The NFL has used an alignment consisting of eight divisions ever since 2002 (nine years now). The winner of each division automatically qualifies for the playoffs.
In this article, I analyze the parity in the NFL using statistical techniques and comparing how many times teams were able to win their division.
I find that by assessing the NFL as a whole, there is statistical evidence against divisional parity. Further, looking at specific divisions, I found that the AFC East and AFC South are the only two divisions which showed statistical evidence against divisional parity. This is because both the Patriots and Colts have won their respective divisions seven of the nine years, which is more of a direct result of the two dominant quarterbacks Tom Brady and Peyton Manning.
Brady and Manning have each been the starting quarterback for their teams all nine years of the study. A slight exception to note would be the 2008 season, which Brady missed due to injury. The Patriots did not win the AFC East in 2008.
The table below shows the number of divisional championships for each of the 32 teams.
|AFC East||Number of Divisional Championships|
Just using the eye test, it is clear that some divisions have had much more parity than others. I have used a statistical test called the Chi-Square Goodness of Fit test to analyze these results. The test uses a formula to assess the difference between what has actually happened and what we would expect to happen if the divisions had perfect parity.
Since each division has four teams and the study covers nine years, the expected number of divisional championships for each year is 9/4 = 2.25. The further each team's actual number of divisional championships is from 2.25, the higher that team's contribution to the chi-square statistic. Each team's contribution is calculated using the formula:
(Actual championships - Expected champions)^2 / Expected Championships
The total Chi-Square statistic is then calculated by summing each team's contribution. Treating all of the NFL teams as one group, the total chi-square statistic is 53.33. The Chi-Square distribution (with 31 degrees of freedom, one less than the number of teams) shows that this result is statistically significant using a five percent cutoff. This essentially means that the experience data behind the test would happen less than five percent of the time due to random chance if in fact the NFL was balanced with respect to division championships. Five percent is a typical cutoff for significant tests in statistics.
Looking at the teams by divisions (in eight groups of four), the only divisions that produce a statistically significant chi-square statistic are the AFC East and AFC South. The Chi-Square statistic for the AFC East is 13.67 and for the AFC South is 14.56. Using the Chi-Square distribution (with three degrees of freedom now since we are only looking at four teams), both of these divisions show significant evidence against parity.
Note that since there are eight divisions to look at, it is often a common practice to reduce the five percent significance cutoff to five percent/ eight = 0.625 percent. The AFC East and AFC South both show statistical evidence against parity even when using this more strict significance cutoff. None of the other divisions show strong enough evidence against parity even at the five percent significance cutoff.
So why has there been no parity in the AFC East and AFC South? The answer is pretty simple; the Colts and Patriots have won nearly all of their respective division's championships in this period, and the other teams in their divisions have not been able to win many. It is probably reasonable to extend the explanation to the fact that two outstanding quarterbacks have played for the Colts and Patriots during the entire duration of the study. It appears that superstars can and do have an effect on parity in the NFL.
If the study was able to extend over a longer period of time (which would be much more difficult due to the change of divisional alignments), it would be interesting to see how and if the divisional parity levels out after superstar player's careers come and go. Another important advantage of a study covering a longer period of time is that the data would be much more reliable.
In performing Chi-Square tests, statisticians usually like most of the expected counts to be above five. Due to only nine years in the study, all of the expected counts above were 2.25 (clearly less than five). The results of the test still seem to be meaningful though when put into context.
It would also be interesting to compare the divisional parity in the NFL to other professional sports and additionally to assess the parity in the NFL and other leagues by assessing the number of Super Bowls or equivalent championships won.
Stay tuned for further articles.