Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 310 - Making Finite Population Inferences from Nonprobability Samples
Type: Topic-Contributed
Date/Time: Wednesday, August 11, 2021 : 3:30 PM to 5:20 PM
Sponsor: Survey Research Methods Section
Abstract #317246
Title: Measures of Selection Bias in Regression Coefficients Estimated from Nonprobability Samples
Author(s): Brady T. West* and Roderick Joseph Little and Rebecca R Andridge and Philip Boonstra and Erin Ware and Anita Pandit and Fernanda Alvarado-Leiton
Companies: University of Michigan and University of Michigan and The Ohio State University College of Public Health and University of Michigan and Institute for Social Research, University of Michigan-Ann Arbor and University of Michigan and Institute for Social Research, University of Michigan-Ann Arbor
Keywords: Selection Bias; Non-Probability Samples; Linear Regression; Probit Regression; Non-Ignorable Selection; Polygenic Scores

Selection bias in survey estimates is a major concern, particularly for non-probability samples. Recent developments have provided survey researchers with model-based indices of the potential selection bias in estimates of means and proportions computed from non-probability samples that may be subject to non-ignorable selection mechanisms. To our knowledge, there are currently no systematic approaches for measuring selection bias for regression coefficients, a problem of great practical importance. Generalizing these recent developments, we derive novel measures of selection bias for estimates of the coefficients in linear and probit regression models. The measures leverage auxiliary variables available for the population of interest that provide information about the variable being modeled when conditioning on the predictors of interest. After reviewing the conceptual background, we present results from a simulation study that demonstrates the effectiveness of the measures. We then review results from two independent applications of the measures to data from real surveys. We conclude with links to easy-to-use R functions for computing the measures and suggestions for practice.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2021 program