Since the date for the 2010 Apple cup was set for December 4th many loyal Cougs have had fantasies of swirling snows greeting the Huskies as they take the field this year at Martin Stadium. Thoughts of Jake Lockers' NFL-bound hands so cold they creek like rusty hinges have certainly delighted me. Surely a frozen playing field would be advantage cougars! Right? Well, being a grad student in the sciences who is desperately trying to avoid actual work, I decided to see if this has historically been true.
Below is graph of the point differential (WSU's point total minus UW's point total) plotted against the temperature at kickoff for the last 21 years. The trend line is what is known as a LOWESS (Locally Weighted Scatterplot Smoothing) regression, and the shaded area around the regression is the 90% confidence interval determined by 1000 bootstrap replicates (in other words I can be 90% sure that the "correct" regression falls within that shaded area)
It appears as if there is a trend for the Cougars to perform better in colder Apple Cups. That being said, the relationship is very weak , in fact there is a reasonable chance there is no relationship (the probability that the trend is not 0 is about 42.3%). In a scientific setting we would never try to interpret trends this week. Nevertheless, in the last 21 years the Cougs are 4-3 in the coldest 7 Apple Cups and 1-6 in the warmest 7 Apple Cups. With an historical average temperature of 31 degrees in Pullman on December 4th, it hardly seems as if it hurts the Wazzu's chances.
Finally let me admit some faults in this analysis: Really there is a lot more at work in the final point differential than simply temperature. A rigorous analysis would normalize the data for other factors such as the records each team brought into the game. That could potentially strengthen the suggested trend (or completely obliterate it). Additionally, it is generally colder in Pullman than Seattle in November; this could simply be reflection of home field advantage rather than having anything to do with temperature.
This short analysis was really just a way of avoiding "real" data analysis I should be doing. Did anyone find this interesting in the least, or even read it? If so I may occasionally post up a graph or two.