"The video game series “Football Manager” has a racism problem" - press article

[endtime]

https://qz.com/1636057/football-manager-2019-has-a-racism-problem/

Quote

Football Manager is one of the most popular sports video games in the world. The 2019 version was a chart-topping hit in the UK, where the game is most widely played. Previous versions of the game have consistently sold more than a million copies each.

Football Manager allows fans to simulate the experience of managing a professional soccer team at a level of extreme detail, with hundreds of thousands of players from clubs in more than 100 leagues from 50 countries coded into the game. In addition to choosing your squad and strategy, you also have to deal with the media and the emotions of players. Unlike in most sports games, gamers don’t actually play the matches, though they can watch them and make strategic adjustments during the game. Football Manager is considered so realistic that it has been used by English Premier League teams for scouting, and by managers to improve their skills.

The game appears to have a racism problem.

In addition to rating players for physical skills like agility, dribbling, and strength, players and team staff are scored on non-physical attributes like sportsmanship and loyalty. Ratings are given on a 1-20 scale, with 20 being best. The game uses a huge network of scouts to generate these ratings for hundreds of thousands of players and staff across the world, but the final ratings are approved by the game’s approximately 100 head researchers, according to the Telegraph.

Among the many physical attributes assigned to players and staff, including things like hair color and length, one measures the darkness of their skin, with 1 being the lightest and 20 the darkest. For example, Russian forward Alexandr Kokorin is a 1 and Senegalese midfielder Badou Ndiaye is a 20.

A Quartz analysis of Football Manager’s data found that players and staff with darker skin tend to have worse ratings on non-physical attributes. Using data found in the Football Manager Editor database for 2019, we examined the share of players and staff with each skin color number who received a rating of 10 or above on sportsmanship, temperament, professionalism, and loyalty—that is, those who rank in the top half of scores for these non-physical attributes. Only players and staff who had been with their national team for at least 25 matches were included in the analysis, to assure that it was based on prominent people. The data include about 4,800 players and coaches.

As the chart below shows, on each attribute, players and staff with lighter skin are more likely to receive a score of 10 or better. (The relationship was first pointed out by Football Manager player Evan McFarlane on Twitter.)

To test whether this wasn’t just inter-country differences, or specific to team staff, we also analyzed the professionalism ratings of more than 900 players on English Premier League team rosters who are currently making more than £1,000 ($1,270) per week (thereby excluding young academy players who are unlikely to appear in a Premier League match). Again, we find that lighter skin players tend to receive higher scores on non-physical attributes: 72% of players assigned a skin tone of 1-5 received a professionalism score of 10 or better, compared with just 55% of players assigned a skin tone of 16-20.

The results of this analysis won’t come as a surprise to black players like English forward Raheem Sterling, who has spoken out about the way black players are treated by the media and fans. Sterling recently called for tougher penalties against teams whose fans shout racist abuse at players, a regular occurrence, suggesting it should impact that team’s place in the standings. In this context, it is perhaps not surprising that racist attitudes seem to have bled into soccer’s most popular management simulation game.

Quartz contacted Sports Interactive, the company that develops Football Manager, for comment and received no response.

Thoughts? 

[JeffDogg]

And I thought my life was useless

[Lord Rowell]

Certainly does not suggest that Football Manager has a racism problem.

However, the statistical analysis, if correct, does beg some wider questions about society, perceptions and other things as a whole.

I'm surprised this and maybe some other kinds of analysis haven't been done sooner, given the vast amount of data.

[decapitated]

I'm pretty sure there isn't any conscious racism nor institutional racism at SI themselves. They've always been a very progressive company, so I can't see that being the case. I am shocked at those graphs, but I would imagine there's a logical explanation that isn't "football manager is racist" like the article states as I simply can't believe that's true.

 

 

[decapitated]

Theres talk on Reddit that the survey QZ did might have been structurally flawed.

 

Quote

The experiment just simply isn't a good experiment. There is no control with it. They only look at players and staff with 25+ international caps. They didn't look in to what players/staff have their ratings auto-generated, they didn't look at all people that have manually-entered data. Maybe what they found is absolutely true, and that for players with over 25 caps, darker skinned players have worse hidden attributes. However they don't mention anything about the possibility of that just being true only from their sample size. For all we know, for players with between 1-25 caps their findings might be reversed, and white players have worse hidden attributes.

They offer no counter points, they offer no suggestions for what might be causing these statistics. They just see "from our small sample size, dark people have worse hidden attributes than light people". It's a biased article that claims that Football Manager's researchers are racist based on an incredibly poor analysis that does nothing but say that their very small sample size shows that darker skinned players on average have worse hidden attributes than white players.

 

[JeffDogg]

That study is so poor and biased, what tells us they didn't carefully chose the players? What tells us they didn't invent numbers?

