Development and Validation of Decisional Balance and Processes of Change Inventories for Heavy, Episodic Drinking

Abstract

The transtheoretical model of behavior change indicates that people who successfully change problem behaviors move through a series of stages of readiness to change. This model has been successfully applied to a wide range of health-related behaviors, but only limited work has been devoted to adapting this model to the cessation of heavy episodic drinking. The goals of the study were to develop measures of two of the key constructs of the model: a decisional balance inventory measuring the pros and cons of alcohol consumption and a processes of change instrument assessing techniques used to change heavy drinking behavior. Three hundred and eighty-two college students completed questionnaires containing these two instruments as well as measures of quantity and frequency of alcohol intake, demographic information, and problems associated with alcohol use. Measurement development was conducted on these two scales using a split-half cross validation procedure and following sequential methods. Results were quite satisfactory for the decisional balance inventory. One of the stopping criteria (MAP) suggested a three component solution, but two other stopping criteria (PA, Scree) suggested only one. The two factor solution was retained, due to empirical and theoretical evidence. Cronbach's coefficient alpha and SEM modification indices were then examined to reduce the item set. Confirmatory analysis using SEM to test alternative models for this reduced item set resulted in the acceptance of an uncorrelated two factor model and a 16- item scale measuring the pros and cons of alcohol use. This scale demonstrated concurrent validity by its strong association with measures of quantity and frequency of alcohol use, problems associated with alcohol use, and stage of change. A processes of change instrument was also developed. Initial principal components analysis revealed only seven factors using the MAP criteria and five factors following the parallel analysis criteria. The scree procedure suggested only one factor. Structural equation modeling (SEM) was then used to improve this solution since the factors were expected to be highly correlated and higher order factors were hypothesized to exist. Several rounds of reevaluation of theoretical fit and concomitant item evaluation were performed, using Cronbach's coefficient alpha and SEM modification indices to inform decision making. The result was a 30 item scale that fits the data adequately. Confirmatory results concurred with this solution. This scale was judged to have marginal concurrent validity by its relationships with measures of quantity and frequency of alcohol use, problems associated with alcohol use,