Soccer Referees Do Favor Home Teams, Study Shows

[Footballer]

A fellow referee emailed me this. Just wanted to share.

http://www.sciencedaily.com/releases/2006/10/061030120410.htm

[uniteo]

I call BS

Not having seen the particulars regarding methodology I can't imagine how one could disentangle the effects of a home team playing differently from the reaction of the ref, for example...methodologically it would all be guesswork

[refmike]

Not quite sure what the above post means but consider that the home team and the home ref have worked together a few times and they know what he is likely to call, while the visitors do not. They are likely to get more calls till they figure it out (if they do figure it out). This is not favoritism, just a fact of life.

[what a goal!]

You would think that the article is full of crap.
I am pretty sure that every ref that is around would have been told to be 100% neutral. They would be professional enough to know the rules.

[DerbyRam54]

It can work the other way, at least at lower levels. Here at the grassroots level, my O-30 women's team had the same local ref for 4 out of our 6 home matches. Nice enough chap, but over time I became convinced that, probably at a subconscious level, he was trying to show that he wasn't a homer. He seemed to go out of his way to penalise minor infractions (7 seconds to release the ball was my favourite, first occurrence, midway through the first half, peep, IDFK) I never said anything about it because that's not my style, but I did suggest to our captain that she have a word with the assignor, it's not a good idea to keep putting the same ref on the same matches if you can avoid it IMHO.

[Leighs_babe]

I heard from a FIFA AR that he was working with a FIFA referee from the Islands and he said to them remember the HOME team gets all the close calls. And if you were to watch the match between houston and pachacua then you would believe it or the Jam match the other night on FSC.

interesting

[KMJvet]

I call BS

Not having seen the particulars regarding methodology I can't imagine how one could disentangle the effects of a home team playing differently from the reaction of the ref, for example...methodologically it would all be guesswork

They way they describe the methodology in lay terms in the article it seems like did a multivariate analysis. The used the data from lots of games so it should have good statistical power.

[jkc313]

Having a sound background in statistics and experimental design, I looked at this paper back when it came out and can assure everyone it's hogwash. The authors came up with "special equations" and tried to make allowances for such things as home teams play more aggresively before large crowds. Well, Duh! There was no methodolagy, no experimental design and no statistical anaysis done in any meaningful way. I hope the authors weren't paid for this garbage. I'd be embarassed to have my name associated with this paper.

[njref]

Having a sound background in statistics and experimental design, I looked at this paper back when it came out and can assure everyone it's hogwash. The authors came up with "special equations" and tried to make allowances for such things as home teams play more aggresively before large crowds. Well, Duh! There was no methodolagy, no experimental design and no statistical anaysis done in any meaningful way. I hope the authors weren't paid for this garbage. I'd be embarassed to have my name associated with this paper.

There is no way to determine whether the home team getting a higher % of calls is due to the crowd's effect on the referee or the crowd's effect on the teams, assuming that the crowd is a factor statistically. It is guesswork at best.

[USSF REF]

Yeah... but if you want to be a professor you have to get published ;) Funny to me that this guy (assuming he wants to be a professor or is one) used football as his topic - perhaps to persuade a journal jury, could there be a lot of football fans on one of those some place in the world? I wonder...

[ur_land]

Having a sound background in statistics and experimental design, I looked at this paper back when it came out and can assure everyone it's hogwash. The authors came up with "special equations" and tried to make allowances for such things as home teams play more aggresively before large crowds. Well, Duh! There was no methodolagy, no experimental design and no statistical anaysis done in any meaningful way. I hope the authors weren't paid for this garbage. I'd be embarassed to have my name associated with this paper.

jkc, I think that you are thinking about another paper (I've seen a few papers on referees and bias on SSRN). This paper has not been published yet, so it's a little difficult to say whether it's methodologically sound (although if you do have a copy, pm me b/c I would love to read it).

And there are ways to control for different styles of playing (for example, controlling for fouls called). If they controlled for differences in fouls called between home and away teams (which would take into account difference in playing style between home and away teams), any differences in cautions and dismissals that remained would be more likely due to referee bias. It also looks like they attempted to control for differences in aggressiveness in other ways as well, but not having read the paper I'm not sure what they did.

And from recent research in social psychology over the past ten years or so, there's pretty incontrovertible proof that unconscious biases and influences can affect our conscious behavior (if anyone cares enough, I can provide copious citations from my grad school readings). But one of the things that can overcome unconscious biases is knowledge (and vigilance) about the bias, a goal to act counter to the bias, and constant practice and feedback. For that reason, we referees should welcome research like this (assuming its methodologically sound) that lets us know about potential pitfalls we did not previously know about.

[njref]

And there are ways to control for different styles of playing (for example, controlling for fouls called). If they controlled for differences in fouls called between home and away teams (which would take into account difference in playing style between home and away teams), any differences in cautions and dismissals that remained would be more likely due to referee bias. It also looks like they attempted to control for differences in aggressiveness in other ways as well, but not having read the paper I'm not sure what they did.



You assume that cards should be in proportion to the number of fouls. I don't think that assumption is valid. One could equally well hypothicate that the effect of a home crowd is to infuriate or confuse the visiting team and lead them to commit a higher percentage of fouls worthy of a card, or calm the home team, or maybe embarass the home team into not doing stupid things that earn cards. Or maybe the crowd helps the play of the home team, and teams who are winning get fewer cards. Then the additional cards awarded to the visitor team would be due to harsher play, not referee bias.

You are right, I haven't read the paper either, so maybe they are smarter than I am and have figured out a way to control for team play. But I don't see how that could be. Until then, you have the old adage: "Correlation does not mean causation."

I always knew that someday my degree in sociology would be of some use....

[ur_land]

Until then, you have the old adage: "Correlation does not mean causation."


