STRATEGIES FOR COMPARING MULTIPLE-SITE EVALUATIONS UNDER NONEQUIVALENT DESIGN CONDITIONS

The Center for Substance Abuse Prevention (CSAP) currently funds 251 c ommunity partnerships for planning, coordinating, and implementing com munitywide prevention efforts. Partnership evaluations consist of sing le-site evaluations without controls. The possibilities of experimenta l and quasiexperimental designs to demonstrate the overall effectivene ss of the partnership approach are limited under such circumstances. N onexperimental cross-site comparisons offer a promising alternative, t he focus of which will be hypothesis generation from correlational fin dings and methods development. One hundred forty-four partnership eval uation plans were examined to determine process, outcome, and impact m easures that will have promise for cross-site studies. Results of this content analysis revealed that 23 process, 19 outcome, and 18 impact categories of measurement were available for cross-site analysis. Pair ing process measures (independent variables) with outcome and impact m easures (dependent variables) yielded a number of possible analyses of interest that, if pursued, hold the promise of significantly advancin g plausible hypotheses about how partnerships operate to achieve chang es in substance use and abuse and outcomes important to these ends.