Getting more "bang" for your performance measures "buck"

Funders of intervention programs and services at the federal, state, and lo cal levels are requiring greater accountability via the use of performance measures. Data from these measures are usually reported descriptively for t he entire program. It is suggested here that performance measure data that are already bring collected could become much more useful, at low additiona l cost, by further analyzing the relationships among the measures, especial ly across time. By using logic models to show expected relationships among several types of measures, and then analyzing variability among program del ivery units for those measures, evaluators could provide evidence for the h ypothesized connections among program delivery, outputs, and desired outcom es. Such analyses also would help to link two major purposes for performanc e measurement-program improvement and accountability to the public-which ar e often viewed as contrasting or even incompatible uses.