lOMoARcPSD|22015503
Regression Project Write-up
Introduction to Probability and Statistics (Liberty
University)
lOMoARcPSD|22015503
Catherine Morales
December 5, 2017
Regression Project
I chose to look at football stats because I spent the entire weekend watching the
Conference Championships (ACC, SEC, Big 10, etc.) with my sister’s family. We are huge
Clemson fans, so there is always discussion about how Dabo Swinney is simply the best coach
ever. One reason Clemson has become a football powerhouse is definitely due to his leadership,
and I wondered if that fusion of coach and team was a trend in other top 25 teams. Since the
playoffs and National Championship have yet happen this year, I looked at the final rankings
from last year, and the years that those coaches had spent with the team.
In my calculations, I found that while there was a weak negative correlation between a
team’s rankings and the years they had had their coach, there was not enough to suggest a
relationship. The main outlier was Utah, and I found that if you reduced Utah to 3 or 4, the
correlation bumped to almost 50%, and the test statistic was large enough to suggest a
relationship. This only goes to show that there are many, many factors that lead to the success
of a football team, and one thing that may slightly influence that is the experience of the head
coach.
For my regression model, I used the rankings as the x-values, and the head coach’s
seasons as the y-values. I did this so that when you punch in a team’s rankings, you get an
approximation of the years their head coach has been around. Since the correlation coefficient is
so low, and the coefficient of determination is very small (0.072), the equation the line it forms
are not very accurate to the data. Also, because I used every number between 1 and 25 as my x-
values, there are no number within the region to test the equation with. However, when you
look at the scatter plot, you can see the slight downward trend in the data.
