Arsenal News
Arsenal News & Transfers
As featured on NewsNow: Arsenal newsArsenal News 24/7

Arsenal News, Only Arsenal, Blogs, Transfer News


August 2017
« Jul    

There is a clear regional bias among the refs. The EPL Referee review analysis



Publication on July 20th: Woolwich Arsenal, the club that changed football.

The book that re-writes the Arsenal story.



There is a clear regional bias among the refs.

By DogFace and Walter Broeckx

Untold Arsenal has a team of qualified referees who have reviewed more than 40% of the EPL games from last season. The reviews themselves were based on full match video footage with the advantage of video technology features such as slow motion and pause.

By reviewing those 155 games we have made a database of more than 7000 decisions that have been judged by our panel of dedicated and qualified referees.

The numbers you will see are based on those decisions and those reviewed games.


After having reviewed in depth refereeing for clubs across the regions we can compare and contrast  based on the numbers we have found so far in the EPL Referee review. If you want to find all the articles so far in this series you can go to this page

We will try to see if we can find a difference between the refs and the regions by putting a few numbers next to each other and see if we can find  links between regions or against certain regions.

So I first made a table with the different refs and the average positive or negative bias they have shown in the games from the different regions they did.

Walton MID -3,30 9,50 -0,67 -4,50 -0,83
Clattenburg NE -0,78 1,50 -1,00 0,93
Jones NW -0,31 -10,00 3,00 -3,00 1,57
Taylor NW -0,38 -7,00 1,75 12,00 -20,00
Webb YOR 0,64 4,50 5,50 5,00 -5,36
Atkinson YOR 1,59 -1,94
Probert SW 1,29 0,67 -4,25 -1,33
Friend MID 1,86 -4,50 -12,50 2,13
Dean NW 2,27 -2,40 -1,50 -1,00 -0,39
Dowd MID 2,65 0,33 0,00 6,00 -7,25
Oliver NE 3,20 3,33 3,50 -5,50
Marriner YOR 3,60 -1,00 -4,00 -3,72
Halsey HER 4,08 -4,00 -8,00 1,93
Foy NW 3,25 -0,67 -0,25 -5,36
Mason NW 6,63 -4,00 4,75 0,50 -8,00
Attwell MID 7,00 14,00 -2,25 -8,25
Swarbrick NW 8,00 4,00 -6,25
Moss NE 5,50 -5,50


A negative number (-) means that their bias was against the region that is named on top of the column.

And the higher the number (positive or negative) the higher the shown bias. If we take ref Attwell as an example, maybe you will remember that he was demoted and sent back to the first division last season. And if we look at our numbers we see that in the games he did he had big swings in his bias numbers.  The Midlands was enormously favoured as was the North West. When they demoted him they gave as a reason that he messed up with the big decisions too much. And well that seems to be what our numbers also have to say about him.

What is interesting is that great while (obviously) we don’t have the power to demote refs, the PGMOL can so, and given that our findings coincide with theirs, I wonder if they will also demote a few others refs if they compare their numbers and ours.

What PGMOL did not refer to was the massive positive bias for the Midlands that occurred under Attwell – who is from the Midlands. So can we see now if we can find a similar pattern with other refs? That is why we put in the region they come from behind the name of the referees. At first I wanted to put Halsey in the North West region but I have been told he is from Hertfordshire but now living in Bolton. So I went for his place of birth in his case.

A next step I took was having a look at how many positive or negative refs a region had to suffer in their games. Of course suffering a positive bias maybe a wrong choice of words.

Positive refs 75% 38% 67% 42% 22%
Negative refs 25% 54% 33% 58% 78%


So here are the numbers for each region. A team from the North West has a 75% chance of getting a positive bias in a game. A team from the North East has only a 38% chance of getting a positive bias when they play.

Teams from the Midlands are better off as they also have a positive bias chance of 67%. The other region (the artificial “region” that incorporated Norwich and Swansea, as they don’t really fit into any region) has a slightly negative bias chance in their games.

And if we look at the London region as a whole we see that they only have a chance of 22% to get a positive bias.

