Simple ways to make the results of exercise science studies more informative

Abstract

Objectives: To demonstrate some alternate ways of presenting and analyzing pretest-posttest control group designs relative to what is commonly done in exercise science. An emphasis is placed on using simple examples and avoiding statistical jargon to enhance readability for exercise scientists. Design & Methods: To examine some concerns with how within subject figures illustrate data, statistics to interpret when ana lyzing pretest-posttest control groups designs, how to analyze studies involving three time points or those including a third factor, and values to use when testing assumptions of statistical tests. Results & Conclusions: To improve interpretation of data, researchers assessing pretest-posttest control group designs should report the change score and variability of the change score as opposed to only reporting pre-test and post-test variabilities. When performing a 2 × 2 (group by time) mixed ANOVA the interaction term is the only statistic that needs to be interpreted and no follow-up tests are necessary. When assessing a third time point, the most informative follow-up tests to a significant 3 × 2 (time by group) ANOVA involves performing all three 2 × 2 (time by group) interactions to keep the within subject nature of the data. When including a third factor (in addition to the time and group variables), researchers may wish to compute change scores to eliminate the factor of time and allow for the change to be directly assessed. When examining the assumptions of normality and homogeneity of variance, it is important that the change scores meet the assumptions as opposed to the pre-test and post-test measures.