By Adam Kilgore
Washington Post Staff Writer
Friday, December 8, 2006
Southern California's loss to UCLA on Saturday ensured that the decision on which school would face Ohio State for college football's national championship would generate controversy. Two one-loss teams, Southeastern Conference champion Florida and Big Ten runner-up Michigan, felt they had strong cases to claim the No. 2 perch in the Bowl Championship Series ratings, the spot previously held by USC.
Charged with making the determination were six computer formulas, a poll of current coaches and the Harris Interactive poll, a second-year participant in the BCS formula and the creation of the world's 12th-largest and fastest-growing market research firm, the company's Web site boasts.
When Florida leapfrogged Michigan in the final ratings and claimed a spot in the Jan. 8 BCS championship game, the expected outcry ensued, and the scrutiny of voters and their ballots became more intense than ever. This is particularly true of the Harris Interactive poll, which replaced the Associated Press poll in the BCS formula after the AP asked to not be included in the process following the 2004 season, when an undefeated Auburn team was left out of the championship game in favor of USC and Oklahoma, which also were unbeaten.
The Harris Interactive poll's first season as part of the BCS resulted in little controversy, because only USC and Texas finished the regular season unbeaten. This season offered no such clean solution, and Michigan fans predictably targeted much of their partisan ire at Harris Interactive's eclectic smattering of former coaches, players, administrators and media members who vote in the poll.
Trying to come up with a broad scope of opinion, the Harris Interactive poll utilizes 114 voters, among them an ice cream company executive, a former college president and Boomer Esiason. Some of the ballots were as unusual as the makeup of the group. All but seven of the voters ranked Ohio State first and Michigan and Florida either second or third.
Two ranked Florida No. 5. Four ranked Michigan No. 4. One voter, former Washington State and Iowa State coach Jim Walden, ranked Florida No. 1 ahead of Ohio State, Wake Forest at No. 7, Louisiana State No. 11 and Texas Christian No. 14.
While those votes could be scrutinized because the final poll was made public, they could not be questioned before they helped decide which teams would play for the BCS title.
"Realistically, you have 114 individuals who have some opinion about the teams," said Jim Quilty, the Harris Interactive vice president who manages the poll. "What we have to do, if any part of the panel has rankings that are counterintuitive as to how they may have previously ranked a team, is monitor the situation. But we don't question their opinion. Inherently in market research, everyone has a bias."
Some voters wore that bias on their sleeve: Former Rice quarterback Bert Emanuel voted his alma mater No. 25. It was the lone vote the 7-5 Owls received in either the Harris Interactive or the coaches poll.
Some voters tried to eradicate bias from their poll: Former Greensboro News & Record sportswriter Larry Keech created a system dependent only on records that placed the undefeated teams first, teams with one-loss behind them, and so on. Teams with equal number of losses were differentiated based on the records of the teams to which they lost.
By squeezing out the subjectivity, Keech became the only voter to put Boise State at No. 2 and one of two who ranked Florida No. 5.
Robert Lawless, a former college president at Tulsa and Texas Tech, arrived at placing Florida fifth by a different measure. Lawless created mathematical formulas on which he based his decision.
Lawless, who is retired, said he might not vote next year because he said he spent 20 hours a week contemplating his ballot.
All of the study led Lawless, the lone pollster without a direct correlation to college football, to a peculiar conclusion: He had Ohio State, Michigan and Wisconsin as his top three teams, which made him the only voter to rank the Badgers No. 3.
"I would say over the years, [Big Ten Commissioner] Jim Delaney and I had enough battles that I would never, if I had any bias in me, then I wouldn't vote the top three out of the Big Ten," Lawless said. "But that's the way it came out."
Harris Interactive takes an active approach in voter turnover, using what Quilty called "panel management." By either voters leaving or voters being removed, it tries to replace 10 to 40 percent of the voters after each season.
Voters are selected from a pool of candidates recommended by conference officials. Each conference sent a list of 30 candidates to Harris Interactive, which then randomly selected 10 of those. A handful of recommendations from independent teams also are randomly sifted and chosen, leading to a total of 114 voters.
The process resulted in some interesting panelists, to be sure. Whether or not they are the 114 people best suited to choose the national championship can be debated, like anything else in college football.
"That's subjective, really subjective," Quilty said. "I'm not sure how we would define who the best 114 voters are. It's like asking who the second-best football team out there is."
Staff writer Eric Prisbell contributed to this report.