Research That Matters (January 17 - 20, 2008)

Purpose: There are many studies applying the Transtheoretical Model (TTM) constructs to study intentional behavior change in unrestricted settings with consistent results. However, there is evidence that restrictive settings can affect the measurement and structure of the TTM constructs. Substance abuse treatment in restricted criminal justice (CJ) settings continues to rise. Understanding how the change process, as depicted by the TTM, interacts with restricted settings is critical to the development and evaluation of effective treatments in these settings. Secondary analysis was conducted on TTM data collected from three studies. Project CARE (n=187) recruited inmates in a substance abuse treatment program in a large urban county jail. Project Check-In (n=116) recruited offenders mandated to a Therapeutic Community in a state prison facility. The third study, Project MATCH, was a large multi-site alcohol treatment matching study with two recruitment arms, aftercare (n=724 at pre-treatment and 650 at post-treatment) and outpatient (n=912 at pre-treatment; 844 at post-treatment). Methods: This study, conducted in two phases, tested for non-invariance of measures of the TTM constructs for problematic alcohol users in treatment. Phase I compared two restricted and one unrestricted group entering treatment and Phase II compared one restricted and two unrestricted groups at the end-of-treatment. In a Profile Analysis (PA), subscale mean profiles for each of the TTM constructs for alcohol were compared across restricted and unrestricted settings at each time point. Using structural equation modeling techniques, the factor structure of a pretreatment TTM model (readiness, confidence, temptation) across the restricted settings (jail and aftercare) and the unrestricted setting (outpatient) and an end-of-treatment model (readiness, confidence, temptation, and processes of change) across the restricted setting (prison) and the unrestricted settings (aftercare and outpatient) were examined. Results: In the Profile Analysis, TTM subscale mean profiles for the restricted settings were found to differ from the unrestricted settings (p<.05). The tests of invariance, at pretreatment and end-of-treatment found the overall TTM models to be invariant but found certain constructs to be non-invariant. At pre-treatment, the restricted settings had equivalent factor structure on the confidence and temptation construct models, but the structure was non-invariant between the restricted and unrestricted settings. The difference in the factor structure across settings was specifically in the factor loadings for the social situations subscale of these measures (p<.001) across the restricted and unrestricted settings. At post-treatment, the factor structure was found to be equivalent for the behavioral processes for the two restricted settings and non-invariant across the restricted and unrestricted settings. The difference in the factor structure across settings was specifically in the factor loadings for the counter-conditioning and stimulus control subscales (p<.001) and on the model covariance between the behavioral processes and temptation (p=.018). Implications: Findings indicate that, overall, the TTM model is robust and explanatory of behavior change, even for inmates whose behavior is restricted. However, it is critical that we recognize the differences in the factor structure for the constructs (i.e. controlling stimuli) affected by the restricted settings and that new methods of measuring these constructs are needed.