Sample size determination using appletsView Presentation *Russell V. Lenth, University of Iowa
Keywords: power, sample size, practical significance
In order for a study to be successful, it must be adequately powered so that an effect large enough to be scientifically relevant is likely to be detected once the data are collected. We will discuss how to use some interactive sample-size applets developed by the author, in the context of two scenarios: a two-sample comparison, and a split-plot experiment. Discussion will include the importance of pilot data, the equal importance of independent scientific input, and how to use that information to determine a reasonable sample size.