UEFA avoiding racism at the 2020 Euros? Bayes-ically impossible
If we were to draw a Venn diagram of everything that statistics can provide a useful framework to examine (along with narrative, context, and case studies, of course, because we’re not heathens here), everything that is political, and sports, we’d have one very large circle. And what better way to ring in the new decade than taking a moment to come full circle on that circle and use statistics to examine politics in sports?
Late last year, UEFA issued an incredibly bold statement about the upcoming Euros: Despite us encountering at least one incident of racism in Europe every week for as long as we can remember, the powers-that-be seem confident that the power of football at the continental scale will overcome any bigotry that national associations have failed to stamp out.
"In our experience, the Euro has always been a very festive event, at least within the stadiums," UEFA Vice President, Giorgio Marchetti, said in a press conference. "We are confident that this particular atmosphere will take priority over stupid and sometimes criminal things that unfortunately from time to time happen in football, and we never want to see in our sport."
These “stupid and sometimes criminal things” Marchetti dares not speak the name of are prevalent in every single league and in every single fanbase. National and international media and the leaders of several UEFA countries definitely aren’t helping the matter. But still, perhaps Marchetti and UEFA are right. To know for certain, let’s use statistics to infer whether or not racist incidents are likely to occur during the 2020 Euros.
Because we’re working with a smaller dataset, because I have a lot of questions about UEFA’s priors, and because there’s never a bad time to use the method, we’ll be taking a Bayesian approach to this problem (my sincerest apologies to Thomas). Specifically, we’ll rely on prior information about racist incidents in European football to guide our inferences about whether UEFA’s claim will hold true.
Let’s start with the hypothesis put forward by UEFA and Marchetti: The festive atmosphere of the Euros (condition 1) combined with fans only resorting to bigotry when they have nothing better to do (condition 2) will help ensure that racism doesn’t mar the tournament.
To test this hypothesis, we implement the following Bayesian formula:
Or to put it another way, the probability of racism occurring during the Euros (the “posterior”, if you want the Bayesian term for it) is equal to the probability of our “priors” (the conditions laid out by Marchetti), multiplied by the “evidence” (the probability of racism occurring at a sporting event)), divided by the probability of the Euros having a festive atmosphere and fans that have better things to do than be racist.
To properly solve this formula, we’d need to pick a prior distribution, talk about posterior distributions, and dig deeper into likelihood functions. This author would have to spend days of their life collecting data on every racist event at a football game in the last decade or so, which sounds like the worst possible way to ring in the new year. Instead, let’s focus on the highlights that provide our considerable prior knowledge:
From what we know about Euros in the past, UEFA is definitely on to something when they assume there’ll be a festive atmosphere surrounding the competition. In general, major tournaments tend to see far more festivity than regular games. Will the festive atmosphere be diluted by the competition being spread out over 12 different countries? Perhaps! But we have no data to adjust our Festivity Quotient here, so we’ll assume it has no effect.
This is where UEFA loses us. It’s very rare to hear an argument that putting something mildly entertaining in front of people will mitigate their desire to dehumanise people who don’t look like them. It also goes without saying that if people could be distracted from acts of racism (and other forms of bigotry) by festive events, marginalised people wouldn’t ever stop throwing parties. Nor does UEFA seem to believe that the magical restorative powers of international soccer atmosphere apply to other forms of social discord. They have, for example, taken steps to prevent certain nations involved in political or military disputes from facing each other in a tournament setting, rather than relying on the festive atmosphere to do its thing.
Condition 1 and Condition 2
Of course, Marchetti believes it’s a combination of conditions 1 and 2 that will mitigate any racism during the Euros. Which is especially perplexing as pretty much every racist incident inside a football stadium seems to occur while football is actively being played. To take this point to its logical conclusion, UEFA is claiming that the festive atmosphere (condition 1) is somehow both separate from, and more entertaining than, the actual sport all these fans are paying to watch. And that somehow, the combination of these two variables is all that’s necessary to combat racism.
Thinking about this for more than a second gave this author a headache, but that doesn’t necessarily mean UEFA haven’t thought their conditions through. Basic common sense and overwhelming evidence that distracting festivity doesn’t mitigate racism, however, does support that idea that UEFA may be wrong here.
The lack of data collection involved in this entire exercise means our evidence is purely anecdotal. However, we can definitely conclude that racist incidents are occurring with depressing regularity in leagues across Europe. We could also probably make a conjecture that fans involved in these incidents are emboldened by the political atmosphere in many European countries at this point.
In Bayesian statistics, there’s absolutely no issue with choosing uninformative priors: a prior that acknowledges that we do not have a lot of data or experience with a particular problem. What Marchetti and UEFA have here, however, is more of a wilfully uninformed prior, having made several incredibly bold claims with absolutely no regard for the data around them. And they’re using these priors to make a prediction that absolves them from any kind of planning or pre-tournament work that might protect players and fans of colour from any potential racist abuse during the Euros.
Given that the political atmosphere in Europe is clearly seeping into most stadiums, UEFA might want to consider re-examining their priors here so they can arrive at a more realistic prediction of whether or not it’s likely that we’ll see racist events and behaviour during the 2020 Euros.
Of course, if we look at our priors on how well they’ve dealt with racist behaviour in the past, the likelihood of that happening seems pretty low.
Header image courtesy of the Press Association