Optimal Allocation of Interviews to Baseline and Endline Surveys in Place-Based Randomized Trials and...

Area of Science:

• Econometrics
• Survey Methodology
• Experimental Design

Background:

• Place-based trials often use independent cross-sectional surveys for pre- and post-intervention measurements.
• This design involves selecting separate random samples for baseline and endline surveys within geographic clusters.

Purpose of the Study:

• To determine the optimal allocation of survey resources between baseline and endline data collection.
• To maximize the precision of the estimated average treatment effect in place-based experiments with a fixed budget.

Main Methods:

• Formalizing the survey allocation problem as an optimization task.
• Developing formulas for optimal resource distribution based on key statistical parameters.
• Illustrating applications with real-world survey data from Africa and a digital political advertising experiment in Texas.

Main Results:

• The optimal share of interviews for the baseline survey is consistently less than 50%.
• This optimal share increases with the total number of interviews per cluster.
• It also increases with the intracluster correlation and the correlation between baseline and endline measures.

Conclusions:

• The study provides a framework for optimizing survey sample allocation in place-based trials.
• Findings offer practical guidance for researchers planning experiments to enhance the precision of treatment effect estimates.
• Optimal allocation strategies can improve the efficiency of resource utilization in impact evaluations.