New figures show a third of all referee decisions were wrong before last season

by Tony Attwood

Last season we looked at the inter-relationship between fouls, tackles and yellow cards in Premier Legaue clubs.  Untold did it because, as with so many other topics, this is something that is an absolute no-go land for the media and football commentators.

Even after we showed how bizarre and extraordinary the Leicester City figures for tackles, fouls and yellow cards were in the first half of last season, still no one made a comment – except one Leicester supporter who tried to explain things away by the performance of one player.  Unfortunately his generalised claim (that Leicester went down hill after their tackle king was injured) did not accord in any way with the dates of that injury or of Leicester’s decline.

From a logical point of view we would expect the clubs that tackle the least should get a smaller number of fouls against them, and a smaller number of yellow cards.  That is true, for example of Liverpool.  They were 16th in the tackles table, but committed the fewest fouls and got the lowest number of yellow cards.

But there were anomalies.   Arsenal were 13th in the “tackles undertaken” league, and 10th in the fouls league – which sounds like a fair situation.  Except they were top of the yellow card league.

Of course one can argue that Arsenal’s fouls are twice as dirty as Leicester – but that seems unlikely since Leicester committed more tackles than anyone else and even the best tactician is not that perfect a tackler.

After we published our analysis, Leicester rapidly went downhill in terms of position in the league, and number of tackles that went flying in, sinking from being chief challengers to Liverpool in the PL table, down to a still creditable fifth, although winning just two of their last ten games.

It was all very curious, so we published.   No one in the media showed any interest.  Usual story.

And there it stopped until Footcharts.co.uk published their fouls per yellow card table for the whole football league, from 2005 to 2020.

In my table below I have taken representative seasons up to 2013/14 because there is little variation between them, beyond the general trend showing that teams needed to commit fewer and fewer tackles to get a yellow card.  From 2013/14 onwards all seasons are included.

What the figures show is that from the start, away teams committed fewer fouls than the home team to get a yellow card.   This might be considered a bias by referees who were influenced by the home crowd as both the ghost games and “silent monitor” research has shown.  Thus the conclusion has been – remove the crowd noise and the referees’ judgements change dramatically.  Add the crowd noise and referees are biased by the crowd towards the home team.

Season	Home fouls per yellow	Away fouls per yellow	Difference	% greater chance of card
2004/5	9.9	7.5	2.4	32%
2008/9	8.4	6.7	1.7	25%
2010/11	7.3	6.0	1.3	14%
2013/14	6.9	5.8	1.1	19%
2014/15	6.9	6.9	0	0
2015/16	6.3	6.2	0.1	2%
2016/17	6.9	6.7	0.2	3%
2017/18	6.6	6.1	0.5	8%
2018/19	6.6	6.2	0.4	2%
2019/20	8.0	10.3	-2.3	-28%

But as the table shows in column four (“Difference”) the number of additional fouls that the home team can commit and get away with before getting a yellow card has been declining.  At the start the home team could commit 2.4 additional fouls before a yellow card.  By 2014/15 everything had evened up.

Now of course these figures don’t take into account the behaviour of the players.  Did the away players clean up their act?  Did the home players get more aggressive?  It seems unlikely.

But something has to explain why the home team could commit 2.4 more fouls before the yellow card came out, back in 2004/5 whereas ten years later than number had reduced the zero.  The most likely explanation is that the referees had woken up to home team bias.

However we’d never had football without fans before.   And last season it was the away teams that got away with more 28% more fouls before getting a yellow!   And when we ask why, the most obvious reason is twofold.  First there was no screaming abuse from the home crowd, the linespersons and referees were not influenced by the crowd.  Away team defenders found they could get away with more and do.

Second, while it seems home teams are used to getting the benefit of referee bias, this sudden change was a liberation for away teams who rose to the occasion.

What’s more, lockdown only affected 13 games (in Arsenal’s case), so for this difference to show across an entire season’s figures, shows just how dramatic the change in referee behaviour was.

Now in his recent article Walter pointed out that the PGMO were claiming that VAR had added another 15% accuracy to referee decision making, thus suggesting that prior to VAR (which as we know PGMO resisted for an extra year after most of the rest of Europe had installed it) PGMO had an accuracy rate of 83%, rather than the 98% they were claiming.

But this new information suggests that this figure is unlikely to be true.  Of course one could argue that once the crowd was taken away from the stadia, away teams managed to make many more of their tackles be executed within the laws of the game, while the home team were not affected, but this seems very unlikely.

Rather along with the evidence from experiments in which referees watch matches on screens with and without crowd noise, it seems that the referees were previously influenced by the home crowd.  Take them away and the referee is more focused on the game.   Which means it was not that the referees were 83% accurate before, but rather than additionally we have to take into account around a 20% crowd influence.

That takes PGMO referees down to a pre-VAR pre-covid level of 63% accuracy.

Which means that we could expect about a third of all refereeing decisions to be wrong prior to the absence of crowds and the absence of VAR.   Quite a thought.