Portal de Periódicos da CAPES

FIRTHLOGIT: Stata module to calculate bias reduction in logistic regression

2008; Federal Reserve Bank of St. Louis; Linguagem: Inglês

Joseph Coveney,

Income, Poverty, and Inequality

The module implements a penalized maximum likelihood estimation method proposed by David Firth (University of Warwick) for reducing bias in generalized linear models. In this module, the method is applied to logistic regression. Others, notably Georg Heinze and his colleagues (Medical University of Vienna), have advocated the method for use under conditions of complete and quasi-complete separation, in which conventional maximum likelihood fails in obtaining finite estimates.