Evaluating health programs: Where do we find the data?

Abstract

This paper examines the data requirements for impact evaluation in the light of recent federal efforts to improve the evaluation of health programs. Typically, new health programs are launched before there is sufficient time or administrative interest in devising appropriate evaluation methodology. Therefore, any evaluation of such a program must be designed retrospectively. We argue that the best and typically only sources of data appropriate for such evaluations are pre-existing administrative statistics and archival records. The potential and limitations of these data sources are reviewed.