The final article in our series on data and its interpretation is a short piece entitled "Momentous Sprint at the 2156 Olympics?". I have included 3 files here: the actual article by Tatem et al. (2004), the data file, and a note on the methods used by the authors. The question that I would pose for your consideration is...do you agree with the authors' assertion that the evidence suggests that the gap between male and female Olympic sprinters is getting smaller and may disappear within the next 40 or 50 years? Can you critically analyze the statistical methods used by the authors and offer some alternative interpretation of the data?
Wednesday, February 25, 2009
Momentous Sprint
The final article in our series on data and its interpretation is a short piece entitled "Momentous Sprint at the 2156 Olympics?". I have included 3 files here: the actual article by Tatem et al. (2004), the data file, and a note on the methods used by the authors. The question that I would pose for your consideration is...do you agree with the authors' assertion that the evidence suggests that the gap between male and female Olympic sprinters is getting smaller and may disappear within the next 40 or 50 years? Can you critically analyze the statistical methods used by the authors and offer some alternative interpretation of the data?
Subscribe to:
Post Comments (Atom)
I for one as not convinced of the logic that the scientists use. It would seem logical to assume that as women's times get faster so will the men's. Secondly, as drug testing gets more stringent and severely enforced I am not sure doping will present the huge problem that it is today. And some clarification please. 10.57 +- .232 is going to be slower than the men's time of 9.73+-.144 right?? Yet the scientists attest that the women's time will eclipse the men's.
ReplyDeleteIan stated that "It would seem logical to assume that as women's times get faster so will the men's" but look at the slopes of the men's adn teh women's time and notice that they cross, suggesting faster timers for women in the future.
ReplyDeleteIan also said "And some clarification please. 10.57 +- .232 is going to be slower than the men's time of 9.73+-.144 right??", but he is misunderstanding the argument. The times he quotes are the projected times for the 2008 Olympics...the lines cross in the year 2156, which is the date suggested for a faster women's time than mens.
Back to the drawing board...read this article and looka t the methods and the date very carefully...where are the real flaws in the argument?
Ok, so the part about the Beijing Olympics...I thought Usain Bolt decimated the previous record. It would seem that we haven't reached our potential. I mean look at Phelps, yes it's another sport, but look at all the unbreakable records that he broke. I think both male and female athelets haven't reached the limits of their potential.
ReplyDeleteNot sure if I am on the right track, but it seems it doesn't make sense to see a predictable linear progression here. The times for 2004 adn 1984 female sprint only differ by 0.04 seconds. The times for the 1968 and 2004 men sprint differ by only 0.1 seconds. I know the authors admitted that their analysis does not take into account timing accuracy, which this may or may not be the result of, but an inconsistency in accuracy of this kind should alone discourage an analysis like this one.
ReplyDeleteI agree with anthony given the varabality of the times, there are several occurances of the time for a given year being slower than a previous year this coupled with the fact the time depends on the physical capacity of the runner I do not think they can accurately predict the winning time or that women will eventually run faster then men.
ReplyDeleteAnthony and Victoria seem to be making an argument for the invalidity of the proposed statistical relationship based on teh degree of variability that they see in the data. But this critique is not really on point...remember that any statistical argument will have some error or variability. That is why we use standard deviations or CV's...to demonstrate the variability of a relationship. In a linear regression, we can plot the regression line, which involves "point" estimates of teh slope and the Y-intercept (like the mean, for example). But variability can be described by "confidence intervals" around those point estimates. SO variability is to be expected, and the degree of variability is not the weak point in the statistical argument that I am looking for. Any other ideas out there?
ReplyDelete