Generalized P Value Testing In Panel Data Model

In this paper, we consider the testing and confidence interval problems of unknown parameters in Panel data models by using the concepts of generalized p-value and generalized confidence interval.Panel data model is an important type of linear mixed effects model which is widely used in fields such as economics ,medical science, biology and computer science. The testing problems of Panel data models have received considerable attention in the literatures. But for some problems we can't establish exact tests and confidence intervals becuase the the variance components are unkown. The generalized p-value and generalized confidence interval due to Tusi and Weerahandi are nimble and effective approaches to solve the testing problems with nusiance parameters. This paper studies the testing and confidence interval problems of unknown parameters in Panel data models and establishes the exact tests and confidence intervals based on the concepts of generalized p-value and generalized confidence interval.For the regression coeffients ,we cosider the testing and confidence interval problems under several important cases, and obtain the exact tests and confidence intervals based on generalized p-value theoretical method. we also discuss the invariance of these tests and confidence intervals under scale transformation. Results of simulation shows generalized p-value test has good powers. For the variance components, we discuss the testing and confidence interval problems of single and arbitrary linear combination and establish exact tests and confidence intervals. We also discuss the invariance of these tests and confidence intervals under scale transformation. The powers of these tests are also obtained by simulations.The methods of obtaining exact tests and confidence intervals based on generalized p-value and generalized confidence interval are feasible when there are nuisance parameters. These methods are convenient to compute and easy to apply to small sample problems.