Department of Mathematical Sciences
|Thursday, April 14, 2016
|1:15pm to 2:15pm
|Xiaojie Du, Binghamton University
|Inference in Linear Regression Models: An Empirical Likelihood Approach
An empirical likelihood approach with an increasing number of estimated constraints is derived for inference about the intercept and slope of the linear regression model of which the error variable has an unknown symmetric density. A Wilks’ Theorem for the resulting empirical likelihood is obtained. Confidence regions and tests for the parameters are developed based on Wilks’ Theorem.