Statistical Power Analysis for Multiple Regression/Correlation: A Computer Program

Publisher Website
Google Scholar
Abstract
A micro-computer program has been developed which enables researchers to compute statistical power for analyses performed by multiple regression/correlation (MRC). The program's structure mirrors the logical application of MRC as a data analytic system, to wit: It allows the user to group variables into sets which will be employed as predictors; it allows the user to determine power for the cumulative impact of any set(s) of variables, and also for the unique impact of any set over and above other sets. The program features a spreadsheet-like interface; the user enters the number of variables included in each set, the proportion of variance hypothesized to be explained by that set, the sample size, alpha, and whether Model I error or Model 2 error will be used in the analyses. The program then reports the effect size (f) and the value of power corresponding to these parameters.

This publication has 1 reference indexed in Scilit: