Performance of Some Yang and Chang estimators in Logistic Regression Model.

Abstract

In logistic regression models, the maximum likelihood (ML) method is always one of the commonly used to estimate the model parameters. However, unstable parameter estimates are obtained as a result due to the problem of multicollinearity and the mean square error (MSE) gotten cannot also be relied on. Several biased estimators has been proposed to handle the issue of multicollinearity and the logistic Yang and Chang estimator (LYC) is one of them. Likewise research has also made us to understand that the biasing parameter has effect too on the value of the MSE. In this paper we proposed seven LYC biasing estimators and they were all subjected to Monte Carlo simulation studies and Pena data set was also used too. The result from the simulation study shows that LYC estimators outperforms the Logistic Ridge Regression (LRR) and the ML approach. Furthermore, application to Pena real data set also conform to the simulation results

Keywords

• logistic regression
• multicollinearity
• biased estimators
• maximum likelihood
• simulation
• mse