On estimation in multinomial logit model for a panel study

Abstract

In this paper, we introduce a new method, which provides the estimator with smaller mean square error (MSE) than traditional method. The suggested method retains the simplicity of traditional method and enables us to obtain a smaller MSE for the predicted estimator. In the procedure of prediction, we use the information of sample size for each survey. To understand the efficiency of our suggested method compared to the traditional one, we use a series of surveys taken in the Taiwan presidential election of 2000 as an empirical study. This empirical study is carried out in order to better understand the performance of the suggested method's estimator compared to the estimator of the traditional method.

Keywords:

• Multinomial logit model
• panel study
• voting percentage