The NFL Has a Race-Related Ejection Problem

Data show that NFL referees disproportionately eject Black players from games.

Posted Jun 25, 2020

Imagine you’re walking up to a supermarket. As you approach, you see a Black man wearing a mask standing near the automatic double-sliding door entryway to the store. Near his pocket, in his right hand, he’s holding what looks like a metallic object. 

What do you think he’s holding?

Same scenario, same question, only what if the man was White?

In an experiment investigating racial bias, 31 participants first saw a picture of either a Black male or White male face. The experimenter then told participants that they’d be shown an object and that the goal of the experiment was to quickly and accurately identify what this object was [1]. 

In some scenarios, the participant would be shown a gun, in other scenarios the participant would be shown a tool (like a wrench). Non-Black participants were faster to identify that the object was, indeed, a gun when they first saw a picture of a Black face. 

What’s more, when researchers conducted the same experiment with a timed deadline, participants mistakenly identified tools as guns more often when paired with a Black (vs. White) face.

Automatic Processing

Humans’ brains have adapted a rapid-response system for purposes of survival. That feeling of  “oh s*** is that a snake?” is an example of the rapid, unconscious automatic processing system at work. 

This system is efficient at producing results, but the speed at which it works can often lead to errors. No, it’s not usually a snake. And no, the black man standing outside of the supermarket is not holding a gun; he’s holding a pair of pliers. 

This automatic processing system isn’t inherently racist, but racial stereotypes can easily corrupt the system. Humans aren’t built to be racist, but we are built to categorize information into chunks, especially for groups of “other” people [2]. We stereotype because it saves energy, makes sense, and at one time in our history, likely saved lives. 

When you pair an error-filled response system with a chunking process that is built on oversimplification, you get knee-jerk interpretations that are informed, at least in part, by a racially biased society.

Knee-Jerking in the NFL

Football, though an incredibly violent sport, does have behavioral guidelines in place. Unethical behavior includes: endangering another player’s safety, fighting, unsportsmanlike conduct, and other acts that are deemed “detrimental to the league.” 

While there are more obvious violations of the rules like taunting or coming into contact with a referee, there are borderline calls that are left up to the judgment of the game’s officials. 

An infraction of the rules is not always clear in a split second, nor is the intention of a given player always clear. Did he intend to lead the tackle with his helmet or did his momentum cause him to collide headfirst? Premeditation is also a factor.

Through the cacophony of human bodies and violent collisions, a referee’s job is to determine ethical from unethical, right from wrong, often in an instant. Sound familiar? The situation facing NFL refs is pretty similar to the experiment or the grocery store scenario — they are charged with making quick, split-second decisions.

Sure, sometimes the full team of seven referees confers to render a final judgment about particularly nuanced plays, but even this slightly more considered team judgment doesn’t get much time. Fans in the stadium, viewers at home, coaches, and players all get impatient quickly. 

And while the team of officials has time to make a more accurate call, bias has already crept in, as the flagging official has already deemed a given action as “unethical.” Even these seven-person discussions and video reviews are all a result of an expert witness — a perception of an event made in an instant.

Ejections Aren’t Equal

Ejections are the ultimate punishment for a player’s unethical behavior during a game. While referees make mistakes from time to time, the league’s own data suggest that refs make accurate calls a whopping 97% of the time

So, if the refs are that damn good, ejections should reflect accurate, unbiased judgment over time with larger and larger samples of data, right? 

Of the 61 ejections in the NFL between 2017 and 2019, there shouldn’t be any inconsistencies when looking across positions, teams, ages, and...race, right?

Data taken from The Institutive of Diversity and Ethics in Sport (TIDES) 2019 report show that the percentage of Black players in the NFL has remained largely the same since 1992. The average percentage of black players in the NFL across a 29-year span between 1991-2019 was 66.5%, ranging between 58.9% and 69.7% year to year [3]. White players averaged 30.8% during the same span and ranged between 27.4% and 36%.

If all things were equal, then, Black players should constitute roughly 66% and white players roughly 30% of all ejections between 2017 and 2019. In reality, 95% (+28.9%) of all ejections were to black players, and only 1 white player was ejected in that timeframe [4]. 


The most accurate conclusion from these data is likely not as simple as, “the NFL is racist.” Indeed, this is not meant to be an indictment on the NFL or all NFL referees — but this isn’t meant to be an excuse either. 

“To err is human,” does not pardon true racist intentions, nor does it pardon inherent, automatic biases in human judgment. Awareness, however, is a perfect starting place for dialogue and change. 

Sadly, it’s not just the brain’s rapid automatic processing that is the victim of racial bias. The brain’s slower, more considered, controlled processing also seems to produce some unsettling results in the NFL. We’ll explore this more in our next article. 


[1] Payne, B.K. (2001), Prejudice and perception: The role of automatic and controlled processes in misperceiving a weapon. Journal of Personality and Social Psychology, 81, 181-192.

[2] Abrams, D. & Hogg, M.A. (2004). Metatheory: Lessons from social identity research. Personality and Social Psychology Review, 8, 98-106.

[3] Asian/Pacific Islander, Latino, and Other constitute the remaining demographics.