Poll Pull

Think you know what it takes to be a top-ranked college football team? Think again.
In a research paper engagingly titled “Whoa, Nellie! Empirical Tests Of College Football’s Conventional Wisdom’, Ohio State University’s Trevon Logan examines three pieces of so-called “convention wisdom’ about what it takes for a college football team to be ranked highly in national polls. Those beliefs include:
  • It is better to lose early in the season rather than later (because a team’s later performance will weigh heavily in how others view a team)
  • A team is rewarded for playing strong opponents (having a “strong’ schedule)
  • A large margin of victory is necessary to obtain or retain a high ranking (because such victories receive “style’ points as they reflect a team’s dominance)
To test these propositions, the paper relies on a data set of week-by-week AP poll results for 25 of the most prominent college football teams over a 25 year period to see if what “everybody knows’ turns out to be true empirically – because, as the author notes, “…the bases for these assertions are anecdotal…and the plural of anecdote is not evidence.’
Instead, the research finds that, rather than being penalized, teams are actually rewarded for losing late in the season. Indeed, teams that lose late in the season are re-ranked higher (the week following the loss) by roughly 3/4 of Associated Press poll voters than they would have been if they had lost early in the season. He goes so far as to suggest that the results support a notion that late season losses leave teams with less ground to make up – that, in essence, an early loss might well remove an otherwise “worthy’ team from consideration altogether, while a late season loss might be discounted.
Similarly, defeating strong opponents does not yield any apparent advantage in terms of ranking – but LOSING to strong opponents….helps. As for that margin of victory? Well, it matters, but only if you are on the embarrassing side. The paper notes that, “while winning by large margins does not confer any ranking advantage (despite numerous claims to the contrary), losing by a blowout hurts, and losing to a strong team does not soften the blow.’
The paper acknowledges that the meaning of these findings for scheduling and conference alignments in the future is unclear. Since strength of the opponent is not a factor, teams may be better served by avoiding games against strong opponents throughout the season, he suggests – though it may also be helpful to schedule those games against strong teams late in the season. On the other hand, as the paper notes, college football schedules are decided many years in advance (the full schedules for the next three seasons are already posted for most major college football programs).
The research paper is available online at http://www.nber.org/papers/w13596

The paper contains some interesting anecdotes about college football. Ever since Rutgers defeated Princeton (6-4) on November 6, 1869 college football has been a central part of both sports and collegiate folklore. However, even into the early 20th century, many prominent teams had no official coach, and most of those who coached did so on a part-time basis. Indeed, football was a particularly controversial feature on some college campuses due to its violent nature. As early as 1905 President Theodore Roosevelt was threatening to ban the sport based on its brutality (more than 100 players had been killed in the 1905 season), which led to the creation of the National Collegiate Athletic Association (NCAA) to devise rules for the game.