminal thinking data was not related to age (t (34.9) = 0.37, p = .711) or education (t (24.9) = -0.20, p = .841); however, participants with higher initial levels of criminal thinking (t (137.2) = -1.94, p = .054) and shorter sentences (t (63.8) = 3.43, p = .001) were somewhat more likely to have missing post-intervention criminal thinking data.

Planned Missing Data

Of those who completed both pre- and post-intervention criminal thinking data, disciplinary infraction data were collected on 48 (24.4%). These were the first 48 participants to complete TCC. Due to the intensive labor required to obtain the data, disciplinary infractions were not collected for subsequent participants (i.e., planned missing data). To assess whether the first 48 participants who completed the TCC intervention were representative of the total sample, we conducted Welch's t-tests for unequal variances comparing those with and without disciplinary infraction data. Missing disciplinary infraction data was not related to initial levels of criminal thinking (t (73.5) = 1.43, p = .157), changes in criminal thinking (t (77.6) = -0.62, p = .536), age (t (193) = -1.43, p = .156) or education (t (64.4) = -0.50, p = .618); participants with longer sentences (t (131.4) = -2.54, p = .012) were more likely to have missing disciplinary infraction data.

Results

Descriptive Statistics

Pre- and post-intervention criminal thinking descriptive statistics are reported in Table 1. The coefficient alphas for the PICTS-SV scales are also provided and suggest satisfactory internal consistency. Among the subsample with data on disciplinary infractions, pre-test counts averaged 1.92 (SD = 1.46) and post-test counts averaged 0.48 (SD = 0.92).

Table 1. Descriptive Statistics of the PICTS-SV Scales.

Pre-test Post-test
Scale Items M SD α M SD α
General Criminal Thinking 56 123.02 27.12 .92 114.16 26.72 .92
Proactive Criminal Thinking 32 65.89 15.44 .87 62.81 15.34 .87
Reactive Criminal Thinking 24 56.95 14.64 .90 51.35 13.67 .88
Open in a new tab

Note. Items = number of items; M = mean; SD = standard deviation; α = coefficient alpha

Mean Level Changes in Criminal Thinking and Disciplinary Infractions

We first examined whether TCC participants' criminal thinking and disciplinary infractions changed from pre- to post-intervention using paired-samples t-tests with listwise deletion (see Table 2). The unbiased4 effect sizes, Cohen's d (i.e., standardized mean differences) and rpb (i.e., point-biserial correlation), were used to quantify the amount of change.

Table 2. Pre-post Analyses.

Scale n MD r t d d 95% CI rpb rpb 95% CI
General 193 -6.16 .63*** -3.73*** .27 [.14, .39] .13 [.07, .19]
Proactive 196 -2.12 .61*** -2.19* .16 [.03, .28] .08 [.02, .14]
Reactive 194 -4.15 .63*** -4.82*** .34 [.22, .47] .17 [.11, .23]
Infractions 48 -1.44 .19 -6.34 .90 [.49, 1.31] .41 [.26, .57]
Open in a new tab

Note.

*

p < .05;

**

p < .01;

***

p < .001. General = general criminal thinking; Proactive = proactive criminal thinking; reactive = reactive criminal thinking; infractions = disciplinary infractions; MD = unstandardized mean difference; r = correlation of pre and post scale scores; t = t-obtained value; d = unbiased Cohen's d; rpb = unbiased point-biserial correlation of time and scale scores.

As shown in Table 2, results indicate all facets of criminal thinking and disciplinary infractions significantly decreased during the intervention. The criminal thinking effect sizes are small to moderate based upon Cohen's (1988) heuristics. Reactive criminal thinking decreased over twice as much as proactive criminal thinking. Considering minimal staff time was required for the intervention, these effect sizes are meaningful. The disciplinary infraction results showed a large effect size that is over three times that of general criminal thinking.

Individual Differences in Changes in Criminal Thinking and Disciplinary Infractions

Analysis plan

To understand whether certain individuals benefit more from the TCC intervention, we assessed the degree to which there are individual differences in changes in criminal thinking and disciplinary infractions. The criminal thinking scales demonstrated strong stability coefficients from pre- to post-intervention, while disciplinary infractions did not (see Table 2). To test predictors of change, we created change scores (post-intervention – pre-intervention) for the criminal thinking variables and disciplinary infractions. To examine how age, years of education, and length of sentence related to changes in general, proactive, and reactive criminal thinking, as well as disciplinary infractions, we conducted linear regressions using the change score as the criterion.

Initial level of criminal thinking or disciplinary infractions was included as a control in each regression analysis. This addresses the practical problem of regression to the mean, whereby during repeated measurements, relatively high or low observations on the first measurement are likely to be followed by less extreme scores upon subsequent measurement (Fitzmaurice, 2001). For example, to determine whether age predicts changes in criminal thinking, pre-intervention criminal thinking was entered as a control variable, age was entered as a predictor, and the difference score reflecting changes in criminal thinking served as the dependent variable.

