Beyond the RCT: Some Overlooked Challenges to Statistical Inference in Human Health Research

J. Michael Oakes, PhD
* Candidate for Director Position in the Biostatistical Design and Anlaysis Center
Assistant Professor
Division of Epidemiology & Community Health
Minnesota Population Center

Monday, March 19, 2007
12:15 pm
Moos 1-450 Moos Tower
Minneapolis Campus

Abstract:

The double-blind randomized clinical trial (RCT) is a superior design for answering many questions involving pharmaceutical effects. But the design is inappropriate for most other questions involving human health, especially those involving social behavior such as health disparities and clinical practice. Some reasons for this, such as ethical and blinding constraints, are well understood. But others, such as social stratification, social interaction, and contextual effects, typically remain overlooked. Accordingly, this talk will address some inferential challenges posed by such phenomena. Practicable methods for improving inference beyond the RCT, such as group-randomized trials and matched sampling, will be explained and advanced.


For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html