The use and abuse of control groups in program evaluation

Abstract
The selection of a control or contrast group is of utmost importance when designing methods to evaluate programs.