Abstract: Meta-analysis is a statistical methodology to combine information from diverse studies to reach a more reliable and efficient conclusion. It can be performed by either synthesizing study-level summary statistics (SS) or modeling individual participant-level data (IPD), if available. However, it remains not fully understood whether the use of IPD indeed gains additional efficiency over SS. In this talk, we discuss the relative efficiency of the two methods under a general likelihood inference setting. We show theoretically that there is no gain of efficiency asymptotically by analyzing IPD, provided that the random-effects follow the Gaussian distribution and maximum likelihood estimation is used to obtain summary statistics. Our findings are confirmed by simulation studies and a real data analysis of beta-blocker treatment effect for myocardial infarction. This is a joint work among Dungang Liu, Xiaoyi Min and Heping Zhang.

Bio: The Statistics Seminar speaker for Wednesday, October 17, 2018, is Ding-Geng Chen, the Wallace H. Kuralt Distinguished Professor and director of the Consortium for Statistical Development and Consultation (CSDC) in the School of Social Work, and is jointly appointed as a clinical professor in the Department of Biostatistics at the UNC Gillings School of Global Health. He is an elected fellow of American Statistical Association.

As a professor in biostatistics, he is interested in developing biostatistical methodologies in clinical trials, meta-analysis, Bayesian statistics and their applications to public health. As a professor in social work, he is interested in developing Bayesian social and health intervention research, cusp catastrophe modelling, statistical causal inferences, propensity score and structural-equation models (SEM). He is PI/Co-PI for several NIH R01 research projects in biostatistical methodology development and public health applications.

Talk: TBA