Tis the season for college football. The Bowl Games are upon us. Teams are ready. Ready for what? Well, in the immortal words of Montana Tech coach, Bob Greene: "I really feel like our team is ready to go hit individuals from another institution of higher learning."
But to determine which college footballs team gets to hit which college football teams requires one of the strangest formulas in the history of sport. Bowl games are assigned via BCS ranking or supposed to be assigned by BCS rankings or are mostly-unless greed becomes a factor-assigned by BCS rankings. Of course, once the universities get involved, greed always becomes a factor.
This year, for example, the Outback Bowl selected 7-5 Florida over 9-4 South Carolina or 8-4 Mississippi State, despite the fact that both teams beat Florida in Gainesville (SC also won the SEC East), but because Florida's rabid fan base would "hypothetically" buy more tickets than either SC or MS, greed won out.
Anyway, that's another story.
In this story, since 2000, those rankings—known officially as the BCS standings—are comprised of three components: the USA Today Coaches Poll, the Harris Interactive College Football Poll and an average of six computer rankings, with each component counting one-third toward a team's overall BCS score.
Sometimes this leads to great games (and this year's championship between Oregon and Auburn—on ESPN January 10 at 8:30—should be a doosy). Sometimes not. One thing that is certain, how the teams end up in those games is far from fair.
Let's set aside the six computers. The general thinking about the combination of the Harris poll (which polls former coaches, former players, former athletic administrators and some active media members) and the coaches poll (which polls 59 active coaches) is that they are, while perhaps not quite fair and balanced, at least accurate enough to avoid too much argument.
Take the Harris poll. According to Harris Interactive, their poll has been designed to be a statistically valid representation of all eleven FBS Conferences and independent institutions. But, as ESPN BCS analyst Brad Edwards—whose job it is to predict the polls—recently told me, that's only if you don't consider psychology a factor.
What part of psychology? Well, there's our confirmation bias for starters.
Confirmation bias is an innate tendency for people to favor information that confirms their preconceptions or hypotheses regardless of the veracity of the information. As a result, people gather evidence and recall information from memory selectively, and interpret it in a biased way.
Edwards says confirmation bias is especially prevalent in how media voting impacts the Harris poll. "It's sort of a joke around ESPN," he says. "Most media members are asked to make a national championship prediction before the season starts and, if they are voters, they'll fight to keep that prediction alive for as long as it can be justified. In other words, if there are several undefeated teams or several one-loss teams, and one or two teams must be separated from that bunch at the end of the season, many voters will ignore what their eyes see and instead try to fulfill their preseason prophecy by elevating their team of choice."
Of course, as these biased media members started to agree with one another, the "availability cascade" also came into effect. This is a self-reinforcing process in which a collective belief gains increasing credence the more it's repeated in public. At the start of the 2010 season, most media members deemed Alabama the power to beat, an idea they defended late into November, even Alabama started losing games.
Meanwhile, the Coaches' Poll suffers different issues.
"When the coaches rank teams they look at their record and their championships," says Edwards, "because that's where coach's job security and performance-based incentives come from. The media spends much more time analyzing schedule strength, but coaches don't get evaluated on how difficult their schedule is. Their job is to win against whatever schedule is put in front of them."
And this is the same criteria that impacts their choice of ranking. This means there's still a bit of confirmation bias coming into play (as no one beside the coaches really cares about conference championships), but there's also the self-serving bias (the tendency to evaluate ambiguous information in a way that's beneficial to self interests) and a huge anchoring problem.
Anchoring is a tendency to rely too heavily (or "anchor") on one piece of information when making decisions—which is what happens when the coaches rely too heavily on success in the conference championships when deciding end-of-season rankings (which decides who goes to which Bowl).
Worse, because the coaches poll publishes these final votes (while every other vote taken during the season is anonymous) the "herd instinct" comes into play. Just like it sounds, this instinct makes one go along with the crowd. It's visible in the coach's preference for established football superpowers (like the SEC) over upstart outsiders (like the WAC), and in their reliance on conference championships.
And this is just the beginning. The full list of our cognitive biases runs for pages and a great many of them impact college football. The fact of the matter is Daniel Kahneman-—who first discovered our cognitive biases—won a Nobel prize because he figured out that there are certain situations where humans are terrible at assessing probabilities. The BCS rankings are an assessment of how probable it is that a team will beat another—and this is exactly the kind of situation that erodes our judgment.
In fact, once you factor cognitive biases into the college football rankings, then the only conclusion possible is that the BCS polls are less an assessment of competitive skill and more an ink blot test for the predilections of the American sport establishment. And not even a very good ink blot. Look at it this way, Brad Edwards is good at his job because he understands psychology—but would ESPN really need his analytic insight if the BCS was anywhere close to accurate?