The Mathematics of Behavioral Change

Abstract

This paper describes a new method of calculating the failure rate of programs involving behavioral change. The method overcomes a threat to construct validity implicit in the previous calculation method. Using it, one can estimate the failure rate as well as the fraction of program participants who are ultimately expected to remain successful. We illustrate the use of the method with data from a cohort of parolees. However, it should be of interest to those engaged in evaluating other programs that attempt to change individual behavior.