Posted: August 9, 2007

Biostatistics methodology yields an objective Tour de France ranking

The world’s biggest bike race, the Tour de France, is also one of the most statistically confusing events in sports. Consisting of 20 individual timed stages, the race ultimately awards just one winner, though winners of individual stages and of specific intra-race competitions are also recognized with colored jerseys and with a system of points which can be subtracted from riders’ overall times. The complexity of the system — it’s possible to win the overall race without ever winning an individual stage — makes it extremely difficult to rank today’s riders against those who have raced in previous years.

Though a number of such rankings have been used, they each rely on a subjective system which works by assigning a weight to each victory, with the highest weight allocated to more important criteria (such as the yellow jersey that represents the best overall time) and lower weight to less important criteria (such as a stage victory). The problem is that choosing different weights results in different rankings.

By borrowing statistical methods developed for biological problems — such as those used to determine the degree to which a particular substance evokes an immune response or the impact certain side effects have on a patient’s tolerance for a drug — a Rockefeller University biometrician has developed a system of ranking Tour de France competitors that does not assign any particular weight to these criteria.

Working with economists from Catholic University of Leuven and Tilburg University in the Netherlands, Knut M. Wittkowski, director of biostatistics, epidemiology and research design at The Rockefeller University Hospital Center for Clinical and Translational Science, analyzed 54 years’ worth of data using a methodology that assumes certain criteria are more important than others, but does not quantify how much more. It then compares each cyclist with all other cyclists and ranks them accordingly.

The results yield an objective comparison between cyclists that suggests that Lance Armstrong and Bernard Hinault are tied for the top spot even though Armstrong won first place seven times and Hinault won just five. The methodology can also be used to predict how current racers will ultimately compare to lifetime leaders.

μStat Multivariate Statistical Analysis Tool
1953 to 2007 Tour de France Rankings



newswire@rockefeller.edu
Contact: Zach Veilleux (212) 327-8982