Monday, November 07, 2005

Bad Analysis

I came across an example of exceptionally bad analysis in Matthew Zemek's "Weekly Affirmation" at College Football

When Cal lost to Oregon in a sloppy football game, USC’s strength of schedule dipped appreciably. A team that had been 5-0 and leading UCLA by 12 points with nine minutes left in regulation had suddenly spiraled to 6-3 before the showdown with the Trojans next weekend. When the Canes and Hokies kicked it off in Blacksburg, I was thinking to myself, “If USC doesn’t blast Cal out of the water—really crush them—on Nov. 12, one has no choice to dock the Trojans for their strength of schedule, just as I docked Texas for its strength of schedule.” I didn’t expect Cal to take this kind of a nosedive, but it happened, and that’s part of the larger body of evidence I have to deal with. My mind has to remain open when new evidence presents itself.

He continues:
Later in the day, the landscape changed even more against USC. When UCLA—a team supposedly worthy of contending for the Pac-10 title as USC’s final regular-season opponent—lost so decisively and pathetically to Arizona, I had no choice but to internally downgrade USC’s schedule strength still more.


Objectively speaking, these two games have absolutely no net effect on USC's strength of schedule because USC plays all four teams. The loss by Cal, USC's Nov. 12 opponent, was a win for Oregon, USC's Sept. 24 opponent; just as UCLA's loss was a win for Arizona, who USC beat on Oct. 8. Therefore, the results of these two games for these four teams cancel each other out in USC's final schedule strength.

Subjectively speaking, the results of these two games should also be a wash. Whereas presumed USC victories over Cal and UCLA will now be less impressive, the past victories over 8-1 Oregon (in Eugene) and Arizona should now appear more impressive. If anything, the results of these two games bolster the evidence supporting USC's current reputation because USC has already beaten the victors of the Cal-Oregon and Arizona-UCLA games, but has not yet played the losers of those two contests.

In addition to Zemek's poor interpretation of the results of those games, he also makes the erroneous assumption that a team's strength is partly determined by its strength of schedule. I'm not saying that strength of schedule is not a legitimate factor of a team's BCS ranking (I'm not saying it is, either, but that's a separate issue). However, for those who believe the Pac-10 is a weak conference (myself not included), one only has to look at USC's non-conference performance over the past 3 years for proof that this assumption about schedule strength is suspect.

Zemek prides himself on his sound analysis of college football. Elsewhere in this piece, he writes, "good college football journalism concerns the process of making sense of events after they happen," and, "it’s making sense of events as they occur, and especially after they occur, that counts." I couldn't resist calling him out.

No comments: