[sdenton4]

The most intriguing 'graf in the article for me was this:

'After scouring the results of nearly 100,000 marathon finishers, Sandra Hunter, a professor of exercise science at Wisconsin’s Marquette University, made an interesting—if not intuitive—find. The more men there are in a race relative to the amount of women, the bigger the performance gap between those genders. “If you had one female for every twenty men, the likelihood that that female is going to be the best . . . compared with the best male in that age group is pretty small,” says Hunter.'

Which suggests it has a lot more to do with the statistics of outliers than anything else. Elite performance is signal (training!) plus noise (daily variation, environmental variance, Athena rooting for you, etc), which can overwhelm the signal on any given day. But each participant is also a random draw on the /signal/ variable as well. Get more people in the event, and you get more draws on the signal variable. Get more people on race day, and you get more chances for outliers on the noise variable.

[geargrinder]

I'm not exactly sure what you are saying. Elites are outliers by definition. Elite runners are mostly elite because of genetics. Then they train hard to realize their potential. 99% could never become elite in traditional running events, no matter how much, how hard or how smart they train. The top 1/10 of 1% of the general population is more likely the definition of elite in running.

[sdenton4]

Outlet statistics work differently from measurement of means. The distribution of the largest draw from a collection of draws from a normal distribution depends heavily on the size of the sampled population.

Consider each runner's skill, training, etc as a sampled variable. Then the top score in the sample depends heavily on the population size. Comparing the best draw from two equivalent groups of different sizes is thus going to favor the larger group. And this sounds like what they observed in the study.