Final Update for College Football Performance as Affected by Distance from Home

The moment of truth: is college football performance affected by distance from home? As far as I can tell, the answer is no.

I have used R for easy data visualization and statistical manipulation. The two statistical metrics I will use are correlation coefficient and p-value. Correlation coefficient attempts to measure the degree to which two variables are related. It ranges from -1 to 1, with -1 indicating a very strong negative relationship, and 1 a very strong positive relationship. P-value is a way to assess how likely the relationship in the data is caused by randomness. Typically, a p-value less than .05 is considered significant, that is, not due to randomness. The first relationship I tested for was the simplest: Distance vs. Average Yards per Attempt.


As you can see, there is no clear relationship here. The correlation coefficient, -.03, affirms that there is no correlation between the two variables. The p value, 0.47, also indicates that there is no relationship here.

Next, I looked at Distance vs. (Average Yards per Attempt) / (Rating). I used this as a basic scaling for the skill of the player—to try to control for the quality of the recruit.


Again, the data points do not create any real pattern. The correlation coefficient is -.04, and the t value is 0.4

Adjusted Yards per Attempt does ignore rushing yards, which is a large part of many quarterbacks’ production. My metric for “value”, which adds in rushing yards per game and considers the total number of games the quarterback played for the school, has a stronger correlation with rating than simple adjusted yards per attempt.

This is rating vs. adjusted yards per attempt. The correlation coefficient is .26, which is a weak relationship, but the p-value is effectively zero, so the relationship is certainly real.


This is rating vs. value. The correlation coefficient is .31, so it is not a much stronger correlation, but it is stronger. The p-value is also effectively zero.


This is distance vs. value. The correlation coefficient is -.03, and the p-value is 0.47. Still no relationship


This is distance vs. (value) / (rating). The correlation coefficient is -.04 and the p-value is 0.43.


Perhaps simply dividing value by rating is not the best way to scale value to rating. So, I took the model regression for rating and value to create an expected value for each rating. Then I subtracted expected value from real value to create Value Above Expected.


Yet again, there is no real relationship. The correlation coefficient is -.03 and the p-value is 0.45.


I feel comfortable concluding that distance from home has no effect on a college football player’s performance. Obviously, there are some assumptions being made, like that quarterback performance would be similar to all player’s performance. However, I can think of no reason why quarterbacks would have a different effect from distance from home.


As promised, here is a folder that contains my spreadsheet with quarterbacks, as well as yearly recruit data.