Model Selection For Longitudinal Data Based On The Quadratic Inference Function Under A Simple Order Restriction

In the present paper.we review the GEE and QIF for marginal regression model of Longitudinal data. We also introduce Bayesian information criterion based on the quadratic inference function (BIQIF).which has better performance in the marginal regression model for Longitudinal data under unrestriction. Following BIQIF,we propose an new information criterion for model selection under a simple order restriction, and call it Order Restriction Information Quadratic Inference Function (ORIQIF). Furdermore,we proof the consistency of ORIQIF. Lastly,we demonstrate the performance of ORIQIF via three simulation studies.