Can Completers,Non-Completers,and Non-Starters of Community-Based Offending Behavior Programs be Differentiated by Internal Treatment Readiness Factors?

Abstract

This research aimed to determine whether completers, non-completers, and non-starters of community-based offending behavior programs could be differentiated by their levels of internal “treatment readiness.” The Corrections Victoria Treatment Readiness Questionnaire (CVTRQ) measures offenders’ Attitudes and Motivation, Emotional Reactions, Offending Beliefs, and Efficacy which, according to the Multifactor Offender Readiness Model (MORM), are internal dimensions of an offender’s readiness to engage with treatment. Participants were offenders who had been court-mandated to attend a community-based cognitive skills offending behavior program. There were no significant differences between groups in respect of the CVTRQ total score. After controlling for risk of reconviction, however, the Self Efficacy construct differentiated program non-starters from program completers, while the Emotional Reactions construct differentiated program non-completers from program completers. In conclusion, the CVTRQ failed to differentiate program completion groups with the same success as elsewhere.

Keywords

offending behavior programs treatment readiness attrition program non-completion community corrections

Introduction

Within the National Probation Service of England and Wales nearly 8,000 offenders were required to attend an accredited offending behavior program as part of their sentence in 2016-2017 (Ministry of Justice, 2017). Evidence from a substantial body of research demonstrates the effectiveness of completion of community-based cognitive skills programs on reconviction rates (Hatcher, McGuire, Bilby, Palmer, & Hollin, 2012; Hollin et al., 2008; van Voorhis, Spruance, Ritchey, Listwan, & Seabrook, 2004); however, the literature has also highlighted high levels of attrition from such programs (Hatcher et al., 2012; Hollin et al., 2008; Hollis, 2007; Palmer et al., 2007; van Voorhis et al., 2004) and the associated poor outcomes of those who do not complete (Hatcher et al., 2008; McMurran & Theodosi, 2007; Palmer et al., 2007).

Attrition, defined as the failure to commence and/or complete the program to which an individual is referred, is not a new problem. In the United Kingdom, Cann, Falshaw, Nugent and Friendship (2003) reported a total rate of cognitive skills program non-completion of 11% across their prison sample, while in community settings, Hollin et al. (2008) reported that nearly half of those sentenced to a program failed to commence and a further quarter commenced but failed to complete. Olver, Stockdale and Wormith’s (2011) meta-analysis of attrition reported an overall attrition rate (including preprogram attrition) of 35.8%, with in-treatment dropout at 27.1%. These figures rose, however, when considering community programs only (39.3% and 31.5% respectively).

In a clinical sense, the concern regarding attrition from programs is obvious; those individuals who meet the criteria, and hence are identified as having a need for intervention, fail to benefit from the intervention. Considered in this context, attrition thus results in a violation of the risk and need principles (Andrews & Bonta, 2010) and hence maximal treatment gains cannot be obtained. However, the picture may not be this simplistic. McMurran and Theodosi’s (2007) meta-analysis concluded that while completion of programs was seen to be effective in reducing reoffending (d = 0.11), non-completion was found to be associated with elevated levels of reoffending (d = −0.16) with community samples performing worse than prison samples (d = −0.23 and −0.15 respectively). Similar results were reported by Hollin et al. (2008) in their large-scale study of community-based general offending behavior programs. Program non-completers were twice as likely and non-starters more than twice as likely as the no treatment comparison group to be reconvicted, even after controlling for any differences between the groups in terms of age, risk of reconviction, offense type, number of previous convictions, and follow-up time. Comparable results have also been reported in North American evaluative research (e.g., van Voorhis et al., 2004) and when utilizing matched comparison groups rather than statistical control (e.g., Hatcher et al., 2008).

Empirical evidence for the reasons for the “non-completion effect” is currently lacking and warrants further investigation. One possible explanation is that iatrogenic effects result from program dropout which may act to increase the risk of further offending. Such effects could include resulting emotional distress (either from the process of dropout itself or from the interruption to a therapeutic process at a disadvantageous point) or increases in antiauthority and antisocial attitudes (McMurran & Theodosi, 2007; Olver, Stockdale, & Wormith, 2011). Any impact seen here is, of course, likely to depend on the reason(s) why the individual failed to complete the program. Further research is necessary to unpick such reasons and their potential impact on subsequent outcomes. An alternative explanation of the non-completion effect is that the existence of unmeasured pretreatment differences between the non-completer and comparison groups influence both program attendance and reconviction outcomes. One such factor may be how ready the individual is to engage with, and hence benefit from, the intervention offered.

Proposed by Howells and colleagues (e.g., Howells & Day, 2003; Ward, Day, Howells, & Birgden, 2004), treatment readiness is a concept which the authors claim is crucial to program engagement, and presumably therefore, attendance. Treatment readiness is defined as “the presence of characteristics (states or dispositions) within either the client or the therapeutic situation, which are likely to promote engagement in therapy and that, thereby, are likely to enhance therapeutic change . . .” (Ward et al., 2004, p. 647). As such, the concept is broadly conceptualized; treatment readiness is concerned with the interaction between the offender, the intervention, and the organizational or contextual factors relating to the offenders and the intervention.

