Two Statistical Inference Methods Within Marginal Models: GEE And QIF

With the development of modern science and technology, longitudinal studies are more and more widely applied in psychology,biology and medicine.This paper discussed two statistical inference methods within marginal models:GEE method and QIF method. We compared the impact of the selection for working correlation on the estimation through the simulation studies.Simulation results show that:For normal data, when the sample size is large,if the working correlation structure is misspecified,the QIF method is more efficient than the GEE method,if the working correlation is correct,the two methods are almost equivalent.For binary data, the results of the two methods are almost the same.