Reyes-Gastelum, D., Pierce, S.J., & Van Egeren, L. (October 2012). Obtaining Balanced Intervention and Control Groups in a Multi-site Cluster Randomized Trial: Minimization as an Alternative to Randomization in the Head Start on Science Program. Demonstration session presented at Evaluation in Complex Ecologies: Relationships, Responsibilities, Relevance, the 26th annual conference of the American Evaluation Association, Minneapolis, MN. This PDF file contains the PowerPoint slides and notes used for the presentation.
Demonstration Session Abstract: This intermediate session will demonstrate how we assigned classrooms to intervention and control groups for a longitudinal, multi-site cluster randomized trial of an early childhood science education program called Head Start on Science (HSOS). We will cover some of the methods available for the group assignment (i.e., randomization) process, the advantages and disadvantages of the minimization method, and the circumstances under which it may outperform pure randomization. This method uses an algorithm to explicitly balance groups with respect to a collection of covariates during the group assignment procedure. In addition to discussing covariate selection and demonstrating how to use the MinimPy software (Saghaei & Saghaei, 2011), we will also discuss how and why we incorporated various constraints on the group assignments into our procedures. Our goal is to increase audience members’ knowledge of methods, tools, and issues related to group assignment in experimental studies.#ClusterRandomizedTrial #StatisticalMethods #QuantitativeMethods-TheoryandDesign #HowTo #Allocation #MinimPy #2012Conference #Randomization #Minimization