Lawrence H. Moulton and M. Ashraf Chaudhary. Department of International Health, The Johns Hopkins University School of Public Health, 615 N. Wolfe St., Baltimore, MD 21205, 410-955-6370, lmoulton@jhsph.edu

Community-randomized trials typically have small number of randomization units. Even with stratification schemes, such as in pair-matched studies, one can be the victim of “bad luck,” obtaining a treatment allocation that is substantially unbalanced with respect to one or more baseline covariates. We suggest highly constraining the design through restriction to those allocations that meet specified criteria. We present some simulation results that investigate departure from nominal Type I error under severe correlation situations. A SAS® program is described that carries out the designated restrictions and displays key information regarding the effects of the restrictions.

Learning Objectives: Close inspection of the poster will enable the viewer to

• Understand the risks involved in randomization with small numbers of experimental units
• Learn how to use relevant covariates to constrain the potential treatment allocations
• See how a SAS(R) program can perform a highly constrained randomization

Keywords: Statistics, Community Research

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.