An attempt to appear educated resulted in the opposite.

Why on earth would you suggest running a multivariate analysis given the data? You have no idea what you're talking about. No statistician would ever look at that data and say, "Hmmm, we definitely need to try using MANOVA with this." Suggesting that he should look for other explanatory variables implies that there was an explanatory variable in the data to begin with. There wasn't.

He (correctly) never said anything about correlation. You did! He also never concluded anything. He speculated which is completely fine.

All this data is saying is, "Hey, this thing is very different from everything else. In fact it is behaving so differently that it is nigh impossible that it is not being affected by something unusual. Period."

People are free to speculate as to why this is happening. Also, the reason for him examining pre-2007 and post-2007 is not hard to understand. There is a sudden increase in the Patriots' plays-per-fumble in the data. It's interesting.