Sunday’s New York Times had an article about the passing game in the NFL. Apparently The article is full of statistics and sure to please sports nerds like me. The face of football games is changing, with more frequent shorter passes (with a high probability of success) replacing longer passes (with a lower probability of success). As the article notes, the 2008 NFL season boasts the most points per game, the highest pass completion percentage, the lowest interception percentage, and the highest passing rating in recent history. Click on the image below to view some of the statistics.
I found this article really neat. I am not sure if the game is really, fundamentally changing, although the statistics show a clear trend that is steady over time. I have always wondered if football is really in steady state. In steady state, the offense and defense would be balanced each year, producing similar statistics. I’m not sure that is a good assumption. There are very few exceptional athletes that can play football at a professional level. What if there is a shortage of good defensive players? (Or maybe the would-be defensive players are slowly leaving the defense to do other things). Maybe the safeties and corners (defensive players that prevent pass completions) have gotten really good, which has encouraged shorter passes? Maybe Tom Brady’s injury is enough to skew passing statistics?
There have been things that change how the game is played. Instant replay, for example, has resulted in more accuracy in calling big plays and in more turnovers. Increased fines for rough tackles and late hits (to prevent injuries) may cause more tentative defenses. Neither of these factors is central part of football, but both would be correlated with more frequent, shorter passes (correlation is not causation!) Just some food for thought.
Update: After looking at the distributions for the parameters in the article, I noticed that all of the distributions overlap big time (variance is an important part of the story). I ran one-way ANOVA tests on the number of attempts per game, completions per game, passing yards per game and sacks per game from 2001 – 2008 using Yahoo! Sports data. I did not find any statistically significant p-values. There are probably better statistics tests to run, but I just covered ANOVA in my class, and I want to prove to any of my students who may be reading that you can really use the statistical methods we cover is class.