FROM THE CHFF CREW: This is great, folks. Two Trolls going at it, their stubby little legs and arms flailing at each other, clumps of back hair being tossed all around. Last week, reader Mike Stickles argued that "establishing the run" is overrated. Reader Mark Wald counter-trolled, saying that Stickles (and the CHFF) failed to look at the importance of rushing attempts. Well, Stickles is back, with a little pig-skin-kwan-do of spreadsheets and Cold, Hard Football Facts. Look at those chubby little fists of facts fly!
From Mike Stickles:
I read the critique by Mark Wald in your "Troll slams CHFF over 'Establish the run' findings" article. He had a valid point and it deserved an analytical response, so I went back to the stats I'd collected, plus dug up a few more, and prepared one. He's right about rushing attempts correlating strongly with winning, but the Cold, Hard Football Facts are right that teams rush to sit on a lead, not to get a lead. Here's what I came up with (a spreadsheet with the summary data is attached):
Recently, Cold, Hard Football Facts reader Mark Wald made the assertion that number of rushing attempts is more important than rushing yards or yards-per-carry. Since my previous analysis of the numbers neglected to consider rushing attempts, it seemed only right to correct that oversight, and in the process see if the Cold, Hard Football Facts can answer the question of which comes first - rushing or winning?
Do teams win because they run the ball more, or do they run the ball more because they are winning? If establishing and maintaining a running game has a correlation to winning football games, we would expect first-half rushing attempts to correlate with winning about as well as second-half rushing attempts. On the other hand, if winning teams usually have more rushing attempts because they're trying to sit on a lead, we should see second-half rushing attempts correlate far more strongly with winning than first-half rushing attempts.
A close look at the data clearly shows the latter: teams run because they are winning. They do not win because they are running.
The whole game
Mark's primary assertion - that total rushing attempts correlate very strongly with winning - is definitely supported by the Cold, Hard Football Facts. In looking over the 218 games analyzed for the previous article, we see that teams which ran more often than their opponent were 171-42 (.803). Looking at number of rushes: 
  • Teams which ran fewer than 20 times in a game were 5-58 (.079)
  • Teams which ran 20-24 times in a game were 19-70 (.213)
  • Teams which ran 25-29 times in a game were 44-51 (.463)
  • Teams which ran 30-34 times in a game were 68-25 (.731)
  • Teams which ran 35-39 times in a game were 48-11 (.814)
  • Teams which ran 40 or more times in a game were 34-3 (.919)
Very impressive stuff. But still, the question is - which comes first, the running or the winning?
The first half
Again, if running more often causes teams to win more often, we should expect to see that correlation show up in the first-half numbers. But when we look, there's not much there to see.
Over those same 218 games, teams which ran the ball more often than their opponents in the first half were 115-88 (.567). This does support Mark's assertion that more rushing attempts is more important than more rushing yards (115-100, .535) or more yards per carry (107-108, .498).
But having more passing attempts has nearly the same effect as having more rushing attempts (114-90, .559).
Plus, more rush attempts does not produce as impressive an effect as more passing yards (137-78, .637) or more yards per pass attempt (148-70, .679).
Looking at the actual number of rushes in the first half shows the same picture:
  • Teams with fewer than 10 rushes were 27-50 (.351)
  • Teams with 10-12 rushes were 48-54 (.471)
  • Teams with 13-14 rushes were 42-35 (.564)
  • Teams with 15-16 rushes were 40-36 (.526)
  • Teams with 17-18 rushes were 30-21 (.588)
  • Teams with 19 or more rushes were 31-22 (.585)
There is a correlation to winning football games here, but it's not a very strong correlation (stronger than the trend for yards per carry, but weaker than the trend for rushing yards), especially when compared to the number of rushes for the whole game. And we can see a similar (though slightly weaker) trend in terms of passing attempts:
  • Teams with fewer than 13 pass plays were 35-41 (.461)
  • Teams with 13-15 pass plays were 53-54 (.495)
  • Teams with 16-18 pass plays were 52-54 (.491)
  • Teams with 19-21 pass plays were 41-40 (.506)
  • Teams with 22 or more pass plays were 37-29 (.561)
The data seem to suggest that it's not running more often that's important in the first half, but simply having more plays, period, regardless of whether you run it or throw it. Looking at the number of overall plays run in the first half bears this out:
  • Teams with fewer than 24 plays were 17-34 (.333)
  • Teams with 24-27 plays were 38-51 (.427)
  • Teams with 28-31 plays were 57-50 (.523)
  • Teams with 32-36 plays were 67-58 (.536)
  • Teams with 37 or more plays were 39-25 (.601)
The correlation for total plays (more plays = more winning) is stronger than for either pass plays or running plays alone, though it's still not as impressive as the trends for first-half passing yards or yards per pass attempt. Now let's look at the second half numbers.
The second half
Teams with more rushing attempts than their opponents in the second half went 179-29 (.861) over the 218-game span of the study. In terms of actual number of rushes:
  • Teams with fewer than 10 rushes were 5-91 (.052)
  • Teams with 10-12 rushes were 19-59 (.244)
  • Teams with 13-14 rushes were 26-35 (.426)
  • Teams with 15-18 rushes were 65-24 (.730)
  • Teams with 19-21 rushes were 47-8 (.855)
  • Teams with 22 or more rushes were 56-1 (.982)
Clearly, the rushing that correlates with winning occurs in the second half, not in the first. And that strongly suggests that teams are not rushing to gain the lead, but to sit on a lead they already have. To put what we suspect to the test, let's look at how play selection changes depending on whether or not teams have the lead.
Analyzing play selection
A review of the play-by-plays of last week's games (2007 week 6) shows 1693 plays (excluding spikes, kneeldowns, and aborted snaps). Of these, 726 (42.9%) were runs. Here is the breakdown for each quarter, divided up by whether or not the team has the lead (we'll ignore periods when the score is tied):
Team has the lead:
  • 1st quarter: 26 runs, 32 passes (42.8% runs)
  • 2nd quarter: 48 runs, 76 passes (38.7% runs)
  • 3rd quarter: 89 runs, 82 passes (52.0% runs)
  • 4th quarter: 99 runs, 74 passes (57.2% runs)
Team is behind:
  • 1st quarter: 38 runs, 46 passes (45.2% runs)
  • 2nd quarter: 79 runs, 114 passes (40.9% runs)
  • 3rd quarter: 79 runs, 109 passes (42.0% runs)
  • 4th quarter: 50 runs, 171 passes (22.6% runs)
There is very little difference in play selection in the first half. However, in the third quarter, teams with the lead begin to run more often; in the fourth quarter, teams with the lead run even more often, while teams which trail all but abandon their running game. Clearly, what we suspected is indeed true - teams run more to sit on a lead, not to get a lead.
Teams gain a lead by any means necessary, running or passing. Once they have the lead, they run in an effort to hold on to it. Once a team falls behind, they pass to play catch-up.
Bottom Line
Maybe it's not totally clear what "pundits" mean when they talk about "establishing the run." They could mean run the ball more often, run the ball for more yards, or run the ball more effectively, as evidenced by yards per attempt.
But whatever criteria they use, they talk about "establishing the run" as if it has a high correlation to winning football games. It does not. The Cold, Hard Football Facts clearly show that, on the contrary, winning that leads to more running.
They also show, as evidenced in the previous article, that passing effectively is more likely to lead to victory than "establishing the run," no matter how you want to define this term.