With studies that dumb, it's far from unthinkable. People with too much time on their hands will do the stupidest things just to get attention.

[s1111]

 

1 hour ago, endtime said:

As the chart below shows, on each attribute, players and staff with lighter skin are more likely to receive a score of 10 or better. (The relationship was first pointed out by Football Manager player Evan McFarlane on Twitter.)

It looks like the editor has been updated since these tweets were posted to remove the ability to see a nation's ratings for these personality attributes, as I can't see them there any more and I've looked through them before. Someone at SI realised how embarrassing it was.

A lot of the analysis on the players is flawed (e.g. doesn't taken into account the players who have a 0 set for a random attribute) but it wouldn't be surprising if there was subconscious racism in the way players are assessed because there is in real football. According to some commentators every black player is a powerful beast that relies on their physicality to make up for their lack of intelligence, every black midfielder is the new Patrick Vieira, Naby Keita and Ngolo Kante are very similar players because they are small and black despite having totally different styles and skills etc.

[RBKalle]

I'm actually surprised it didn't pop up a few years back already...

The "national personality traits" are a timebomb because apparently they "insitutionalize" racist-ish perceptions. While in reality they're mere tools the GAME needs to populate its gameworld with a decent level of approximation and fidelity to the original db.

At a first, superficial glance, low Loyalty for this or that nation can feel arbitrary and even offensive, but it's the nature of the game! If your only chance to make it as a professional football you MUST leave your country at the earliest age possible, you're naturally going to be less inclined to be "loyal", while you'll need to "adapt" more. Conversely, a British player can easily develop loyalty to a club he's played for since he was a child, while his need to be more flexible and open to change aren't as high.
Ditto for "unambitious" players coming from fourth-tier European countries where football is at best a part-time gig. Is it "racist" to assume an Andorran player won't be as Ambitious as his Brazilian counterpart who'll need to make it in order to earn a better life?

Professionalism, Temperament and Pressure are a bit of a mixed bag, but again, when the general conditions to have a pro career are tougher, it's understandable that players can in general be on the more "volatile" side.

I'd say it's more stereotypes than racism anyway... Kinda like when you think of Brazilian midfielder, you think one of the hundred super-skilled guys, not the likes of Dunga or Felipe Melo... And if you hear "Scandinavian player" you immediately go for tall, big, hard-working, not-so-technical, player good for mid-EPL level. And it's not true all African players are the next Vieira or Yaya Touré... There's also the powerful striker or the pacey forward who can run like Bolt but can't hit a barn door (basically, Oba Martins).

Anyway, that "study" shows a complete lack of understanding of what FM is. And a noticeable bias.

[99]

@RBKalle you wrote everything I was thinking myself and commenting with my dad here while reading the article. So needless to say, I second your post.

[DementedHammer]

Ditto. I couldn't have put it better. 

[Lord Rowell]

I think that while it is useful to point out limitations of the analysis, I don't think its wise to rubbish it.

Even academic studies have limitations and flaws.

I do suspect, as has been said and was also implied by myself earlier, that any trends in the data are likely unconscious bias, sociological reasons and so on. As a wise person once said (I wish I could remember who) if you were raised in the west and don't have any racial bias, then the education system (I'd add media in more modern context) has failed.

Self-reflection is good

One thing I think we should all refrain from is discussing specific examples of players, coaches etc. because that then in turn could be used against the researchers. This is a big, big, wide issue in society.

[RBKalle]

2 ore fa, s1111 ha scritto:

Someone at SI realised how embarrassing it was.

Why is it embarrassing?

It's a TOOL, a set of rough guidelines to populate the gameworld with fictitious players. Also, IIRC, those values were the median range, not a fixed one. So you'd still get a Faroese guy with 20 Ambition, despite the national average being much lower. Not because Faroese people are inherently lazy and unmotivated, but because being a footballer there is neither a top priority nor a rewarding choice.

If the ability to edit those traits has been removed for fear of being called out, it was a questionable move that takes a little tool away from the editing community.

Are some of the national attributes a byproduct of social bias? Likely so, but what's the alternative? A 100% randomised process of creation? That'd destroy the gameworld and the game. A harmless, politically correct "every nation has Average Average values in the 9-12 range"?

For the same logic, is it "racist" to assign an African country a lower youth rating and reputation? It's based on real-life data (performances, amount of top players) but you'd easily say all those are results of institutionalized racism that has prevented 100 new Salahs to become worldclass players. [and it's a debate that actually exist about African football's lack of progress contrary to expectations and to relative success in youth competitions].

I think those people are simply attacking an "easy" and well-known target to get attention, while completely missing the point. Their intentions may be good, but we're closer to the usual "GTA games are fueling violence, misogyny and so forth" sensationalism.

