Random Effects Probit and Logit: The Right Marginal Effects for the Right Econometric Specification

12 Pages Posted: 13 Apr 2017

See all articles by James R. Bland

James R. Bland

The University of Toledo, Department of Economics

Amanda Cook

Bowling Green State University - Economics

Date Written: April 12, 2017

Abstract

Random effects probit and logit specifications are common when analyzing economic experiments. Stata's fitted values from these estimations, however, appear to fit data poorly compared to their pooled counterparts. This is entirely due to Stata reporting the median predictive value, when practitioners expect the mean predictive value. This propagates into marginal effects calculations, which may lead researchers to under- or over-state the economic significance of results. We demonstrate a simple procedure for calculating the correct predictive mean and mean marginal effect in Stata.

Keywords: Random effects, probit, logit, marginal effect, prediction

JEL Classification: C23, C24, C87

Suggested Citation

Bland, James R. and Cook, Amanda, Random Effects Probit and Logit: The Right Marginal Effects for the Right Econometric Specification (April 12, 2017). Available at SSRN: https://ssrn.com/abstract=2951889 or http://dx.doi.org/10.2139/ssrn.2951889

James R. Bland (Contact Author)

The University of Toledo, Department of Economics ( email )

Toledo, OH 43606
United States

HOME PAGE: http://https://sites.google.com/site/jamesbland/

Amanda Cook

Bowling Green State University - Economics ( email )

Bowling Green, OH 43403
United States

HOME PAGE: http://https://sites.google.com/site/amandacatherinecook/

Here is the Coronavirus
related research on SSRN

Paper statistics

Downloads
253
Abstract Views
1,155
rank
132,375
PlumX Metrics