Furthermore, we tested whether changes in criminal thinking predicted changes in disciplinary infractions. Because – due to planned missingness – the disciplinary infraction analyses used a small subsample of participants (n = 48), we tested for potential univariate outliers that could distort average effects. One participant showed excessively large reductions in reactive criminal thinking, exceeding the outlier cutoff suggested by Cousineau and Chartier (2010) for a sample size of 50 (z-score < -3.29). Therefore, the reactive criminal thinking regression analysis was conducted without this participant's data.

Regressions were conducted within a structural equation modeling framework in the R package “lavaan” (Rosseel, 2012). To help correct for systematic bias due to both unplanned and planned missing data, Full Information Maximum Likelihood (FIML)5 was used with all other baseline variables entered as auxiliary variables6.

Findings

Results of the 15 regressions, including the standardized regression weights and their associated 95% confidence intervals, as well as the squared semi-partial correlations (i.e., R-squared change), are shown in Table 3. The correlation between the difference score (i.e., change) and pre-intervention criminal thinking was -0.45 for general criminal thinking and proactive criminal thinking, -0.48 for reactive criminal thinking, and -.82 for disciplinary infractions. All models were saturated and thus have no model fit. Age and years of education significantly predicted changes in all facets of criminal thinking during the intervention. TCC was more effective for older and more educated inmates independent of their pre-intervention criminal thinking. Changes in reactive criminal thinking significantly predicted changes in disciplinary infractions, indicating the two variables changed together7.

Table 3. Prediction of Change Analyses.
Outcome Scale Beta Beta 95% CI sr2
General Criminal Thinking
 Age -.18** [-.31, -.05] .03
 Education -.20** [-.34, -.06] .04
 Length of sentence -.08 [-.21, .05] .01
Proactive Criminal Thinking
 Age -.17* [-.30, -.04] .03
 Education -.21** [-.35, -.07] .04
 Length of sentence -.08 [-.21, .05] .01
Reactive Criminal Thinking
 Age -.15* [-.28, -.02] .02
 Education -.16* [-.30, -.02] .02
 Length of sentence -.07 [-.20, .06] .00
Disciplinary Infractions
 Age -.07 [-.25, .12] .00
 Education -.12 [-.27, .03] .01
 Length of sentence -.07 [-.33, .21] .00
 General Criminal Thinking .10 [-.06, .25] .01
 Proactive Criminal Thinking .07 [-.07, .20] .00
 Reactive Criminal Thinking† .21* [.00, .42] .03
Open in a new tab

Note.

*

p < .05;

**

p < .01;

***

p < .001; each row represents a separate regression;

†

an outlier was removed for the regression analysis. Beta = standardized regression weight; sr2 = squared semi-partial correlation.

To further understand the predicted changes in criminal thinking across various levels of age and education, the number of standard deviations away from zero change was calculated: standardized change (see Table 4). These values are similar to standardized mean differences (i.e., Cohen's d), except the deviation is from zero change rather than average, or mean, change. The values are interpreted similar to Cohen's d.

Table 4. Predicted Standardized Change.
Predictor Age (years) Education (years)
Outcome Scale 25 35 45 10 11 12
General Criminal Thinking -0.18 -0.40 -0.62 -0.11 -0.25 -0.40
Proactive Criminal Thinking -0.07 -0.28 -0.48 0.03 -0.13 -0.29
Reactive Criminal Thinking -0.28 -0.47 -0.66 -0.23 -0.35 -0.47
Open in a new tab

Note. Standardized change can be interpreted similar to standardized mean differences (i.e., Cohen's d).

Based on these calculations, 25-year-old participants showed a very small effect size for decreases in proactive criminal thinking and a small effect size for decreases in reactive criminal thinking. In contrast, 45-year-old inmates showed moderate effect sizes for both proactive and reactive criminal thinking. Less educated inmates who completed two years of high school (i.e., 10th grade) showed essentially no changes in proactive criminal thinking and a small effect size for decreases in reactive criminal thinking. In contrast, adults who completed four years of high school (i.e., 12th grade) showed a small effect size for proactive criminal thinking and a moderate effect size for reactive criminal thinking. We do not report estimated change for more extreme values of age and education due to very small subsamples8. In sum, although the intervention was not as effective for younger, less educated participants, the intervention was effective for older, more educated participants.

