1. 1
Wage Discrimination Amongst NFL Athletes:
Group 3b:
Ryan Williams
Patrick Gill
George Ulloa
Abstract:
The purpose of this study was to determine if there is wage discrimination
between NFL athletes of different race. We examined numerous NFL players, both
2. 2
black and white, from multiple positions, their career statistics and their salaries. Using
data visualization and regression software, our goal was to use the data to determine if
white NFL players are paid more than black NFL players or vice versa. Naturally,
different positions call for different salaries, so we broke down our data by position. We
gathered data for 40 Quarterbacks, 40 running backs and 50 Wide receivers/Tight
Ends, all of whom are actively playing in the NFL. The data collected consisted of many
performance statistics, the respective 2015 annual salaries of the players, as well as the
players’ race and how many years and games the players have played. After visualizing
the data and interpreting multiple regressions, we found no evidence that race plays a
determining factor in what a player is paid. However, we did find that certain positions
are heavily dominated by a specific race. The exact cause for this is unknown, and can
possibly be contributed to multiple factors, and could be a topic of interest for a future
project.
The Question/Hypothesis:
If salaries paid to black and white NFL players of the same position with similar
career statistics vary, then race is a factor of determining what a player is paid. Do white
NFL athletes tend to be paid more? Do black athletes tend to be paid more? Is race a
factor of determining pay? What other factors lead to higher pay?
The Data:
Table 1 Descriptive Statistics
N Mean S.D. Min Max
Salary in mil. per 127 7.49 6.28 0.55 22.13
3. 3
year
Race=1 if Black 127 0.65 0.48 0.00 1.00
Touchdowns 127 68.14 81.93 7.00 535.00
NumAllStarGm 127 1.90 2.45 0.00 14.00
Games Played 127 103.35 51.00 12.00 262.00
Years Played 127 8.54 3.65 1.00 18.00
Total Yards 127 11149.98 11847.51 1097.00 71117.00
Pass Yards 127 6473.98 13199.40 0.00 70446.00
Rush Yards 127 1720.98 2547.48 -1.00 11388.00
Rec Yards 127 2955.02 3417.24 0.00 13648.00
Position=1 if QB 127 0.31 0.46 0.00 1.00
Position=1 if RB 127 0.30 0.46 0.00 1.00
Position=1 if WRTE 127 0.39 0.49 0.00 1.00
Observations 127
The Dependent Variable is the annual salary of a player and the other statistics
are the independent variables. Nine independent variables were chosen to be evaluated
for each player’s career and are as follows:
● GamesPlayed- Career number of games played by an athlete.
● YearsPlayed- Total years played in the NFL by an athlete.
● NumAllStarGm- Career number of pro bowl appearances by a NFL athlete.
● Touchdowns- Number of Touchdowns in a NFL player’s career.
● Race1- The race of the player, black or white was recorded. For the purpose of
the regression software, a player was assigned a “1” if he was black and “0” if
white.
● TotalYards- An NFL athlete’s career yards gained.
4. 4
● PassYards- Career passing yards for a NFL athlete.
● RushYards- Career rushing yards by an athlete.
● RecYards- Career receptions by a NFL athlete.
● Position- Position of a player.
● Salary in mil per year (Dependent variable)- A NFL player’s annual salary
(2015)
Tableau Visualization:
5. 5
https://public.tableau.com/shared/MTF8D98RM?:display_count=yes
This Tableau bar graph enables us to visualize the correlation between the
number of touchdowns both Wide Receivers and Tight Ends completed and their
salaries. In order to make a fair comparison to determine if race played a key factor in
the amount of money a player makes, we observed the number of touchdowns
completed by both a black player (Julio Jones) and a white player (Eric Decker). Both
share almost identical stats (312/306) and are around the same age. However, the data
shows that Jones made almost double the amount of pay than Decker. Our group has
come to the conclusion that other factors such as size, speed, stamina, injury prone,
etc.… also plays a role in accurately determining a player's worth. NFL teams are
6. 6
responsible for making precise assessments of players not only to improve their
organization, but to open more salary space to acquire more players.
Regression Results:
Table 2: Regression Results
(1) (2) (3) (4)
Salary
in mil. per year
Salary
in mil. per year
Salary
in mil. per year
Salary
in mil. per year
Race=1 if Black -
4.806***
-1.944+ -1.659 -0.499
(4.42) (1.69) (1.54) (0.46)
Touchdowns 0.0373*
**
0.0137 -0.0104
(4.27) (0.44) (0.31)
NumAllStarGm -0.0407 0.234 0.277
(0.16) (0.82) (1.00)
Games Played 0.0372 0.0265
(1.27) (0.81)
Years Played -
1.127**
-
1.102**
(2.94) (2.93)
Total Yards 0.0002
33
0.0004
18
(1.13) (1.57)
Position=1 if QB 0
(.)
