Exploring the subconscious bias of referees

Football fans are often quick to accuse the ref of being biased when decisions go against their team. But what if there were a subconscious – if not deliberate – basis for arbitrary bias? New academic research examines identity and bias, and I got to speak to the people behind the study.

At the 2018 World Cup, Switzerland beat Serbia 2-1 in a controversial and ill-tempered group stage match. The Serbian team complained that a German referee was biased in favour of the Swiss team. German is, after all, one of the four languages of multi-lingual Switzerland along with French, Italian and Romansh.

Previous studies have implied that referees in some leagues have discriminated on grounds of race or nationality, but language? Could the Serbian team have a point?

Researchers at the University of St. Gallen, University of Texas and Molde University College analysed data from matches in the top two divisions in Switzerland from 2005-06 season to 2017-18. In these matches, the referee was from the same linguistic region as one of the sides, while the opponent came from a different linguistic area.

For example, FC St. Gallen, Basel and Grasshoppers Zurich come from the German-speaking ‘cantons’, while Neuchâtel Xamax and Servette FC (Geneva) are based in the French-speaking Romandy part of the country.

Of the clubs studied, 23 came from German-speaking areas, 11 were French-speaking and four were Italian. The referee share was 85.8% German, 13.7% French and 0.5% Italian.

How the referee bias study works

The academics used the Modified Causal Forest (MCF) estimator, which uses machine learning to estimate causal effects, controlling for a large number of variables. According to the formula, having a referee from the same area as the home team…

• Increases the amount of red and yellow cards in favour of the home team (i.e. against the visitors) by 0.27
• Impacts the outcome of the game: home sides gain about 0.23 more points (one for a draw, three for a win) than when the referee is from another linguistic group

While the variations may be small, they can have a huge impact on teams’ fortunes and finances. For example, at the end of the 2018-19 Swiss Super League season, four teams finished on the same number of points but only three could qualify for the UEFA Europa League – FC Lugano, FC Luzern, FC Thun and FC St. Gallen. The latter lost out due to an inferior goal difference.

Alex Krumer is a professor in the Faculty of Business Administration and Social Sciences at Molde University College and co-author of the study.

“In-group favoritism based on the division of people into groups is likely to be one of the more primitive human instincts that developed during the evolutionary process,” Krumer tells me. “Therefore, behaviour that can serve the interest of one’s group is likely to be an inherent feature of humans.”

This is the case, even for a wealthy, politically stable country like Switzerland.

“For example, despite all the wealth and happiness in Switzerland, there is still some tension between people from different linguistic areas. Our study shows that football referees assign significantly more yellow and red cards to teams from a different linguistic area,” Krumer adds. “If differences in languages is an issue in a country like Switzerland, then this is a particular concern in less prosperous countries such as Moldova, Ukraine and even Spain, where language-related debates were much more violent. So, results on the football pitch may represent much bigger problems in society, as football is likely to be a mirror of society.”

Are football authorities aware of this apparent subconscious bias?

“One practical implication is that football authorities will be aware of such a problem,” Krumer explains. “This is important since proper feedback training to referees may reduce the bias in the long term. In the short term, the simple thing is to avoid assigning referees from the same linguistic area (or other common feature) as one of the teams.”

You can read the research in further detail here.

Disclosure: The author works part-time with the University of St. Gallen’s PR agency, which is how he came across the research