Game Attendance

It has long been my thesis that the more successful the program, the better the attendance. So, I'm going to geek out on you. It's the cost of having a Cap'n who's into statistics.

I decided to test my theory. The results that I am posting are pretty informal - more in the category of exploratory rather than confirmatory. If I were doing this for publication or for a client, I would never submit such "sloppy" work. But the results are pretty interesting. I hope you enjoy thinking about it. 

Data: I got the data for the 2023 from the NCAA website. After looking at it I chose two variables for simple regression analysis: For the dependent variable I chose average percent of stadium capacity. I think this variable somewhat controls for differential among school size, alumni, and college town population. ACU, for example, has a 12,000 seat capacity. If we average 8,000, for example then our % capacity is 66.7.  Tarleton has a 24,000 seat capacity so it takes them 16,000 to have 66.7 capacity. Since they have 4-5 times more students and alumni, I think it's a fairly good comparison.

The independent variable is winning percentage. I also looked at total wins but there was basically no difference in the results.

I took a couple of steps to smooth the data...I eliminated the Ivy League. They have very large legacy stadiums but have de-emphasized football. Penn, for example, plays in 70,000 seat Franklin Field in Philadelphia. Then I eliminated all stadiums smaller than 5000 capacity. The Patriot League, for example, has several 2-3000 capacity stadia. 

So, what are the results:

With a sample size of 110, there is a .35 positive correlation between winning percentage and the percent of your stadium that is filled. That is quite a high correlation. The stats mantra is "correlation is not causation" but there is a very strong relationship between winning and attendance.

The prediction equation of the simple linear regression is significant to .0002. It suggests that if an FCS team wins 10 of 12 games for example, average attendance could be predicted at 68% of stadium capacity plus and minus error. For nerds like me, the R-squared is a low .12 suggesting that the error is mostly random and not very related to winning percentage. If 68% seems low, consider that throughout all FCS programs, capacity averages roughly 57%. (Last year ACU was at 59% this year so far its nearly 74%.

Are you asleep yet? If I were to mess around more, I might run a multiple regression and try to identify other meaningful variables but I doubt it would give much more info. So my take away is that the increase of attendance capacity above the mean is definitely influenced by winning but maybe only by about a third. History, tradition, location, money etc all play a part as well.