3 Statistical Power Analysis
Research Methods in Psychology
I. FOUNDATIONS OF RESEARCH ISSUES
Published Online: 26 SEP 2012
Copyright © 2013 John Wiley & Sons, Inc. All rights reserved.
Handbook of Psychology, Second Edition
How to Cite
Rossi, J. S. 2012. Statistical Power Analysis. Handbook of Psychology, Second Edition. 2:I:3.
- Published Online: 26 SEP 2012
Statistical power is the probability that a statistical test will correctly reject the null hypothesis. This chapter presents the concept of statistical power and several closely related techniques and topics. There have been numerous calls in recent years for the increased use of statistical power analysis and other techniques as alternatives to the traditional practice of null hypothesis significance testing and the use of p values as the principal method of establishing the evidence for an effect. Detailed discussions will be provided on the three main study design factors that determine power, sample size, effect size, and alpha level, as well as on several additional important design factors, including covariates, repeated measures, unequal group sample sizes, the number of groups or predictors in the analysis, and within-group dependency in cluster randomized designs. Very brief coverage along with suggested readings for more information is provided for a number of other issues, including missing data, multiple dependent variables and multivariate analytical techniques, structural equation modeling, meta-analysis, and accuracy in parameter estimation. The problems associated with the use of “rules of thumb” for sample size estimation in multiple regression analysis are discussed as well as the use of power in the interpretation of nonsignificant results, including a critical discussion of the concept of retrospective power analysis. Methods for calculating power are covered, but the primary emphasis is on examining how various aspects of study design can be manipulated to enhance the power of a study.
- statistical power;
- effect size;
- sample size;
- null hypothesis significance testing paradigm;
- Type I and Type II errors;
- retrospective power