P.S. Real-life players' traits are assessed according to real-life events... Plenty of unprofessional, lazy, unambitious white players out there as well.

[Constantine]

So this is basically cherry picking data to stir some s*** up??

[endtime]

33 minutes ago, RBKalle said:

It's a TOOL, a set of rough guidelines to populate the gameworld with fictitious players. 

The article is referring to data on real players in the database, as set by researchers, not to those country-specific guidelines used in newgen creation. 

[RBKalle]

38 minuti fa, endtime ha scritto:

The article is referring to data on real players in the database, as set by researchers, not to those country-specific guidelines used in newgen creation. 

Which makes the apparent decision to scrap that section from the editor even more questionable...

And the data sample still feels a tad arbitrary and too convenient. Not to mention the graphs' labels don't make sense... What do those % mean? Are they on the 1-20 range, so 80% means 16?

I also daresay most of those "shocking values" are likely to be in the low 10s, so nothing overly disrespectful... And for foreign players/new signings, I suppose it's a matter of researchers keeping it average when they don't know enough about the player to go toward the higher or lower end of the scale.

FFS, wanna check how many players with LOW (and I mean VERY LOW) values for those attributes are there? Wanna bet most of the unprofessional/temperamental gits in English football are white?

[OspreyJ]

1 hour ago, Constantine said:

So this is basically cherry picking data to stir some s*** up??

Happens quite often really

[Lord Rowell]

I think some here are being too quick to dismiss, and criticise the data, OP etc.

It's like we don't want to admit we might, as a world, have issues and bias.

SI will have their own analysis of their data. I would suspect and hope they are aware of potential unconscious bias in data (if they exist), and would be hence in the first instance issuing guidelines to researchers. If those biases don't exist then all good

I am extremely disappointed at some responses here though just to defend and dismiss, pretend there are no problems in the world and that if there are, then somehow the FM database is immune to them. Well sorry that's very unlikely.

[Lucas]

18 minutes ago, Lord Rowell said:

I am extremely disappointed at some responses here though just to defend and dismiss, pretend there are no problems in the world and that if there are, then somehow the FM database is immune to them. Well sorry that's very unlikely.

All I can see is people approaching this (as is true with research) in very different ways, so I think its slightly unfair to say people are being defensive and dismissive based on their own interpretations.

You're using your own interpretation to say, there's a lot of different factors and how we should be open to thinking about it, while at the same time dismissing people for putting forward their viewpoints. 

To ask for objectivity and then get subjective and disappointed at some of the responses strikes me as a bit counter productive for what you're trying to argue at the same time. 

If you're going to argue for open minds, then probably worth using that rather than make sweeping generalisations at people discussing it because you disagree with what they are saying.

[JeffDogg]

When you know how studies are made, you learn to not pay attention to them. I've heard people whose job is to do studies, and he basically says that with the limits they have and the deadlines and everything, more often than not the study is way more biased and oriented than explicitly said.

So they probably decided the game was racist, and then matched a study to the "facts"

[alian62]

4 hours ago, JeffDogg said:

When you know how studies are made, you learn to not pay attention to them. I've heard people whose job is to do studies, and he basically says that with the limits they have and the deadlines and everything, more often than not the study is way more biased and oriented than explicitly said.

So they probably decided the game was racist, and then matched a study to the "facts"

That's correct . If your after an answer then you ask the questions to get that answer . IE Do you think cats are more white or predominantly brown , We asked 1000 people with brown cats and they said the answer is brown .

[Zardoz]

It reminds me on a study of a Dutch university about meat-eating.

The study said that meat-eating persons are more aggressive etc. (put in any evil attribute you can imagine). I heard about it and immediately thought that cannot be true. Shortly after the professoress who supervised the data confessed that the whole "study" was a fake..

I do not think that the whole "study" is a fake but I think that it is highly probable that those crusaders fighting racism took the data they needed to back their assumption according to the motto: "There simply MUST BE racism in everything  coming from (old) white men and we simply have to dig until we hit it."

 

[Muttley84]

It's funny how we seem to forget that data correlation even with statistical tests doesnt mean always causality between them. I'll just leave this here:

[Tajerio]

18 hours ago, Lord Rowell said:

I think that while it is useful to point out limitations of the analysis, I don't think its wise to rubbish it.

Even academic studies have limitations and flaws.

I do suspect, as has been said and was also implied by myself earlier, that any trends in the data are likely unconscious bias, sociological reasons and so on. As a wise person once said (I wish I could remember who) if you were raised in the west and don't have any racial bias, then the education system (I'd add media in more modern context) has failed.

Self-reflection is good

This point bears repeating.  I don't think that the disparity is going to be down to conscious racial animus; it's down to most of the researchers living in a world whose dominant cultures have incorporated structural denigration of people of color.

