A third issue was our use of a random-effects model. The group has suggested that a fixed-effects model is preferred; however, a fixed-effects model assumes a common treatment effect in each study, and the variation in the observed study estimates is only because of chance. 6 Under the randomeffects model, the true effects in the studies (in which we have professional, college, and high school football players) are assumed to vary, and the summary effect is the weighed effect average of the effects reported in these different studies. 6 Since there is heterogeneity between the studies, a random-effects model is the better option, even if a more conservative estimate of the results are presented. In this case, the I 2 is approximately 85% (high heterogeneity), meaning that approximately 85% of the variability in the results is because of real study differences, and only 15% is because of chance. 7 What concerns us about this group's stance are 3 issues. First, regardless of statistical significance, there is minimal practical difference in concussion rates between mild altitudes and sea level. The risk difference is less than 1 concussion per 1000 athlete-exposures, and we recently demonstrated that even arbitrary "predictors" (ie, team logo) can appear to have similarly significant effects. 3 Second, there is no physiological basis for these altitudes to influence concussion, so debate over these altitudes should not even exist. 8 Finally, as the epidemiological and physiological data supporting their altitude hypothesis are scrutinized, it is surprising that they now downplay their relevance. The altitude hypothesis, rooted in epidemiology, was a driving influence in developing their anticoncussion jugular compression device, which was designed to mimic the supposed protective benefits of altitude exposure.