# Budapest Lane Changes Hold No Explanatory Effect On Speed

July 30th, 2017 News

Courtesy of Barry Revzin

It is often the case where studies are concerned that positive results get published but negative results do not. Since we caused such a stir a year ago with our coverage of a potential current in the Rio Olympic pool (a positive result), we wanted to make sure to keep up the reporting even in negative results. The good news is: there is nothing to suggest that there is anything amiss in the Budapest World Championships pool.

With the World Championships in Budapest complete, I redid the analysis that I did after the Rio Olympics in August 2016. I followed each swimmer in each 50m event as they progressed through the rounds – from prelims to semifinals and from semifinals to finals, looking at the impact of lane change on speed change. Since these are short sprints, you wouldn’t expect much variance in strategy from round to round. The expectation is that swimmers would get faster from round to round, but there should be no effect of changing lanes – it shouldn’t matter if a swimmer moved from lane 2 to lane 7 (a change of +5) or if they moved from lane 7 to lane 2 (a change of -5).

There are two relevant statistics here:

• R2, which roughly indicates how much of the variability of the response data is explained by the independent variable. An R2 of 100% would indicate that all the variability is explained by the input, an R2 of 0% would mean that none of the variability is – that this chosen variable has no impact on the effect being observed.
• The pvalue. This is the probability that, given that there is no effect, you would expect to see the distribution of data that you do. The lower the pvalue, the less likely that the particular distribution of data comes by chance alone. Typically, the cutoff for publishing is 5% (which allows for a 1 in 20 chance that there is no effect and the data is just noise).

Where this specific analysis is concerned, what we want to see as swimming fans is no effect: an R2 close to 0% and a pvalue that is high enough to not even be worth a second thought. With that in mind, here are the values of those statistics that I see in Budapest as compared to the Rio 2016 Olympics, the Maria Lenk Trophy of 2016 (swum in the same pool as Rio), and the previous World Championships in Kazan in 2015 and Barcelona in 2013 (the other suspected current meet):

 Meet # of data points R2 pvalue Barcelona 2013 190 32.8% 0.0000000000000000628% Kazan 2015 192 1.8% 6.7% Maria Lenk Trophy 2016 32 50.8% 0.000466% Rio 2016 48 46.8% 0.000000837% Budapest 2017 192 0.14% 60.3%

As you can see, the Budapest result is as much as an indicator of no effect as you can get. Lane change holds practically no explanatory effect on speed change, and you’d expect to see this kind of data distribution due to pure chance most of the time. The comparison to Barcelona and Rio is pretty stark. For those more visually oriented, here is a scatter plot of the Budapest data:

That is a whole lot of nothing. Which is great! No effect is the best kind of effect.

0 Followers

Most reacted comment