Monday 9 April 2012

The Tely 10: A case study

I have heard it said, more than once, that if Canada had more people participating in running we would find more talent. The idea -in theory at least- is simple: increasing the number of participants should increase the talent pool. Even if there isn't a perfect correspondence we might expect some degree of correlation. I here use the Tely 10, a 10 mile road race, as a test case for the claim that participation is correlated with quality results. I will first outline the race's history, which comes into play in this analysis.

The first Tely 10 was held in 1922 and except for the WWII era (1939-1945) has been held every year since. The race is a downhill, point-to-point course, and has remained mostly unchanged (with just a few exceptions). The race is located in the city St. John's on the island of Newfoundland. The location is remote; the next-largest city will take you over 20 hours of driving to reach. Because of this remoteness, in-province talent prevails almost every year (multi-year local winners include current reigning champion Colin Fewer, Olympian Paul McCloy, and Harold St. Croix). Judging from the 2011 results I estimate at least 95% of the competitive talent is from the island. This race seems immune to traveling racers. With first prize being 750$, the potential money is comparable to the cost of travel. Furthermore Newfoundland's population has remained stable for the past 40 years; from 1971 onwards it has been oscillating between 510,000-580,000.

Population of Newfoundland for 60 years, between 1951-2011, as recorded by Statistics Canada 
What I want to show now is that the number of quality Tely 10 runners has been largely independent of the total participation. I define 'quality' as men who can run under one hour (I am neglecting the women's half of the race, which began its competitive era in the 1970s). This 1:00 cutoff is arbitrary but convenient as some early years only recorded times for those who ran under one hour. I acquired data using Joe Ryan's book The Tely 10: A History of Newfoundland's Premier Road Race (2002), the NLAA website, and the Tely 10 page itself. St John's weather is notoriously unpredictable, however the race has been traditionally held in July (early editions ranged from May to October). Temperatures for the race are usually mild. Wind and rain are often part of the race, but from my own five racing experiences I would deem the weather conditions 'fair' and the air quality always good.

Here is a stacked plot of runners over and under the 1 hour mark for the years 1922 to 1977. For example in the year 1977 there were 29 finishers: 12 under the 1:00 mark and 17 above.


In the above plot we see that both in annual participation and sub one-hour performances the numbers are relatively stable. Hence the first 50 years of the Tely 10 (1922 - 1977) involved only a dozen or so men, of which about 2-10 would run a fast (sub one hour) race time. Though the race was popular to watch (or listen via radio) the numbers remained low. This was perhaps due to the perceived danger in distance running; participants were required to have an on-site doctor's approval before racing and the race itself was not closed to traffic. Combining these early years the correlation between number of participants and sub one-hour performances was somewhat positively related:


In 1978 participation reached stratospheric levels. The growth is obvious; when I plot the racing years 1978 to 2011 the number of men (and women too) entering the race went from two dozen to well over a thousand (there were 1197 male finishers in 2011):


The sub one-hour runners are difficult to view because of this growth so here is a log plot of the same graph:


Notice that the blue column, i.e. sub 1 hour performers, remains relatively stable during the post jogging boom despite the enormous growth. The correlation between total participation and quality results is quite weak; a tenfold increase in numbers did not result in any more competitive race finishers, and actually fewer in some of the years listed:


Inevitably some degree of correlation exists between total participation and elite performance. With the late 1970s running boom came a surge of more sub 1-hour competitors and faster winning times:

Sub 1:00 performances (blue bars) in the Tely 10 for all years held, 1922-2011 plus winning times (red dots)
Why was there a jump in the attendance after 1977? There is no single reason for the Tely's sudden rise in popularity. We can rule out prize money (introduced in 2008), Newfoundland's population growth (relatively constant), or out-of-province attendance (relatively small). One explanation is that jogging became widespread in North America around the time of Jim Fixx's The Complete Book of Running (1977), though here the cause and effect is difficult to separate. The psychology of running changed: after the boom running was no longer an elite endeavor, was now even seen as healthy; races such as the Tely were re-interpreted as all-comer events. Many mass-participant races, including the Ottawa and New York City Marathon, began in the 1970s. I use the phrase "active spectators" to describe the surge of popularity of racing among the non-competitive, though competitive runners also increased in their numbers.

Despite the ever-growing popularity of the Tely 10, the year with the most sub 1:00 runners was 1983 (with 33 dipping under the 1 hour mark) and from a total of only 163 male participants. In the early 1990s the number of participants increased dramatically into the hundreds but the number of fast runners actually went down. The fastest time ever run for the ten mile course was by Paul McCloy in 1985 with a world-class finish of 47:04 (at the relatively young age of 21). Overall winning times have not improved since the 1980s (staying in the 50-52 minute range). By contrast the competitive aspect of the race reached a low point in 2000 with only seven sub 1:00 runners out of a field of 587 (though also an especially warm year). Recently the number of competitive runners has almost -but not quite- reached the quantity of quality seen in the 1980s. This is perhaps reminiscent of the Canadian men's marathon performances during the same interval: the marathon boom in the 1980s was followed by a lull in the 1990s, then a resurgence in the late 2000s. What could be the underlying connection, if any?

One correlation I found to hold was between the winning time and number of sub 1:00 performances, though a plateau is apparently reached beyond ten or so runners. More 'fast' runners often makes for better winning times, but only up to a point.


What has transpired during the last 35 years of the Tely 10 is more complex than what any one graph could show. For instance the 80s competitive boom was largely but not entirely supported by young runners. It could be argued that the novelty of running among many baby boomers was a contributing factor. Total participation in the Tely further climbed in the 1990s yet the number of quality performances dropped. Perhaps health aspects of running were oversold and young eager talent looked elsewhere for more rebellious or novel sports. Or perhaps young runners in the 1990s were racing too early. These causal explanations oppose one another yet claim the same end result. The latest years of the Tely are showing an improvement. Is it due to the addition of prize money or will local performances decline again in future years?

Update: Below we can see the percentage of younger, (potentially) competitive runners, i.e. 20-39, has remained steady at 30-40% of the total male participation during the past 15 years. Age categories for earlier years are unavailable.

[End of update]

Some combination of both popularity and competitiveness is required to achieve quality achievement in any sport, yet neither is perfectly correlated. This idea helps to explain why a relatively small population can achieve a world-class level of talent. The complex relationship between sub 1:00 performances and total runners implies increasing total participation may not in fact yield better results, and in some cases is negatively correlated. If Kenya and/or Ethiopia doubled their participation numbers would their performances on the world stage improve, or might they remain unchanged or even falter? There are few guarantees. I hope these graphs guide the curious reader towards the complex realities in identifying factors related to elite performance, especially when trying to predict quality results by participation numbers alone.



Bonus plot:

I didn't want to neglect the women's half of the race entirely. Below you can see how women, who did not participate in the Tely 10 prior to 1969, went from a single competitor that year to an overall dominant status:

Just like with the men's side the women's best time was posted during the 1980s: Nicola Will ran ten miles in 55:47 in 1986. Only one other competitor, Kate Vaughan in 2010, has come within a minute of this time (56:36).