Position=1 if RB -2.787
(1.54)
Position=1 if WRTE 0.979
7. 7
(0.44)
Constant 10.60*** 6.283*** 10.36**
*
10.45**
*
(12.13) (5.73) (7.37) (6.06)
Observations 127 127 127 127
R
2 0.135 0.315 0.416 0.467
Adjusted R2 0.128 0.298 0.387 0.431
Wald Chi2
Prob > chi2 0.000 0.000 0.000 0.000
F 19.527 18.865 14.237 12.913
Log lik. -
403.793
-
388.974
-
378.874
-
373.079
Absolute t statistics in parentheses
+
p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001
The data shows that race is not a significant factor in determining a player’s
salary. We have already thought of modifications that we could make to our data set to
see if the results could potentially change. First, a much larger dataset could be
gathered, encompassing all starting players in the NFL. Also, games played and years
played can skew the data. Naturally players in the NFL longer are going to have higher
stats than those players who are knew. Thus, if a dummy variable was created for the
mean average of games and years played for example, the data could potentially show
if race plays more of a statistically significant role in determining salary. However, with
outside knowledge of the NFL and the players’ stats and salaries combined with our
regression results, it does not seem that race plays a factor in determining salary.
Conclusion
8. 8
The regression in our study has shown that the race variables are not statistically
significant to the dependent variable of salary by race. There is more research to be
done in the future concerning discrimination in pay--that may justify our observations
more accurately. Like all sports, production is the most important factor to success so in
theory a players pay should reflect his on the field performance fairly accurately.
Because contracts often span 3-4 years, a player’s current contract may reflect his
expected production from earlier years and may not be a proper representative of the
player’s current performance. This gap between contract negotiations helps explain our
observation that player’s stats may not equal their salary.
Contracts are also given out for expectation of future production, not for past
production. This is important to understand because while a player like Peyton Manning
has all-time great stats, he was not as highly paid in our dataset because his play was
expected to decline (he retired the year after this data was observed, confirming his
decline in play).
One area where race may still be a factor is the actual categorizing of player by
position. On a position by position comparison players are generally paid based on
performance, wide receivers being paid similarly to other wide receivers, and running
backs salary comparable to other running backs. Initial observations show a high
correlation of specific races to specific positions, such as a high number of white
quarterbacks, and a majority of running backs and wide receivers being black.
Preliminary research has already been done on how player’s positions are chosen, and
there is some evidence that subconscious racial stereotypes play a part--such of blacks
being “physical specimens” and whites playing more cerebral positions with leadership
9. 9
roles on the field. One article caught our attention during our research, and it discusses
how black NFL players are often “stacked” in roles where there is less thinking and
more physical play, where “signal callers” who are often paid more such as middle
linebackers centers and QB’s are more likely to be white (An Amazing Specimen). This is a
very interesting statement primarily because our data matches the race to position
issue, and could definitely be the focus point of a future study.
Works Cited
ARTICLE #01
Author: Bigler & Jefferies
Title: “An Amazing Specimen”: NFL Draft Experts’ Evaluations of Black Quarterbacks
Pub Year: 2008
Journal: Springer Science
DOI https://docs.google.com/a/uconn.edu/file/d/0B8Pi8hVz-G35NHJUejUzQ3hTVzQ/edit
ARTICLE #02
Author: Gius & Johnson
Title: Race and compensation in professional football
Pub Year: 2010
Journal: Applied Economics Letters
10. 10
DOI
https://www.researchgate.net/publication/24069700_Race_and_compensation_in_professional_football
ARTICLE #03
Author: Lawrence M. Kahn
Title: The Effects of Race on Professional Football Players' Compensation
Pub Year: 1992
Journal: Industrial and Labor Relations Review
DOI http://www.jstor.org/stable/2524836?seq=1#page_scan_tab_contents
ARTICLE #04
Author: David J Berri
Title: Race and the Evaluation of Signal Callers in the National Football League
Pub Year: 2009
Journal: Journal of Sports Economics
DOI http://jse.sagepub.com/content/10/1/23.full.pdf
ARTICLE #05
Author: Jomills Henry Braddock II, Eryka Smith and Marvin P. Dawkins
Title: Race and Pathways to Power in the National Football League
Pub Year: 2012
Journal: American Behavioral Scientist
DOI https://docs.google.com/a/uconn.edu/file/d/0B8Pi8hVz-G35RWNyT1luTDZYQ1k/edit