Completely agree with your entire post, njref; and until we read the paper, we have no grounds for evaluating the (good and not so good) methodological, statistical, and theoretical aspects of the study.

[jkc313]

jkc, I think that you are thinking about another paper (I've seen a few papers on referees and bias on SSRN). This paper has not been published yet, so it's a little difficult to say whether it's methodologically sound (although if you do have a copy, pm me b/c I would love to read it).

And there are ways to control for different styles of playing (for example, controlling for fouls called). If they controlled for differences in fouls called between home and away teams (which would take into account difference in playing style between home and away teams), any differences in cautions and dismissals that remained would be more likely due to referee bias. It also looks like they attempted to control for differences in aggressiveness in other ways as well, but not having read the paper I'm not sure what they did.

And from recent research in social psychology over the past ten years or so, there's pretty incontrovertible proof that unconscious biases and influences can affect our conscious behavior (if anyone cares enough, I can provide copious citations from my grad school readings). But one of the things that can overcome unconscious biases is knowledge (and vigilance) about the bias, a goal to act counter to the bias, and constant practice and feedback. For that reason, we referees should welcome research like this (assuming its methodologically sound) that lets us know about potential pitfalls we did not previously know about.

The date on this piece is october or november. You may have copious research available but I will bet virtually none of it is sound statistically. Don't think I'm picking on psychology, I'm a veterinarian and virtually all reserach done until very lately has been junk science filled with phrases like "this shows a strong correlation between a and b" which is intended to confuse people into believeing there exists a cause and effect relatonship. You see that phrase in any paper you better look and see if there is a regression analysis. My favorite correlation of all time was one calculated by a Statistician to show how silly correlations are. He found an almost perfect correlation between serial killers and haveing drank milk at one time or other during their life. What a suprise since almost everyone in the US has drank milk at least one time! Yet, some will say that drinking milk causes one to become a serial killer. Make sure any analyses are statistically significant at least at the .05 level. If not, they mean nothing. The fact that this article mentions the author invented things to overcome problems with his data is a giveaway.

[rca2]

I agree that its baloney. Statistically significant compared to what? A soccer game is not a series of coin tosses with equal chances of heads or tails. You would have to make assumptions that are absurb.

[Caesar]

The conclusion has come from an economics weenie who has tried to reduce football to a set of equations. I think that sort of speaks for itself.

[USSF REF]

Why not do a qualitative study on the content in match reports... or a long form interview with players, I mean players will be happy to point out the correlation that the home teams get all the calls....

lol. bogus study.

[NJ Ref]

Statistics is a way to lie using numbers….and that has been proven at a relatively high confidence level!

[ur_land]

We can stop carping on hypotheticals; the author sent me a copy of his paper. I haven't worked my way through it yet, but if you want a copy, please pm me with your email and I'll send it to you.

You may have copious research available but I will bet virtually none of it is sound statistically.

I'll take that bet any day of the week.

I'm a veterinarian and virtually all reserach done until very lately has been junk science filled with phrases like "this shows a strong correlation between a and b" which is intended to confuse people into believeing there exists a cause and effect relatonship.

I agree with this--a correlation tells you absolutely nothing about causation. But....

You see that phrase in any paper you better look and see if there is a regression analysis.

No. The presence of a regression analysis tells you nothign about caustation. Any statistical test, no matter how sophisticated, can only tell you (and I'm simplifying here) if there is a significant relationship between variables or a significant difference between groups on a varaibale. What gives you information about whether you can make a valid causal inference (based on your statistical results) is the experimental design. The gold standard is the randomly-assigned experiment (as in the case of classic medical trials and in the psychological research you dismissed above). You can't do a randomly assigned, double blind experiement with referees and soccer games, so you have to resort to quasi-experiemtal analyses, which utilize statistical controls. Maybe he did a good job of this, maybe he didn't. I'll decide that once I've read the paper.

Make sure any analyses are statistically significant at least at the .05 level. If not, they mean nothing.

Even if p is <.05 the analyses still may mean nothing. All p<.05 means is that there is only a 5% chance that the result we obtained wasn't a false positive (i.e., a 5% chance that the relationship we found is meaninglss and was just a random error).

The fact that this article mentions the author invented things to overcome problems with his data is a giveaway.

It is a worrying sign. But economists often formulate creative workarounds to deal with inadequacies of the experimental design and to create better controls to allow for a better chance of finding a causal relationship (see Freakonomics for examples of some especially creative and statistically sound economics scholarship). Maybe he was sloppy, maybe he wasn't. I'm going to withhold judgement until I actually have a chance to read the paper.

[intechpc]

Without reading the actual report it's hard for me to cast judgement on this study one way or the other. Although, from the article it seems there are outside influences that were not considered, I can't say for sure. However, I don't think it really matters how accurate or inaccurate the report is. What matters is how the FA, FIFA, USSF et al react to it.

Personally I feel that we as referees should always have an open mind to the possiblity that there are sub-concious factors that affect our rulings. Certainly when refereeing within my hometown league I have to be especially concious of the possible bias toward my home town teams. This brings about a renewed focus on my part to be particularly objective.

Now if the associations take this report as in impetus to educate referees on ways to ensure their objectivity in a match, how can that be bad? My feeling is as long as the focus is on general objectivity regardless of team disposition then it can only be to our benefit. However, I do think it would be a mistake if the associations were to take a look at this specifically from the perspective of favoring home teams. That in my opinion could be subject to fault since there are other sources of bias that would go ignored and if there is inaccuracy in this study that inaccuracy would now become a part of the game.

My opinion of this is the associations need to use this as simply a reason to re-focus referees on how to handle ALL potential sources of bias and ensure their general objectivity across the board. In that way this study would have some true value to the game.