My Track Record

Back in August, I fielded a query about high-altitude racing from fellow Seattle runner Henry Wigglesworth. Now Henry has another inquiry — a “question for the ages,” as he refers to it in an email. He writes:

If you had to quantify the various components of a runner’s performance on a given day (and assume for the purposes of this query that you must), what value, expressed as a percentage, would you assign to each of the following? (1) natural ability; (2) training (a.k.a. “unnatural or acquired ability”); (3) lifestyle issues (does runner have an infant at home, work a highly stressful job, or is otherwise limited by non-running activities); (4) X factor (usually described as “I just felt good/bad that day, dunno why”). I realize that these categories overlap a bit (especially #2 and #3), and the question is not susceptible to the same scientific rigor that usually distinguishes your analysis, but I am curious nonetheless about what you think.

Henry, as much as I’d like to give you a set of percentages that add up to 100%, I can’t. These percentages depend on so many factors (listed below) that any guess at “average” values would not be especially meaningful, in my opinion.

The answer depends on one’s definitions of “natural ability,” “training,” etc. You correctly note the overlap of training and lifestyle issues, but distinguishing between training and natural talent is just as hard. Not only do our genes influence how fast we can run without any training, they also affect how much exercise we choose to get (Frederiksen & Christensen, Scand J Med Sci Sports 13: 9, 2003) and how our body responds to it (Gagnon et al., Med Sci Sports Exerc 29: 1448, 1997; Larsen et al., Scand J Med Sci Sports 15: 48, 2005). If a person’s genes give him/her greater-than-normal enjoyment of and improvement from training, that advantage represents both talent and training — training-related talent, if you will — confounding any attempts to separate the two.

The answer depends on the type of race. I haven’t seen any definitive data on this point, but personal experience and intuition suggest that performances in longer races are less dependent on raw talent. Similarly, talent should matter somewhat less in events where technique is important, like the steeplechase and technical trail races.

The answer depends on the population of concern. When we talk about race performance, we’re often most interested in individual differences. The reasons for those differences depend on which people are being compared. Consider the example of two longtime friends who’ve worked out together every day for years. Since both are essentially doing the same training, differences in performance could be attributed largely to differences in talent. At the opposite extreme, identical twins presumably have the same amount of talent, so differences in their performance should be due entirely to training and other non-genetic factors.

Given all of the above considerations, it’s not surprising that different studies offer different estimates of the extent to which genes (i.e., talent) affect performance. A review by Rupert (Comp Biochem Physiol A Mol Integr Physiol 136: 191, 2003) shows that the percentage of variation in aerobic performance has been reported to be as low as 35% and as high as 87%, with a rough average of around 50%.

A few attempts have also been made to estimate the percentage of performance variation explained by training experience, frequency, volume, and pace. These estimates probably overestimate the influence of training per se because of the difficulty of disentangling training from talent. (For example, apart from the comments above, training pace depends partly on talent.) In any case, these estimates range from less than 25% to over 80% (Bale et al., Br J Sports Med 20: 170, 1986; Karp, International Journal of Sports Physiology and Performance 2: 72, 2007). In one of these studies, training mileage and years of training predicted about 45% of the variation in the marathon PRs of female U.S. Olympic Trials qualifiers, but no significant fraction of the variation in the males’ PRs (Karp, 2007).

The answer depends upon the individual. The percentages cited above are for variations in performance within a population. This isn’t quite the same thing as explaining the performance of any one particular person. Since variables like maximal oxygen consumption (VO2max) tend to improve least with training in people whose VO2max is already naturally high (Saltin et al., Circulation 39: VII1, 1968; Wenger & Macnab, J Sports Med Phys Fitness 15:199, 1975), we can assume that some runners are fast “mostly” due to natural talent. On the other hand, there are those who are extremely slow if remaining sedentary, but extremely responsive to training; in them, the influence of talent alone would appear to be much smaller.

The answer depends on the day. Most of this essay has focused on talent and training because, in general, those are the two most important determinants of performance. The impact of a lifetime of training is quite large, for example, compared to the 1-6% improvement that one can get from an optimal taper (Mujika & Padilla, Med Sci Sports Exerc 35: 1182, 2003). But here too there are exceptions. If you stubbornly insist on starting a race despite being sick enough to belong in a hospital bed, your performance that day might reflect the severity of your illness more than anything else, talent and training included.

Thus the short answer to the original question is: it depends. I’m sorry the truth is so complicated, Henry.