Correcting for Multiple Comparisons

The large number of null hypothesis significance tests in the present study – 19 – indicates that the risk for false positive (i.e., Type I) errors is high. When only one hypothesis test is conducted with an alpha level of .05, the probability of an incorrect rejection of the null hypothesis is 5%; this probability increases as the number of hypothesis tests increases9. Because controlling for the false positive rate via the Bonferroni correction reduces statistical power, we controlled for the false discovery rate instead via the Benjamini-Hochberg (B-H) correction (Benjamini & Hochberg, 1995). The false discovery rate ensures, of the hypothesis tests that were statistically significant, the probability of an incorrect rejection of the null hypothesis is 5%. The false discovery rate is attractive due to maintaining statistical power while still preventing a large number of erroneous significant effects. However, the procedure has a greater probability of Type I error rates compared with the traditional Bonferroni correction. With this new criteria, the following effects are no longer statistical significant: mean changes in proactive criminal thinking, age and education predicting individual changes in reactive criminal thinking, and changes in reactive criminal thinking predicting changes in disciplinary infractions. Accordingly, these effects should be interpreted with caution as they may be due to random sampling error.

Discussion

We tested the effectiveness of a self-administered, cognitive-behavioral intervention targeting criminal thinking for inmates in segregated housing. Taking a Chance on Change (TCC) was designed to provide treatment to those who are unable to participate in traditional group cognitive-behavioral treatment. The 197 inmates who had pre- and post-TCC data on average spent half of a year to complete part or all of the eight TCC units. It was hypothesized the intervention would reduce general criminal thinking over time as measured by pre- and post-intervention assessments. Furthermore, reactive criminal thinking was hypothesized to decrease more so than proactive criminal thinking due to the intervention's focus on psychological stress and anger. Disciplinary infractions were also expected to decline following participation in the intervention as a result of changes in criminal thinking. The results confirmed these hypotheses with small to moderate effect sizes.

The curricular focus on emotion regulation skills may explain why more substantial reductions in reactive criminal thinking were found compared to proactive criminal thinking. Whereas reactive criminal thinking is associated with poor impulse control as a result of strong negative emotions such as anger, proactive criminal thinking is associated with more purposeful and calculated criminal behavior (Walters, 2007). TCC teaches skills such as problem solving, effective communication, anger management, and stress reduction, which may have increased inmates' ability to regulate strong negative emotions and subsequently reduced impulsivity. As such, inmates may have become less prone to reactive criminal thinking patterns. Proactive criminal thinking has been more closely associated with callous and unemotional personality traits (e.g., psychopathy) that are potentially less responsive or non-responsive to intervention (Walters, 2008). In addition, 11 of the 32 proactive criminal thinking items from the PICTS-SV and original PICTS refer to past criminal history (e.g., I have helped out friends and family with money I got doing crime). Due to their historical nature, these items therefore cannot capture any change due to the intervention, resulting in less change in proactive criminal thinking overall.

The reduction in reactive criminal thinking may also explain the reduced rate of disciplinary infractions following TCC participation. Participants who evidenced greater reductions in reactive criminal thinking from pre- to post-intervention, but not proactive criminal thinking, also accrued fewer disciplinary infractions over time. However, the effect was not significant after the B-H correction, partially due to the limited statistical power resulting from the small subsample of participants with disciplinary infraction data. As disciplinary infractions are often reactive in nature (Toch & Adams, 2002; Walters & Geyer, 2005), they may be due to deficiencies in areas such as impulse control and the ability to regulate one's negative emotions, which are specifically targeted by TCC. Future research should replicate an association between changes in criminal thinking and criminal behavior in response to the TCC intervention using meditational analyses to confirm the present study's results are not due to sampling error.

In addition to overall reductions in criminal thinking, two demographic characteristics, age and education level, predicted greater responsiveness to the intervention as evidenced by changes in criminal thinking. Specifically, older and more educated participants decreased in criminal thinking more than younger and less educated participants. Sentence length did not predict changes in criminal thinking. The results are more conclusive for general and proactive criminal thinking as the effects with reactive criminal thinking were non-significant after controlling for the false discovery rate.

Given the TCC intervention requires only an eighth grade reading level, on average, and participants had completed eleven years of education on average, it is unlikely that those with less education had trouble understanding the material presented in the intervention. Because TCC is self-directed, it may be akin to completing homework assignments in school. As such, participants with more years of education may be more accustomed to and/or comfortable completing homework-like tasks, and therefore more engaged and invested in the intervention.

Regarding age, the age crime curve theory suggests many young adults involved in criminal activity eventually discontinue offending behavior when they enter middle or late adulthood (Farrington, 1986). Similarly, in the absence of an intervention, older jail inmates evidence greater reductions in criminal thinking compared to younger inmates, who show more modest reductions (Tangney et al., 2015). In contrast, younger jail inmates tend to become more connected to the criminal community during the period of incarceration (Folk et al., 2015), which suggests their tendency to become further embedded in the criminal subculture during incarceration. It could be that older inmates are more motivated to change their lifestyle and thus make use of the intervention whereas younger inmates may be more invested in their criminogenic beliefs to prepare to return to the incarcerated criminal community after segregation.