The concept of treatment readiness has been incorporated into the Multifactor Offender Readiness Model (MORM) by Ward et al. (2004). The MORM highlights dimensions relating to the offender—cognitive, behavioral, volitional, and personal/social identity factors—in addition to the context—circumstances, location, opportunities, resources, support, and program/timing factors—which combine to produce the level of treatment readiness. The authors argue that “it is likely that offenders with low readiness across multiple areas will be those that … do not complete treatment” (Ward et al., 2004, p. 668). This article seeks to test this argument through the evaluation of the preprogram treatment readiness of community-based program non-starters, non-completers, and completers.

To assess those offenders who may be lacking in treatment readiness, Casey, Day, Howells, and Ward (2007) have developed the Corrections Victoria Treatment Readiness Questionnaire (CVTRQ) which is based on the MORM and measures four internal components of treatment readiness: Attitudes and Motivation, Emotional Reactions (to their own offending behavior), Offending Beliefs, and Efficacy (or their perceived ability to engage). As the tool is limited to the assessment of internal offender characteristics, the authors’ state that it

does not seek to evaluate the MORM but rather those aspects of the model that might be reliably assessed . . . to inform professional assessments of readiness in a way that is both cost- and time-effective and might be particularly valuable in the routine assessment of suitability for programs. (Casey et al., 2007, p. 1428)

The developers state that the CVTRQ displays high levels of discriminant and convergent validity and has been found to predict offender treatment engagement and performance, giving evidence of predictive validity. As such, the CVTRQ has been endorsed by Casey et al. as a tool that can “play a valuable role in the assessment of offenders who are being considered for rehabilitative treatment” (p. 1427). Such assessment prior to treatment aims to improve the selection process through the reduction of inappropriate referrals and inform the allocation of resources to enhance low treatment readiness prior to program commencement (Burrows & Needs, 2009; Casey et al., 2005).

The aim of this article therefore is to evaluate the preprogram treatment readiness of subsequent program non-starters, non-completers, and completers of a community-based program attrition within England and Wales. The research aims to determine whether those who fail to start or dropout from their allocated program could have been identified, and hence targeted for work to enhance their treatment readiness, at the preprogram stage through their CVTRQ scores. It is hypothesized that (a) there will be significant differences between the completion groups in their levels of internal treatment readiness measured by the CVTRQ; completers will score higher than both non-completers and non-starters, and (b) it will be possible to predict program completion group (completer, non-completer, non-starter) from their CVTRQ internal readiness factors.

Method

Design

This study aims to investigate the treatment readiness of program completers, non-completers, and non-starters of community-based court-sanctioned offending behavior programs. As such, data were collected from offenders prior to the commencement of the program to which they had been referred to by the courts. Information on the offenders’ risk of reconviction was also collected and statistically controlled for where appropriate.

Participants

All participants were offenders sentenced by the courts to a community order which contained a requirement to attend one of seven community-based offending behavior programs delivered within one Probation Area in England and Wales. As such, all participants were deemed by the Probation Service to meet the entry criteria for the program to which they were sentenced. The programs included in the study were the Controlling Anger and Learning to Manage it program (CALM), the Drink Impaired Drivers program (DID), the Enhanced Thinking Skills program (ETS), the Women’s Acquisitive Crime program (WAP), the Offender Substance Use program (OSAP), the Cognitive Skills Booster program (CSB), and the Low Intensity Alcohol Module (LIAM). All programs have been accredited by the Correctional Services Accreditation Panel and hence conform to the principles of effective program design (McGuire, 2004).

Data were collected from 68 offenders convicted of a diverse set of offenses. These were categorized using the England and Wales Home Office offense category coding scheme (see Smith & Hoare, 2009) into violence against the person (42.6%), robbery (1.5%), burglary (1.5%), offenses against vehicles (2.9%), other theft offenses (10.3%), fraud and forgery (2.9%), criminal damage (1.5%), drug offenses (2.9%), and other miscellaneous offenses (33.8%). The mean age was 28.07 years (standard deviation [SD] = 7.77) with a range from 19 to 55 and their gender was predominantly male (67 male, 1 female). Participants had mean OGRS-3 risk of reconviction scores (Howard, Francis, Soothill, & Humphreys, 2009) of 62.78 (SD = 19.56) with a range from 11 to 92 (see below for more information on OGRS-3).

Of the 68 participants, 31 (45.6%) completed the program, 16 (23.5%) commenced but did not complete the program, and 21 (30.9%) failed to commence the program to which they were sentenced.

Materials

Corrections Victoria Treatment Readiness Questionnaire (Casey et al., 2007): The CVTRQ measures the internal characteristics of treatment readiness and has been developed in line with the Multifactor Offender Readiness Model (Ward et al., 2004) which proposes the contribution of both internal and external factors to treatment engagement.

The questionnaire comprises 20 questions which measure the four constructs of Attitudes and Motivation (6 items, α = .84), Emotional Reactions (6 items, α = .79), Offending beliefs (4 items, α = .73) and Efficacy (4 items, α = .60) (Casey et al., 2007). Attitudes and Motivation refers to the offender’s attitudes and beliefs about programs and a desire to change (e.g., “treatment programs are rubbish” and “I want to change”) while the Emotional Reactions variable measures emotional responses to their own offending behavior (e.g., “when I think about my last offence I feel angry with myself” and “I feel guilty about my offending”). Offending beliefs measures the individual’s beliefs about personal responsibility for offending (e.g., “I am to blame for my offending” and “when I think about my sentence I feel angry with other people”) and Efficacy refers to the individual’s perceived ability to participate in programs (e.g., “I am well organized”).

