We hear it every year and it's as true now as it's been for years: the preseason poll must go.
It's innocuous on its own, of course, and nobody wants to be the person that declares an immediate cessation to the practice of guessing which teams are better than the others going into the year.
That's a black belt in fun-hating and opinion totalitarianism, right?
But if the preseason poll existed in a bubble, we wouldn't be having an issue. It's that the poll creates a bias, one that fades over the course of the year but persists until the end of the regular season—and it can have serious effects on the course of a year.
There are anecdotal instances of this, though how much weight those deserve is debatable.
A strong case could be made—and was, by CBSSports.com—after the 2011 regular season that Oklahoma State, not Alabama, merited the second BCS championship game spot to face LSU. And yes, Alabama was buoyed by a preseason No. 1 ranking that year, one that kept it ranked third even after the regular season loss to LSU.
But Oklahoma State didn't come out of nowhere that year.
Remember, Oklahoma State started that season ranked ninth, behind only Texas A&M in terms of Big 12 teams—and A&M summarily folded over the course of the season.
There may have been unfair factors that kept Oklahoma State out of contention for the BCS Title, but a preseason bias doesn't appear to be among them.
And for every anecdote like that, you can go in the other direction. Oklahoma was the preseason No. 1 last year. Georgia was No. 19. OU finished 9-3, while UGA was 10-3.
Georgia at No. 18, Oklahoma at No. 19.
So yes, by the end of the regular season—which is the only time they truly matter—polls are generally a meritocracy.
A look at the last five years of preseason polls and their final regular season counterparts, however, uncovers a slight but noticeable preseason bias.
We selected the final regular season polls because those can affect the bowls that teams are invited to, whereas the postseason polls are just the number next to the team at the end of the year. We're looking for consequence here.
With that in mind, there are two most important factors in finishing in the top 10 before bowl season. First, win ten games. Second, begin the season in the Top 10.
Of the last five final regular season AP polls (since that's what the invaluable collegepollarchive.com has in its archives), 49 of the 50 teams that populated the top 10 won at least 10 games.
The lone exception was 9-3 Florida in that horror show of a 2007 season. So that's the most important factor.
The issue of preseason rank is a bit thornier, however. Here's a look at those same final regular season top 10s and the teams in them, with the teams separated into three categories: Preseason Top 10, Preseason 11-20, and Everyone Else. And even with 12 or more games to flush out the biases of the preseason, we see an easier road to the Top 10 for those that started there than those that didn't.
|2007-2011||Teams||Avg. Rank||Avg. Wins||Avg. Losses||Win %||#1s||Unranked Teams|
|Everyone Else||13||6.4||11.38||0.85||93.0%||1||A lot, obviously|
Quick note: you'll noting that the average ranks are all 5.0 or higher. That's because the average rank of a Top 10 is not 5.0, it's 5.5. Add up the numbers 1-10 and divide by 10. Yep! You learned something today!
To put the chart more bluntly: teams that began the season in the top 10 had the easiest road to staying in the Top 10 by the end of the regular season, and teams that started outside the top 20 had the hardest.
Now, this analysis is a bit crude and doesn't take into account things like schedule strength, the timing of the teams' losses or conference affiliation. An SEC team is going to get preferential treatment to a WAC team. That's just how it goes.
Still, if teams that started the year in the top 10 were subjected to the same voter scrutiny as teams that started outside the top 20, we'd likely have some different-looking polls by the end of the year. And when the BCS is only one-third computer rankings and two-thirds dubious polls, perceptions and biases can have a significant effect on BCS rankings, which affect BCS bowl bids, which affect BCS bowl money, which affects the very health of the conferences themselves. So this is a big deal.
That all said, it's worth remembering one simple point: teams that start the season in the top 10 are generally very, very good football teams, and on a macro scale they're better than the 10 teams below them. Nearly half of the teams ranked from 11-20 to start the last five seasons ended up out of the Top 25 completely by the time the regular season was over; for top 10 teams, that rate is essentially cut in half.
So make no mistake: polls generally reflect quality and results. They're not pure hokum. They are, however, absolutely prone to preseason bias and this is a major flaw.
That's why we're looking forward to the playoff's proposed system of a selection committee examining teams' overall resumes when deciding the four championship contenders. The more information that's available, the less prone we are to relying on assumptions and biases (take another look at that CBSSports.com post linked above for a perfect example), and the better our choices will be.
And step one toward removing those very biases is to stop conducting the preseason poll. Results, people. Results, results, results. Pay attention to the games at hand, not your last poll, and you'll be a better pollster for it.