We initially hypothesized those with longer pending sentences would be less likely to change because they may be more willing to enhance connections to the criminal community in preparation for a longer period of incarceration. Length of prison sentence was unrelated to changes in criminal thinking. One likely explanation for our finding is that, because inmates completed TCC at different points during their sentences, irrespective of sentence length, they had different times between TCC completion and release. Those anticipating release in the near future may be more motivated to change to better integrate into the community at large. As such, the time between TCC completion and release, not simply sentence length, may be the most relevant information to assess. Regrettably, we did not have access to this data in the current study.

Limitations and Future Directions

The present study utilized a pre-post design without a control group, allowing for several threats to validity (Cook and Campbell, 1979). First, participants were not randomly selected from a pool of possible participants (i.e., selection effects). Although some exclusion criteria existed (see methods section), participants were largely self-selected, likely resulting in self-motivated inmates. Second, the experience of restrictive housing may have increased or decreased criminal thinking and disciplinary infractions without the presence of an intervention (i.e., history and maturation effects). Third, participants can change in settings where participants receive little attention and are energized to change by even minimal attention (i.e., novelty effects, Shadish, Cook, & Campbell, 2002). The psychological changes may have been due to simply participating in an intervention program, regardless of its content. Other threats to validity are also present in pre-post designs and can be found in Cook and Campbell (1979) and Shadish and colleagues (2002), Future research on TCC's efficacy should include a randomized control group in order to minimize threats to validity. An active control group would allow for researcher's to minimize novelty effects as well. This design would also allow researchers to test if content acquisition from TCC modules mediates the observed effect sizes.

Given that TCC was less effective for younger and less educated inmates, it is possible the intervention content is not developmentally appropriate for these subpopulations. Future work might consider modifying the intervention content.

Disciplinary infraction data were only available for 17.6% of the total sample (n = 48), decreasing the statistical power to find an association between changes in criminal thinking and changes in disciplinary infractions. Future research should replicate the disciplinary infraction findings with a larger sample, allowing for greater statistical power.

Clinical Implications

Mental health professionals and correctional counselors who struggle to deliver meaningful interventions to inmates in restrictive or segregated housing may find promise in the initial results reported here. The challenge of encouraging positive cognitive and behavioral change in this population cannot be overstated. Current TCC results indicate the program may be an effective means of providing structured support for positive behavioral change using limited staffing resources. Although TCC is not intended to be a mental health intervention and is not designed for inmates with serious mental illness, there are a large number of inmates in long-term segregated/restricted housing who need help reducing criminogenic thinking and improving their behavior. Some of these inmates have the capacity to benefit from TCC and self-propelled psychoeducational programs. In a correctional environment that can be characterized by hostility, dismissiveness, and disengagement, provision of self-guided criminogenic reduction programming may uniquely address the risk, needs, and especially responsivity principles in this population (Andrews & Bonta, 2010).

Results showed demographic differences in intervention responsiveness. For maximal utility, TCC should be targeted to older and more educated inmates. Inmates in their early twenties or without a high school diploma showed smaller reductions in criminal thinking compared to their older and more educated peers (see Table 4). Given that some criminologists argue less educated and younger inmates are at greater risk for offending (Andrews & Bonta, 2010; Farrington, 1986), it may be worthwhile to explore new content to help the TCC intervention be more demographically appropriate. Alternative criminal thinking interventions are needed for this vulnerable subgroup of inmates.

Inmates who completed TCC did so because they wanted to, not because the programming was required. As such, TCC may function best for inmates who already recognize that their criminal lifestyle is no longer working for them and are ready to change – or to take a chance on change. Readiness to participate in TCC requires some degree of maturity; completion of TCC requires predictable support from mental health professionals who deliver the program. The weekly staff “check-ins,” structured by the TCC worksheets and the inmate's response to the material, provide important opportunities for validation of prosocial discourse and behavior. For some inmates in long-term restrictive housing, these interactions, combined with the psychoeducational material, have yielded significant benefit.

Conclusion

The TCC intervention is one of the first to tackle criminal thinking in one of the most challenging inmate populations: those in restrictive or segregated housing. By utilizing self-directed study at an accessible reading level, the intervention is uniquely suited to a correctional setting where staff and monetary resources are limited and security concerns often act as a barrier to treatment delivery. Although further research is needed to determine the effectiveness of TCC in reducing recidivism, the reductions in criminal thinking and disciplinary infractions evidenced in the current study suggest this is a promising intervention and mode of treatment delivery.

