Site Search

Colloquium - Dept. Mathematics & Statistics presents Dr. Leilei Zeng of the Dept. of Statistics and Actuarial Science, University of Waterloo

Add this event into your calendar using the iCAL format
  • Fri, 08/12/2016 - 1:00pm




Title: Methods of Clustered Longitudinal Binary Data

 

Abstract:

Clustered longitudinal data feature cross-sectional, serial, and possibly mixed types of associations existing among responses 

at different time points, from different subjects within the same cluster. The methodological challenges arise from modelling association structures as well as dealing with missing data. When the marginal mean and the association of the binary responses are of primary interest, a composite likelihood based approach is developed to carry out robust estimation with minimal model assumptions that leads to valid inferences for a wide variety of missing data problems. Alternatively, if interest lies in the change in the binary response over time and the association among the changes, transition models and corresponding estimation methods based on GEE2 and alternating logistic regression are developed. 

The design of clustered randomized longitudinal trials based on transition models is also considered. Specifically we derive methods for calculating the minimum number of clusters required to achieve certain power to detect a cluster level treatment effect



Dina Labelle
mthsta1@uwindsor.ca
(519)253-3000 ext.3016