Can we find a reason for this? Well I tried to see if we could compare this with how many refs there are in each region. And because of there being a few refs from Yorkshire I had to add an extra column now. And we will start with the same table but add a few important lines:


Positive refs 0,75% 0,38% 0,67% 0,42% 0,22%
Negative refs 0,25% 0,54% 0,33% 0,58% 0,78%
Number refs 3 6 3 4 1
% of total refs 16,67% 33,33% 16,67% 22,22% 5,56% 0,00%
% or refs no Yorkshire 40,00% 20,00% 26,67% 6,67% 0,00%

So what you can see now is that the refs from the North West are the highest in number. They form 33.33% of all the refs in the PL. And the teams from the North West have the highest chance to receive a positive bias.

The next highest represented area are the Midlands with 4 refs or 22.22% of the total of all the refs. And we can also see that the teams from the Midlands are also the second highest in having the chance of a positive bias.

And if we then turn our attention to London we see that they have 0 (zero) refs in the PL and they also have the lowest chance of getting a positive bias in their games.

As there were no Yorkshire teams in the PL but 3 refs from that region I also did the same without the Yorkshire refs: we can assume that they divided their bias fairly over all the teams and the regions.

But the only conclusion we can reach from doing that is the same conclusion we already made from including the Yorkshire refs:-

The more refs you have in your region the higher the chance of receiving a positive bias when you are from that region. It can’t be a coincidence that the order of having the most refs in the region is the same as having the highest chance of a positive bias.

The order is clear and is the same:

1. North West

2. Midlands

3. North East

4. Other

5. London




Even though Mike Riley said there isn’t bias amongst PL referees we first of all now have proven that there was a definite home bias in our previous articles.

We then have done a series of articles when we had a look at each region.

And now we have shown that there is another form of bias that we can find in our numbers.

We now have shown that there is a link between how many refs there are in a region and how the teams from that region get a positive bias. The more refs in a region the higher the chance of a positive bias.  This can be seen the clearest in the North West region.

So if we want to have a fairer league it seems that the influence of the North West should be brought down and that there should be some refs from the London area in the PL. If not the bias in favour of teams in from the North West from some refs will make it more difficult for teams from other regions to compete on a level playing field.

And instead of sticking their heads in the sand the PGMOL and Riley should look at the numbers and do something about it. They should take immediate measures to reduce the impact of any possible bias.

A first step should be to make sure that every ref can only have a 6 point impact on one team over the whole season. Meaning they can only send a ref for two games to each team in the season. This is a first and absolute necessary step to take. Not doing this simple step is almost a criminal act against fair play from the PGMOL.

But of course the PGMOL has to realize the possible bias that exists amongst refs. And I see no sign that they are even willing to try and realize there is bias amongst refs. The statement Riley made last season about there being nothing wrong brings fear to my heart.  The first step to improve things is to realize that something is wrong.

It is the duty of a ref to bring fair play to each game, and it is the duty of the PGMOL to take care that fair play is there for every team. Not doing this is bringing the game in disrepute. You know a thing that can get managers and players banned. So if the PGMOL does not change this, they should be banned.

This concludes the regional analysis. From now on the next step will be investigating the teams via a survey of each team in the PL. And we will try to see how the refs did when we reviewed the games from those teams.

So stay with us in the coming weeks as we will have more interesting numbers and results and new extra graphics!






10 comments to There is a clear regional bias among the refs. The EPL Referee review analysis

  • CB

    As always, very revealing/interesting.

    If any of your readers have contacts in the media then they should let them know these findings and what PGMOL say. I have and will let you know if anything comes of it.

  • Mick

    This is compulsive reading and well done to you for presenting your findings in such a clear manner. I dread to think of what the percentage bias figures would be when teams from London play teams from the NW, especially if the match is in the NW or a ref from the NW is officiating. It would seem that the London teams need a miracle to get a result under such circumstances as they are effectively a goal or two down before the game has even started.

