Increasing Power & Decreasing Costs: A New Method for Drawing High-Quality National Probability Samples of U.S. Citizens

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

This project conducts experiments to assess the potential of a new, high-quality probability sampling method for studies such as the American National Election Study, the General Social Survey, and other face to face surveys. It is designed to enable the survey research community to improve data collection, increase statistical power, yield high response rates, and reduce costs per completed interview. The project builds on a NSF funded Workshop (SES-102394) that convened ten leading experts on survey methodology and sampling to discuss the future of the American National Election Study. The proposed innovations combine a dual frame sampling technique with self-completed surveys (on a computer) in most cases. The study consists of three experiments to assess the feasibility and success of the self completed surveys and provides two implemenentations of the dual-frame sampling design. The first experiment assesses the difference in survey modes between face to face and self complete. The second experiment investigates the optimal level of incentives that respondents require in order to complete surveys of varying lengths in order to observe a response rate comparable to that currently achieved by the American National Election Study. The third experiment is a "proof of concept" roll-out of the new sampling method and self-complete mode, which will be informed by the results of the previous two experiments. This will provide a parallel comparison to the American National Election Study pre-election face-to-face and internet surveys that will be fielded at the same time. If this project is successful, it could yield significant cost savings for face to face surveys while at the same time providing more powerful statistical results.