[chockablock]

Thanks this is a much more mathematically reasonable analysis. I especially like the idea of using the #fumbles made by Pats' opponents as a control for the weather (although that confounds the Pats' defense's ability to generate fumbles).

The OP gets credit for noticing an unusual trend in the data, but does all kinds of weird seat-of-the-pants analysis that I found pretty hard to even look at.

E.g. Sorting the teams by fumble frequency, and then running a regression on the ranked list (!!?); using plays/fumble instead of fumbles/play; reporting stats with the null hypothesis that the Pats fumble rate is drawn randomly from other teams (instead of asking what the likelihood is of an extreme value at least that high).