I also think that while the suggestions RBKalle has offered for the disparity in the country average ratings have some grounding in reality, it is also almost certainly the case that those ratings are also influenced by negative stereotypes.

[enigmatic]

I do wonder why they haven't simply done a regression of skin tone versus (non-zero) professionalism score rather than this silly "top half, bottom half" split and wonder if that's data mining for controversial conclusions or because they don't know how to do statistics...

There's an issue with selection of the "25 caps and over" bracket too (especially when FM has a "reputation" score much more attuned to balancing players of similar renown) - there are a number of countries you can get 25 caps for without actually being a professional footballer (or a remotely adequate one) and very few of them have many pale skinned footballers.

That said, I wouldn't be altogether surprised if there's some truth in it. Frankly, the "personality" ratings in FM for most players are based as much on researchers' subconscious perceptions and random number generators as on hard data - we don't get to watch their personalities for 90 minutes each week and I'd much prefer a "default" personality adjusted by manager tendency-like variations for stuff which is actually reasonably well established via media commentary  - "learns foreign languages quickly", "does not like changing clubs", "will drop divisions for first team football", "expects high wages", "trains intensively", "dislikes media attention" etc.

That said, many of those subconscious perceptions like the average Peruvian being a lot less professional than the average German are driven by researchers actually from the same nation and ethnic group as most of the players they rate...

[RBKalle]

2 ore fa, Tajerio ha scritto:

This point bears repeating.  I don't think that the disparity is going to be down to conscious racial animus; it's down to most of the researchers living in a world whose dominant cultures have incorporated structural denigration of people of color.

20 ore fa, Lord Rowell ha scritto:

any trends in the data are likely unconscious bias, sociological reasons and so on.

Oh please!

It's a world where we still judge people of the same colour just because they're from two cities over, speak a weird-sounding dialect, listen to music we don't like, dress in a certain way, eat different food at a different time of the day etc...

Of course we all have bias and they affect how we view the world around us.

But do those bias make a FM researcher go "oh, X is black, so he must be a tad more unprofessional and temperamental than Y, who is white"? Or are those judgments based on what a competent and dedicated follower of his club knows about the players' professional life?

To me it's so straightforward it's not even worth keeping on debating... Which is the most likely?

a) every FM researcher, not just the EPL ones (because let's not forget newly signed players initially retain their old values, assigned by the old club/league's researcher) is an undercover alt-right agent who's sneaking in racist messages by assigning lower attributes to non-white players

b) every FM researcher is assigning those values according to his knowledge of his club's players, based on on-pitch performances, training ground news, reports etc.

So unless you also think every negative piece of news involving a non-white footballer is a racist fabrication, the debate should be pretty much over by now.

BTW, take Zlatan (who hilariously is as white as Paul Scholes in that odd research...). Is it disrespectful to assign him low Loyalty, subpar Sportsmanship and Temperament (while Ambition and Professionalism are top)?
Is it racist to give Diego Costa low, low figures for Sportsmanship? Wanna talk about Joey Barton? What about Cassano and Balotelli? Two goofballs whose only difference is the skin tone...
All those traits are as "factual" as you can get in a subjective field like rating anything...

Last, but not least, what about Brazilian researchers giving low attributes to Brazilian players? Are they all white and possibly grandchildren of some German immigrants? Or, again, it's basically OBSERVATIONAL evaluation leading to So and So getting lower values because of something that happened IRL?

Quote

I also think that while the suggestions RBKalle has offered for the disparity in the country average ratings have some grounding in reality, it is also almost certainly the case that those ratings are also influenced by negative stereotypes.

Again, most stereotypes come from real-life situations people experienced and then applied to most members of that group.

Do you take the typical "Brazilians are ambitious, flamboyant but erratic" as a derogatory generalization? And why and when did it begin to be such a commonplace that Brazil is now synonymous with a specific brand of attractive, albeit not always consistent or profitable, football?
Was it racism and jealousy? Or was it that people were so in awe with Brazilian players (I guess it stems from the Garrincha, Didí, Vavá, Pelé years) those traits become some sort of national calling card.
Had the first significant contact with Brazilian football involved the likes of Roque Junior, Felipe Melo, Jo and Fred, the general opinion of the average Brazilian footballer would be much worse.

And the same goes for all the possible national characterizations around. Some have even changed over the years... and some will change soon.

Spain's recent success turned them from "poor man's Brazil, perennial chokers" to "tiki-taka machines". Italy are instead still seen as defense-obsessed cheaters, despite having been leaky for a solid decade. But will it last?

Others, where diversity is more common, aren't easy to pinpoint... Is Belgium about big powerful strikers of African descent, or is it about smaller cheeky tricksters? What about Switzerland? Even France has become tougher to pigeonhole

So I don't really thing those football stereotypes are even an indicator of how people from a specific country are perceived. It's just about football. Other sports have similar, but different, scenarios.
 

