Accuracy dominates bias and self-fulfilling prophecy [pdf]
internal.psychology.illinois.edu
internal.psychology.illinois.edu
"Interpersonal expectations relate to social reality primarily because they reflect rather than cause social reality. This is the case not only for teacher expectations, but also for social stereotypes, both as perceptions of groups, and as the bases of expectations regarding individuals. The time is long overdue to replace cherry-picked and unjustified stories emphasizing error, bias, the power of self-fulfilling prophecies, and the inaccuracy of stereotypes, with conclusions that more closely correspond to the full range of empirical findings."
I'm a layman when it comes to this but I'd be very interested in any refutations of this paper.
I doubt there can be refutation in the sense that "its conclusions are obviously wrong".... the catch is how you interpret/act on those conclusions.
Yes, stereotypes become stereotypes because they are, or at least recently were, "largely true" (in the statistical sense - i.e. not always but "too often to ignore"). E.g., in my country, a lot of the petty criminals are gypsies - it would be foolish to reject that truth. However, the stereotype "gypsy = criminal" is still incredibly harmful (despite being grounded in truth), both at the individual level (producing harm to innocents) and at the group level (doesn't solve the problem, only contributes to making it worse).
The bigger problem is that people are ill-equipped to handle these thoughts in the brain simultaneously: that a stereotype can be largely true, and yet it's harmful/a bad idea to act instinctively based on the information it provides. The correct responses to a stereotype are complex/should be based on reason - those based on instincts/reflexes are very often wrong. But it's so hard to resist the instincts, that often it's simpler to just dismiss the stereotype as "false".
Of course if you are able to accurately take probabilities and uncertainty into account when making decisions, you can make good use of observed correlations; but most people tend to be better at dealing with deterministic rules (I certainly am) and need to consciously remind themselves not to overestimate the accuracy of a stereotype.
I try to treat everyone as an individual, but when I don’t have the opportunity to do this I have no choice but to fall back on my life’s experience. Sometimes this means I avoid people I wouldn’t if they looked or acted differently, but I have no desire to be a victim to prove a point.
> I have no desire to be a victim to prove a point.
This is not a dig against you. Humans - myself included - are incredibly flawed (while being incredibly awesome at the same time). I just find it helpful to keep this in mind. Personally, applying the stereotype as you say is a very reasonable (even "rational") choice. I do it too. But, at the same time, we should recognize the limitations of it - both you and I have probably treated a gypsy unfairly. That's ok, I'm fine with that, I've given up hope that I can be perfect - but, I still find that it helps to remind me that I'm not. And what I'm doing is not helping them. I don't know how I could _personally_ behave better (without harming myself) - but, that only means that I don't believe a solution can be found at the "personal" level. I believe "more of the same" will only work to reinforce the stereotype - so at a policy level, I believe that policies are needed to actively help them raise above their conditions (yes, maybe even positive discrimination is justified - at least, it's _something_ ), while at the same time working to fight against the symptoms (prevent the formation of "ghettos"/ their isolation in roma-only communities).
But, I digress. This was about our own weaknesses, and how the stereotype, even when grounded in truth, can be harmful, because of the positive feedback loop it creates (which is very hard to break). In particular, a stereotype tells you that e.g. gypsies used to be petty criminals (in large proportions), but it does NOT tell you that they have to be so in the future. It does not tell you the "why" - maybe it's coded in their DNA. But what if (more likely, IMO) it's actually coded in our social norms? Because those, we can change! It takes effort, and time, but it's doable.
I don’t have any useful suggestions other than the observation that when Roma immigrate to countries like the USA or Australia they completely changed their culture and have been absorbed into the general population.
You cannot be a Gypsie/Roma in North America!
Their culture involves roaming in caravans from town to town, possibly doing bits of labour, but mostly running scams. Kids don't go to school, they live their lives as they please.
In France, I saw a Roma Caravan pull up on a local farmers property my French friend indicated that if the farmer complained - he would get a physical threat on his person and property (though I am aware that property rights / routes are different in Europe).
