Covariate Selection for Generalizing Experimental Results: Application to Large-Scale Development Program in Uganda

TL;DR

This paper introduces a covariate selection method for generalizing experimental results to a target population when covariate data in the population is limited, demonstrated through a large-scale development program in Uganda.

Contribution

It proposes a novel algorithm to identify separating covariates necessary for valid generalization under data constraints, requiring rich experimental data but limited population covariates.

Findings

The method successfully estimated the population average treatment effect in Uganda.

It enables generalization when traditional methods fail due to limited population covariate data.

The approach incorporates researcher-specific constraints on measured variables.

Abstract

Generalizing estimates of causal effects from an experiment to a target population is of interest to scientists. However, researchers are usually constrained by available covariate information. Analysts can often collect much fewer variables from population samples than from experimental samples, which has limited applicability of existing approaches that assume rich covariate data from both experimental and population samples. In this article, we examine how to select covariates necessary for generalizing experimental results under such data constraints. In our concrete context of a large-scale development program in Uganda, although more than 40 pre-treatment covariates are available in the experiment, only 8 of them were also measured in a target population. We propose a method to estimate a separating set -- a set of variables affecting both the sampling mechanism and treatment effect heterogeneity -- and show that the population average treatment effect (PATE) can be identified by adjusting for estimated separating sets. Our algorithm only requires a rich set of covariates in the experimental data, not in the target population, by incorporating researcher-specific constraints on what variables are measured in the population data. Analyzing the development experiment in Uganda, we show that the proposed algorithm can allow for the PATE estimation in situations where conventional methods fail due to data requirements.