nMAX: Practical Sample Size Planning for Multiple Regression, Measured Variable Path Models, and Beyond (Poster 8)

• In Event: Innovations in Statistical Power, Intervention Planning and Design, and Causal Effects

Abstract

Planning studies with sufficient statistical power is critical for ensuring statistical and inferential integrity. However, given various practical constraints, a priori power analysis has become an exercise in self-delusion to convince oneself (and granting agencies) that a low sample size will still provide enough power. Even the most sincere attempt at sample size planning is fraught with the fundamental challenge of setting values for not only the focal parameters for which statistical tests are planned, but also each of the peripheral parameters. Acknowledging the practical challenges in conventional power analysis, the current methodological innovation work presents a new framework that aims to (1) restore appropriate conservatism and robustness to the sample size planning process, and (2) greatly simplify that process.

Authors

• Gregory R. Hancock, University of Maryland

• Yi Feng, University of California - Los Angeles