Racism and Coaching
This newsletter was prompted by the firing of two National Football League (NFL) head coaches over the past two weeks. Both coaches are Black, and some observers have attributed their firing to racial discrimination.
You don't need to know anything about football to follow what I've written here. I will be discussing NFL coaches simply to illustrate how statistics can play a role in calling out racism.
When statistics aren't needed
Typically, we don't need help from statistics to identify racial bias. We just listen to the victims. We examine institutional policies and practices. We reflect on our own attitudes. And we look at the words and actions of the perpetrators. (Case in point: Jon Gruden’s resignation last October as head coach of the Las Vegas Raiders after numerous racist, homophobic, and misogynistic e-mails he wrote between 2011 and 2108 came to light. Stats on exactly how many of those comments Gruden made, or how many he averaged per month, weren't reported, presumably because stats like that wouldn't add much to what his comments already show.)
When statistics are indispensable
In some cases, statistics are essential for determining that racism has occurred. The 2013 NFL Concussion Settlement provides a disturbing example.
The settlement itself didn't concern discrimination. Rather, in response to lawsuits on behalf of more than 4,500 players, the NFL agreed to spend roughly a billion dollars supporting former players who suffer from neurological conditions attributable to the sport. One provision of the settlement is free testing in order to determine whether former players should receive financial compensation; about a third of these players who've submitted test results have already received awards. In 2020, attorneys for two players alleged that award rates for dementia are two to three times higher for white players than for Black ones. Although the NFL hasn't released stats disaggregated by race, the attorneys’ estimates indicate a strong racial disparity. Indeed, it soon came to light that raw scores on the dementia tests administered to ex-players are adjusted through a process informally referred to as "race-norming".
Race-norming means, in effect, that cut-off scores for dementia are lower for Black players than for white ones. Thus, for each test, there will be a subset of relatively low scores which would make test-takers eligible for compensation if they're white but not if they're Black. (Further details are given in Appendix 1.) In June 2021, the NFL released a statement pledging to eliminate race-norming in their settlement program, but it’s still unclear how and when that will happen.
In sum, this is a case in which racial discrimination was identified through one statistical observation (more white players receiving compensation) and accounted for by another one (test scores of Black and white players being treated differently).
When statistics provide context
I've noted that in many cases we don't need stats to call out racism, while in some cases stats are essential to the process. Here's a third scenario: Stats sometimes provide important context for how and why racism occurs. This is illustrated by the firing of the two Black NFL head coaches this month.
The two coaches in question are David Culley (fired by the Houston Texans on January 13) and Brian Flores (fired by the Miami Dolphins on January 10).
NFL coaches are fired – or resign under pressure – on a fairly regular basis. For example, eight head coaches who began the 2021-22 season are already gone. That's exactly 25% of the entire group. But Culley and Flores were both Black, and the NFL currently has only one other Black head coach (Mike Tomlin). Thus, one could say that since the start of the 2021-22 season, 66.7% of Black head coaches have been fired.
Although the two team owners said nothing about race when describing why they fired their head coaches, and although there aren't any "smoking guns" (e.g., e-mails in which an owner makes racial comments about his coach), some journalists have argued that racism contributed to the firings. In the following sections I'll argue that there's no clear evidence for (or against) the notion that either firing was racially motivated. Then I'll show how statistics and other evidence reveal systemic racism in the NFL, which is an essential part of the context in which the coaches were fired.
I'll be responding here to a January 22 article published in the The Atlantic, because this is one of the most prominent outlets where racist treatment of Culley and Flores has been alleged, but my comments apply to similar articles that have been written about the two coaches.
David Culley
The Atlantic writer claimed that David Culley was fired simply because he's Black. The writer's main evidence was that Culley had only served as a head coach of the Texans for one year. But this in itself isn't proof of racism. Since 1994, 17 head coaches have been fired after one year or less, while others have resigned under pressure within a year. In Culley's case, his team performed miserably last season (4 wins, 13 losses), Culley himself stated publicly that the team's poor record was the reason he got canned, and the Texans' general manager indicated disagreements with Culley about their vision for the team. The Atlantic journalist discounted these details, arguing that Culley should've had a pass because he wasn’t able to field his two best players last year. In short, the journalist argued that Culley deserved to continue as Texans' head coach, but he was fired because he's Black. No supporting evidence was provided.