Acknowledgments

This research was supported in part by Grant #1F31DA039620 from the National Institute on Drug Abuse. Disclosure: Authors Sharen Barboza and John Wilson receive a salary from MHM Services, Inc. which created the Taking a Chance on Change program. The program is used exclusively within correctional systems where MHM Services, Inc. has been contracted to provide mental health services to inmates. There is no direct monetary arrangement with regard to the specific program. The Taking a Chance on Change program, along with many other treatment programs and tools, is available to all systems where MHM Services, Inc. provides care. Our clients can choose to have MHM staff deliver any of our programs or not. There is no additional cost or benefit to MHM Services, Inc. or the authors, for clients to use Taking a Chance on Change or any of our programs. While the authors do not believe this relationship creates a bias in regards to studying the effectiveness of the program, we felt obligated to share this information.

Footnotes

1

A Lexile analysis of the TCC was conducted using an online software program (Simpson, 2014). Content was analyzed according to five different grade level formulas: Flesch Kincaid Grade Level, Gunning Fog Score, SMOG Index, Coleman Liau, and ARI.

2

Some demographic information (range: 16% - 34%) was missing due to inmates failing to complete the demographic questionnaire or the demographic information being unavailable to researchers through institutional records.

3

The Sentimentality subscale is no longer included in any composite score because it failed to load onto a general criminal thinking factor in two large-scale studies (Walters, 2014; Walters, Hagman, & Cohn, 2011).

4

Because Cohen's d is systematically and upwardly biased, a correction is made to provide a more accurate effect size (i.e., Hedges g; Hedges, 1981).

5

FIML is widely accepted for longitudinal analyses because it decreases bias and increases statistical power compared with listwise deletion when data are missing at random or missing completely at random (Graham, 2009; Enders & Bandalos, 2001). Although confirmation that post-intervention missing data is missing at random or completely at random is not possible, including initial levels of criminal thinking and the demographic variables in each structural equation model is likely to account for a significant portion of the missing data mechanisms (Collins, Schafer, & Kam, 2001).

6

Auxiliary variables are variables included within a structural equation model to help account for the missing data mechanisms, in turn reducing bias in parameter estimation (Enders, 2008).

7

When the reactive criminal thinking outlier was included, the reactive criminal thinking regression coefficient was no longer significant (Beta = .13; 95%CI = [-.06, .31]). Regression diagnostics indicated the outlier had the largest influence on the regression line of any data point (i.e., Cook's distance = 0.38; DFbeta = -.97), potentially introducing coefficient bias.

8

For example, there were only 4 participants 20 years-old or younger and only 7 participants with an 8th grade education.

9

If all 19 hypothesis tests were completely independent the study-wide false positive rate would be 1 – (1 - .05)19 = .62 (Tabachnick & Fidell, 2001). When hypothesis tests are highly dependent, as in the present study, the rate is much lower. Unfortunately, an exact rate is never known due to the unknown dependency parameter.

