Using predictions and marginal effects to compare groups in regression models for binary outcomes

Methods for group comparisons using predicted probabilities and marginal effects on probabilities are developed for regression models for binary outcomes. Unlike approaches based on the comparison of regression coefficients across groups, the methods the authors propose are unaffected by the scalar identification of the coefficients and are expressed in the natural metric of the outcome probability. While they develop their approach using binary logit with two groups, they consider how their interpretive framework can be used with a broad class of regression models and can be extended to any number of groups.

The dataset groupcompare-hrs1.dta was created using the HRS files:

1. Health and Retirement Study public use dataset for 2006 file h06f2b.dta. Produced and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740). Ann Arbor, MI.
2. RAND HRS Data, Version P file rndhrs_n. Produced by the RAND Center for the Study of Aging, with funding from the National Institute on Aging and the Social Security Administration. Santa Monica, CA.

Contributed projects based on HRS public data are provided by researchers who want to share their work with the research community. Researchers interested in contributing their own products are invited to contact us at hrsquestions@umich.edu. HRS does not produce or support these products and is not responsible for their content or use. They are provided here as a service to the research community.