Power Analyses for Detecting Effects for Multiple Coefficients in Regression

Abstract

I present applications of Stata to conduct power analysis for multiple regression. Simple power analyses are provided through the Stata command power and the user-written command powerreg. However, tools for power analysis for multiple regression coefficients are limited. I present strategies for detecting significant regression coefficients in a single study (for example, power estimates for two or more coefficients). I also address the power for detecting multiple effects (that is, the probability that all coefficients are statistically significant) in the same sample through the Monte Carlo approaches afforded by the simulate command.