Powering Up for the Head Start on Science Program: Using Power Analysis to Plan the Sample Size Required for a Multi-Site Cluster Randomized Trial

This intermediate session demonstrated how we conducted an a priori power analysis for a longitudinal,multisite cluster randomized trial of an early childhood science education program, then later revised it to accommodate budget changes suggested by the funder without compromising the viability of the study. We covered how the research questions, design, and analysis plan informed the power analysis approach; the software we used; and what the input parameters required actually represent. Then we discussed how we used both pilot data and relevant literature to choose sensible values for those inputs; the potential impact of varying those inputs; the assumptions we had to make; and how we accounted for likely consent rates and levels of attrition.

We emphasized throughout how the audience members may apply the process to planning their own largescale evaluation studies to increase the ability to document results and improve competitiveness of proposals.

For more information, contact: pierces1@msu.edu