BCS & College Football Coaches Poll Shenanigans – Bad Voters or Bad System?
Some of you might not care about college football, but I'm always interested in a story about people gaming a system and what drives that behavior, whether that's in healthcare (such as not reporting errors) or in sports. We can criticize the individuals involved all we want, but what they are doing is usually perfectly rational and understandable, given the design of the system.
This article caught my eye the other day (it also appeared in USA Today): “Coaches Poll votes, revealed: Big 12 stifles Northern Illinois.”
One input into the “BCS” (Bowl Championship Series) rankings, which determines which teams play in the highest profile bowl games, is a poll of certain college coaches. These coaches work such grueling schedules, they can't possibly watch enough college football to make informed choices about how to rank teams (even given that it's a subjective ranking on which teams they THINK are best)… these rankings are often based on reputation as much as anything (much like the US News rankings of hospitals, actually).
One problem with the Coaches Poll is that some of the coaches directly benefit from their own rankings. When a single coach's ranking is an outlier, it raises suspicion of self-serving votes.
Oklahoma coach Bob Stoops found his team “on the bubble” – possibly into a BCS bowl game, as the second representative from the Big 12 conference (since they weren't the conference champion, they weren't guaranteed a spot). The other possible bubble team was Northern Illinois University. The NIU Huskies could only be selected for a BCS game (instead of Oklahoma) if they were ranked in the top 16 spots.
So, Stoops voted Oklahoma #6 (an outlier, since the consensus ranking was #11). Stoops also ranked NIU #24, another self-serving vote, as the consensus placed NIU at #16, meaning they go to the Orange Bowl, not Oklahoma.
Other Big 12 conference coaches also tried manipulating the system by either voting Oklahoma #6 or ranking NIU #24. The conference of the Orange Bowl representative gets a share of the payout (rumored to be $18 million or more). They clearly had a financial incentive to do so. But, the manipulation didn't work.
Among the Big 12 voters other than Stoops and [Texas Tech coach Tommy] Tuberville [who voted Oklahoma #11 and NIU #16] , the voting went as follows: Briles had Oklahoma sixth and Northern Illinois 19th; Holgorsen had OU ninth and NIU 24th; Rhoads had OU sixth and NIU 20th.
As one columnist said, “Stoops' coaches poll strategy didn't work, but you can't blame him for trying.”
Coaches from other conferences, without a direct financial incentive, voted differently.
Northern Illinois was listed on all ballots, but the only coach outside the Big 12 to have the Huskies at No. 24 or lower was Michigan's Brady Hoke, who voted them 25th while voting Oklahoma No. 11. Vanderbilt's James Franklin had Northern Illinois 23rd and Oklahoma 11th.
Maybe Brady Hoke is buddies with Bob Stoops or just thinks that a smaller school like NIU shouldn't be in a BCS bowl game.
Coaches from NIU's conference (the MAC) didn't vote in such a selfish way:
The MAC's six voters – including Northern Illinois' Dave Doeren – cast their ballots in a way that was not as helpful to the Huskies as the Big 12's votes were to the Sooners.
Do we call the Big 12 coaches' voting “unethical” (putting the individual responsibility on them) or do we blame the system and the creators of the coaches poll and its role in the BCS? Was the Big 12 more “sophisticated” (or devious) in their attempted manipulation or was it a coordinated effort directed by the league office, perhaps? Or, are the MAC coaches and Tuberville more ethical and honest?
Do the designers of measurement and ranking systems in other industries have to take action to prevent opportunities for manipulation?