I am not sure whether to call this “bad science” or to be generous and call it “trivial science”.
There is a new paper in PLOS ONE (open access) from “researchers” at Boston University:
Effects of Warming Temperatures on Winning Times in the Boston Marathon by Abraham J. Miller-Rushing, Richard B. Primack, Nathan Phillips and Robert K. Kaufmann, PLoS ONE 7(9): e43579. doi:10.1371/journal.pone.0043579
The only conclusions of any little value in this paper are that
- Boston marathon winning times fail – so far – as a proxy for global warming temperature rise, and
- if global warming increases Boston temperatures by about 5.8°C there is a 95% chance that the effect on winning time may be detectable!
Of course they could simply have reported that the variability of temperatures on race day drowned out the effects – if any – of any global temperature change. But that would have been too simple, too truthful and would not have helped towards publication.
The findings of this so-called “research” are:
- Higher temperatures and higher headwinds on the day of the race increase winning times for the marathon. Who would think otherwise? I would not have thought that weather conditions on any other day than race day would have much impact. Collecting this data is mildly useful but it is all readily available. Trivial.
- If temperatures do not increase relative to temperature variability on race days, the effects of warming on marathon times may not be detectable. However, at some point temperature increases may be large enough to affect marathon times. Amazing! So what exactly was the point of this work? Trivial.
- In summary, despite the well-known effect of temperature on marathon performance, we found that warming trends in Boston have not caused winning times to slow over time because of high variability in temperatures on race day. So race day temperature overrides any effects – due to global warming or anything else – on average annual temperatures. Obvious and Trivial.
- …. our models indicate that if race starting times had not changed and average race day temperatures continue to warm by 0.058°C/yr, a high-end estimate, we would have had a 95% chance of detecting a consistent slowing of winning marathon times by 2100. If average race day temperatures warm by 0.028°C/yr, a mid-range estimate, we would have had a 64% chance of detecting a consistent slowing of winning times by 2100. So your model says that if global warming increases race day temperatures above and beyond the natural daily variability then you have some chance of detecting this effect in the winning times! By modelling high rates of global warming (5.8°C per century) you can force the race day temperatures to show an underlying increase such that there is a 95% chance that the effect on winning times could be detected. The Boston marathon itself provides no evidence of global warming so far. Wow!! This is not just trivial but borders on “idiot science”.
The effects of global warming clearly cannot be detected in the results of the Boston marathon. In the paper their Figure 2 and Table 1 are fairly trivial but of passing interest. The rest of their modelling efforts (input and output) are just garbage.
Open diamonds represent men’s times from 1933–2004. Closed circles represent women’s times from 1972–2004. Women’s running times improved rapidly in the first 14 yr of women’s participation in the marathon. From 1983 to 2004, the differences between men and women’s winning times held relatively constant at an average of 15 min 47 s.
Table 1. Regression results showing effects of temperature and wind on winning times in the Boston Marathon.
Abstract: It is not known whether global warming will affect winning times in endurance events, and counterbalance improvements in race performances that have occurred over the past century. We examined a time series (1933–2004) from the Boston Marathon to test for an effect of warming on winning times by men and women. We found that warmer temperatures and headwinds on the day of the race slow winning times. However, 1.6°C warming in annual temperatures in Boston between 1933 and 2004 did not consistently slow winning times because of high variability in temperatures on race day. Starting times for the race changed to earlier in the day beginning in 2006, making it difficult to anticipate effects of future warming on winning times. However, our models indicate that if race starting times had not changed and average race day temperatures had warmed by 0.058°C/yr, a high-end estimate, we would have had a 95% chance of detecting a consistent slowing of winning marathon times by 2100. If average race day temperatures had warmed by 0.028°C/yr, a mid-range estimate, we would have had a 64% chance of detecting a consistent slowing of winning times by 2100.
On balance I shall just classify this as Trivial Science bordering on Bad Science
The k2p blog 