The small numbers problem

Getting good results out of a statistical analysis requires a large sample size. The so-called small numbers problem masks variability in the data.

Huh?

If there are only a few observations, and they're really close together; there isn't much difference between them. Think about it this way, there are about 9 million people in the Chicago region, with probably a huge difference in preference for what to put on a Maxwell Street Polish. If you ask three friends for their preference, you would hardly be able to claim that their responses accurately represent the diversity of tastes for Maxwell Street Polishes for 9 million residents of the Chicago Region. Therefore, you need to get a larger sample. As your sample (n) increases you get a better and better representation of the variability of taste for the entire population, and at a certain point they essentially become the same. In other words, for some stuff there's no reason to do a census. It's not that we don't care what people think; it's that statistically speaking, no new information would probably be added.

This brings me to the real point of this post: The Bowl Championship Series. I have a problem with the BCS, and before you start shouting for a playoff, finish reading the post! I have a problem with the BCS because it relies on such a small sample size to judge each team. Each team plays around 12 games in the Football Bowl Subdivision - hardly a large enough sample size to determine a teams performance variability. Therefore, it's really easy to trick a system that relies on such a small "n" to think that your a good team. In comparison, NCAA Division I Men's Basketball teams play about 30, which is funny because 30 is usually the sweet spot for statically useful samples - when your "n" hits 30 you can usually assume that your sample variability is pretty close to the population. The logical response, for College Football: "Well we obviously have to increase the number of games teams play by instituting a playoff." I can hear Kirk Herbstreit's whiny voice now. Does anyone see the problem with this?

I do.

The average rating for a 2010 bowl game was about 3.89. The max was 15.29 for the BCS Championship between Oregon and Alabama (BCS Ratings). Let's contrast this with Men's Basketball. The average television rating for a NCAA Tournament game was 6.00 (Tourney Ratings). If we assume just a 1 for 1 product switch here, the TV folks (read ESPN) would gain an instant 2 point bump on average by switching to a playoff format. But average would probably be a bad way to look at this. If I exclude the BCS Bowl games, the average rating for a "Run-of-the-mill" Bowl drops an entire point, so implementing a playoff could conceivably DOUBLE the ratings for games people aren't really watching. See where I'm headed? There's a reason so many football pundits are advocating for a playoff, and it ain't fairness or equity. It's dollars and cents. They're capitalizing on our deep desire for what can be billed as a clear, decisive, and just outcome, so they can make more money. They're telling us that a consensus national championship is what we want.

The problem with this whole picture is that it forgets that college football players are STUDENT-ATHLETES. The season historically ended in mid-November so that these STUDENT-ATHLETES could take their finals in early December. Now we have a college football season that ends in early December and a bowl season that actually extends into the Spring Semester for these kids (Remember…these are just kids in their early twenties!), and with a playoff how much longer into January will we stretch the season? All of it justified by the need for a consensus national champion, simply so a select few networks can increase their ratings, improve their ad-buys, and make more money. I'm not exonerating the BCS here either. This program is all about funneling dollars as well. The key question is where is the money being funneled. The problem, the real problem with the BCS, is that ESPN is getting cut out of lots of cash under the BCS. Ultimately the network folks don't really care who wins the National Championship.

In my opinion, if you want to fix the College Football National Championship problem, the whole paying the players problem, and reduce cost of college across the board, then all profits from any aspect of collegiate athletics should be immediately put back into the University or College. Not the athletic department, not ESPN, not boosters. It should be given to students. Do it and you'll see people come up with a solution in a jiffy. We'll get a national champion and not get robbed in the process. The real problem is in our almost pathologic need to know "who is the best" and the ability of a small group of people to capitalize on that need. Both of these come at the expense of a bunch of college kids - athletes or not.