Self-reported responses are made on a 5-point Likert-type scale which ranges from 1 (strongly disagree) to 5 (strongly agree). The maximum score for the constructs of Attitudes and Motivation and Emotional Reactions is 30 and the minimum score is 6. The Offending Beliefs and Efficacy have a maximum score of 20 and a minimum score of 4. A total internal readiness score is generated by summing the four construct scores. High scores on the total and subscale scores represent high treatment readiness. Casey et al. (2007) have proposed, based on receiver operating characteristic (ROC) curve analysis, that total scores of 72 or over may indicate a readiness for treatment (see Casey et al for more information).

Offender Group Reconviction Score-3 (OGRS-3; Howard et al., 2009): is a predictor of reconviction which is calculated utilizing six static factors relating to age, gender, and criminal history. The tool provides a risk of reconviction score ranging from 0 to 100. The higher the score obtained on the measure, the higher the predicted risk of reconviction. The OGRS-3 tool has been shown to be an effective predictor of proven reoffending; it has an AUC of 80% for all offenders and 84% for prisoners only (Howard et al., 2009).

Case Recording and Management System (CRAMS): The Probation Area’s CRAMS database was used to retrieve file information held on the participants. The data collected from this system were the participants’ date of birth (from which their age was calculated), their OGRS-3 score, the program to which they had been sentenced, and their index offense.

Procedure

Permission to undertake the research was gained from the Probation Area participating in the study and ethical approval was received from the University of Leicester’s Psychology Research Ethics Committee.

All participants who attended their routine Program Induction Meeting with a member of the Probation Area’s Programs Team during the study period were invited to take part in the research. The Induction Meeting occurs after sentencing but prior to the offender being referred to a particular delivery of the program. At the meeting, informed consent was sought from potential participants. It was made clear that participation was not a compulsory part of their probation order and taking part, or indeed declining to take part, would not hold any consequences for progression on their sentence. The consent form outlined the confidential nature of the research and explained that their CVTRQ questionnaire data would be anonymised once it was matched with their file data (see below) and hard copies of the questionnaires would be destroyed. If any of the participants had literacy problems, the questionnaire was read to them by the Program Team member of staff. Once the CVTRQ had been completed, participants were debriefed and provided with a debrief sheet which outlined the research aims and provided contact details of appropriate support should it be required.

The additional data were collected from the CRAMS database. The data retrieved from this database are described above. At the end of the study period, CRAMS was used to identify whether the offender had completed or dropped out of the program. Those offenders who had failed to attend any sessions of the program were categorized as a “non-starter,” those who commenced but failed to complete the program were categorized as a “non-completer,” and those who completed the program were categorized as a “completer.”

Analysis Strategy

Initially, the program completers, non-completers, and non-starters were compared in respect of their age, risk of reconviction, and index offense. Previous research has indicated that program completers are likely to be younger (Hollin et al., 2008; Nunes & Cortoni, 2006a; Palmer et al., 2008; Palmer et al., 2009; Robinson, 1995; Turner, 2006; van Voorhis et al., 2004; Zanis et al., 2003) and lower risk than program dropouts (Browne, Foreman, & Middleton, 1998; Craissati & Beech, 2001; Hollin et al., 2008; Nunes & Cortoni, 2006b; Palmer et al., 2008, 2009; Turner, 2006; Wormith & Olver, 2002). One-way analyses of variance were undertaken to determine whether these findings were replicated within this sample.

The analysis then used a one-way ANOVA and a MANOVA to assess whether the program completion groups differed in respect of the CVTRQ total and construct scores respectively, before progressing to determine whether CVTRQ total and construct scores were predictive of program completion/attrition. Given that the aim was to ascertain whether CVTRQ scores add anything to the prediction of attrition, over and above the information already used by services to select offenders for programs, where differences between the groups were identified in respect of age, risk, or index offense, these variables were entered into the regression model to control for the influence that they have on the outcome. Two multinomial logistic regressions were undertaken with the completion status variable (completer, non-completer, non-starter) comprising the outcome variable; the first utilized the CVTRQ total score and the second the CVTRQ construct scores as predictors. The total score and construct scores were not entered into the same model due to the assumption of independence.

Results

Univariate Analyses

Table 1 presents the mean ages and OGRS3 scores by group (completer, non-completer, non-starter). Kolmogorov–Smirnov tests indicated no violations of the normality assumption and the Levene’s test also indicated no violations of the assumption of homogeneity of variances for either age or OGRS3.

Table 1.

Age and OGRS-3 Scores by Group (Completer, Non-Completer, Non-Starter).

	Completers (N = 31)	Non-completers (N = 16)	Non-starters (N = 21)
Age	29.48 (8.18)	26.19 (8.80)	27.43 (6.13)
OGRS-3 score	55.58 (21.15)	65.44 (16.88)	71.38 (15.29)

Note. Values are represented as M (SD). OGRS-3 = Offender Group Reconviction Score-3.

