RESPONSE

Abstract

Dear Editor-in-Chief: We interpret Williams’ original demonstration (3) to mean that observed reductions in mortality risk due to changes in fitness (1) might be, rather than must be, an artifact of measurement error. We subsequently showed that measurement error alone does not explain all changes in fitness (2), casting doubt on the premise of his demonstration. In his letter to the Editor-in-Chief, Williams now presents an alternative hypothesis that is based on the premise that fitness changes do affect mortality risk. This alternative hypothesis is purported to substantiate his earlier hypothesis. Yet Williams notes there is little power to distinguish the two hypotheses, a point with which we agree. Whatever the merits of his new alternative hypothesis, we remain persuaded that while it is conceivable that measurement error might explain our observations, they are unlikely to do so. The published mean treadmill time difference of the Aerobics Center Longitudinal Study men who went from the unfit-to-fit category was 4.23 min, while the mean for those who went from the fit-to-unfit category was −4.87 min (2). The 9.10-min mean difference between these two groups, which were the focus of the Williams paper, is substantial from both physiological and statistical perspectives. Andrew S. Jackson, P.E.D. Department of Health and Human Performance, University of Houston, Houston, TX James B. Kampert, Ph.D. Carolyn E. Barlow, M.S. Centers of Integrated Health Research, The Cooper Institute, Dallas, TX James R. Morrow, Ph.D. Department of Kinesiology, Health Promotion, and Recreation, University of North Texas, Denton, TX Timothy S. Church, M.D., Ph.D. Steven N. Blair, P.E.D. Centers of Integrated Health Research, The Cooper Institute, Dallas, TX