Whether or not someone deserves to be an NFL coach is the kind of topic that's endlessly chewed over in sports broadcasting, social media, local bars, etc., because it's a matter of opinion. Meanwhile, it's a matter of fact that the Texans' owner runs a multimillion dollar business and chose to fire one of her most important employees. Maybe it was a good decision, maybe not. Maybe it was racially motivated, maybe not. But if you believe that Culley deserved to keep his job for longer than a year, that opinion isn't by itself evidence for racism.
Brian Flores
On the surface, the case of Brian Flores looks suspicious, because he was fired after leading the the Dolphins to their first back-to-back winning seasons in two decades. Although acknowledging that Flores had a conflictual relationship with his boss and some key players, the Atlantic writer noted that "plenty of other coaches with strong personalities keep their job," and concluded that Flores' termination must've therefore been racially motivated. This is bad logic. Empirical studies have shown that regardless of race, some head coaches with "strong" (i.e., difficult) personalities keep their jobs, while others don't. The mere fact that a "strong" coach is fired doesn't itself demonstrate anything one way or another about racism.
The problem with addressing topics like this anecdotally is that every example can be met by a counterexample. (NFL fans: Do you remember Marty Schottenheimer? In 2007, Schottenheimer, who's white, was fired by the San Diego Chargers after leading the team to a 14-2 record, the best record in the NFL that year, owing to conflict with his boss and other personnel.) Meanwhile, in the 17 days since Flores was fired, new details have emerged about conflicts with team personnel, failure to communicate with subordinates, and so on. Here again, I'm not saying that racial discrimination can be ruled out as grounds for his firing. I'm just saying that the mere fact that he was fired after two good seasons isn't by itself strong evidence of racism. (At the same time, as I discuss in Appendix 2, there's evidence that Black head coaches in general are more likely than white ones to be fired after a winning season.)
Systemic vs. individual racism
Racism operates at many levels. An undergraduate of color who fails a test may not have been discriminated against by the instructor, yet racial discrimination could have contributed to the grade via test content (e.g., racially biased questions), psychological processes (e.g., stereotype threat), and/or sociocultural factors (e.g., a relatively weak pre-collegiate educational experience owing to the impact of racism on American public schools). Analogously, a long history of racism in the NFL may have contributed to the demise of Culley and Flores, even if the team owners who chose to fire them weren't racially biased. In short, we need to keep in mind the distinction between systemic racism (which has shaped institutions like the NFL) and individual racism (which exists in institutions like the NFL, and which may or may not have influenced the decisions to fire Culley and Flores).
Racism and NFL coaches
In a January 21 article for The Conversation on lack of diversity in the NFL, George Cunningham, a Professor of Sport Management at Texas A&M, provides helpful context for what happened to Culley and Flores.
Cunningham's point of departure is that at the beginning of this year, only three of the 32 NFL teams had Black head coaches. Now only one team does, in spite of the fact that roughly two-thirds of NFL players are Black.
Cunningham notes that lack of racial diversity in any group can best be understood through attention to individual, organizational, and systemic factors. Statistics can help reveal how each factor does or doesn't come into play.
At the individual level, studies have shown no significant differences between Black and white NFL head coaches with respect to win-loss records and other key variables. More broadly, evidence indicates that the paucity of Black head coaches is not attributable to racial differences in performance.
In Cunningham's view, organizational and systemic factors are highly influential. For instance, in both college and professional football, Black coaches are significantly less likely to be hired for the lowest-tier coaching jobs, and significantly less likely to be promoted to roles that most commonly lead to head coaching positions (i.e., offensive and defensive coordinator), even though data indicate no racial differences in skills. This tells us that regardless of the extent to which racism impacts the hiring of head coaches, racial discrimination has diminished the number of black candidates with the best resumes.
Finally, the impact of systemic racism on NFL coaching is illustrated by the Rooney Rule, first instituted by the league in 2003, which required that at least one black candidate be interviewed for any vacant head coaching position (and which, as of 2021, requires at least two ethnic minority candidates to be interviewed for such positions). The very existence of this rule indicates a problem, as does the fact that it hasn't worked. Although the number of Black candidates has increased since 2003, studies show that the proportion of Black head coaches hasn't increased since then. There's only one now, almost 20 years later. (That number might increase to two, since Brian Flores is interviewing with the New York Giants today, but two out of 32 still wouldn’t be very impressive.)
