One of my all-time favorite writers is a guy named Stewart Mandel who covers college football for Sports Illustrated, but his latest ranking of the 34 college bowls games included a line that struck me as a personal challenge. Keep in mind that Mandel, like most sports writers, knows more about statistics than your average journalist. These folks live and breathe numbers, and then make a living by wrapping prose around them. But when Mandel dropped the C-bomb in his article (correlation), he crossed into my turf.
Here's what he said: "...I think you will find that the date a game is played no longer bears any correlation to the quality of its matchup."
No correlation? As in R-squared equals 0.00? I couldn't believe it. So I decided to put the two data series side by side and test Mandel's hypothesis that there is no correlation between quality and timing of college bowl games. The "game quality" data series was easy enough to create because Mandel gives us a simple 1-34 ranking that most observors would agree with (say with 90 percent correlation). The "timing" series I assembled pretty quickly by creating a number for each game that represents the days before the BCS championship game, which will be held on January 8, 2009 between the University of Oklahoma and the University of Florida. The chart below shows you that in fact, there is some correlation, 0.265 to be exact. So is Mandel wrong? Well, you could interpret this result to say that he is 73.5 percent right. Not too shabby for a sports journo, eh? He's still not correct, but the man has a very good point.
The chart shows each of the 34 games as dots on the grid. On further analysis, I decided to highlight the key BCS games (Fiesta, Rose, Sugar) as well as the Poinsetta Bowl. Why Poinsetta? Because this clash between 10-2 TCU and undefeated Boise State is played 16 days before the championship - one of the earliest - yet is ranked the #5 game by Mandel. The chart also highlights the four other games to be played on New Year's Day in addition to the Rose Bowl. Those are the BCS Orange Bowl, Outback Bowl, Gator Bowl, and the great SEC-Big10 undercard Capital One Bowl (which I believe contender Florida lost last year to this year's hapless Michigan).
For many decades, January 1 was the day of days for college football. This was the day that all the great games were played, when the Cotton Bowl was one of the biggies, and when Notre Dame was sure to be on the tube. With the advent of the BCS, college football stretches the greatest games out. The powers that be have not yet caved into demand for a playoff - and for that I am glad. Every other sport has caved into playoff mania, and in many of them there are more teams that make the playoffs than don't. In college football at least, every game matters.
Congrats to all the teams playing bowls this year, even Louisiana Tech (7-5) vs. Northern Illinois (6-6). You guys are doing a lot better than the poor kids who played in the Apple Cup (still meaningful!).
Disclaimer: there are two reasons college football is relevent to growthology.org readers. First, if its big enough for President-elect Obama to get involved, it's big enough for us. Second, my son and I are about to create a new blog for fun based on college football and statistics -- an entrepreneurial venture of sorts. I'm hoping the experience teaches him some math skills (how many other 8th graders are learning econometrics?) and also some Internet skills. Will keep you posted.