[Human Performance 
Labs and The Counsilman Center for Swimming Science]

Needless to say, there has been considerable concern by swimmers, coaches, parents, and the governing bodies of swimming, as to the effects of the new “fastsuits” upon swim performance. While legality is a complex problem for the sport, it may very well be that the performance enhancing aspect to this question may be answered by the individual athlete. Another approach, however, is to look at the past performances of American athletes at the U.S. Olympic Trials, arguably the most important meet of any athletes career, second only to the Olympic Games.

By using the performances of the top 16 or top eight athletes from as far back as 1968, an “expected” or “forecasted time” can be calculated with surprisingly close accuracy. This assumes that nothing has changed over time within the various events, such as rule changes etc. As it turns out, swim times do not always improve to the same extent year after year. As a sport becomes mature, the yearly improvements get smaller and thus the line describing the improvements is not linear. Thus, a second relationship can better describe how times improve, namely a power curve or one whose slope decreases as time progresses. The linear and power curve generated by using all of the data back to the late sixties is found in the files referred to as comprehensive forecasts.

Another approach which seems to be even more accurate is based upon the fact that recent times (the last 12 years or so) are likely to be more accurate as a means to forecast current results than are the older results. These recent forecasts are less likely to overshoot or undershoot the actual times and have thus been used as our primary means to compare the current Olympic Trial results.

Finally, when all of the events have been completed in the 2000 US Olympic Trials, a tally can be made of those events in which the forecasts were not met or rather were exceeded. This requires statistical analysis and will result in a generalizable conclusion. In this case, whether or not the new suits had any measurable effect upon the performances of the swimmers as a sample of the larger competitive swim population.

During the trials, attention has been paid to all swimmers making the semi finals and finals in terms of recording which suit the swimmer wore and which manufacturer supplied it. A further complex analysis will be performed against seed times and resultant performances as a means to verify the general conclusion.

Similar steps will be taken at the 2000 Olympic Games, given that the researchers have access to Australia and the swimming venue.

Objective: To provide accurate forecasts for the swim times in each event at the 2000 Olympic Trials.

Rationale: Based upon the performances of United States athletes at the Olympic Swimming Trials between the years 1968 through 1996, statistical analyses allow prediction of the times for each event at the 2000 Trials.

Assumption: That no significant event or cataclysmic change has occurred within the sport which has effected how swimmers train or how swimmers perform in high level competition.

Hypothesis: It is hypothesized that the forecast models may fail as a function of the swim apparel worn in the 2000 trials. Fabrics and apparel surface area may influence swim times if anecdotal and manufacturers claims are born out.

Procedures: To examine questions of the improvement of swimming performance over the years, we have analyzed the times for the eight finalists for each swimming event of the 1968 Olympic Trials and each of the trials held since, including those of 1996. This data has been subjected to several statistical models with the purpose of predicting the average of the times that will be swum in the current competition.

Improvement in performance has characterized swimming over this period. To better understand this, we first plotted the average (mean) times by event for each of the trials. These plots support the general conclusion that improvement has been occurring in each event, but that the improvement has not necessarily been linear, that is in a straight line. The data tend to form a shallow curve, with improvement from trial to trial being greater during the earliest years (1968 to 1972) than in most recent years.

Both linear and nonlinear (power) analyses were run to gain an insight as to the model that might best predict the current performances. Since the times during the earlier years were considerably slower than in more recent trials, another analysis focused only on those more recent trial times, ones with similar in average time to that of the 1996 event.

A series of graphs representing the data and the forecasted times form the various analytic procedures follows.

Go back to the 2000 Trials Website
.


The Human Performance Labs are part of the
Kinesiology Department in the School of HPER.

URL: http://www.indiana.edu/~kines/
Comments: stagerj@indiana.edu
Copyright 2000, The Trustees of Indiana University
Original Web Design by Dakin Burdick
Web Maintenance and Redesign by E. Jeannette Silvers
Last updated: 9 Feb. 2001