References

  1. Andrews DA, Bonta J. The psychology of criminal conduct. 5th. New Providence, NJ: LexisNexis Matthew Bender: 2010. [Google Scholar]
  2. Batastini AB, Morgan RD, Kroner DG, Mills JF. Escaping the Cage: A mental health treatment program for inmates detained in restrictive housing. Unpublished treatment manual, Department of Psychology, Texas Tech University; Lubbock, Texas: 2014. [Google Scholar]
  3. Benjamini Y, Hochberg Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of Royal Statistics Society. 1995;57:289–290. [Google Scholar]
  4. Bickle J. An intermediate outcome evaluation for the thinking for a change program. Columbus, OH: Department of Rehabilitation and Correction; 2015. Retrieved from http://www.drc.ohio.gov/web/Reports/Eval_ThinkingforaChange.pdf. [Google Scholar]
  5. Blud L, Travers R, Nugent F, Thornton D. Accreditation of offending behaviour programmes in HM Prison Service: ‘What Works’ in practice. Legal and Criminological Psychology. 2003;8:69–81. [Google Scholar]
  6. Bush J, Glick B, Taymans J. Thinking for a change: Integrated cognitive behavior change program. Washington D.C: National Institute of Corrections, U.S. Department of Justice; 1997. revised 1998. [Google Scholar]
  7. Caudy MS, Folk JB, Stuewig JB, Wooditch A, Martinez A, Maass S, Taxman FS. Does substance misuse moderate the relationship between criminal thinking and recidivism? Journal of Criminal Justice. 2015;43:12–19. doi: 10.1016/j.jcrimjus.2014.11.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cohen J. Statistical power analysis for the behavioral sciences. San Diego, CA: Academic Press; 1988. [Google Scholar]
  9. Collins LM, Schafer JL, Kam CM. A comparison of inclusive and restrictive strategies in modern missing data procedures. Psychological Methods. 2001;6:330–351. [PubMed] [Google Scholar]
  10. Cook TD, Campbell DT. Quasi-experimentation: Design and analysis issues for field settings. Boston: Houghton Mifflin; 1979. [Google Scholar]
  11. Cousineau D, Chartier S. Outliers detection and treatment: a review. International Journal of Psychological Research. 2010;3:58–67. [Google Scholar]
  12. Cuadra LE, Jaffe AE, Thomas R, DiLillo D. Child maltreatment and adult criminal behavior: Does criminal thinking explain the association? Child Abuse and Neglect. 2014;38:1399–1408. doi: 10.1016/j.chiabu.2014.02.005. [DOI] [PubMed] [Google Scholar]
  13. Cunningham MD, Sorensen JR. Nothing to lose? A comparative examination of prison misconduct rates among life-without-parole and other long-term high-security inmates. Criminal Justice and Behavior. 2006;33:683–705. [Google Scholar]
  14. Davidson RJ, Putnam KM, Larson CL. Dysfunction in the neural circuitry of emotion regulation-a possible prelude to violence. Science. 2000;289:591–594. doi: 10.1126/science.289.5479.591. [DOI] [PubMed] [Google Scholar]
  15. Disabato DJ, Folk JB, Wilson J, Barboza S, Daylor J, Tangney J. Psychometric validation of a simplified form of the PICTS for low-reading level populations. Psychopathology and Behavioral Assessment. 2015 doi: 10.1007/s10862-015-9533-7. Advanced online publication. Retrieved December 22, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Enders CK. A note on the use of missing auxiliary variables in full information maximum likelihood-based structural equation models. Structural Equation Modeling. 2008;15:434–448. [Google Scholar]
  17. Enders CK, Bandalos DL. The relative performance of full information maximum likelihood estimation for missing data in structural equation models. Structural Equation Modeling. 2001;8:430–457. [PubMed] [Google Scholar]
  18. Farrington DP. Age and crime. Crime and Justice. 1986;7:189–250. [Google Scholar]
  19. Fitzmaurice G. A conundrum in the analysis of change. Nutrition. 2001;17:360–361. doi: 10.1016/s0899-9007(00)00593-1. [DOI] [PubMed] [Google Scholar]
  20. Folk JB, Blasko BL, Warden R, Schaefer K, Ferssizidis P, Stuewig J, Tangney JP. Feasibility and acceptability of an impact of crime group intervention with jail inmates. Victims & Offenders. 2015:1–19. doi: 10.1080/15564886.2014.982777. (ahead-of-print) [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Folk JB, Mashek DJ, Stuewig JB, Tangney JP, Moore KE, Blasko BL. Changes in jail inmates' community connectedness across the period of incarceration. 2015 doi: 10.1080/01639625.2018.1442652. Manuscript submitted for publication. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Gendreau P, Goggin CE, Law MA. Predicting prison misconducts. Criminal Justice and Behavior. 1997;24:414–431. [Google Scholar]
  23. Golden LS, Gatchel RJ, Cahill MA. Evaluating the effectiveness of the National Institute of Corrections' “Thinking for a Change” program among probationers. Journal of Offender Rehabilitation. 2006;43:55–73. [Google Scholar]
  24. Graham JW. Missing data analysis: Making it work in the real world. Annual Review of Psychology. 2009;60:549–576. doi: 10.1146/annurev.psych.58.110405.085530. [DOI] [PubMed] [Google Scholar]
