Divided We Poll, United We Pool: An Application of Multiple Imputation for Multiple Surveys

• Activity Number: 439 - Contributed Poster Presentations: Survey Research Methods Section
• Type: Contributed
• Date/Time: Wednesday, August 10, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Jui-Chung Allen Li* and I-Ting Tsai and Fang-Yu Rona Hu
• Companies: Quanthon Corporation and Quanthon Corporation and Quanthon Corporation
• Keywords: survey nonresponse; multiple imputation; opinion poll; survey data quality; telephone survey; government statistics

Abstract

We apply Gelman, King, & Liu’s (1998) “Multiple Imputation for Multiple Surveys” in a novel approach to reduce halfway refusals in lengthy telephone surveys. The questionnaire is divided into subsections. One or multiple selected subsections are presented to a particular respondent. Collected data are pooled together by multiple imputations, as described in Gelman et al. (1998). We illustrate the implementation with two examples: the 2019 Survey of Women’s Needs (sponsored by New Taipei City Government) and the 2021 Survey of Public Satisfaction with Government Performance (sponsored by Yunlin County Government). Our approach yields lower nonresponse rates than the earlier wave(s) of the same survey. We will report a detailed comparison of our approach and the traditional approach using data from the 2022 Survey of Public Satisfaction with Government Performance sponsored by the Yunlin County Government.

Reference: Gelman, A., King, G., & Liu, C. (1998). Not asked and not answered: Multiple imputation for multiple surveys. Journal of the American Statistical Association, 93(443), 846-857.