Bireprob: An Estimator for Bivariate Random-effects Probit Models. (March 2016)

• Record Type: Journal Article
• Authors: Plum, Alexander
• Is Part Of: Stata journal. Volume 16:Number 1(2016)
• Page Start: 96
• Page End: 111
• Publication Date: 2016-03
• Subjects: st0426 -- bireprob -- bivariate random-effects probit -- maximum simulated likelihood -- Halton draws; Statistics -- Periodicals; Statistics -- Computer programs -- Periodicals; 001.422
• DOI: 10.1177/1536867X1601600111
• Languages: English
• ISSNs: 1536-867X

Abstract

I present thebireprob command, which fits a bivariate random-effects probit model.bireprob enables a researcher to estimate two (seemingly unrelated) nonlinear processes and to control for interrelations between their unobservables. The estimator uses quasirandom numbers (Halton draws) and maximum simulated likelihood to estimate the correlation between the error terms of both processes. The application ofbireprob is illustrated in two examples: the first one uses artificial data, and the second one uses real data. Finally, in a simulation, the performance of the estimator is tested and compared with the official Stata commandxtprobit.