Powerful and robust designs for longitudinal multilevel studies

Shirley Ortega Azurdy PhD

Project at: Faculty of Health, Medicine and Life Sciences, Maastricht University

Project financed by: The Netherlands Organisation for Scientific Research (NWO)

Supervisor: Prof. dr M.P.F. Berger

Project running from: 1 September 2004 – 1 March 2009

Summary:
The use of powerful (optimal) designs for multilevel data with time-structured measurements will result in a high probability of finding a true effect with a minimum sample size. However, an optimal design for multilevel models dpends on the best fitting (correct statistical model). Since the best fitting model is unknown before data collection, it is highly relevant to try and find a robust design that remains highly powerful for alternative models. The aim of this research proposal is to find such designs. A second aim of this study is to provide social and behavioral scientists with general and robust guidelines for the required sample sizes at different levels of their multilevel design. These guidelines will help reserahcers to reduce the costs of data collection..

Date of defence: 1 July 2009

Title of thesis: Robust designs for longitudinal studies