Statistical power analysis with microsoft excel: normal tests for one or two means as a prelude to using non-central distributions to calculate power

Abstract

This article presents statistical power analysis (SPA) based on the normal distribution using Excel, adopting textbook and SPA approaches. The objective is to present the latter in a comparative way within a framework that is familiar to textbook level readers, as a first step to understand SPA with other distributions. The analysis focuses on the case of the equality of the means of two populations with equal variances for independent samples with the same size. This is the situation adopted as case 0 by Cohen (1988), a pioneer in the subject, to develop his set of tables and so, the present article can be seen as an introduction to Cohen's methodology applied to tests based on samples from normal populations. We also discuss how to extend the calculation to cases with other characteristics (cases 1 to 4), similarly to what Cohen proposes, as well as a brief discussion about the advantages and shortcomings of Excel. We teach mainly in the area of business and economics, which determines the scope of our analysis.