Cognitive biases in football punditry, por Chris Dillow. Este é daqueles textos que nos obriga a pensar. Por um lado, é óbvio que o comentário futebolístico é animado pelos enviesamentos cognitivos mais gritantes. Por outro lado, é difícil imaginar como seria possível comentar um jogo de forma minimamente razoável. «A equipa foi goleada, mas é estatisticamente inevitável que resultados destes aconteçam de vez em quando, pelo que é prematuro fazer uma avaliação da qualidade intrínseca do plantel» é tudo o que um adepto não quer ouvir. Para serem minimamente razoáveis, os comentários teriam de ser absolutamente inodoros. Talvez uma boa dose de irracionalidade seja o preço a pagar para podermos gozar o jogo.
Football might be the world’s most popular sport, but moaning about co-commentators runs it a close second: so far, Clarke Carlisle, Andy Townsend and Phil Neville have all trended on Twitter and not in a good way.
One reason – or at least justification – for this contumely is that their judgments are often clouded by numerous cognitive biases; I’ve long suspected that Daniel Kahneman could have re-written Thinking Fast and Slow by taking examples solely from football punditry.
Here is an (incomplete) inventory of these biases:
1. Hindsight bias. How often after a player shoots wide, does Andy Townsend say he could have passed instead? It’s easy to know the right thing to do when you’ve seen what went wrong. This is especially true when discussing defences. Pointing out that the defensive shape was wrong after it conceded a goal is not good enough; the question is: did that same shape prevent many other goals? Americans call this Monday morning quarterbacking: it’s a shame there’s no English equivalent.
2. Outcome bias. Results shape our perceptions of performance. For example the Netherlands’ 5-1 defeat of Spain looks like a tonking. But their third goal might have been disallowed; their fourth was due to an unusually bad goalkeeping error and their fifth caught a demoralized side on the break. It’s a cliche – because it’s true – that football is a game of small margins, but if those margins all drop in your favour you can achieve a great result without a proportionately superior performance.
3. Misperceiving randomness. There’s always one team at a World Cup that does surprisingly well. This, though, is only to be expected. A lot of teams have a small but reasonable chance of getting to, say, the quarter-final. Across 32 teams, one of those chances is likely to turn up – just as we are likely to win a prize if we buy enough lottery tickets.
4. The hot-hand fallacy. This is a tendency to see “form” where none really exists but merely a run of luck. Take, for example, a player who scores a goal every other game – roughly Suarez’s and van Persie’s record but less that Dzeko’s. Over a 50-game career, such a player has a 50-50 chance of scoring six goals in six games. Such a run could well be enough to win him a golden boot and legendary status. A variation of this fallacy is the “commentator’s curse” – when a commentator praises a player only to see him shank a pass horribly. What happens in such cases is that an in-game run of luck suddenly stops.
5. Bayesian conservatism. Once we’ve got an idea that a team or player is rubbish, we interpret evidence to back up this idea; this is seen in co-commentators repeating themselves. My prior is that Diego Costa is a nasty cheating traitor who will therefore fit in well at Chelsea; I doubt this will change.
6. Selective perception. This is closely related to conservatism. It’s our tendency to notice evidence that supports our priors more than evidence that contradicts it. So if we have an idea that someone is a good player, we’ll spot his good points more than his bad; this is especially likely because so much of what players do (or don’t do) happens when they don’t have the ball; do they make good runs, or track back well?
7. Over-reaction. For example, Uruguay looked poor last night. Is this because they are a genuinely mediocre side (as their qualifying recordsuggests) or is it because they had an off-day? We can’t tell for sure on the basis of 90 minutes. If we do so, we might well be over-reacting.
8. Overconfidence. Everyone thinks their opinions on football are the correct ones, in part because we underweight our vulnerablity to all of the above; this is the Dunning-Kruger effect. One example of this is that many of us are backing Brazil to win the cup. But in fact, although they are favourites the smart money thinks there’s a 75% chance they won’t win.
Of course, cognitive biases aren’t necessarily wrong; indeed, they persist precisely because they sometimes (often) lead us to the right conclusion. And sometimes, they can cancel out; for example, Bayesian conservatism offsets over-reaction. My point is merely that one reason why pundits so often annoy us is that they are prone to such biases – and, of course, we are morely likely to see such biases in others than in ourselves.