Quote

Self-reflection is good

Sure it is, as long as it's not coming from debatable assumptions, logical leaps and various half-baked analysis.

[enigmatic]

1 hour ago, RBKalle said:

But do those bias make a FM researcher go "oh, X is black, so he must be a tad more unprofessional and temperamental than Y, who is white"? Or are those judgments based on what a competent and dedicated follower of his club knows about the players' professional life?

Per my point, given that a competent and dedicated follower of his club knows very little about a player's professional life (values other than professionalism are easier to rate tbf) it's not difficult for biases to creep in. Hell, I'd struggle to rate my coworkers on a 20 point professionalism scale, never mind a large group of people I've mostly never met

Sure everybody knows that Messi is super professional and Balotelli isn't and definitely aren't racist for pointing it out, but deciding which player gets 9 and which player gets 11 is a lot more pure guesswork (plus whatever heuristics research teams use, which might have little to do with race but aren't particularly factual either).

[enigmatic]

FWIW their data collection is so bad if I was SI's PR guy I'd be asking them to write a retraction and apology...

 

I filtered the 19.3.0 database for players with skin tone 20 (the blackest black) that had professionalism in the range 1-10 and in the range 11-20. 

That gives me 42 very black guys with below average professionalism, and 58 very black guys with above average professionalism. So actually, the majority of very black guys are above average, the opposite of what their data implies (60%, as opposed to about 40% on their graph). 

Of course, they restricted this to internationals with more than 25 caps.

Problem is, there's only one current player with more than 25 caps recorded as having the blackest skin tone

So where did their data come from? Well if we start including non-players, we get four non-players with above average professionalism including a couple of English club legends and a couple of more obscure ex players. And three people with below average professionalism. That's ...  the opposite of what they're claiming

So how do they get their claim that most of the blackest [former] international regulars in the database are below average. Because they've included the international regulars who are not rated at all for their professionalism in the "below average professionalism" column? These guys are obscure players for St Vincent and the Grenadines, Barbados, Kenya, Zimbabwe Mozambique and Malawi, so I don't think the problem is the researchers being too racist to rate them either! 

It's going to be a similar story with other skin tones. There are 20 professional and 11 unprofessional regular internationals with skin tone 19 - the colour of Wanyama, Benteke and Bolasie. And plenty more Burkinabe and Sudanese internationals whose professionalism is unknown. There just aren't that many white players from nations so obscure their regular internationals' personalities aren't rated...

You can probably imagine I can't be arsed to do this for the whole dataset. But it's not very difficult if you have the data editor.

[laurentius82]

The length social justice warriors are willing to go these days to find "racists" to lynch... To suggest that something like this would reflect a racist agenda at SI is just beyond ridiculous.  Fully agree with RBKalle on everything he has said. 

38 minutes ago, enigmatic said:

Problem is, there's only one current player with more than 25 caps recorded as having the blackest skin tone

It hurts my scientist's soul that something like this can be published in the guise of a statistical analysis. And in the end, what's the point of SJW nonsense like this, other than trying to collect points in the Wokeness Olympics  by unveiling "problematic" people who don't even themselves realize that they are actually nazis? 

[RBKalle]

2 ore fa, enigmatic ha scritto:

Per my point, given that a competent and dedicated follower of his club knows very little about a player's professional life

As "professional life", I mean all the aspects that can help them to assess the 8 Hidden Mental Traits (Ada, Amb, Loy, Pre, Pro, Spo, Tmp, Cont).

Anyone who gets to be a FM researcher should have enough knowledge (first- or second-hand) about his club's players to be able to come up with relatively accurate values for the entire squad. Clearly, assessing a 34yo veteran is easier than doing so for the Academy graduate or for the exotic import who joined 6 months ago and has played a handful of unremarkable games.

I also think most researchers, when in doubt, will keep it average. Going with a 10-12 value you can't go wrong and it won't "break" the player.

And, as per your last post, the data sample looks iffy. I've also checked some data in the editor and their figures simply don't add up...  In the end, it's a lot of BS about nothing, as the sampled data is in the >10 range. Which mean ALL those players are above average anyway...

[enigmatic]

"Relatively accurate" still involves a lot of guesswork and very different research norms (Some German researchers like to give personalities to reserve team youngsters they don't know enough about to guess their pace or strength, the Man Utd one isn't even willing to hedge a bet on whether some of the club's frequently profiled stars are ambitious or not, presumably on the basis they've never been approached by a bigger club. There are valid arguments for both approaches). And yes, from a gameplay perspective guessing average ratings is better than leaving players unrated for the RNG to turn into a psycho. But if you're guessing it's very easy to vary based on stereotype.