As such, two one-way ANOVAs were undertaken. No significant differences were observed between the groups in respect of their ages (F [2, 65] = 1.056, p = .354, η² = 0.03). However, the groups did differ in respect of their OGRS3 risk of reconviction scores (F [2, 65] = 4.757, p = .012, η² = 0.12). Post hoc Scheffe tests indicated significant differences between the non-starters and completers (p = .014) with non-starters being at a higher risk of reconviction than the completers.

Analysis of CVTRQ

The internal consistency of the CVTRQ with the current sample was found to be good (Cronbach’s alpha: α = 0.821). The intercorrelations between the total and construct scores are presented in Table 2. As expected, the constructs are all positively and significantly correlated.

Table 2.

CVTRQ Total and Construct Intercorrelations.

	Total CVTRQ	Attitudes and motivation	Emotional reactions	Offending beliefs	Efficacy
Total CVTRQ	—	.795**	.784**	.695**	.584**
Attitudes and Motivation		—	.450**	.521**	.350**
Emotional Reactions			—	.327**	.272*
Offending Beliefs				—	.302*
Efficacy					—

Note. CVTRQ = Corrections Victoria Treatment Readiness Questionnaire.

p < .05. **p < .01.

Table 3 presents the mean CVTRQ scores by completer, non-completer, and non-starter groups. Again, Kolmogorov–Smirnov tests indicated that there were no violations of the normality assumption and the homogeneity of variance assumption was also met for each of the variables. As such, a one-way ANOVA was undertaken with the CVTRQ total score as the dependent variable in addition to a MANOVA with the construct scores comprising the dependent variables. In both analyses the independent variable was “completion group” (completer, non-completer, non-starter). In respect of the MANOVA, the assumption of homoscedasticity was met and hence the MANOVA was progressed.

Table 3.

CVTRQ Total and Construct Scores by Group (Completers, Non-Completers, and Non-Starters).

	Completers	Non-completers	Non-starters
Total score	77.50 (11.78)	74.40 (8.92)	77.00 (14.27)
Attitudes and Motivation	24.66 (3.43)	22.94 (3.04)	25.42 (4.80)
Emotional Reactions	21.84 (6.51)	23.20 (4.40)	22.17 (4.94)
Offending Beliefs	15.93 (3.42)	14.31 (3.20)	16.25 (3.23)
Efficacy	15.00 (3.19)	13.63 (3.12)	12.90 (2.45)

Note. Values are represented as M (SD). CVTRQ = Corrections Victoria Treatment Readiness Questionnaire.

The ANOVA incorporating the CVTRQ total score did not detect any significant differences between the groups: (F [2, 55] = 0.338, p = .715, η² = 0.01). However, the MANOVA returned a significant difference between the groups in respect of the multivariate combination of the four construct variables (F [8, 104] = 2.249, p = .029, η² = 0.15) using Wilks’ Lambda. Assessment of the univariate analyses determined that the only construct score to differ significantly between the groups was the Efficacy variable: (F [2, 55] = 3.624, p = .033, η² = 0.12. Post hoc Scheffe tests indicated significant differences between the non-starters’ and completers’ Efficacy scores with non-starters scoring lower on this construct than the completers. All other univariate differences were nonsignificant: Attitudes and Motivation (F [2, 55] = 1.539, p = .224, η² = 0.05, Emotional Reactions (F [2, 55] = 0.430, p = .653, η² = 0.02) and Offending Beliefs (F [2, 55] = 0.190, p = .190, η² = 0.06).

Multinomial Logistic Regression

Two multinomial logistic regression analyses were performed to determine whether it was possible to predict completion group from the CVTRQ variables. The first analysis was undertaken with the CVTRQ total score while the second utilized the CVTRQ construct scores. Within both analyses the OGRS3 risk of reconviction variable was entered into the model to control for differences observed between the completion groups.

The analysis containing the OGRS3 and CVTRQ total score variables produced a model which represented a good fit of the data, χ² (110, N = 68) = 116.183, p = .325, using a deviance criterion. Comparisons of the log-likelihood ratios for models with and without predictors, however, showed a nonsignificant improvement with the addition of the predictors, χ² (4, N = 68) = 7.836, p = .098. Correct classification rates were 84.6% for the Completers, 6.7% for the Non-completers condition, and 35.3% for the Non-starters condition. The overall correct classification was 50%, which represents an increase of 16.67% over by-chance accuracy. Analysis of the contribution of individual predictors to the model with and without each predictor revealed that the OGRS3 variable significantly predicted the outcome, χ² (2, N = 68) = 7.144, p = .028, but the CVTRQ variable did not, χ² (2, N = 68) = 0.603, p = .740. Table 4 displays the odds ratios and beta values.

Table 4.

Multinomial Logistic Regression of Program Completion as a Function of OGRS3, and CVTRQ Total Score.