Can you imagine a caravan pulling up on a Farmer's property in rural Alabama? And threatening violence? They would be shot by the owner, or instantly run out of town by the local cop who could give a rats behind about whatever their 'cultural status' might be.
The EU institutionally and systematically enables the 'Roma culture' - which is in and of itself nearly impossible to make civilized in the terms that we understand. (Not necessarily saying anything against the culture other than we don't know how to 'make it work' as the previous commenter observed)
How does the EU provide for the education of Roma children as they roam through Europe? Health care? Taxation? Etc. etc..
I understand the impetus for the EU to protect culture, and also - you know - history Nazism and all that - but the cultural situation may be 'implacable'.
Dissolving their cultural identity and putting them into a suburban school getting accounting degrees might be the only way we know how to 'fix the problem'. Which of course comes at the cost of cultural dissolution.
Not as a society or culture in general. Ignoring prevalence of poverty, crime, prejudice, illiteracy of one's own group (and note an individual belong to multiple group and some choose to identify with one or more in particular) is usually not beneficial.
But it will benefit an individual if they say are not involved in crime and want to apply for a job, school or move up the corporate ladder or interact with a neighbor, and others don't automatically apply stereotypes to them based on their skin color, race, nationality, language spoken etc.
What the paper actually argues is to measure accuracy of the stereotypes instead of calling them wrong. If it is say 90% accurate it is not on the person hiring to fix this perception.
It it's similar with black people in tge US and crime, however there are known institutional causes of it, not really a self fulfilling prophecy either. (They also apply to so-called white trash and result in the same, but in different states and therefore it is only a specific subgroup that is stereotyped instead is skin colour. Similar about Puerto Ricans or Mexican immigrants.)
On the other hand, whenever people are casually talking about 'darker skinned faces', in Romania meaning gypsy, then we're talking about something else. These are the situations where people are just promoting a gratuitous sense of pride and superiority for its own sake.
I was once robbed by Gypsies/Roma in a rather spectacular way.
The whole family was involved.
Given the fact that these folks roam around the continent willy nilly, it's rather difficult for them to be actually employed. Maybe in the past, as specialized workers?
Point being - thievery and corruption may not just be a more common artifact of their culture it may be an existential and critical part of their culture.
'Running scams' may be deeply integrated into their culture, such as it is a common custom. You go 'to school' from a young age. They might 'learn how to rob' from a young age - because it's their livelihood.
Not only this - robbing people, and a culture of robbing people from a young age, implies a deeply problematic view of others by people in that culture, it has to be deeply cynical.
To be clear: I'm not saying any of this is absolutely true. Someone needs to write the paper.
I'm saying that some negative attributes are a fundamental part of specific cultures. Maybe it wouldn't be too hard to think of attributes of our own culture which are not so great.
But because this would just be a pop-culture holocaust - it could never possibly be discussed, let alone researched.
I guess sometimes the interests of peaceful and polite culture trump the darker truths we may sometimes want to shield from ourselves.
But the whole gypsy thing was meant to be just an example of how a stereotype, while being grounded in truth, can also be harmful due to the instinctive response that it generates (that's because I believe the solution is to integrate them in the regular societies and thus destroy most of their toxic "cultural heritage"; but the prevailing response in our societies is to avoid them as much as possible, which mostly helps to perpetuate the problem).
My grandmother had some interesting encounters with Roma when she was younger. Apparently, the only time people locked their doors in town was when the Roma caravans rolled through (otherwise, you'd risk them entering your house uninvited, etc, etc, which did happen to her at least once). However, they were apparently also quite good at certain kinds of labor, such a replating (industrial?) mixing bowls with tin, a job she hired a "gypsy king" for, who assembled some of the younger Roma in his group for the task.
I.e., mobility is not mutually exclusive with having so-called marketable skills or providing services.
I feel however that these kinds of jobs are going to be far and few between as everything becomes professionalized.
How so? You just state this as a given, and in fact, that also could be seen as a stereotype.
