Age, race times, and statistics.

Insights into road racing can turn up in unexpected places, even in a technical journal such as the American Statistician (1998, Vol. 52, No. 3, pp. 205-210). A recent article used a branch of statistics known as extreme-value theory to model the relationship between age and finishing time among the top five finishers of each single age in the citizens' division of the 1995 Bolder Boulder 10K.

The model estimates that the fastest times tend to occur at about age 28 for both men and women, that 20-year-olds are just about as fast as 40-year-olds, and that the best runners in their sixties were comparable to the rising stars in their early teens. (That's a relationship that might make older runners grin with pride.) Among runners over the age of 50, each additional year of age was associated with another 30 to 40 seconds in the 10K.

This statistical model does not predict how any individual's performance will change over time. Runners not yet close to their individual potential are apt to find that the effects of training outweigh those of age. Even as you age, you might decrease rather than increase race times. What's more, the life experience--training and racing history--of today's 60-year-olds may be quite different from what present-day 30-year-olds will have undergone by 2028. Increases in the number of competitive runners who stay in the sport may lead to sharp improvements in winning times for older age groups. Meanwhile, let it provide some satisfaction that 60-year-old runners share speed and performance potential with teens, and 40-year-olds with those half their age.

(AR&FA Member David Kenny is a statistician at the Biostatistics Center of The George Washington University, and hopes to break a three hour marathon this year.)