	B	SE of B	Wald	Exp (B)	95% CI for Exp (B)
Non-starters vs. Completers (ref)
Intercept	−4.122	2.765	2.222
OGRS3	0.052	0.022	5.682*	1.053	[1.009, 1.099]
CVTRQ Total	0.005	0.028	0.026	1.005	[0.950, 1.062]
Non-completers vs. Completers (ref)
Intercept	−0.585	2.644	0.049
OGRS3	0.022	0.019	1.338	1.023	[0.985, 1.062]
CVTRQ Total	−0.017	0.029	0.365	0.983	[0.929, 1.062]
Non-completers vs. Non-starters (ref)
Intercept	3.537	2.940	1.448
OGRS3	−0.030	0.024	1.586	0.971	[0.927, 1.017]
CVTRQ Total	−0.022	0.030	0.537	0.978	[0.923, 1.037]

Note. OGRS-3 = Offender Group Reconviction Score-3; CVTRQ = Corrections Victoria Treatment Readiness Questionnaire; CI = confidence interval.

p < .05.

The second multinomial logistic regression utilized the CVTRQ construct scores as predictors alongside the OGRS3 control variable to predict completion group (completer, non-completer, non-starter). This analysis also produced a model which represented a good fit of the data, χ² (104, N = 68) = 99.576, p = .604, using a deviance criterion. Comparisons of the log-likelihood ratios for models with and without predictors showed a reliable improvement with the addition of the predictors, χ² (10, N = 68) = 24.444, p = .007. Correct classification rates were 73.1% for the Completers, 40.0% for the Non-completers condition, and 58.8% for the Non-starters condition. The overall correct classification was 60.3%, which represents an increase of 26.97% over by-chance accuracy. Analysis of the contribution of individual predictors to the model with and without each predictor revealed that the OGRS3, χ² (2, N = 68) = 6.121, p = .047, and Efficacy, χ² (2, N = 68) = 7.312, p = .026, variables significantly predicted the outcome. The other variables entered into the model did not significantly predict the outcome: Attitudes and Motivation (2, N = 68) = 3.801, p = .150), Emotional Reactions (2, N = 68) = 5.313, p = .070), Offending Beliefs (2, N = 68) = 2.021, p = .364). However, the Emotional Reactions variable did significantly differentiate the Completers from the Non-completers. Table 5 presents the odds ratios and beta values.

Table 5.

Multinomial Logistic Regression of Programme Completion as a Function of OGRS3, and CVTRQ Construct Scores.

	B	SE of B	Wald	Exp (B)	95% CI for Exp (B)
Non-starters vs. Completers (ref)
Intercept	−4.052	3.172	1.632
OGRS3	0.052	0.024	4.572*	1.054	[1.004, 1.105]
Attitudes and Motivation	0.183	0.122	2.271	1.201	[0.946, 1.525]
Emotional Reactions	0.031	0.075	0.169	1.031	[0.890, 1.195]
Offending Beliefs	0.000	0.127	0.000	1.000	[0.780, 1.282]
Efficacy	−0.353	0.143	6.119*	0.702	[0.531, 0.929]
Non-completers vs. Completers (ref)
Intercept	−0.240	3.036	0.006
OGRS3	0.032	0.220	2.106	1.032	[0.989, 1.077]
Attitudes and Motivation	−0.047	0.114	0.174	0.954	[0.763, 1.192]
Emotional Reactions	0.175	0.085	4.205*	1.191	[1.008, 1.408]
Offending Beliefs	−0.169	0.123	1.873	0.845	[0.663, 1.076]
Efficacy	−0.167	0.136	1.516	0.846	[0.649, 1.104]
Non-completers vs. Non-starters (ref)
Intercept	3.812	3.247	1.378
OGRS3	−0.20	0.025	0.656	0.980	[0.932, 1.029]
Attitudes and Motivation	−0.231	0.130	3.142	0.794	[0.615, 1.025]
Emotional Reactions	0.144	0.090	2.573	1.155	[0.969, 1.377]
Offending Beliefs	−0.169	0.145	1.365	0.844	[0.636, 1.122]
Efficacy	0.186	0.145	1.644	1.205	[0.906, 1.602]

Note. OGRS-3 = Offender Group Reconviction Score-3; CVTRQ = Corrections Victoria Treatment Readiness Questionnaire; CI = confidence interval.

p < .05. ****p < .001.

Analysis of the odds ratios indicates that with each unit increase in Efficacy scores the odds of being in the non-starter group compared to the completer group decrease by 30% (p = .013). In addition, with each unit increase in Emotional Reaction scores the odds of being in the non-completers group compared to the completer group increase by 16% (p = .040).

Discussion

This study builds on the research of Casey et al. (2007) who assessed the utility of the Corrections Victoria Treatment Readiness Questionnaire in predicting offender treatment engagement and performance. This research aimed to determine whether the CVTRQ can be utilized to predict which offenders would fail to complete their court-sanctioned community-based cognitive skills program.

The sample, taken from one probation area in England and Wales covering seven cognitive skills programs, consisted of 45.6% program completers, 23.5% program non-completers, and 30.9% program non-starters. This rate of program dropout is similar to the community-based figures reported by Hatcher from 2005/2006 (2009: 43%, 15.4%, and 41.6% respectively) and represents an improvement on those reported by Hollin et al. from 2001/2002 (2008: 28.2%, 23.3, and 48.5% respectively). It is likely that improvements in program implementation and offender engagement practices over time are responsible for these changes (although these are merely hypotheses). When compared to the meta-analytical findings from community samples (Olver et al., 2011), the current sample featured lower rates of in-program attrition (23.5% vs. 31.5%) but higher rates of preprogram dropout (30.9% vs. 7.8%). Whether this disparity is due to sample variation (e.g., the meta-analysis included offenders referred to sex offender programs which are not represented within the present article) or procedural/offender processing differences would be interesting to explore in subsequent research.