> people are ill-equipped to handle these thoughts in the brain simultaneously
Actually, the evidence is that people are very well equipped to handle these thoughts simultaneously: they use the stereotype only as long as they have no further evidence.
A. At a personal level: do I need to go into details how being presumed "by default" to be non-trustworthy and potentially a criminal/thief can seriously impact your lifetime opportunities?
B. At a group level: the instinctual response to criminals is to avoid &"isolate" them; this creates ghettos. It doesn't help that they also have a tendency to auto-isolate themselves. I fail to see how this response is any of help towards actually solving the problem (well, short of the "final solution", I guess, if you're willing to seriously consider that).
>do I need to go into details how being presumed "by default" to be non-trustworthy and potentially a criminal/thief can seriously impact your lifetime opportunities?
Yes, please do. When you do, take into account that "per default" means "as long as no other information is available", which only applies to people that you are completely unconnected to and have never interacted with.
Also, please try to incorporate this into a model of the real world with tradeoffs. So for example weigh my/your/his/her right not to be victimized, and everyone's duty to take care of themselves and others, with the right of someone not to be viewed with suspicion.
> the instinctual response to criminals is to avoid &"isolate" them
Citation needed.
> tendency to auto-isolate themselves
Citation needed. Aren't you stereotyping here?
> help towards actually solving the problem
What is "the problem" you refer to? What would you see as a solution?
I'm not sure whether there's something you actually disagree with - or you just want to "be right" by creating unreasonable burdens of proof. If so - here: you're right, I'm just a random dude making baseless affirmations, please ignore me.
> Aren't you stereotyping here?
Yes I am. I thought you're a supporter of the idea that stereotypes are true more often than not?
[edit] And to clarify, because it seems to be needed: by 'harmful', I mean harmful to the victims of the stereotype. If you discriminate someone, you're not necessarily harmed - but the other one may be. This isn't a "slavery was harmful to ranch owners" argument - it's not even a "slavery was harmful for the US society" argument. It's just a "slavery was harmful for black people" argument. I'm honestly flabbergasted that someone would honestly demand proof for that sort of statement.
Yes, more often than not. Sometimes they're wrong. For example, the now common stereotype that stereotypes are mostly wrong is wrong.
> Slavery
What does slavery have to do with it, except nothing?
Let's give a more realistic example:
Say there's a woman walking alone at night. She notices a large man walking behind her. She crosses the road. Now she knew nothing of this man, so she just used her stereotypes (large men=possibly dangerous) and the situation (late at night) to inform her decision. Evil stereotype or prudent decision? Has he been "victimized" by the (almost certainly) incorrect assumption of her's that he is a danger?
Analogy - I was hoping that might help you understand what I meant by "harmful" (ie "who is the one being harmed?"). I didn't mean that you were harmed by it.
> Let's give a more realistic example
But I never stated that all stereotypes are harmful, or all create victims, or whatever. I stated that SOME do, and gave a specific example that you challenged. And now you move to a different example to prove.... what?
You still haven't at all expanded how you got from a stereotype about the Romani, which others in the thread have noted is quite useful, to slavery. Do you think slavery was the result of inaccurate stereotypes and implicit bias?
I think you would find it very hard to reconcile that hypothesis with the actual history of slavery.
In some circumstances. While being incredibly harmful to some people/ in some other circumstances. These are not mutually exclusive!! You know, kinda' how atomic energy can be beneficial while at the same time being incredibly harmful. These properties can be simultaneously true.
> You still haven't at all expanded
But I did... again, in the message you are replying to. Here, I'll try again to convey the same message, without analogy this time:
Do you believe that the stereotype "gypsy = criminal" would/should be perceived as "largely accurate and largely beneficial" for a (honest) person of Roma ethnicity? Because that's what you were challenging me to dis-prove.
Slavery was indeed "largely beneficial" to the slave owners. Even to some empires/ nations. Seldom to the slaves. See how "largely beneficial" has multiple facets? That's what I meant, I was hoping that for slavery you wouldn't find it hard to believe that it was NOT largely beneficial, for the slaves themselves.
