Commentary (Muscle Performance and Swimmers‪)‬

I would like to commend the authors on their work. This study to a large extent mirrored the published work by Cools et al, (1) albeit in a different population. Unique to the present work, however, is the investigation of 2 types of training programs on muscle-performance characteristics in adolescent swimmers. Specifically, after isokinetic pretesting of the scapular protractors and retractors, swimmers performed a supervised program of either strength training or endurance training for 12 weeks, followed by a posttest. The authors did not identify differences in muscle performance between groups after the intervention for any of the dependent measures. Furthermore, the authors reported that muscle endurance, as a function of the fatigue index (FI), decreased between the 2 measures. Identifying differences between groups and among multiple interventions is influenced by several factors, including variance among the groups, sample size, and the effect size(s) of the intervention(s). (2) In this study, the authors acknowledge that intersubject variance (large SDs) may have contributed to the inability to detect a difference between groups in isokinetic muscle performance measures after the intervention programs. Although this may be a contributing factor, it is my opinion that the inability to identify group differences was minimally influenced by this factor. Additionally, even though the group sizes were small, larger group sizes in this situation would likely have done very little to increase the chance of identifying a statistical, let alone clinically important, difference between groups. The underlying reason the authors were unable to detect differences between groups was probably a function of the similarities in intervention programs. This statistical information could have been ascertained by performing a prospective pilot study. Armed with the information from an a priori pilot study, the authors could have calculated the sample size necessary to identify a statistical difference or realized that the training programs were so similar that further pursuit of the study was not likely to demonstrate differences. Arguably, there are times when identifying no differences between interventions is clinically important. In these cases, when the investigators expect to find no difference, it becomes even more critical to conduct a power analysis. At this juncture, given that neither a pilot study nor power analysis was performed, it would have been helpful to the reader had the authors reported the actual effect sizes ([[eta].sup.2]).