In line with the extant literature (e.g., Hatcher, 2009; Hatcher et al., 2012; McMurran & Theodosi, 2007) program non-starters presented with higher risk of reconviction scores than program completers (although no such difference was seen between completers and non-completers). This finding presents an ongoing conundrum for correctional service providers; the offenders who, according to the principles of risk and need (Andrews & Bonta, 2010), display the greatest necessity for programming and require the largest doses of intervention are most difficult to engage and are most likely to fail to commence interventions. There are indications within the literature, however, that if these individuals are successful in completing their interventions then the gains are significant (Hatcher, 2009; Palmer et al., 2008, 2009). More investigation is required to determine how services can increase levels of intervention completion among high risk offenders.

In assessing the utility of the CVTRQ scores in predicting attrition, the scale’s total score did not fare well: the three groups’ total scores were remarkably similar (ranging from 74.4 to 77.5) and, even after controlling for differences in risk, this variable did not predict group membership (completer, non-completer, non-starter). Such a finding, in isolation at least, would suggest that internal treatment readiness factors (as measured by the CVTRQ) are not associated with program attrition. Further examination of the findings suggests, however, that such a picture is not completely accurate; differences between the groups emerged on examination of the individual treatment readiness construct scores. The largest difference—that relating to the Efficacy variable—provided evidence that program completers had more preprogram confidence in their ability to participate in the program than those who subsequently failed to commence the program. The factors that may influence offenders’ perceptions of efficacy in relation to program completion are not clear from this research and would benefit from further investigation. It could be possible, however, that services could increase completion rates through the removal of obstacles from the paths of those offenders whose report low efficacy in relation to program participation.

An alternative interpretation could be drawn in relation to this finding, however. The CVTRQ was completed by all offenders on their induction to the Programs Unit. It is possible that some offenders, having received their sentence and spoken within their Offender Manager (and others) about the program to which they have been sentenced, decided prior to completion of the CVTRQ that they did not wish to participate in the available intervention. As such, rather than reflecting a reduced sense of self-confidence, the non-starters’ Efficacy scores could merely reflect their decision that they would not be engaging with the program. While the distinction between these two explanations may be subtle, the practical response required from practitioners should vary: while the former may indicate the necessity for self-efficacy enhancing work to build confidence and encourage the offenders’ beliefs in their own abilities, the latter would require work which stresses the potential benefits of the program.

Inspection of the scores obtained on the Attitude and Motivation variable may aid in decision making here; those offenders who do not see the potential benefits of engaging with an intervention, and have therefore decided at the preprogram stage that they will not participate, are likely to score lower on this variable. Contrary to the alternative position, however, the non-starters did not return significantly lower scores to the program non-completers or completers on this variable, perhaps indicating that the issue here is not one of conscious disengagement or negative perceptions of the benefits of the program but is instead low self-efficacy on the part of the offender. Such low confidence in their own ability to engage with the intervention may stem from previous negative experiences in similar, such as educational or vocational, environments; indeed, a qualitative investigation of offenders’ preprogram expectations revealed a concern that the program environment would resemble previous educational experiences and that this would be a barrier to engagement with the program (Hatcher, in preparation).

Whereas failure to start the program seems to be related to the self-efficacy construct, there is some indication that failure to complete the program may be linked to the offenders’ management of emotional responses relating to their own criminal histories. The Emotional Reactions variable differentiated the program completers and non-completers (after controlling for differences in risk levels). Non-completers reported feeling more ashamed, guilty, angry, and upset about their offending histories and their status as an offender than program completers. Given that most in-program dropout occurs within the early stages of the program (Hatcher, McGuire, Palmer, & Hollin, 2011), this finding would suggest the need for preprogram work focusing on the offenders’ cognitive appraisals of their offending behavior prior to their referral to their allocated cognitive skills program. Such work would benefit from monitoring and evaluation to determine its impact on reducing within-treatment attrition.

The aim of this article was to determine whether it was possible to use the CVTRQ to differentiate program completers from program dropouts from community-based offending behavior programs. While interesting observations have been made concerning two of the construct variables (Efficacy and Emotional Reactions) and their associations with preprogram and within-program dropout, the CVTRQ failed to predict program attendance/dropout with the same success reported when predicting treatment engagement (Casey et al., 2007). In evaluating these findings, the small sample size and resulting lack of power mean that any conclusions should be considered tentative, however, one possible interpretation of the findings is that the scale does not generalize from its original sample to the present one. It is also possible that the CVTRQ predicts engagement with the program contents (i.e., identifying those who will gain most from their participation in the program) but is less useful in identifying those at risk of failing to commence of completion interventions.

