Department of Statistics, The Ohio State University
Statistics and Biostatistics Colloquium Series

A case-control design is a useful approach to detect association between a genetic marker and a complex disease. The case-control data can be displayed in a 2x3 contingency table (for a bi-allelic marker). To analyze this 2x3 table, the trend test (1 degree of freedom) or Pearson's test (2 degree of freedom) are simple and often employed. The trend test depends on the underlying genetic model; otherwise it is not robust to misspecification of the model. Pearson's test is robust but it has higher degree of freedom. The maximum of three trend tests respectively optimal for the four common genetic models (MAX) is more "efficiency robust" than a single trend test and Pearson's test. Here, we present two-stage analysis strategy. In the first stage, we select the underlying genetic model using data. Then, in the second stage, we test association based on the selected model. Simulation studies and applications to real data are presented to demonstrate that the new strategy is preferable than the existing ones (trend test, Pearson's test, and MAX) for case-control genetic association studies.

Meet the speaker in Room 212 Cockins Hall at 4:30 p.m. Refreshments will be served.