General Tests For Group Comparisons
When measuring persons in different groups (control-treatment) the object is often to test mean differences between groups. The well-known t-test can be used for independent continuous observations, but in practice we often observe more complex data. The data can be categorical (e.g., binary or ordinal), and can be correlated (e.g., longitudinal, cross-sectional), which violates the assumptions of the t-test. A more general approach is examined that can control for the correlated data structure, while testing differences across groups.
Profile of the student
Interested in novel statistical techniques and the statistical software R.
Contact Jean-Paul Fox (CU B331, OMD).