Another explanation is that the focus of the CVTRQ is too narrow to provide a useful measure of treatment readiness. The MORM is clear that treatment readiness is a function of dimensions relating both to the offender (cognitive, behavioral, volitional, and personal/social identity factors) and to the context (circumstances, location, opportunities, resources, support, and program/timing factors) and research into the offender perspectives of treatment engagement support this contention (Sturgess, Woodhams, & Tonkin, 2016). Unfortunately, the CVTRQ neglects to measure contextual factors and, despite professing to measure the internal component of readiness, the contribution of volitional and personal/social identity factors are also missing from the tool. As such, the extent to which this tool measures the concept of treatment readiness, or even internal treatment readiness, as outlined by the MORM, must be questioned.

Research assessing the role of contextual or process factors in program engagement and attrition certainly requires further attention. Research has shown that the inappropriate allocation of individuals to programs, determined through an incongruence between program selection criteria and the individuals’ risk profiles, impacts on both attrition and reconviction outcomes (e.g., Hatcher, 2009; Hatcher et al., 2012, Palmer et al., 2008, 2009). Furthermore, there are indications that factors such as waiting time from court order to program commencement (Turner, 2006), levels of staff support and encouragement (Mason & Adler, 2012), and preprogram motivational work (e.g., McMurran, 2009) also influence attrition rates from correctional treatment programs. In trying to reduce attrition rates, it is thus incumbent on service providers to not only consider internal readiness factors, but also how organizational and process factors might, either directly or through interaction with individual factors, work to undermine treatment readiness.

The proponents of the MORM are clear in their opinion as to how intervention services should use information relating to the prediction of treatment readiness. Rather than use indications of treatment readiness as selection tools for deciding who should and should not receive treatment, Day and colleagues, conceptualize treatment readiness as adaptable and hence the responsibility of the service provider to manipulate:

the primary responsibility for modifying low levels of readiness lies with the practitioner and the service in which s/he works. In other words, it becomes the treatment provider’s duty to find ways to deliver interventions that meet the individual’s needs, and discourages the tendency to view low levels of motivation or compliance as pathological. (Day, Howells, Casey, Ward, & Birgden, 2007, p. 22)

Hence, when working with offenders who are resistant to treatment, the authors have proposed that the indications of treatment readiness provided by tools such as the CVTRQ should “serve as . . . heuristic model(s) for clinical decision making” (Ward et al., 2004, pp. 665-666). These decisions may involve modifying the client, the treatment program, or the setting to increase readiness and hence offender engagement (Beyko & Wong, 2005). This perspective, of course, concurs with the responsivity component of the Risk-Needs-Responsivity model (Andrews & Bonta, 2010; Andrews, Bonta, & Wormith, 2011) which suggests that successful interventions are those which match the style and mode of intervention to the offenders’ abilities and learning styles.

While the authors of this article concur wholeheartedly with this view, such tools need to demonstrate their utility with respect to predicting treatment engagement/readiness/completion before they can be used to guide the allocation of resources in the manner described. At present, it cannot be concluded that this is the case with respect to the CVTRQ. While Casey and colleagues have provided evidence of the utility of this tool within one context, this article concludes that the CVTRQ has not usefully differentiated program completers, non-completers, and non-starters of a community-based cognitive skills program for offenders.

Footnotes

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.

ORCID iD

Ruth M. Hatcher

References

Andrews

D. A.

Bonta

(2010). The psychology of criminal conduct (5th ed.). Cincinnati, OH: Anderson Publishing.

Andrews

D. A.

Bonta

Wormith

J. S.

(2011). The Risk-Need-Responsivity (RNR) model: Does adding the Good Lives Model contribute to effective crime prevention? Criminal Justice and Behavior, 38, 735-755.

Beyko

M. J.

Wong

S. C. P.

(2005). Predictors of treatment attrition as indicators for program improvement not offender shortcomings: A study of sex offender treatment attrition. Sexual Abuse: A Journal of Research and Treatment, 17, 375-389.

Browne

K. D.

Foreman

Middleton

(1998). Predicting treatment drop-out in sex offenders. Child Abuse Review, 7, 402-419.

Burrows

Needs

(2009). Time to contemplate change? A framework for assessing readness to change in offenders. Aggression and Violent Behaviour: A Review Journal, 14, 39-49.

Cann

Falshaw

Nugent

Friendship

(2003). Understanding what works: Accredited cognitive skills programmes for adult men and young offenders (Home Office Research Findings No. 226). London, England: Home Office.

Casey

Day

Howells

(2005). The application of the transtheoretical model to offender populations: Some critical issues. Legal and Criminological Psychology, 10, 157-171.

Casey

Day

Howells

Ward

(2007). Assessing suitability for offender rehabilitation: Development and validation of the Treatment Readiness Questionnaire. Criminal Justice and Behaviour, 34, 1427-1440.

Craissati

Beech

(2001). Attrition in a community treatment programme for child sexual abusers. Journal of Interpersonal Violence, 16, 205-221.

10.

Day

Howells

Casey

Ward

Birgden

(2007). Treatment readiness: An overview of Australasian work. Issues in Forensic Psychology, 7, 21-25.

11.

Hatcher

R. M.

(2009). An investigation of attrition from community-based offending behaviour programmes (Doctoral thesis, University of Liverpool, UK). Retrieved from http://research-archive.liv.ac.uk/1220/

12.

