Improving the Design of Cluster Randomized Trials


Carl Westine—IDPE Student, WMU
Wednesday, March 12, 2014
Location: 4410 Ellsworth Hall
Time: Noon – 1pm

Evaluators and researchers rely on estimates of parameter values including effect sizes and variances (unconditional or conditional) to appropriately power cluster randomized trial designs.  Individual disciplines, including education, increasingly test interventions using more complex hierarchical linear model structures.  In order to improve the design of these studies, researchers have emphasized the development of precise parameter value estimates through meta-analyses and empirical research.  In this presentation, I summarize recent research on empirically estimating design parameters with an emphasis on intraclass correlations and the percent of variance explained by covariates, R2.  I then demonstrate how these parameter values are becoming increasingly accessible through software.  Using Optimal Design Plus, I show how evaluators and researchers can utilize these parameter value estimates to improve the design of cluster randomized trials.

 

No Comments

No comments yet.

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.