Comparisons of Theoritical Binary Models

Abstract

The purpose of this paper is to discuss the theory of the logit model, probit model and a linear probability model. Comparison of the model in terms of parameter $ \ beta_k $ performed using maximum likelihood and weighted least squares method. Several other test comparisons are shown as multicollinearity, residual test, hypothesis testing, goodness-of-fit and diagnostic data as well as remote data influential. Keywords: logit model; probit model; linear probability model; maximum likelihood method; method of Ordinary Least Square distributions; Weighted least squares method.