In sum, statistics help describe the context in which two Black NFL head coaches were fired. Stats can't tell us to what extent, if any, either firing was racially motivated. But they help show that the firings took place in a context of systemic racial discrimination, and they help explain how discrimination led to such a dearth of Black NFL head coaches in the first place.
Appendix 1: The Heaton norms
Heaton norms, dating back to 1991, adjust raw scores on several neuropsychological tests (mostly simple tests related to memory and learning) according to demographic variables such as the race/ethnicity, age, educational level, and gender of test-takers. Conceptually, the assumption behind these adjustments is that if you want to accurately evaluate an individual's performance, you have to choose an appropriate comparison group.
There's nothing controversial about adjustments like this for certain kinds of variables. Take physical growth. For any given age, pediatricians compare individual girls to the general population of girls, and individual boys to the general population of boys. This isn't sexist; it just allows for a more accurate monitoring of individual growth, and it helps know when to be concerned.
Controversies arise when neuropsychological test scores are adjusted on the basis of variables like race/ethnicity. The original purpose for doing so was to protect members of certain groups from being misclassified. For instance, when the mean score for Blacks on a particular test is lower than the mean score for whites, test manuals often call for the performance of the individual Black test-taker to be compared to the distribution of scores for Blacks, not the overall distribution of scores. Otherwise, some low-scoring Black individuals would be misclassified as scoring below the cut-off for a neurological impairment. (Some studies show that the rates of purported impairment decrease by more than half for Black test-takers when the reference group is other Black individuals rather than the general population.)
Although it's important to prevent misclassifications of low scorers, the use of Heaton norms (and others like them) is controversial for a number of reasons, including misapplied statistical adjustments, biased test content, the assumption that "Black" or "African-American" represents a single, undifferentiated category, inequities in the educational experiences of racial/ethnic minorities, and so on. The issues are complicated and nuanced. Here, in oversimplified terms, is how I would sum it up: If we use tests that are biased against any particular group, we shouldn't use statistical methods to correct for the bias. We should change our approach to assessment.
Appendix 2: Racial discrimination after winning seasons?
Sportswriter David Berri has noted that Black NFL head coaches are more likely than white ones to be fired after a winning season (which is simply defined as more wins than losses). Berri looked at data from 1978 through 2017 and found that during this time period, 16 teams with winning records (3% of the total) fired their head coach, and 4 of those coaches are Black.
At the time Berri examined the data, there had only been 17 Black head coaches in the history of the NFL who completed at least one season, and so 23.5% of them (4 out of 17) had been fired after a winning season. However, 174 white head coaches had completed at least one season, and only 6.9% of them (12 out of 174) had been fired after a winning season.
In sum, although NFL head coaches are rarely fired after a winning season, this occurs over three times as often among Black head coaches as among white ones. This looks like racial discrimination, and it's consistent with other evidence of both systemic and individual racism in the NFL. At the same time, I would suggest caution in interpreting the results, because it's a very small sample. Case by case, there may be other reasons why a coach was fired. Perhaps race played a role in the firing of three of those Black coaches, for instance but not the fourth one. In any event, the sample is small enough that the case of Brian Flores merits independent consideration. Yes, Flores was a winning coach, and yes, he was fired in a climate in which institutional and individual racism are prevalent. But this is circumstantial rather than direct evidence that racial discrimination motivated his firing.
Appendix 3: Demographic footnotes
—Throughout this newsletter, I use the term "Black" rather than "African-American", because that's the term most commonly used in mainstream sports journalism. I can't speak for how the coaches themselves identify. One of the fired coaches, Brian Flores, was born in the U.S. to Honduran parents and is occasionally described as "Afro Latino", but journalists seem to most commonly refer to him as one of the NFL's "Black" head coaches.
—This newsletter may seem to imply that if an NFL head coach isn't Black, he must be white. Generally, that's accurate, but there is currently one Latino head coach (Ron Rivera) and there have been two others in the past. Along with the historically small number of Black NFL head coaches, I'm not aware of other racial/ethnic minority representation at this level. On the whole it's a very white group.