Title: The Use of Weights in Multilevel Models: Theoretical Concerns and Practical Solutions.

 

Abstract: The use of Multilevel Models (MLM) is on the rise in the social sciences. Their utility and flexibility to a wide variety of problems have made them of interest to investigators who are increasingly faced with complex data. Complex data structures that have been built upon sampling designs have traditionally been handled using the weighting of observations. In MLM, this is not the case. The use of weights in MLM will be discussed with attention paid to the theoretical risks and gains one should be aware of when deciding to use weights. A simple simulation study will also be presented to demonstrate practical methods of evaluating the need and or effectiveness of weights in the Multilevel context.