Sunday, November 7, 2010

2010 Week 10 College Football Poll Analysis; "Only one of Boise/TCU" bias

Now is the time of the year when we get to microanalyze the polls to look for agendas, inconsistencies, and generally how the voters may be trying to manipulate the system.  The easiest way to do this is to look at the changes week to week and if they make sense given the results.

First, the Coaches Poll.  The top-4 stayed the same but there was some swapping and changes in the vote totals.

Oregon lost 1 point, garnering 50 first place votes and 1463 points, down from 51 and 1464 last week.  That is 12 points short of a perfect 1475 so at least a few of 9 voters to didn't vote them #1 have them lower than #2.  Given that my computer has their schedule strength just about equal to Auburn's (#38 and #39 respectively right now) and quite a bit better than Boise State or TCU (#66 and #62), and that they have dominated their opponents, I find it hard to understand how someone could have them lower than #2, but they do.  The first place vote they lost seems to have gone to TCU which one can understand given their dismantling of Utah, but Oregon dismantled a much better Stanford earlier in the year too.  But Oregon's poll result is reasonable.

Auburn stayed with 4 first place votes, but gained 5 points from 1384 to 1389.  Given that they played a I-AA team, and a 5-4 one at that, it is hard to understand that change, but it must be due to someone having Auburn behind a Utah or perhaps even Alabama and those teams dropping force a move up of Auburn a spot here and there.  If that is the reason, I laud the voters who had the guts to have Auburn that low.  But if it is because someone moved them ahead of either TCU or Boise State as a result of this weeks games where TCU and Boise State handily beat much better competition, shame on them.

TCU went from 1292 and 1 first place vote to 1340 and 2.  This is the biggest change and is just about an indication that 48 of the 59 voters moved them up 1 spot.  Perhaps some of the movement was due to someone having them behind Utah or Alabama, but it is more likely that based on their performance folks jumped them over Boise State.  Perhaps justifiable, but if one thought Boise State was better going in, how did their handling a good (and just outside the top-25) Hawaii team change their minds?

Boise State was the loser this week dropping from 1361 and 3 first place votes to 1330 and still 3.  This clearly shows that TCU was bumped ahead of Boise State as the reason TCU moved up.  Again, perhaps justifiable but questionable.

Now to the AP poll.  Again, the top-4 teams are all there, but there has been some changing of order which is questionable.

Oregon kept their 49 first place votes but lost 3 points from 1487 to 1484.  This is 16 points short of perfect so again a few of the 11 voters have them worse than #2.  See above for how that is hard to justify.

Auburn makes the move to #2 while keeping the same number of first place votes (2) and keeping the same number of overall points (1396).  So this move was due to voters not giving Auburn anything new, but dropping Boise State.

TCU kept their 2 first place votes and gained 41 points from 1350 to 1391.  Meanwhile, Boise State kept their 7 first place votes but went from 1403 to 1366 points, a drop of 37.  Since Auburn didn't improve their point total, it would seem that 18 or so voters that had Boise #2 and TCU #4 just swapped them.  Yes, TCU deserves kudos for handling Utah, but Boise State just did the same to a good Hawaii team.  Hard to understand this drop as it seems voters are saying they have to anoint one of Boise State or TCU their #2 team but can't drop Auburn to keep both high.

So, are voters manipulating the system?  I don't think we see week to week evidence of that, but there does seem to be a bias that says only one of Boise State or TCU can be voted high and there was simple swapping of the two in several cases.  It will be interesting to see what happens the next few weeks, if voters start to raise or even worse drop teams just to try to manipulate the system to keep a certain team out in the BCS calculations.