Guest speaker for the Psychology Colloquium Series this week is Dr Kevin Bird of the Department of Psychology, University of New South Wales. His presentation is titled "MANOVA abuse: what it is and what can be done about it."

Abstract: Surveys have shown that the vast majority of psychological researchers who ‘run a MANOVA’ on their multivariate data do so as the first stage in a sequential testing procedure, where standard univariate F and t tests are used to complete the analysis. Researchers who do this presumably believe that the MANOVA test provides protection against inflation of the overall familywise error rate. I present new analytical (as distinct from Monte Carlo) results which show that the MFWER (maximum familywise error rate) from the standard three-stage sequential procedure is usually grossly inflated. I also present new single-stage, two-stage and three-stage procedures which use different critical values of F and t tests, thereby controlling the MFWER. Finally, I discuss the conditions under which each of these procedures is appropriate.

When: Wednesday 5 September, 2007
Time: 4:00 pm
Where: Room 1015, 10th Floor, Mathews Building

Please click here for a detailed listing of future colloquia.