  25. Hedges LV. Distribution theory for Glass's estimator of effect size and related estimators. Journal of Educational Statistics. 1981;6:107–128. [Google Scholar]
  26. Lowenkamp CT, Hubbard D, Makarios MD, Latessa ED. A quasi-experimental evaluation of Thinking for a Change: A “real-world” application. Criminal Justice and Behavior. 2009;36:137–146. [Google Scholar]
  27. Marsee MA, Frick PJ. Exploring the cognitive and emotional correlates to proactive and reactive aggression in a sample of detained girls. Journal of Abnormal Child Psychology. 2007;35:969–981. doi: 10.1007/s10802-007-9147-y. [DOI] [PubMed] [Google Scholar]
  28. Maruna S, Mann RE. A fundamental attribution error? Rethinking cognitive distortions. Legal and Criminological Psychology. 2006;11:155–77. [Google Scholar]
  29. Mears DP, Bales WD. Supermax incarceration and recidivism. Criminology. 2009;47:1131–1166. [Google Scholar]
  30. Metzner JL, Fellner J. Solitary confinement and mental illness in U.S. prisons: A challenge for medical ethics. Journal of the American Academy of Psychiatry and Law Online. 2010;38:104–108. [PubMed] [Google Scholar]
  31. Miller S, Seirup A. Managing and treating inmates in long-term segregation: Implementation of Escaping the Cage. Poster presented at the 3rd North American Correctional and Criminal Justice Psychology Conference; Ottawa, Canada. 2015. Jun, [Google Scholar]
  32. Mills JF, Kroner DG, Hemmati T. The measures of criminal attitudes and associates (MCAA): The prediction of general and violent Recidivism. Criminal Justice and Behavior. 2004;31:717–733. [Google Scholar]
  33. NIC (National Institute of Corrections) Cognitive interventions program: Think National Institute of Corrections Information Center. Washington D.C: U.S. Department of Justice; 1996. [Google Scholar]
  34. Orobio de Castro B, Merk W, Koops W, Veerman JW, Bosch JD. Emotions in social information processing and their relations with reactive and proactive aggression in referred aggressive boys. Journal of Clinical Child and Adolescent Psychology. 2005;34:105–116. doi: 10.1207/s15374424jccp3401_10. [DOI] [PubMed] [Google Scholar]
  35. Pattavina A, Taxman FS. Planning for the future of the U.S correctional system. In: Taxman FS, Pattavina A, editors. Simulation strategies to reduce recidivism. 3-20. New York, NY: Springer; 2013. [Google Scholar]
  36. Ragatz LL, Anderson RJ, Fremouw W, Schwartz R. Criminal thinking patterns, aggression styles, and the psychopathic traits of late high school bullies and bully-victims. Aggressive Behavior. 2011;37:145–160. doi: 10.1002/ab.20377. [DOI] [PubMed] [Google Scholar]
  37. Robinson D, Porporino FJ. Programming in cognitive skills: The reasoning and rehabilitation programme. In: Hollin CR, editor. Handbook of offender assessment and treatment. Chichester: Wiley; 2001. pp. 179–193. [Google Scholar]
  38. Ross RR, Fabiano EA. Time to think: A cognitive model of delinquency prevention and offender rehabilitation. Johnson City: Institute of Social Sciences and Arts; 1985. [Google Scholar]
  39. Rosseel Y. Lavaan: An R package for structural equation modeling. Journal of Statistical Software. 2012;48:1–36. [Google Scholar]
  40. Ruxton GD. The unequal variance t-test is an underused alternative to Student's t-test and the Mann–Whitney U test. Behavioral Ecology. 2006;17:688–690. [Google Scholar]
  41. Sadlier G. Evaluation of the impact of the HM Prison Service Enhanced Thinking Skills programme on reoffending outcome of the surveying prisoners crime reduction (SPCR) sample. Ministry of Justice Research Series 9/10. 2010 Retrieved from http://www.justice.gov.uk/downloads/publications/research-and-analysis/moj-research/eval-enhanced-thinking-skills-prog.pdf.
  42. Sánchez HG. Suicide prevention in administrative segregation units: What is missing? Journal of Correctional Health Care. 2013;19:93–100. doi: 10.1177/1078345812474638. [DOI] [PubMed] [Google Scholar]
  43. Shadish WR, Cook TD, Campbell DT. Experimental and quasi-experimental designs for generalized casual influence. Boston, MA: Houghton Mifflin; 2002. [Google Scholar]
  44. Shields IW, Simourd DJ. Predicting predatory behavior in a population of young offenders. Criminal Justice and Behavior. 1991;18:180–194. [Google Scholar]
  45. Simourd DJ. The criminal sentiments scale-modified and pride in delinquency scale psychometric properties and construct validity of two measures of criminal attitudes. Criminal Justice and Behavior. 1997;24:52–70. [Google Scholar]
  46. Simourd DJ. The criminal attitudes program: User's manual (CAP) Kingston, Ontario, Canada: Algonguin Correctional Evaluation Services (ACES); 2007. [Google Scholar]
  47. Simourd DJ, Olver ME, Brandenburg B. Changing Criminal Attitudes Among Incarcerated Offenders Initial Examination of a Structured Treatment Program. International Journal of Offender Therapy and Comparative Criminology. 2015:1–19. doi: 10.1177/0306624X15579257. (ahead-of-print) [DOI] [PubMed] [Google Scholar]
  48. Simpson D. The Readability Test Tool (Software) 2014 Available from http://read-able.com/