I had a guy who's still in FM19 as a player as a flatmate for a couple of years, and I've still got no ****ing idea what actual score I'd give his professionalism or adaptability or loyalty... he gets low single digits for ambition  

 

But yeah, the actual survey is utterly rubbish. Not difficult to do it properly and find that top division reputation (120) very black current players are 75% above average professionalism. Which is still not quite as professional as some of the players at that level with some of the whiter skin tones but it's a bit less sensational

The weird thing is if they were remotely competent they'd probably actually have found a skew due to the kind of normalisation exemplified by those cheesy and weirdly prescriptive regen personality biases by nation that used to appear in the editor (must admit that whilst I found the idea that Belgians are massively controversial quite interesting, I'd have been much more interested in actually editing the regen position and playing style biases of each nation....)

[Federico]

3 ore fa, enigmatic ha scritto:

FWIW their data collection is so bad if I was SI's PR guy I'd be asking them to write a retraction and apology...

If your analysis could be confirmed officialy, I (Sports Interactive) would sue that magazine for diffamation.

If I remember correctly, Sports Interactive was the first Software House in the videogame industry to introduce the sexuality outing in a sport-game context.

And now they come out talking about racism bringing not proofs but a poor research and some populism. In my opinion accusing a Company publicly is some serious stuff, not something to ask only an apology for.

[sporadicsmiles]

They chose players with 25+ international caps. this is going to bias the game automatically because it is not a fair sample. International teams from predominantly white countries have significantly better players than those from predominantly black countries. Even the lower levels of Europe has better players than mid tier Africa. The data is not controlled to CA. And CA is hugely important in the attribute distribution in FM. Either the data they presented needs to be normalised against CA (divide by average CA) or you need to compare players within a defined CA range. They also do not give us the sample size, which is an absolute no-no for studies like this. They should also present some measure of the spread of the data, as this will tell you whether skin colour is being dragged down by a lot of low CA players. 

There is also a fundamental flaw that could come into this from not understanding how the grading system in FM works. A rating of 0 does not mean they have a zero rating. The ratings run from 1-20. If you leave the value as 0, they get a random number assigned when you create the game. This is a lack of knowledge. I am not going to recalculate their data, unless someone can give me an idea how to quickly extract the relevant data from the database. Look at the breakdown of the data used for this article. I show the total number of players, the total players if you exclude those with no data (I chose loyalty at least 1 to filter) and the percent of players included.
 

Skin Tone   		Total      	 total-excluded			% included
20			9			2				22.2
19			55			30				54.5
18			219			130				59.3
17			138			79				57.2
16			189			118				62.4
15			113			72				63.7
14			103			61				59.2
13			126			75				59.5
12			148			107				72.3
11			71			45				63.4
10			80			31				38.8
9			173			120				69.4
8			141			95				67.4
7			300			206				68.7
6			137			120				87.6
5			307			248				80.8
4			78			64				82.1
3			107			89				83.2
2			143			134				93.7
1			25			21				84.0
unknown		262			22				8.40

Assuming the authors of the article did not understand that 0 does not mean 0, but "pick me a number when you set up the game), this pretty much explains their trend. In fact, if you plot it, it looks identical to every figure in this article. It even replicates the dip at skin tone 10. So We can pretty much conclude that the authors of the article did not understand the database values. So for skin tone 20, 88% of the data is 0, which drags the average down (falsely). The lower the number of included data, the more zeros and the lower the average. This is why people should not present data they do not understand. I would have loved to get something like this in peer review, I enjoy ripping apart bad science.

Anyway, I decided to look at some data that is normalized to CA as well. I took all current players who have a number between 1-20 for their loyalty, professionalism, temperament and sportsmanship between CA 143 and 147. I calculate the average and standard deviation for each skin tone value. This gave me 238 players. Which is a small sample size, but I did not want to do more for time's sake. The results are shown below (nobody of skin colour 20 was in this range).

This, by the way, is how to present data like this. Error bars are standard deviation, so you get an idea of the spread of the data. The only one that may show a negative trend with skin colour  is loyalty, but it is not statistically significant. If you wanted to control this further, you could do this for different ranges of CA, to show consistent results.

I would quite like to contact the authors of this article about their bad methods. I am all for talking about racism, and for pointing it out when and where it occurs. Football definitely has a problem with racism, but this is not it. Such crappy misuse of data (whether purposeful or accidental) really gets my goat!!!

[RBKalle]

9 ore fa, enigmatic ha scritto:

"Relatively accurate" still involves a lot of guesswork and very different research norms.

As said, I don't know which guidelines are in place for researchers and for "unknown" data.

