During the 2018 World Cup, Zito Madu pointed out the racially coded language commentators used to describe a match between Poland and Senegal, which didn’t line up with what he saw on the field. A typical article claimed that “‘Poland struggled all game against the pace and physicality of Senegal,’ which is an absurd line for anyone who watched the game,” Madu wrote in SB Nation. It felt like an example of a widespread tendency to focus on Black players’ “pace and power” while praising white players for things like intelligence and work ethic.
But how would the same game have looked to viewers who literally couldn’t see race?
Sam Gregory was working for the Canadian data provider Sportlogiq a couple of years ago when Toronto FC director of analytics Devin Pleuler came to him with an idea. The company’s broadcast tracking technology can capture how players move their limbs and reproduce their stick-figure skeletons in a two-dimensional render. If Gregory’s Sportlogiq colleagues and Pleuler showed the same clips to different viewers as either a video or an anonymized animation, they could measure how attitudes toward race and gender affect how we see soccer.
GREGORY ET AL.
The resulting paper, “Pace and Power: Removing unconscious bias from soccer broadcasts,” caused a stir when they presented it at last month’s New England Symposium on Statistics in Sports. Of the 47 sports fans who watched a two-minute clip of the World Cup TV broadcast, 70 percent said that Senegal, whose players were all Black, was “more athletic or quick.” But of 58 others who saw an animation of the same two minutes without knowing which teams they were watching, 62 percent picked Poland, whose players were all white, as the more athletic side.1 The physical advantages that supposedly defined the African team’s style of play disappeared as soon as their skin color did.
Gregory sees it as a good sign that audiences who watched the video and the stick figures didn’t show significant disagreement on the survey’s other three questions, about which team was more technically skilled, tactically organized or physical.2 “The fact that we got such similar results suggests that at least people were able to tell what’s happening from the renders,” he said.
The athleticism flip-flop offers a new kind of evidence of a prejudice that affects how Black players of every nationality are perceived. For decades, researchers have documented media stereotypes of African players as “‘powerful,’ ‘big-thighed,’ ‘lithe of body,’ ‘big,’ ‘explosive,’ and like ‘lightning,’ attributes that were to be contrasted with ‘the know-how that England possess.’” As Belgian forward Romelu Lukaku, who is Black, told The New York Times, “It is never about my skill when I am compared to other strikers.” Now, for the first time, researchers have a way to isolate how race influences direct perceptions of the game.
Gregory hopes that measuring unconscious bias will be a step toward changing conversations about Black athletes. “Last year there were all these discussions around Black Lives Matter, and there were player protests,” he said. “Obviously the issues off the pitch were more important than the issues on the pitch, but it does feel like even when that conversation was happening, there was very little discussion about racial bias in the way we talk about players.”
The study also examined attitudes toward gender by showing viewers a pair of two-minute clips, one from the American top-flight National Women’s Soccer League and another from League Two, the English men’s fourth tier. Even though the NWSL draws more fans to games, its average player earns about a quarter as much as the average player in League Two. Gregory and Pleuler were curious whether this “clear gender pay gap” could be explained by a difference in the quality of the soccer shown on TV, as some have argued.
People who watched the broadcasts said that the men’s game was “higher quality” by a 57 percent to 43 percent margin. Those who saw the renders with genderless stick figures preferred the women’s match, 59 percent to 41 percent. The results weren’t statistically significant across a small sample of 105 mostly male respondents, but Pleuler believes the line of research is promising. “I think these results are suggestive that your average soccer fan can’t tell the difference between something that does have a large investment level and the women’s game, which does not,” he said.
Unconscious biases don’t just color the way fans and media talk about sports. They could hurt players’ earnings and career prospects, not to mention cost points on the table for teams that aren’t spending their money wisely. Gregory, who’s now the director of analytics at Inter Miami, thinks body pose data might help clubs check their player evaluations. “Scouting to me is the obvious one,” he said. Skin tone has previously been shown to correlate with differences in Football Manager ratings, which clubs sometimes use in their real-life scouting process. Anonymized renders could help determine whether pro scouts exhibit similar biases.
“The idea is that over time, hopefully people will realize that this is a source of bias and they’ll be able to change it within themselves,” Gregory said. “None of us are unbiased in anything we do, so I think a big part of challenging bias is acknowledging it.”