Control deserves special consideration when measuring program impact

Control deserves special consideration when measuring program impact. Program managers are justifiably reluctant to have their programs' success judged in terms of outcomes over which they have less than total control.

Is it fair, for example, to judge the effectiveness of a rehabilitation program by its clients' subsequent behavior even though other factors besides the program also influence how clients behave? If one takes the position that performance measures should be developed only for those events over which a program has total or near-total control, one will find few events that can be measured outside the program's direct outputs.

Yet these direct outputs (e.g., X number of accidents investigated. Y number of students completing educational programs. Z number of polluting installations cited for non-compliance with emission standards) do not go far in answering the questions raised by a program's varied constituent groups.