Olympic Research Update by the Counsilman Center

For those of you who are interested.. in statistics, we can make an educated guess (of an unknown mean) with a certain amount of confidence based on our sample data. A 95% confidence interval is a range of values (plus or minus 2 standard deviations) between which we are 95% confident the actual mean will lie. So, if an event is more than 2 standard deviations faster (or slower) than our predicted mean, then we can say there is a significant difference between the actual and predicted means. Of course, there is always chance for error and, on occasion, we conclude that the two means are different when in reality they are not.

Our predicted means are based on the previous 7 Olympics Games and, in many cases, are very strong. That is, our ability to predict the mean of the final heat in many swimming events is strengthened by the fact that we can account for (in many cases) up to 90% of the variance. What this means is, although the rate of improvement of the top 8 swimmers in each event over time is not linear, it is predictable. Theoretically, if there are a number of events in the 2008 Olympic Games that fall outside of plus or minus 2 standard deviations of our predicted mean, then there may be some outside variable(s) working to alter swim performance. Without proper scientific testing, our explanation of any extraordinary change in performance is speculative at best.

So far, with 10 events remaining (5 men, 5 women), we have successfully predicted the mean of the top 8 finalists in two male and 4 female events. Our predictions were more than two standard deviations slower than the actual top 8 mean in 6 men’s and 4 women’s events and, in many cases, were so much faster than our confidence interval that I nearly choked on a chocolate covered cherry (good for recovery)! Compare that with our failed predictions in 2000 (1 men’s event) and 2004 (2 men’s events) and one cannot help but ponder possible explanations for these phenomenal performances. Further, virtually all of the 2008 finalists are below the lower confidence interval. For example, in the men’s 200 fly and 100 freestyle, seven of the eight finalists were faster than the fastest time that would allow for our predictions to be successful. Our failed predictions are not the result of one or two guys being so far ahead of the field, rather across the board everyone is faster.

Along with our predicted means, we have also listed the athletes who had previously swum within out confidence interval. It is reasonable to assume that these athletes have a chance to final and perhaps win. Of course it is possible to have a ‘dark horse’ bust out of the cornfields and win a medal, but if there are many instances when a swimmer ‘off the radar’ wins gold… well we’ll leave that to the scientists. Oh, right… this may be further evidence supporting a substantial shift in the progression of swim performance over time. Of the events so far, Park (South Korea) was not previously within our predicted range, but won gold in the 400 FR and silver in the 200 FR. Also, the women’s 400 FR, 200 FR, and 200 Fly had medalists who were not previously within our predictions.

Finally, to provide further evidence that a bias has been introduced into competitive swim performance, the number of world record performances in the last two years has been much greater than we predicted based on previous years. Averaged over the last 30 years, the month of August has had about 4 world records. So far, 18 world records have dropped in the 2008 Olympic Games. While it is true that previous Olympics have had as many, if not more, world record performances, the suggested bias in 2008 is supported by the fact that that the rate of change (as a function of simply the number of WR occurrences) appears to be slowing at a relatively stable rate within the last 24 years or so. That is, much like the rate of improvement in any given event is declining as our sport matures, the number of world record performances has been getting less and less over time. Until recently. The cause of this temporary irregularity will be speculative, at best, unless reinforced by historical confirmations. Please see the following link fyi.
http://research.indiana.edu/news/stories/0220.html

Regards,
Chris Brammer
Counsilman Center for the Science of Swimming