Calling our wonks: analysis shows passing correlates with winning
I thought this was a good piece in the NYT http://fifthdown.blogs.nytimes.com/2010/08/31/why-passing-is-more-important-than-running-in-the-n-f-l/?hp
Granted this is only for the NFL, but interesting stuff for our own group of statisticians. Burke explains more in the text, but this table of win correlation by competency is contrary to some held beliefs. Not sure if our guys have looked at similar dimensions in regards to winning. It is worth the jump since the table is a bit confusing without the full context. Essentially 8 wins is the base. Competency on offense is rewarded with a win, while defensive incompetency is punished where the example of poor pass defense results in 1.23 losses (normally show both positives where defensive pass competency results in wins, just stylistic).
| Normalized Variable | Coefficient |
|---|---|
| Constant | 8.00 |
| Off Passing | 1.54 |
| Def Passing | -1.23 |
| Off Running | 0.44 |
| Def Running | -0.44 |
| Penalty Rate | -0.25 |
| Off Int Rate | -0.32 |
| Def Int Rate | 0.45 |
| O Fum Lost Rate | -0.43 |
| D Fum Lost Rate | 0.25 |
August 31st, 2010 at 9:36 AM ^
Drop the OT tag! This thread'll be as dead as Molly Ringwald's career post-Breakfast Club!
EDIT: Trying to find that damn Family Guy clip...
August 31st, 2010 at 11:21 AM ^
It was a NFL study and now specific UM note, so I thought it would then be OT. Anyway, I can't keep it all straight. Good Family Guy reference though.
August 31st, 2010 at 9:54 AM ^
Correlation does not imply causation.
August 31st, 2010 at 10:29 AM ^
You are insinuating that there is no causation just because only a correlation was shown. This, of course, is not true!
Causation is often extremely difficult to prove. In fact, I do not think any analysis on this site has ever proved causation.
August 31st, 2010 at 11:23 AM ^
That is all.
August 31st, 2010 at 11:30 AM ^
because success in any major part of the game will correlate with wins, a) because it all adds up and b) because competency in one area of the game will tend to mean competency in others.
August 31st, 2010 at 9:56 AM ^
Touchdowns correlate with winning. No shit? But really, yes, passing correlates to winning to due the big play factor. In order to pass well, you need to run somewhat well.
August 31st, 2010 at 9:59 AM ^
Mike Martz & Leach-beard the Pirate would like to disagree with you. Although Martz's last two stops didn't exactly correlate with wins either.
August 31st, 2010 at 5:40 PM ^
had a hall of fame running back in Marshall Faulk on his "greatest show on turf" teams. Leach's teams were also pretty effective running the ball vs the likes of Texas in 2008.
August 31st, 2010 at 10:50 AM ^
If you're John Madden, making that point would come naturally to you. Also, in football, the team with the most points at the end, is generally going to win. Not sure how it works in other sports, though.
August 31st, 2010 at 10:02 AM ^
I'm not a wonk, but I am an engineer...
Remember this shows correlation, not causality. The NFL is increasingly a passing-dominated league (look at how frequent the 4000 yd seasons are for QBs now) since they are tweaking rules to favor receivers and protect QBs over the last several years.
So I would expect these results. On average, if a team in the NFL can pass, it has a an offense that scores points. If it can't pass, it doesn't have an offense that scores points. Therefore passing competence would correlate logically with winning football games.
I'd expect similar results in CFB just anecdotally since it seems there are more teams running passing spreads and pro-style offenses than three are teams running spread-option-style offenses or option offenses. But there are a good amount of run-dominated offenses out there, so I would think the correlation between passing offense and wins to be a bit weaker than in the NFL while the correlation from run offense to wins is probably a little stronger. Again, because it's a correlation, not necessarily a causation.
August 31st, 2010 at 10:03 AM ^
It's not terribly surprising. The game is much more passer friendly than it used to be due to massive rule changes. They protect QBs and WRs, they allowed blockers to extend their arms, and they call a lot more PI. Over the past 30-35 years, the NFL has consciously made a decision to promote the "exciting" passing game. So basically, this analysis is just validation that the methods they're using to encourage passing are working.
August 31st, 2010 at 10:08 AM ^
Could that be due to the fact that teams with the best QBs are going to throw more? I mean, if there were 10 RBs going for 2000 yards in a season, you would see that running correlated to winning.
August 31st, 2010 at 10:18 AM ^
Seems like you are crying out for some help from Mathlete.
August 31st, 2010 at 10:45 AM ^
Theory: It is much harder to successfully pass the ball than run the ball, in terms of complexity. So a team that can successfully pass is not winning because they pass, they're simply demonstrating their greater competence - and of course, a better team is more likely to win. Correlation, not causation.
August 31st, 2010 at 11:01 AM ^
What do they say about scoring more points than the opponent, does that correlate to winning?
Of course if you're competent in one area it gives you a better chance to win.
August 31st, 2010 at 11:19 AM ^
Old school commentaters perpetuate the idea that you have to emphasize running the ball, and maintaining possession, in order to win. These are outdated thoughts. While those things are relevant and worthy football philosophies, they aren't singular solutions to winning the game as the vast majority of football commentary would have us beleive. Sure, there's no causation here, only correlation; but, so what? All the "pound the rock" crowd has is correlation, too. In fact, they have less of it that the "pass the rock" crowd does.
The discussion of correlation vs. causation is frequently unnecessary when it comes to sports discussions. We're not trying to cure cancer here, folks, and noone is trying to win the nobel prize. Sure, if you're trying to draw a line between GDP and World Cup performance, you might want to include a plausible theory for the relevancy of the independent variable to the dependent variable. However, if you're trying to figure out which guys you should recruit to play football, all you need is correlation. Recruiting ratings are simple correlation, that doesn't mean they are irrelevant.
Guy 1: "Hey, that guy us ridiculously fast, friggin huge, and constantly snapping quarterbacks in two...I think I'd like him on my team."
Guy 2: "No, no guy. Speed and Size only correlate to success on the football field. They don't necessarily cause it."
Yawn.
August 31st, 2010 at 11:22 AM ^
correlates to winning.
You may quote me on this.
August 31st, 2010 at 4:00 PM ^
I don't see how this is applicable to college since the statistics are NFL based. College offenses are way more diverse then the NFL and I'd like to keep it that way.
August 31st, 2010 at 5:46 PM ^
Interesting, but I'm not sure how relevant it is to the college game. NFL defenses are insanely large and athletic compared to their college counterparts, and holes in the line close in the blink of an eye. Most teams can't run the ball that effectively, so they have to turn to the pass. It's easier for run-first offenses to move the ball in college than in the pros.