We all have seen our fair share of baffling attributes, even for well-known players though. So there MAY be a problem with researchers being expected to assess hard-to-know traits for fringe players too, but let's not forget the difference between, say, Professionalism 13 and 16 isn't really as dramatic as the whole pseudo-research claims.

9 ore fa, enigmatic ha scritto:

And yes, from a gameplay perspective guessing average ratings is better than leaving players unrated for the RNG to turn into a psycho. But if you're guessing it's very easy to vary based on stereotype.

Why must it be that?
If it can't be left blank (and in the hands of the RNG, which I suppose depends on the National Traits anyway), the logical option is going for a middle-of-the-road figure. Again, whether it's an Icelandic or a Nigerian player,  if I don't know much about a player in my team, I'd assign him a sensible value based on the little I've seen. And erring on the side of caution is the easier way. So, when in doubt, keep it 10-12.

9 ore fa, enigmatic ha scritto:

I had a guy who's still in FM19 as a player as a flatmate for a couple of years, and I've still got no ****ing idea what actual score I'd give his professionalism or adaptability or loyalty... he gets low single digits for ambition  

Well, you'd have an idea about how well he'd adapt to life abroad (does he travel, does he speaks foreign languages, has he lived abroad already...). Loyalty depends on how many clubs he has changed and Professionalism you can guess it from how well/often he trains either with the team or by himself.

I reckon he's a lower league player, likely a part-timer, so high single digits or low 10s tops seem reasonable if as far as you can tell his career hasn't had remarkable events.

9 ore fa, enigmatic ha scritto:

The weird thing is if they were remotely competent they'd probably actually have found a skew due to the kind of normalisation exemplified by those cheesy and weirdly prescriptive regen personality biases by nation that used to appear in the editor (must admit that whilst I found the idea that Belgians are massively controversial quite interesting, I'd have been much more interested in actually editing the regen position and playing style biases of each nation....)

I maintain those National Traits are a necessary evil and we shouldn't read much into them.

After all, they're only assessing the general tendency for FOOTBALL PLAYERS coming from that nation. And even within this specific niche, the bell curve is still firmly in the middle, with either extremes (psychos and model pros) being residual.
They're tendencies, so it means that a nation with Below Average (that's the label IIRC) Professionalism will likely generate a slightly higher % of players with slightly lower Pro than one with Above Average.

Again, a stereotype? Likely so, but it's limited to professional footballers!

[forameuss]

On 06/08/2019 at 22:11, Constantine said:

So this is basically cherry picking data to stir some s*** up??

For me "studies" like this cloud any real issues.  Incredibly click-baity article that cherry-picks things to suit their own weak argument, when if you really wanted to write an article about racism, you could have done so much better without much effort.

[nie jem frytek]

Using the same, flawed logic, you could make "an argument" for Nobel Prize being racist.

 

Also it's funny, because in the past I wanted to start a thread about Youth Rating in FM, asking if it isn't actually "racism" of some sort. Just think about Andorran league and the team Lusitanos which creates Portuguese newgens. Lusitanos more often than not make the Andorran league their own due to vast superiority of their newgens, feat that is in no way similar to real life occurence.

[Tajerio]

Hadn't occurred to me that they might have treated "0" as "0," instead of "random 1-20," since I know the game.  But if these Quartz types don't and used "0" as "0", well then.  Makes the article exactly the kind of sensationalist argument against racism that set back properly composed critiques.

[sporadicsmiles]

2 minutes ago, Tajerio said:

Hadn't occurred to me that they might have treated "0" as "0," instead of "random 1-20," since I know the game.  But if these Quartz types don't and used "0" as "0", well then.  Makes the article exactly the kind of sensationalist argument against racism that set back properly composed critiques.

If anything, the discussion should centre around the fact that the research on black players is much lower than for white players. The stats the article uses are invalid because they have not discarded the wrong data points, and so they totally missed the point.

[forameuss]

3 hours ago, sporadicsmiles said:

Statistical insight

Ha...so in an article accusing a game of being discriminatory, they've been discriminatory.  Yass.

[sporadicsmiles]

17 minutes ago, forameuss said:

Ha...so in an article accusing a game of being discriminatory, they've been discriminatory.  Yass.

They did what most laymen do when they encounter numbers and data they do not understand. Did some very basic math, used very little reasoning and did not question it because it agreed with their hypothesis. Here their sin was not understanding what the numbers in the database actually mean. I can actually imagine the growing outrage in the author as they saw so many zeros and thought it meant the players involved had no loyalty, or professionalism, etc. You simply cannot try to draw conclusions from data you do not understand. I'm a scientist, and I deal with this fact all the goddam time (and it is maddening when you cannot actually tell if your data supports or rejects a hypothesis).

