Heavy Lifting - thoughts and web finds by an economist
     I also contribute to Division of Labour Load HL's Front Page
Tuesday, September 26, 2006

Home field advantage in college football

UGA barely won over Colorado on Saturday and fell two positions in the USA Today poll and one position in the AP poll, again being penalized for winning, but not by a wide margin, over what was arguably an inferior opponent.

I have been gathering data on the scores of D1-A football scores since 1869. I now have scores for about 67,500 games played from 1869 through 2005. Here's an interesting stat that merits further investigation:
. sum homescore visitorscore

Variable | Obs Mean Std. Dev. Min Max
homescore | 67437 21.47117 16.42789 0 222
visitorscore | 67437 14.18321 13.25764 0 158

. sum homescore visitorscore if !missing(where)

Variable | Obs Mean Std. Dev. Min Max
homescore | 5494 16.6629 14.20199 0 91
visitorscore | 5494 15.37932 13.71055 0 115

The upper panel reports the average home and visitor scores for all games. The lower panel reports the average home and visitor scores for games in so-called "neutral" sites, including bowl games. Notice that the home team, on average, enjoys almost a touchdown in home field advantage. On the other hand the "home" team at a neutral site enjoys almost no advantage.

What gives? First, there is clearly some homefield advantage in college football (at least for most team), but to what extent? Is home field advantage alone worth almost a full touchdown? I doubt it, as the spread isn't on average seven points in favor of the home team. This suggests that there is something else going on.

Perhaps it is scheduling. Teams invite visitors, especially non-conference teams, with hopes of satisfying the blood lust of the local fans. Therefore, the average visiting team might be that much worse than the average home team, in essense there is a selection problem.

The interesting question is how much of the seven point difference can be attributed to team strengths and how much to home field advantage. The interesting counterfactuals are the neutral site games. Perhaps, as in bowl games, the games pit two equally skilled teams against each other, thereby avoiding the selection problem in regular season schedules. Also, the neutral sites might actually be neutral in the sense that there are equal or at least closer to equal representation of the two teams' fans. More like OU-Texas than, say, Georgia-Florida.

As you might imagine, this is an area of investigation on my part.

Interesting stuff. I believe you are absolutely correct in identifying non-conference scheduling as a key issue. I would recommend either delimiting your study to conference games only to control for this or to control for each teams' strength (via power rankings or similar variable) in a multiple regression analysis.
Post a Comment


Le Chai - galerie du vin



Posts that contain Craig Depken per day for the last 90 days.


Heavy Lifting's Main Page
Email Me
Atom Feed

Heavy Lifting

Great Links

Money I Found Today

Heavy Lifting - Firehose style (56k warning)

Recent Posts

- Home field advantage in college football



Site Meter Blogroll Me!


Modified maystar design
powered by blogger