  • Stuart

    So PGMO should either rule out any chance of bias or ensure each season, the number of refs in each season is inversly proportional to the number of teams in that region. I’d say the first is easier to look at.

  • Matt Clarke

    I would the echo the two comments above.
    Sterling stuff.

  • jaroda

    I note in the London Bias report that 4 out of the 5 London clubs received a +ve bias and yet because of the massive Arsenal -ve bias the overall arithmetic mean for the region is -ve.
    If you exclude Arsenal’s figures then the bias unweighted by decision importance is +ve 2.863 and weighted by the games played and unweighted by decision importance +ve 1.789, the corresponding weighted by decision importance are +ve 3.878 and +ve 2.423.
    What this shows cleary is that London teams (excluding Arsenal) receive a +ve bias.(Think back to the Spurs (H) and QPR (A) games v Arsenal, they got the dodgy calls, we didn’t) It is only Arsenal that receives a -ve bias and massively so.
    The first question any outsider is going to ask is how did you acheive impartiality when compiling these figures. Wait… what’s this… this is an Arsenal blog and Arsenal are hardest done by? It is a shame because I believe this is great piece of work and my gut feeling over the season is that Arsenal were robbed in key decisions all season. However, having Arsenal stand out so far, accurately or not, greatly affects the credibility of the study to any other interested parties.
    My alternative suggestion is that the study is recalculated excluding all of Arsenal’s matches and figures to create numbers that are less likely to be affected by our own ‘bias’ towards Arsenal.
    Would this study show the same conclusions as the one including Arsenal? If yes then brilliant. You are truly impartial. If not then it poses more questions.

  • WalterBroeckx

    I understand what you mean and I am fully aware of that remark.

    Later on in this series I will take on that question about us being biased. That article is in fact 95% ready.
    Believe me (for now) I can prove that others have reached the same conclusions and those are not Arsenal supporters.

    About excluding all the Arsenal games is something that I will ask Dogface.
    If there is one person in the universe who could do this, it would be him 😉

    In a way to answer it myself: I have apart from the database that we have built tried to look at the difference in the numbers with our without Arsenal games.
    I only have the numbers with me for the moment of all the games minus 1 but to give you an impression on how short the numbers are to each other I will give it from my file:
    Only total scores in average correct decisions overall:

    All games: 71.37% correct
    Non Arsenal games: 71.91 % correct

    So there is a negative impact from Arsenal games on the final numbers. But I can prove you (will do this later on and not with just our numbers) that Arsenal has been screwed most of all teams by the refs.

    (I could give you other numbers also but this is a number we have used to compile the region numbers so will just stick to that one)

    Maybe will use it in a further article later one in fact. thanks for the idea 😉

  • WalterBroeckx


    the easiest solution is to have as many refs so that each ref can only do two games of the same team in one season. Maximum impact of a ref would be 6 points in that way.
    That would be one gigantic step forward. A step you could compare with men being able to fly to Mars I think ;-)But much easier done…

  • Andrew Crawshaw

    Walter, Dogface,
    My question is why are we subject to do much bias? I believe that we were denied about 20 penalties (net allowing for those against us). Many of those decisions alone cost us points and that in reality we should have been challenging for the title rather than scrabbling for third place. Assuming the bias is institutional rather than individual who is it who gains out of our misfortune and is it financial in a direct sense, or reputational (financial gain being a secondary consequence).

  • bob

    Walter, Dogface,
    This is admittedly simplistic, but surely a question or line of future inquiry that is begged by your striking findings. The Queen Bee of the Hives of Riley, Micky R himself hails from West Yorkshire. Does his selection by the PGMOL board (whoever they are, which is knowable) or whomever, fit these regional findings in any statistically meaningful (or more than coincidental) sense? Not that it must to prove something, but I’d be curious about what, if anything, you guys might see in his ascension.

  • The question of why the bias exists is really interesting. As is the question of why the PGMOL don’t make a few simple changes in order to show that we are wrong.

    Another interesting question is why the PGMOL shut down their web site, as we started to publish all this data.

    I think slowly we are getting to the answers – but it is important to publish all this data first.