The annoying thing about this article is that I took it apart in about 45 minutes of messing. I have the advantage of being a scientist who is used to spotting weird things with data. And I understand what numbers mean in FM. That this data though is then used to support a hypothesis that is false just annoys me. People will read this and think it is true. To make it worse, the article does some number crunching, but does give you a proper method to follow and reproduce the data. And now it is in the public domain, and that is it. I actually emailed the author of the article to tell his analysis are incorrect, and offered to help him correct it to see what actually happens. I do not have a horse in this race, I do not care what the outcome is. I just care that it is the correct outcome.

As you can maybe tell, this irritated me more than it should. Too many people are misusing data in the media at the moment.

[forameuss]

16 minutes ago, sporadicsmiles said:

They did what most laymen do when they encounter numbers and data they do not understand. Did some very basic math, used very little reasoning and did not question it because it agreed with their hypothesis. Here their sin was not understanding what the numbers in the database actually mean. I can actually imagine the growing outrage in the author as they saw so many zeros and thought it meant the players involved had no loyalty, or professionalism, etc. You simply cannot try to draw conclusions from data you do not understand. I'm a scientist, and I deal with this fact all the goddam time (and it is maddening when you cannot actually tell if your data supports or rejects a hypothesis).

The annoying thing about this article is that I took it apart in about 45 minutes of messing. I have the advantage of being a scientist who is used to spotting weird things with data. And I understand what numbers mean in FM. That this data though is then used to support a hypothesis that is false just annoys me. People will read this and think it is true. To make it worse, the article does some number crunching, but does give you a proper method to follow and reproduce the data. And now it is in the public domain, and that is it. I actually emailed the author of the article to tell his analysis are incorrect, and offered to help him correct it to see what actually happens. I do not have a horse in this race, I do not care what the outcome is. I just care that it is the correct outcome.

As you can maybe tell, this irritated me more than it should. Too many people are misusing data in the media at the moment.

But it speaks to the current World, particularly online.  You literally don't have to be correct anymore.  You can write any old rubbish, and if you make it even vaguely believable then there's bound to be at least a few people who agree with you.  Combine that with the hyperbolic nature of interacting with anyone online - pick a side, then take it to the extreme and argue it to the death - then you get stuff like this.

Thing is, I can see why people would believe it.  It looks the part.  Big clickbaity headline, then the average layman will see the data used and think "well, must be true".  

[Federico]

2 ore fa, sporadicsmiles ha scritto:

I actually emailed the author of the article to tell his analysis are incorrect, and offered to help him correct it to see what actually happens. I do not have a horse in this race, I do not care what the outcome is. I just care that it is the correct outcome.

Please let us know when you'll get a reply from him.

I'm sure you'll get one, he looked quite sensible on the argument.

[enigmatic]

6 hours ago, sporadicsmiles said:

They did what most laymen do when they encounter numbers and data they do not understand. Did some very basic math, used very little reasoning and did not question it because it agreed with their hypothesis. Here their sin was not understanding what the numbers in the database actually mean. I can actually imagine the growing outrage in the author as they saw so many zeros and thought it meant the players involved had no loyalty, or professionalism, etc. You simply cannot try to draw conclusions from data you do not understand. I'm a scientist, and I deal with this fact all the goddam time (and it is maddening when you cannot actually tell if your data supports or rejects a hypothesis).

The annoying thing about this article is that I took it apart in about 45 minutes of messing. I have the advantage of being a scientist who is used to spotting weird things with data. And I understand what numbers mean in FM. That this data though is then used to support a hypothesis that is false just annoys me. People will read this and think it is true. To make it worse, the article does some number crunching, but does give you a proper method to follow and reproduce the data. And now it is in the public domain, and that is it. I actually emailed the author of the article to tell his analysis are incorrect, and offered to help him correct it to see what actually happens. I do not have a horse in this race, I do not care what the outcome is. I just care that it is the correct outcome.

As you can maybe tell, this irritated me more than it should. Too many people are misusing data in the media at the moment.

tbf whilst their sampling is particularly bad because they treated unknown variables as a known, the nitpickier side of me doesn't like your narrow and high and strange CA range sample or tiny number of observations for most skin tones either.   

You get more interesting variation by skin tone if you do what their study purported to do (look at what proportion of players of different skin tones are rated higher than the midpoint of the professionalism scale, focusing only on high reputation players) without the sampling issues, but I'm not sure that finding much lower proportions of skin tone 2 people are below average professionalism than skin tone 19 shows what they think it does either (those ultra pale skin tone 1 guys with rep >120 are significantly more likely to have issues with professionalism...)

[TheWednesday]

I've seen it all now. 

[rinso]

what the actual f...  nest there will be a "study" saying FM is homophobic because of the amount of players that "come out" in the game is disproportionate to the amount of players loaded. 

 

What about the odds against having a "son"  ?  that mean FM is anti-kids??

 

utter rubbish