Third-and-percentages: A do-over

This article is part two of a series on third downs in college football, you can find part one here.

Last Tuesday, I published an article detailing how successful teams are at converting third downs based off of the distance to go and the type of plays teams were running. The data I published last week suggested that teams converted third downs at a much higher rate when passing the ball than when running the ball, and not just for third-and-longs either. This trend appeared for any distance greater than three yards to go. I was lucky enough to have my data reviewed by PeterinVA, who pointed out a mistake I had made in the data assumptions. The NCAA records sacks as rushing plays (I have no idea why). So when I tabulated the play-by-play results by play type, all of the sacks were counted as rushing plays. This explains the drastic difference between pass conversion % and rush conversion %; it only counted the passing plays where a pass actually occurred.

So, I have fixed the raw data and identified all of the plays where a sack occured and gotten the correct results, and we can now proceed as planned.

The Data:

I'll say it again and every time after this, thanks to cfbstats.com for providing the play by play data for every season since 2005. This project wouldn't be possible without it. In my first post, I detailed how I got a list of each and every third down by distance in the last five years (2008-2012). From there I looked at how often teams passed the ball at each distance. The following chart outlines the total pass % by distance to go for each year since 2008, with all garbage time plays removed.

(Also, I redefined garbage time by the same standards that Football Outsiders use when I fixed the sack results. So the data is now more consistent. If it isn't apparent yet, I have already classified my first post as a do-over.)

There seems to be a clear rise in how often teams pass on third down until about third-and-6 or third-and-7, and then teams pass at about the same rate for any distance longer than that. The trend does start to slope back towards rushing the ball more as the distance-to-go gets more and more insurmountable. I think this is because some coaches just start to play it safe and run the ball to prevent an interception and set up the punt. The good thing is that there does not seem to be any trend year to year, so I will be using the average of all third downs from 2008-12 for any future reference to pass %.

Now that we have an average passing % on all third downs by distance-to-go, what can we learn about third down conversion % by distance? I'll now combine a chart from my earlier post detailing the conversion percentages by play type. I'll also include how often teams are passing the ball by distance.

Conclusions:

I'm not really sure what value we can gain from this chart. Teams start passing the ball more than half the time at about three yards to go from a first down. While teams quickly pass the ball more than they run the ball, the conversion % is actually pretty stable. It is only until teams have seven to eight yards to go for a first down that the percent of passing plays that gets you a first down is higher than the same percentage for running plays. But, teams are also passing the ball four times out of five at this point. It's possible that when teams run the ball at this distance, they have noticed that a draw play will fool a defense or something like that. The third-and-13 spike is still there, and thanks to all those in the comments of the previous post, I'll be investigating some of the ideas you guys had in the future.

I'm not really sure what direction I will go next with this data, feel free to suggest anything you would like to see and I will continue to update this series as I make progress.