The assessment of offending-related stage of change in offenders: psychometric validation of the URICA with male prisoners

Abstract

Measuring accurately an offender's present stage of change on offending-related needs may be important for determining match with Á and therefore selection into Á rehabilitation programmes. It is also essential testing empirically the dominant model of offending-related behavioural change Á the Transtheoretical Model (TTM). Several self-report questionnaire scales and algorithms have been developed and applied to offender populations to measure stage of change. The paper reports on an investigation into the psychometric properties (internal reliability, and factorial, convergent and concurrent validity) of the University of Rhode Island Change Assessment Scale (URICA), a self-report questionnaire. We compared the URICA with the New Zealand-designed Criminogenic Needs Inventory's structured algorithm for assessing motivation to change in a sample of 260 male prisoners. Both instruments are based on the TTM. Confirmatory factor analysis provided moderate support for the URICA's original four-factor structure. Strong correlations with the Criminogenic Needs Inventory data supported the validity of the URICA, as did the positive relationship between both measures and unconscious self-deception on the Balanced Inventory of Desirable Responding. However, neither measure was susceptible to consciously over-favourable self-presentation biases. The results of this study provide good preliminary support for the reliability and validity of the URICA in the assessment of offending-related stage of change in male prisoners, and in conjunction with other recent research, suggest it may be premature to abandon the adaptation of the TTM to the measurement of crime-reducing change processes.

How do offenders move through the stages of change?

“…The factorial validity of the URICA is even harder to ascertain in offending samples given the paucity of studies in this area. To date, only two studies on general offenders have found support for a four-factor structure (Levesque, Gelles, & Velicer, 2000;Polaschek et al, 2010). Using a domestic violence adaptation of the URICA, an earlier study by Levesque et al (2000) on 258 domestic violence offenders supported this four-factor structure.…”

“…Using a domestic violence adaptation of the URICA, an earlier study by Levesque et al (2000) on 258 domestic violence offenders supported this four-factor structure. The study by Polaschek et al (2010) reported cautious support for the four-factor structure with a non-normed fit index of .96 and a comparative fit index (CFI) of .97, but did not report the value for the root mean square error of approximation (RMSEA). Other studies involving the offending population found support for fewer than four factors (Cohen et al, 2005;Eckhardt, Babcock, & Homack, 2004;Eckhardt & Utschig, 2007;Hemphill & Howell, 2000;Kim & Song, 2009).…”

“…In studies with offenders, the Cronbach's coefficient alphas have ranged between .60 and .93. The subscales that did not show acceptable levels of internal consistency varied across studies: precontemplation = .69 (Cohen, Glaser, Calhoun, Bradshaw, & Petrocelli, 2005), contemplation = .60 (Polaschek et al, 2010) and maintenance = .61 (Austin et al, 2011). Cronbach's coefficient alphas for the SMP URICA-21 in this study were as follows at pre-SMP: full scale = .78, precontemplation = .62, contemplation = .85, action = .80 and maintenance = .77.…”

“…It was hypothesised that confirmatory factor analysis (CFA) would yield strong support for a four-factor structure for the URICA given that there was evidence from earlier studies in non-offending (McConnaughy et al, 1989) and offending samples (Polaschek et al, 2010). Given that the URICA has had limited research with the Correctional population, an exploratory factor analysis (EFA) would be undertaken if the CFA resulted in a poor fitting model to the a priori four-factor structure recommended by Field (2009).…”

“…This may lead to self-report measures of the SoC being susceptible to social desirability bias, thus, compromising the validity of these measures with this population (McMurran, 2009). Nonetheless, some studies have found psychometric evidence that these stage-based measurements were not influenced by social desirability bias (Polaschek, Anstiss, & Wilson, 2010;Tierney & Mccabe, 2004). Some have argued that the controlled prison setting limits offenders' opportunities to engage in attitudes and behaviours from the maintenance stage (Casey et al, 2005).…”