Hatcher

R. M.

McGuire

Bilby

C. A. L.

Palmer

E. J.

Hollin

C. R.

(2012). Methodological considerations in the evaluation of offender interventions: The problem of attrition. International Journal of Offender Therapy and Comparative Criminology, 56, 447-464.

13.

Hatcher

R. M.

McGuire

Palmer

E. J.

Hollin

C. R.

(2011). Dosage, appropriateness of selection, and reconviction amongst completers and non-completers of community based offender interventions in England and Wales. Paper presented at the North American Correctional and Criminal Justice Psychology Conference, Sheraton Centre, Toronto.

14.

Hatcher

R. M.

Palmer

E. J.

McGuire

Hounsome

J. C.

Bilby

C. A. L.

Hollin

C. R.

(2008). Aggression Replacement Training with adult male offenders within community settings: A reconviction analysis. Forensic Psychiatry and Psychology, 19, 517-532.

15.

Hollin

C. R.

McGuire

Hounsome

Hatcher

Bilby

Palmer

E. J.

(2008). Cognitive skills offending behaviour programmes in the community: A reconviction analysis. Criminal Justice and Behavior, 35, 269-283.

16.

Hollis

(2007). Reconviction analysis of programme data using Interim Accredited Programmes Software (IAPS). London, UK: National Offender Management Service.

17.

Howard

Francis

Soothill

Humphreys

(2009). OGRS 3: The revised Offender Group Reconviction Scale. London, UK: Ministry of Justice.

18.

Howells

Day

(2003). Readiness for anger management: Clinical and theoretical issues. Clinical Psychology Review, 23, 319-337.

19.

Mason

Adler

J. R.

(2012). Group-work therapeutic engagement in a high secure hospital: Male service user perspectives. The British Journal of Forensic Practice, 14, 92-103.

20.

McGuire

(2004). Understanding psychology and crime: Perspectives on theory and action. Maidenhead, UK: Open University Press.

21.

McMurran

(2009). Motivational Interviewing with offenders: A systematic review. Legal and Criminological Psychology, 14, 83-100.

22.

McMurran

Theodosi

(2007). Is treatment non-completion associated with increased reconviction over no treatment? Psychology, Crime and Law, 13, 333-344.

23.

Ministry of Justice. (2017). Offender Management statistics quarterly: England and Wales. London, England: Ministry of Justice.

24.

Nunes

K. L.

Cortoni

(2006a). The heterogeneity of treatment non-completers. Toronto, Ontario: Correctional Services Canada.

25.

Nunes

K. L.

Cortoni

(2006b). Estimating risk of dropout and expulsion from correctional programs. Toronto, Ontario: Correctional Services Canada.

26.

Olver

M. E.

Stockdale

K. C.

Wormith

J. S.

(2011). A meta-analysis of predictors of offender treatment attrition and its relationship to recidivism. Journal of Consulting and Clinical Psychology, 79, 6-21.

27.

Palmer

E. J.

McGuire

Hatcher

R. M.

Hounsome

Bilby

C. A. L.

Hollin

C. R.

(2008). The importance of appropriate allocation to offending behaviour programmes. International Journal of Offender Therapy and Comparative Criminology, 52, 206-221.

28.

Palmer

E. J.

McGuire

Hatcher

R. M.

Hounsome

J. C.

Bilby

C. A. L.

Hollin

C. R.

(2009). Allocation to offending behavior programs in the English and Welsh Probation Service. Criminal Justice and Behavior, 36, 909-922.

29.

Palmer

E. J.

McGuire

Hounsome

Hatcher

Bilby

Hollin

C. R.

(2007). Offending behaviour programmes in the community: The effects on reconviction of three programmes with adult male offenders. Legal and Criminological Psychology, 12, 251-264.

30.

Robinson

(1995). The impact of cognitive skills training on post-release recidivism among Canadian federal offenders. Ottawa, Ontario: Correctional Service of Canada.

31.

Smith

Hoare

(2009). Crime in England and Wales 2008/9. Volume 2. Home Office Statistical Bulletin 11/09.

32.

Sturgess

Woodhams

Tonkin

(2016). Treatment engagement from the perspective of the offender reasons for noncompletion and completion of treatment: A systematic review. International Journal of Offender Therapy and Comparative Criminology, 60, 1873-1896.

33.

Turner

(2006). Developing understanding of accredited programmes completions: The role of case-management and barriers to completion (Unpublished report). West Yorkshire Probation Area.

34.

van Voorhis

Spruance

L. M.

Ritchey

P. N.

Listwan

S. J.

Seabrook

. (2004). The Georgia cognitive skills experiment: A replication of Reasoning and Rehabilitation. Criminal Justice and Behavior, 31, 282-305.

35.

Ward

Day

Howells

Birgden

(2004). The multifactor offender readiness model. Aggression and Violent Behaviour, 9, 645-673.

36.

Wormith

J. S.

Olver

M. E.

(2002). Offender treatment attrition and its relationship with risk, responsivity, and recidivism. Criminal Justice and Behavior, 29, 447-471.

37.

Zanis

D. A.

Mulvaney

Coviello

Alterman

A. I.

Savitz

Thomson

(2003). The effectiveness of early parole to substance abuse treatment facilities on 24-month criminal recidivism. Journal of Drug Issues, 33, 223-236.