Figuring for a playoff

I was going to comment on this post at The Big Lead yesterday, but let it slide.  Since The Wiz has brought it up favorably today, though, I thought I’d make one point about a conclusion the author reached there.

Here’s the passage in question:

… Of course, then you get into the argument that before the BCS, the bowls would not always result in a consensus national champion. That’s when the playoff argument comes into play. Which is why I took the liberty of averaging the margins of victory for the Division I-A national championship games of the last 10 (BCS) years and comparing it to the average of the margins of victory of all the Division I-AA national championship games (from 1978). The results: Average margin of victory for D I-A championship games: 14.50 points. Average margin of victory for D I-AA championship games: 12.87 points. Just goes to show that if you want the best chance of having the truly best teams play for the championship, have a playoff. It’s better for everyone and everything…

Well, wait a minute here.  He’s comparing 30 years of data against 10 years of data to make his argument.  What’s the result if you compare identical time periods?

Based on his point spread chart…

… the average margin of victory in 1-AA title games over the last ten years is 17.0.  If I’m not mistaken, that’s actually greater than 14.50.

Now I’m not a statistician, so I have no idea whether a ten year sample (or a thirty year sample, for that matter) is statistically significant.  What I do know is that when you start juggling numbers to justify an agenda, you’re heading into Mark Twain territory.


Filed under BCS/Playoffs, Stats Geek!, The Blogosphere

9 responses to “Figuring for a playoff

  1. kckd

    Something tells me that if the numbers would’ve been opposite, you would’ve made the argument that we need to look at all possible data.


  2. Dude, the numbers are the opposite. I don’t find the exercise convincing one way or the other.


  3. peacedog

    The numbers are not the opposite. You do not come to an “even” comparison by shrinking the larger sample size to fit the small. You just degrade the potential analysis that way. Twain’s quote applies to your own “handling” of the stats moreso than the person who originally made this argument.

    The argument itself is flimsy, there is no question. At best, we can look at it and say “well, we could use more data in both cases, particularly with the tiny sample size for BCS title games”.


  4. pd, the difference here is that I’m not drawing any conclusion from the numbers.

    I certainly wouldn’t argue that the ten year comparison proves that the BCS is better at providing close games than do the 1-AA playoffs. So I’m not sure how Twain’s quote applies to my point.

    I simply think the exercise was a waste of time.


  5. The argument itself is flimsy, there is no question.

    His argument is that closeness of margin is a valid indicator of the quality of the participants in a title game.

    That’s not flimsy, that’s nonsensical.


  6. kckd

    My point is Senator, as I’ve told you many times before, you don’t like the playoff. Instead of taking the good arguments you have and sticking with them, you argue any and everything which in the end weakens your argument overall. You’re right, the article was stupid and silly, so why even try to rebut? I agree on that. Point differential doesn’t amount to a hill of beans. But it amounted to enough for you to put it on your blog and then retort. Stick to that expansion thing and playoffs, that’s by far your best one. The others are just fluff.


  7. Well, as I said, I was going to ignore the post when I first read it. But it’s spreading into the blogosphere where it’s going to become Zookerian noise in the system.

    I know you’re reasonable about this and realize it’s a worthless argument, but others won’t. So I get on my soapbox about it.

    Again, I’m not railing against playoffs here, just dumb arguments in support of playoffs.


  8. dean

    Comparing 10 years to 30 years really shouldn’t matter because it’s an average. However the larger the population the more distorted (good or bad) the mean can become. Still, it’s not much evidence to support or not support a playoff.


  9. Andrew

    I took AP Statistics this year, and as long as the sample size is more than 10 percent of the population, the statistic can be used. That does not mean, however, that a ten year sample is just as relevant as a 30 year sample.