Measures of Selection Bias in Regression Coefficients Estimated from Non-Probability Samples

04/13/2020

∙

by   Brady T. West, et al.

∙

0

∙

share

We derive novel measures of selection bias for estimates of the coefficients in linear regression models fitted to non-probability samples. The measures arise from normal pattern-mixture models that allow analysts to examine the sensitivity of their inferences to assumptions about the non-ignorable selection in these samples. They require aggregate-level population data on auxiliary variables that are related to the outcome but not included in the regression model, but they do not require microdata for the non-selected cases. A simulation study assesses the effectiveness of the measures, and the measures are then applied to data from two real studies.