  49. Steinbuch AT. The movement away from solitary confinement in the United States. New England Journal of Criminal and Civil Confinement. 2014;40:499–533. [Google Scholar]
  50. Stephan JJ. Census of state and federal adult correctional facilities, 2005 US Department of Justice. Office of Justice Programs, Bureau of Justice Statistics; 2008. [Google Scholar]
  51. Tabachnick BG, Fidell LS. Using multivariate statistics. 5th. Boston, MA: Allyn & Bacon; 2001. [Google Scholar]
  52. Tangney J, Blasko B, Moore K, Stuewig J, Folk J, Boren E. Changes in criminogenic thinking across the period of incarceration: Normative changes and predictors of individual differences. 2015 Unpublished manuscript. [Google Scholar]
  53. Tangney JP, Stuewig J, Furukawa E, Kopelovich S, Meyer PJ, Cosby B. Reliability, validity, and predictive utility of the 25-item Criminogenic Cognitions Scale (CCS) Criminal Justice and Behavior. 2012;39:1340–1360. doi: 10.1177/0093854812451092. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Taxman FS, Belenko S. Implementing evidence-based practices in community corrections and addiction treatment. New York, NY: Springer; 2012. [Google Scholar]
  55. Toch H, Adams K. Disturbed-disruptive patterns. In: Toch H, Adams K, editors. Acting out: Maladaptive behavior in confinement. Washington, DC: American Psychological Association; 2002. pp. 99–120. [Google Scholar]
  56. Walters GD. The criminal lifestyle: Patterns of serious criminal conduct. Newbury Park, CA: Sage; 1990. [Google Scholar]
  57. Walters GD. The psychological inventory of criminal thinking styles part I: reliability and preliminary validity. Criminal Justice and Behavior. 1995;22:307–325. [Google Scholar]
  58. Walters GD. The Psychological Inventory of Criminal Thinking Styles: Part III. Predictive validity. International Journal of Offender Therapy and Comparative Criminology. 1996;40(2):105–112. [Google Scholar]
  59. Walters GD. Changes in criminal thinking and identity in novice and experienced inmates prisonization revisited. Criminal Justice and Behavior. 2003;30:399–421. [Google Scholar]
  60. Walters GD. Proactive and reactive aggression: A lifestyle view. In: Morgan JP, editor. Psychology of Aggression. Hauppauge, NY: Nova Science Publishers; 2005. pp. 29–43. [Google Scholar]
  61. Walters GD. Measuring proactive and reactive criminal thinking with the PICTS: Correlations with outcome expectancies and hostile attribution biases. Journal of Interpersonal Violence. 2007;22:371–385. doi: 10.1177/0886260506296988. [DOI] [PubMed] [Google Scholar]
  62. Walters GD. Self-report measures of psychopathy, antisocial personality, and criminal lifestyle: Testing and validating a two-dimensional model. Criminal Justice and Behavior. 2008;35(12):1459–1483. [Google Scholar]
  63. Walters GD. Criminal thinking and recidivism: Meta-analytic evidence on the predictive and incremental validity of the Psychological Inventory of Criminal Thinking Styles (PICTS) Aggression and Violent Behavior. 2012;17:272–278. [Google Scholar]
  64. Walters GD. An item response theory analysis of the Psychological Inventory of Criminal Thinking Styles: Comparing male and female probationers and prisoners. Psychological Assessment. 2014;26:1050–1055. doi: 10.1037/pas0000014. [DOI] [PubMed] [Google Scholar]
  65. Walters GD, Deming A, Casbon T. Predicting recidivism in sex offenders with the psychological inventory of criminal thinking styles (PICTS) Assessment. 2015;22:167–177. doi: 10.1177/1073191114539384. [DOI] [PubMed] [Google Scholar]
  66. Walters GD, Elliott WN. Predicting release and disciplinary outcome with the Psychological Inventory of Criminal Thinking Styles: Female data. Legal and Criminological Psychology. 1999;4(1):15–21. [Google Scholar]
  67. Walters GD, Geyer MD. Construct validity of the Psychological Inventory of Criminal Thinking Styles in relationship to the PAI, disciplinary adjustment, and program completion. Journal of Personality Assessment. 2005;84:252–260. doi: 10.1207/s15327752jpa8403_04. [DOI] [PubMed] [Google Scholar]
  68. Walters GD, Hagman BT, Cohn AM. Toward a hierarchical model of criminal thinking: Evidence from item response theory and confirmatory factor analysis. Psychological Assessment. 2011;23:925–936. doi: 10.1037/a0024017. [DOI] [PubMed] [Google Scholar]
  69. Walters GD, Schlauch C. The Psychological Inventory of Criminal Thinking Styles and Level of Service Inventory-Revised: Screening Version as predictors of official and self-reported disciplinary infractions. Law and Human Behavior. 2008;32(5):454. doi: 10.1007/s10979-007-9117-5. [DOI] [PubMed] [Google Scholar]
  70. Walters GD, Trgovac M, Rychlec M, DiFazio R, Olson JR. Assessing change with the Psychological Inventory of Criminal Thinking Styles: A controlled analysis and multisite cross-validation. Criminal Justice and Behavior. 2002;29:308–331. [Google Scholar]
  71. Wilkowski BM, Robinson MD. The cognitive basis of trait anger and reactive aggression: An integrative analysis. Personality and Social Psychology Review. 2008;12(3):21. doi: 10.1177/1088868307309874. [DOI] [PubMed] [Google Scholar]

RESOURCES