Assessing the Simultaneous Effects of Modeled Interventions

Abstract

We urge the use of modeling before experimentation to judge the effects of possible interventions for dealing with social problems. This article provides two approaches for assessing the separate and joint effects of such intervention models. The value-added approach decomposes effects in a manner analogous to stepwise regression whereas the ANOVA approach decomposes the effects according to the analysis-of-variance model. The two approaches are applied to a model for the achievement of racial equality in an occupational group at a local level. The treatment effects of an affirmative action program and growth in the number of jobs are analyzed with the two approaches