Which Passing Stats Are Important?
I thought I'd take a quick look at this today. I took the statistics of all quarterbacks with at least 6 games played this season. There were 29 players. Then I looked at the Won/Lost Percentages for their teams and compared them to their stats.
Correlation explained. Correlation measures the degree to which two sets of numbers are related. For example, suppose we were to examine the numbers of hours each student in a class spent studying, with their numerical grades on a test. We would expect there to be a fairly high degree of correlation between the two. Not perfect, because some students can do well without studying hard and others just get by with excellent effort. Similarly, if we were to correlate the number of hours students spent watching TV in the last three days before the test, we would expect there to be something of a negative correlation with the grades. This would probably be a of a lesser magnitude though, since there are plenty of other ways to goof off before a test than watch TV, and some students study hard and watch TV.
Correlation ranges from 100% to -100%. It is important to note that correlation is not causation. The classic example in this regard is that of the number of churches in a town and the number of crimes in that town. If you research it, you will find a high degree of correlation, which might lead you to believe that churches cause crime. However, in fact, both the number of churches and the number of crimes in a town depend on a third variable--population. As the population rises, the number of churches and the number of crimes increases.
Negative statistics. There are several negative statistics in a quarterback's record, and as we would expect, these correlate poorly with winning. For example, interceptions correlate at -50%, and times sacked at -14%. If we calculate out the interception percentage we find that it correlates at -55%, and sack percentages correlate at -15%. This basically indicates that interceptions are far more devastating to a team's won/lost record than sacks (not exactly a surprise).
Positive statistics. There are many statistics in a quarterback's record that correlate positively with winning. Starting with just the counted stats, we find that pass attempts correlates very weakly with winning--just 3%. This is not surprising at all; since losing teams pass more often as a percentage of their total plays, but probably get fewer plays on average. Completions correlate much stronger with winning at 29%. Yards gained correlates at 35%, while touchdowns manages 38% correlation.
If we go to calculated stats, the correlations step up a notch. Touchdown percentage correlates at 44%. Yards per attempt correlates at 59%. Completion percentage (which most folks are unimpressed with) correlates at 67%. And the NFL's passer rating formula correlates at 68%.
The fact that completion percentage rates so highly is something of a surprise, but if you look at the league leaders it is hard to deny:
Daunte Culpepper
Ben Roethlisberger
Chad Pennington
Drew Brees
If you had to draw up a list of MVPs for the season thus far, all those guys would have to draw serious consideration. Their teams are a combined 22-7, they've thrown 51 TDs and 14 interceptions between them.
