Parity in College Football - Part II

A little less than a year ago I did a piece on parity in college football inspired by the pitiful performance on top 5 teams against unranked opponents. Looking at the amount of time teams have spent in the top ten, I found real parity in college football ended with the 50's and that, although more teams rotate up at the top now than in the 60's and 70's, those teams that do make it to the top are just as dominant as those from earlier decades.

Parity, of course, can be viewed from many angles, and here I now offer two new views. Using complete team ratings from 1950 to the present, I looked specifically at how much turnover and how much separation there is between teams.

I measure year-to-year turnover using correlation (Pearson's r). The measure compares two sets of numbers to see how consistent they are. In this case, I compare a teams' ratings in 1950 to the same teams' ratings in 1951 to see how much they changed. If they didn't change at all, the value is 1. If they completely flipped over, the value is -1. If they changed randomly, the value is 0. Consequently, a higher correlation means less turnover from year to year and, consequently, less parity in that time period.

The second measure of parity is the standard deviation. The standard deviation is a measure of the average distance from one team to another team. The values can range across the map, but, for our purposes, a higher standard deviation means more separation between teams and less parity.


The dotted lines through the middle are the mean, so if the solid line goes above the dotted line, that time period experienced below average parity (less turnover or greater separation).

Both measures tend to agree that college football is now seeing greater parity than in the early 1990's, and the turnover over the last few years is higher than ever. They also agree that parity was relatively low during the mid to late 80's.

There was little turnover between the mid 60's to mid 90's during which the separation between teams peaked during the late 60's. During the early 70's, a few dominant teams separated themselves from the pack but then were brought back to earth during the late 70's and early 80's, producing a big shift from a very high to a low standard deviation.

This type of parity, though, is not what commentators are usually interested in. They call upsets parity while these measures are able to consider shifts across the board and gives us a little insight about the competitiveness of games across the country. If these trends continue into this season, we could see more teams diving into the top 10 and interesting games, if not great teams, from week 1 to Pasadena.