Back to Annual Meeting Page		American Public Health Association 133rd Annual Meeting & Exposition December 10-14, 2005 Philadelphia, PA

Abstract #106995

Alison Buttenheim, MBA, Department of Community Health Sciences, School of Public Health, University of California, Los Angeles, 176 W Oak Park Dr, Claremont, CA 91711, 909 399 0816, abuttenheim@hotmail.com

Competing for limited funds in the maternal-child health sector necessitates program evaluations that show a positive impact of interventions on the targeted population. Practitioners rarely have the luxury of true experimental data, relying instead on observational data for most evaluations. However, evaluations that seek to demonstrate causal effects without experimental data must account for how programs are placed across communities, how programs are implemented, and how participants select into programs. These sources of variation (called endogeneity or unobserved heterogeneity) can lead to spurious results if ignored in evaluations.

In this study I identify the most common sources of endogeneity in maternal-child health program evaluations, including nonrandom program placement, unobserved community or household effects, and sample selection bias related to individual participation. I then review two straightforward statistical techniques, instrumental variable approaches and fixed-effects models, that address endogeneity in observational data. These solutions, while common in the econometrics literature, are less frequently used in public health program evaluations and can dramatically improve the validity and rigor of causal claims. I present case studies of MCH program evaluations to show the magnitude of potential bias when endogeneity is ignored, and to demonstrate the effectiveness of instrumental variable and fixed-effects techniques.

Learning Objectives: At the conclusion of the session, the participant in this session will be able to

• Identify common sources of endogeneity in MCH program evaluations using observational data and related potential biases in evaluation results.
• Employ instrumental variable and fixed-effects techniques to address endogeneity.

Keywords: Evaluation, Statistics

Presenting author's disclosure statement:

I wish to disclose that I have NO financial interests or other relationship with the manufactures of commercial products, suppliers of commercial services or commercial supporters.