An Evaluation of Substance Abuse Aftercare Program for Homeless Women with Children Using Confounding Variable-Control Design

Abstract

An outcome evaluation of a substance abuse aftercare program for homeless women with children was conducted using confounding variable-control evaluation design. The confounding variables are chosen from pre-treatment and other contextual variables of the clients that are known to have significant influence on the program outcome, but those that could not have been influenced a priori by the client involvement in in-treatment program activity at Transition House (TH). The latter is the independent variable of this evaluation design. The pre-treatment variables are measured by severity of alcohol and other drug (AOD) problems of the clients, their mental health status, age, and their job status before enrollment in the program. The contextual confounding variables are composed of family and social support available to the clients before and during recovery. While applying multiple regression analysis, we were able to explain 50.8% of the total variance in program outcome by four pre-treatment variables. By adding two contextual variables of family and social support, the total variance in program outcome explained is increased to 64.1%. Finally, by adding the degree of client involvement in in-treatment program activity, we were able to augment the total variance of the program outcome to 69.7%. By estimating the changed variance of program outcome by the in-treatment program activity during the final step, controlling for all other variables previously entered, we were able to establish that client involvement in in-treatment program had unique and positive impact on the program outcome distinct from those explained by the confounding variables. The additional variance uniquely added by in-treatment program activity is 5.6% ( p < .001). It has been determined that the degree of client involvement in in-treatment program had positive and systematic impact on the program outcome.