Title：Modeling of Longitudinal Ordinal Responses with Multiple Predictors and Its Applications

Speaker： Xiaohui Wang  Associate Professor

Date：June 16, 2016(Thursday).Time4:00-5:00 PM

Location：Science building E518

Brief sketch of speaker：Dr. Xiaohui Wang obtained a Ph.D degree in Statistics from Texas A&M University (College Station) in 2006. Dr. Wang is an Associate Professor of Statistics in School of Mathematic- al and Statistical Sciences at University of Texas-Rio Grande Valley (UTRGV). She founded the Statistical Consulting Center at UTRGV (formerly known as UTPA) in 2008 and served as the Director since then. Her research interests lie on hierarchi- cal Bayesian modeling, functional data analysis, multivariate analysis, categorical data analysis, semiparametric and nonparametric methods, Markov Chain Monte Carlo algorithms, Statistical computations and Decision making. Currently, Dr. Wang serves as Evaluator for two projects: a $3.1 million NSF ADVANCE grant and a $4.6 million DoD grant. She served as PI, Co-PI or key personal to 30 internal and external grants. Dr. Wang has more than 20 publications in top statistical Journa- ls such as Journal of the American Statistical Association; journals in public health area such as Population Health Management, Pan American Journal of Public Health, Journal of Environmental Health, Rehabilitation Coun- seling Bulletin, Journal of Immigrant and Minority Health; and journals in engineering and applied mathematics area such as IEEE-Geoscience and Remote Sensing Letters and International Jou- rnal of Biomathematics.

Abstract：Longitudinal ordinal responses are often seen in behavioral sciences, public health and medical studies. In Alliance for a Healthy Border program (2006-2008), a chronic disease prevention program through twelve federally qualified community health centers serving primarily Hispanics in communities along the U.S.-Mexico border, survey and health measurements were obtained at three time points via a pre-post study design. Successes of the program were evaluated with dichotomous or trichotomous ordinal outcomes of weight reduction, glycemic control, and physical activity improvement. The ordinal response modeling methods include generalized estimating equation (GEE) approach for cumulative logit models and transitional ordinal modeling with multiple predictors. 
Some statistics/biostatistics career opportunities in the United States will be introduced at the end.