Sampling and Evaluation – A Guide to Sampling for Program Impact Evaluation
ms-16-112-en.pdf — PDF document, 2848 kB (2917086 bytes)
Author(s): Lance P, Hattori A
Program evaluation, or impact evaluation, is a way to get an accurate understanding of the extent to which a health program causes changes in the outcomes it aims to improve. Program impact studies are designed to tell us the extent to which a population's exposure to or participation in a program altered an outcome, compared to what would have happened in the absence of the program. Understanding whether a program produces intended changes allows society to focus scarce resources on those programs that most efficiently and effectively improve people's welfare and health.
The usual objective in program impact evaluation is to learn about how a population of interest is affected by the program. Programs are typically implemented in geographic areas where populations are large and beyond our resources to observe in their entirety. Therefore, we have to sample. Sampling is the process of selecting a set of observations from a population to estimate a chosen parameter—program impact, for example—for that population.
This manual explores the challenges of sampling for program impact evaluations—how to obtain a sample that is reliable for estimating impact of a program and how to obtain a sample that accurately reflects the population of interest.
The manual is divided into two sections: (1) basic sample selection and weighting and (2) sample size estimation. We anticipate that readers might get the most utility and comprehensive understanding from reading entire chapters rather than trying to cherry-pick portions of the discussions within them, as one might with a reference manual. This manual is more like a textbook.
Further, the manual is aimed at practitioners—in particular, those who design and implement samples for impact evaluation at their institution. Our discussions assume more than a basic understanding of sampling and some mathematical skill in applying sampling theory. That said, we are less interested in theory than in its practical application to solve sampling problems encountered in the field. We hope this manual will be a comprehensive and practical resource for that task.
This document is not available in print from MEASURE Evaluation.