Thursday, September 27, 2007

Change Your Argument

My last words (for a while) complaining about polls will be contained in this post. If you don't like that LSU is #2 in the polls instead of #1, stop complaining about that. Instead, complain about how the voters decide how they will vote. It will come off less self-serving, and will avoid the arguments about which team is actually better. Instead of complaining that the pollsters can't see that LSU looks like the better team at this point, complain about the methodology by which the voters decide their votes, because they use silly and indefensible criteria in determining their rankings. Criticize the silliness and the indefensibility, not the results the silliness and indefensibility generate, because in a highly subjective world, the results are defensible even if the methodology isn't.

College football polls are poorly run. I think this is so obvious that it really doesn't require explanation, but I'll try anyway.

College football polls start in the pre-season before any games are played. Teams are "ranked" based on media perception and hype going into the season, along with such niceties as "returning starters" and "coaching". But mostly it's based on hype. It bears only a passing resemblance to actual football analysis.

Cases in point: At the beginning of this year, Auburn was a top 20 team and Michigan was a top 5 team. Both of those teams have looked REALLY REALLY bad.

Once the games start, teams will usually only move down the polls if they lose or play a close game against a team they are expected to blow out. Teams usually move up because the teams ahead of them do something to warrant moving down. Rarely does a team move up simply due to looking really really good. A team ahead of them has to look bad.

Case in point: Louisville, as a top 10 ranked team, gave up 42 points to Middle Tennessee State University, a Sun Belt Conference team that is now 0-4 on the season, yet they did not move down in the polls. Not even an inch. Of course, if the pollsters had been paying attention and thinking about it even a little bit, they would have moved Louisville way down the polls, and the upset Kentucky pulled over them wouldn't look like like such a big deal. Louisville went and put out significant evidence that they are not that good of a team, but it was ignored because they didn't look bad enough to move down, even out of the Top 10. Two weeks later, they're out of the top 25 completely due to a loss to lowly Syracuse

Other case in point: Auburn has looked terrible, and has lost two close games to teams that are generally considered to be inferior. Both games were close and went down to the end, and if Auburn had actually won them, I'm sure they would be in the top 15 now, even though they've looked absolutely awful at stretches.

So, you have a situation where teams are initially ranked based on factors that have little to do with actual football ability, yet those initial rankings won't change in the face of additional evidence unless a team either loses or looks really really bad. Looking merely bad, like Auburn did after during its Week 1 win against Kansas State, won't make you move down. Looking really good won't make you move up unless a team ahead of you does something to look really bad.

It's an indefensible system. And yet, talking heads defend it all the time. Well, maybe they don't "defend" it exactly, but they often spout the indefensible assumptions as if they are gospel. You'll frequently hear something said like, "I'll vote them #1 until someone beats them," or "USC proved it was #1 last night by beating up on Nebraska." In a sport where not all teams play each other and rankings are determined by vote, voters are expected to exercise discretion and use judgment, but they don't. They pretty much just apply a generally agreed-upon set of rules that don't make a whole lot of sense when you really think about them.

They're not saying "USC proved it was better than LSU" or "USC proved more on the field than LSU has proven" because neither is true. I don't think any reasonable, objective person can look at what USC has done on the field this year and at what LSU has done on the field this year and conclude that USC is the clearly better team. We've played 4 teams that are a combined 9-3 in games not against us. Granted, none of those teams are world-beaters, but MSU beat Auburn, Virginia Tech has grinded out wins, South Carolina beat Georgia, and MTSU pushed Louisville. And we didn't just beat them. We ran them off the field in all the games except the South Carolina game, where we sleepwalked through the second half and gave up a late touchdown to make the game look closer than it was. Until that game, we were averaging 45.3 points per game, and we were giving up an average of 2.3 points per game. Those numbers have changed, but they're still gaudy. We've done it with backups getting significant playing time, and without opening up our offensive playbook for two of the games. We held MTSU, the team that put 42 points up on Louisville, to less than 100 yards total on the game. They never even came close to our end zone.

USC has played three games, winning them by an average of 48 to 18. Not bad. Not bad at all, but they haven't dominated like we've dominated, and they haven't played their backups as much. They haven't done it against significantly better competition either.

No, they're not saying that USC has proven to be better than us, or that they've accomplished more on the field. Saying that would be absolutely silly. What they're saying is, "USC didn't do anything to make themselves look bad, so they won't move down." It doesn't matter how good LSU looks. LSU won't overtake USC until USC loses a game or lets an inferior team stay in a game until the end. At that point, if it ever occurs, we'll move ahead of USC regardless of what we've accomplished on the field, unless of course WE have lost a game or let an inferior team stay in a game until the end, at which point OU or Florida will pass us and then pass USC.

This is why I just don't pay much attention to polls. I try not to anyway, but it's quickly becoming one of the most popular topics of conversation on the message boards. I have taken on the task of trying to discourage this kind of poll watching.

I think polls do more harm than good to college football, especially when they're run as poorly as they are. If the polls started later in the season (say, after week 4) or if voters were more willing to make big adjustments in their rankings early in the season, which would treat pre-season polls as the useless instruments they really are, I wouldn't have such a big problem with them. I just don't like that they start out with so little supporting information, yet they aren't subject to significant change until ranked teams lose or look particularly bad.

To compound the problem, these problems persist throughout the season, until one poll becomes a substantial factor in deciding which teams can play for the national championship. The problems in the polls impact the integrity of the National Championship.

And before people wonder, yes, I would say the same thing if LSU was on top and USC was #2. I'd still have problems with the way the polls are run. I thought the same thing when it was Auburn that was getting screwed back in 2004. College football saw 3 major conference teams go undefeated that year, with Auburn shut out of the national championship game despite surviving a tougher schedule than either of the two teams ahead of them. It happened because USC and Oklahoma started the season ahead of Auburn and never did anything to warrant moving down. Auburn could have won every game they played by 40 and still would have ended up 3rd.

So quit bothering about the polls. It's really best if you ignore them as long as possible. If you must criticize them, criticize the flawed principles that underpin them, rather than the results they give. The results are questionable, but defensible, but the principles are ridiculous and indefensible. If you focus your wrath on the poll methodology, perhaps a movement to reform it will build, and you will accomplish a lot more than if you simply say, "LSU is getting screwed."


gerry dorsey said...

the only thing dumber than top 25 polls is recruiting rankings.

uberschuck said...

Another common polling axiom I don't like is the notion that last year's champ should be ranked #1 until they lose. There is an awful lot of turnover in college football, so one season usually leads to significant changes in any team. To evaluate a team based upon what happened the previous season is bogus.