|
Conscientious sample-size analyses are often the cornerstones of effective study planning. Besides promoting wiser allocation of resources and stronger research ethics, the process itself requires the research team (including the statistician!) to delineate, critique, and tighten the specific aims, research questions, design, measurements, and analysis plans. Critically, the members must agree on reasonable conjectures for the "infinite datasets" they envision sampling from. Today's ever-improving methods and software enable us to assess how various sample sizes will support key analyses under various plausible scenarios. This one-day course covers the scientific/statistical principles and general consulting pragmatics for determining/justifying sample sizes for a range of standard hypothesis tests, equivalence tests, and confidence intervals. Although these analyses are frequentist, we show how basic Bayesian concepts help to clarify and shape the planning process for investigators, statisticians, and peer reviewers. Realistic case studies and scientific issues are stressed far more than mathematical and computational underpinnings. While this is not a "SAS training course," SAS 9 procedures and freeware are used in all examples. The typical attendee has had limited experience conducting sample size analyses, but even veterans in this usually encounter useful new ideas, methods, and strategies.
|
= Applied Session,
= Theme Session,
= Presenter