The authors of this study constructed the dataset from a multitude of publicly available team data sources on the internet, information on player arrests from a USA Today database, and reports from the Institute for Diversity and Ethics in Sport, University of Florida. To these data we added confidential data from the U.S. Department of Labor on the numbers of employees and managers working at the team organizations.* We aggregated data for NFL team organizations over a period of 12 years, 2003-2014. We conducted our analyses on the twenty-seven (of thirty-two) NFL team organizations with complete data.
Dependent variable: Log of the number of player arrests (per team, per year)
Predictor variable: Whether or not a team organization had a critical mass of women executives (2 or more vice presidents) in the prior year.
Control variables (team level): Year, region, number of employees, franchise quality, annual payroll, team organization age, team win percent, the proportion of women in management.
For details on the data and analyses, please consult the full paper:
* The first two authors accessed the data on employer EEO-1 filings as unpaid researchers on temporary assignment with EEOC through an Intergovernmental Personnel Act (IPA) agreement. As part of this agreement, they entered into a confidentiality agreement which prohibits disclosure of any individual team organization data.
Player Arrests as an Outcome Variable
In the paper, we acknowledge the limitations of using player arrest data to construct our dependent variable of off-the-job player misconduct, which we analyzed at the team level. These limitations include:
- Arrests do not necessarily indicate misconduct, because the players may not have been found guilty of the charges. However, the final disposition of the arrest is not well documented and therefore unavailable to use in our study.
- Due to nature of the data and small sample size, we were unable to code the arrests in terms of their severity. One issue was that there are no reliable and valid scales for doing so. Another issue was that some players received multiple charges for the same instance. In our dataset, we counted these as one arrest, so it was not possible to classify this wrongdoing in terms of severity.
- As stated in the paper, we believe that the arrests data reflect the effects of racial profiling of players who are racial minorities. Eighty percent of the arrests in the dataset were of African-American players, yet African-Americans comprise only 68% of players in the NFL. These are serious, general deficiencies of the player arrests variable. However, the existence of racial profiling should not impact the study’s examination of NFL team organizations’ efforts to avoid, prevent, and remedy the player arrests that do or may occur. In other words, we believe that our study’s conclusions would hold, regardless of the degree of racial profiling in the arrest data. However, we also acknowledge that this is an empirical question for future study.
For more information on racial profiling, consult these sources:
Horrace, W.C. & Rohlin, S.M. 2016. How dark is dark? Bright lights, big city, racial profiling. Review of Economics and Statistics, 98(2): 226-232. DOI: 10.1162/REST_a_00543
Koch, D.W., Lee, J., & Lee, K. (2016). Coloring the war on drugs: Arrest disparities in black, brown, and white. Race and Social Problems, 8(4): 313-325. DOI: 10.1007/s12552-016-9185-6
NAACP. Useful resources for addressing racial profiling. Retrieved May 20, 2020 (PDF)
Pierson, E., Simoiu, C., Overgoor, J., Corbett-Davies, S. Jenson, D., Shoemaker, A., Ramachandran, V., Barghouty, P., Phillips, C., Shroff, R., & Goel, S. (2020). A large-scale analysis of racial disparities in police stops in the United States. Nature Human Behavior. DOI: 10.1038/s41562-020-0858-1
U.S. Department of Justice. (2016). Investigation of the Baltimore city police department. Retrieved March 26, 2018, from Baltimore Government Website.