A post hoc statistical power analysis and survey of the research published in the Journal of Athletic Training.

No Thumbnail Available
Date
2000
Authors
Tener, Mark
Journal Title
Journal ISSN
Volume Title
Publisher
Middle Tennessee State University
Abstract
Statistical power is the probability of finding a true effect in the population under investigation. When a true effect is found, the researcher rejects the null hypothesis and accepts the research hypothesis. Therefore, statistical power is the probability of rejecting the null hypothesis when it is false.
The purpose of this study was to perform a post hoc power analysis of published athletic training research. A sample of articles was selected from the three most recently completed volumes of the Journal of Athletic Training (vols. 32, 33, and 34) to answer the proposed research questions. The information collected from the articles concerned the use of an a priori power analysis, the report of observed power, the report of observed effect size, the reported level of significance, and the report of nonsignificant results. Specific articles were designated for a post hoc power analysis. Using well-known power tables, the statistical power for each article was calculated assuming the authors attempted to detect small, medium, and large effect sizes.
Of the 77 articles surveyed, two articles contained a report of using an a priori power analysis. Four articles contained a report of observed power. None of the articles contained a report of observed effect size. Forty-seven articles contained a report of nonsignificant results, yet only nine of these articles contained a report of insufficient power as a possible reason. Thirty-six articles were used for the post hoc power analysis. The mean powers were .18, .53, and .75 for the assumed small, medium, and large effect sizes, respectively.
It was concluded that athletic training researchers had a poor chance of finding true effects and rejecting the null hypothesis, unless the researchers attempted to detect large effects. Recommendations included: (1) requiring certain information be reported in research articles; (2) reporting clear, well-understood results statements; (3) researching the possible reasons for the neglect of power among athletic training researchers; and (4) having editors and reviewers of the research help to change article submission policies.
Description
Adviser: Jon MacBeth.
Keywords
Citation