This post is inspired by a work on mine in collaboration with my university colleague Y. Charon, a work we published in New Studies in Athletics a few years back: Comparing the best athletic performances of the two sexes, B. Grammaticos, Y. Charon, New Stud. Athl. 29:4 (2014) 37. The main aim of the article was to introduce a proper method for the comparison of he performances of men and women and the first question one had to answer was "what is the reference performance to be used in the comparison?". It became soon clear that the current world record was far from optimal as reference due to various factors (a combination of exceptional circumstances and/or talent, lower standards in the application of doping control etc.). Among the various analyses presented in that article was one based on the work of Dupuy who claimed that the women’s/men’s world record ratio follows an S-shape curve over time. In his article Dupuy insists on the fact that the S-shaped curve is typical for what he calls the "gender world record ratio" and it is the result of the late authorisation of women to participate in these disciplines at major events. He then goes on to present a simple economic (!) model in order to explain the mechanism.
In our article, while suspecting the premises of Dupuy, Y. Charon and myself refrained from criticising it overtly, since it was peripheral to the theme of our article. Still I was feeling that something was wrong already with the basic assumption and when I was thinking about what became the main topic of the present article I decided to go back to examine that article with a fresh eye. Reading the introduction of Dupuy one encounters a reference to previous studies that have shown that the evolution of world records in athletics are better fitted with S-shaped curves. Dupuy brushes away this fact by maintaining that the origin of this is biological (!) while the S-shape in the "gender world record ratio" has an economic explanation. He then proceeds to show an example of supposedly S-shaped evolution.
Unless what he means by "S-shaped" is not what all scientists mean by S-shaped, his curves are not S-shaped at all. All of them present a dip after the initial plateau. The explanation for this is elementary. What is happening is that both men and women world records follow an S-shaped progression. Simply for the men's record the rapid progression occurs earlier than the one for the women's record. If one takes the ratio of two S-shaped curves shifted in time one obtains a, more or less pronounced, dip after the initial plateau. This is what Dupuy is observing. So, the data of Dupuy himself support the thesis that the world record progression follows an S-shaped curve.
After this very long introduction I come now to the main theme of this article, namely the S-shaped world record evolution. (In the article by Charon and myself we argue that the same pattern is present when one considers some more representative performance, like that of the 50th or 100th performer of the year, but since data at this depth are totally absent for the long periods we consider, we must make do with the only extant data, i.e. world records). What is the origin of the S shape? It is not at all "biological", as Dupuy claims, but rather "sociological". When a discipline is not official it attracts only a very small number of athletes, the others prefer to devote their efforts to disciplines where they can reap the fruits thereof. Once the discipline becomes official (or, already when it is on the verge of becoming it) more and more talented athletes are adopting it leading to a rapid progress. With maturation of the discipline comes a gradual saturation of performances and the progress becomes incremental.
In order to illustrate my point I chose two women's disciplines which were introduced rather recently and for which records before the official introduction date did exist: triple jump and pole vault. Here is the evolution of the triple jump:
The S-shape in this case shows a pronounced saturation which is due to the fact that the record has not evolved since 1995. In the case of pole vault the situation is similar:
Again the world record did not evolve over the last 10 years but this span is much smaller than that of the triple jump. And anyhow the pronounced saturation is present only in the case of the world record. As I showed in my post on triple jump, the in-depth progression is not arrested although one can detect some saturation tendencies. The same holds true in the case of pole vault.
The idea of this article came form some recent bad news. The Olympic champion and former pole vault world record holder Don Bragg passed away on February 16th. In the short article which appeared in the IAAF site there was also a mention of his sister Diane who held the unofficial world record in women's pole vault with 2.59 m, established in 1952. The IAAF site erroneously mentions that this record survived for 17 years. However the record was only equalled in 1969 and was not broken until 1978. I wrote about Diane Bragg in my article on women's pole vault but I did not, at the time, relate the record evolution with the S-shape prediction.
In our article, while suspecting the premises of Dupuy, Y. Charon and myself refrained from criticising it overtly, since it was peripheral to the theme of our article. Still I was feeling that something was wrong already with the basic assumption and when I was thinking about what became the main topic of the present article I decided to go back to examine that article with a fresh eye. Reading the introduction of Dupuy one encounters a reference to previous studies that have shown that the evolution of world records in athletics are better fitted with S-shaped curves. Dupuy brushes away this fact by maintaining that the origin of this is biological (!) while the S-shape in the "gender world record ratio" has an economic explanation. He then proceeds to show an example of supposedly S-shaped evolution.
Unless what he means by "S-shaped" is not what all scientists mean by S-shaped, his curves are not S-shaped at all. All of them present a dip after the initial plateau. The explanation for this is elementary. What is happening is that both men and women world records follow an S-shaped progression. Simply for the men's record the rapid progression occurs earlier than the one for the women's record. If one takes the ratio of two S-shaped curves shifted in time one obtains a, more or less pronounced, dip after the initial plateau. This is what Dupuy is observing. So, the data of Dupuy himself support the thesis that the world record progression follows an S-shaped curve.
After this very long introduction I come now to the main theme of this article, namely the S-shaped world record evolution. (In the article by Charon and myself we argue that the same pattern is present when one considers some more representative performance, like that of the 50th or 100th performer of the year, but since data at this depth are totally absent for the long periods we consider, we must make do with the only extant data, i.e. world records). What is the origin of the S shape? It is not at all "biological", as Dupuy claims, but rather "sociological". When a discipline is not official it attracts only a very small number of athletes, the others prefer to devote their efforts to disciplines where they can reap the fruits thereof. Once the discipline becomes official (or, already when it is on the verge of becoming it) more and more talented athletes are adopting it leading to a rapid progress. With maturation of the discipline comes a gradual saturation of performances and the progress becomes incremental.
In order to illustrate my point I chose two women's disciplines which were introduced rather recently and for which records before the official introduction date did exist: triple jump and pole vault. Here is the evolution of the triple jump:
The S-shape in this case shows a pronounced saturation which is due to the fact that the record has not evolved since 1995. In the case of pole vault the situation is similar:
Again the world record did not evolve over the last 10 years but this span is much smaller than that of the triple jump. And anyhow the pronounced saturation is present only in the case of the world record. As I showed in my post on triple jump, the in-depth progression is not arrested although one can detect some saturation tendencies. The same holds true in the case of pole vault.
The idea of this article came form some recent bad news. The Olympic champion and former pole vault world record holder Don Bragg passed away on February 16th. In the short article which appeared in the IAAF site there was also a mention of his sister Diane who held the unofficial world record in women's pole vault with 2.59 m, established in 1952. The IAAF site erroneously mentions that this record survived for 17 years. However the record was only equalled in 1969 and was not broken until 1978. I wrote about Diane Bragg in my article on women's pole vault but I did not, at the time, relate the record evolution with the S-shape prediction.