Sloppy Science: the athletics edition

The September 30 issue of Nature featured a brief communication with a study comparing men’s and women’s sprinting. The authors concluded that by the 2156 Olympics, the winner of the women’s 100 m dash will win with a faster time than the winner of the men’s 100 m dash. They did this by plotting the winning time of the 100 m men’s and women’s gold medallist for each Olympics, assuming that the times for each would continue to decrease linearly. From this analysis, they saw that the times for women are decreasing at a rate faster than those for men, and therefore, assuming that the linear progression will continue, the women will “catch” the men in 2156, winning in a time of just under 8.1 (!!!) seconds.

It seems as though the authors were being only semi-serious with this study, which is good because there’s no way one can take these results seriously. First of all, they chose just one data point for each year, which is hardly an accurate indication of the improvements of sprinters over time. Therefore, the standard deviation in their results is massive, resulting in a 700 year error in their result, i.e. women could overtake men as early as 2064 or as late as 2788. A more precise study, such as this one uses data from the top ten times for every year, rather than one (and not necessarilty the best) time for every four years. For both sexes, the deviations from linearity are clear. These authors did similar studies for shot put and high jump, and found similar progressions.

By taking a proper sample size, anomolies such as FloJo’s sub-10.50 s run in 1988 (which may have been drug-aided) don’t skew the results. I also found it a bit surprising that the significant deviation from linearity began in the early 1980’s for both women and men. It would appear that the advances in medicine and training has benefited both sexes equally despite the imbalance of funding and prominence between the two for all but the last 20-30 years or so.