So your analogy was "something that is beneficial to one side but not to the other"? Nothing to do with stereotypes at all? Because that is so distant that I don't see how it is an analogy of anything.
So "being stabbed in the gut is not beneficial to the stabbee, so don't you see that stereotypes are harmful"?
We all operate with stereotypes that are largely accurate and largely beneficial, not to single groups (as you seem to imply), but to everyone (obviously different stereotypes etc.). In fact, without stereotypes (=generalizations) we would not be able to function and the world would be a terribly dangerous place. Now are they all upside with no downside? Of course not. But that's a straw man, because what thing is there with no downsides?
And I also don't see that you've actually supported your assertion that the stereotypes you mention are "incredibly harmful" to those negatively affected. Because the research shows that that is not the case, meaning the real data about the individual replaces the stereotype very quickly. Do you have evidence that this is incorrect?
Please quote this research, you're just being obnoxious and I don't understand why. No such research can possibly exist, since it defies the most basic common sense.
http://www.spsp.org/news-center/blog/stereotype-accuracy-res...
"Common sense is the collection of prejudices acquired by age eighteen." -- Albert Einstein
Now, in real life, when is information clear? Have you heard of, say, "blind auditions" for orchestras? I guess the people who developed that (and got results!) were stupid, because if clear and individuating information was present, there would be no problem at all.
[1] your claim: > stereotypes (I) mention are (not) "incredibly harmful" to those negatively affected. Because the research shows that that is not the case
I.e. your claim is basically that research shows people negatively affected by stereotypes are not harmed much by them. There is and can be no such research.
I assumed that you were aware that the burden of proof is on the person making the claim, that "I claim something, prove me wrong" is not how this works. That assumption appears to have been incorrect.
Given that the burden of proof is on you, what I provided was evidence that makes your claim unlikely, specifically the following: "When individuating information is clear and relevant, its effects are “massive” (Kunda & Thagard, 1996, yes, that is a direct quote, p. 292), and stereotype effects tend to be weak or nonexistent.".
So if the effect is "weak or nonexistent" as soon as any individuating information is known, your path to showing "incredible harm" has just become that much more difficult. Of course, you haven't actually presented any evidence whatsoever, so I guess it doesn't really make much of a difference.
Want proof for what I claimed? Stop quoting Einstein, and read the study you linked.
Nope, that would be you. You make a claim and don't back it up. You invert the burden of proof, in a "prove me wrong", which is both dishonest and ridiculous.
I write show (as in illustrate and you claim I wrote proof). So, put up with proof of your claim, or shut up, and don't accuse others of the intellectual dishonesty you yourself are guilty of.
https://www.merriam-webster.com/dictionary/show
And the Einstein quote was aimed at your "proof by common sense" assertion, which is just ridiculous on the face of it.
Unsurprising at all. The cure is either blind auditions or extended trial periods. Or impartially designed tests.
Tricky question is who is supposed to bear the cost.
That is if they problem really is of big magnitude and worth solving in the specific case. Which is done by measuring the magnitude and outcomes rather than detecting presence of prejudice. Then investigating the causes by hard analysis rather than politically biased guesswork.
Making lots of money out of stock trading is rather easy in the presence of clear and relevant individuating information about all entities that are being traded. So is any decision making. That does not mean that in real life, "decision making" or "stock trading" are simple problems that we shouldn't be concerned with.
The irony is that there is strong prejudice that every prejudice and bias is harmful and in a major way. This actually remains to be seen.
To follow from your example, investors in many index funds do not have access to individuating information yet are making profits. Which means their perception of the market is at least somewhat accurate. But some investors who do have access to special information (and are not caught as doing insider trading) cash out much bigger. Often they do no better or worse too. You get too actually evaluate outcomes of each investment (or hiring) system.
> Yes, please do. When you do, take into account that "per default" means "as long as no other information is available", which only applies to people that you are completely unconnected to and have never interacted with.
Not OP, but it's easy to come up with examples: in Spain it's very probable that after interviewing two candidates with otherwise the same qualifications, the one without the gipsy accent will get the job offer over the one with a gipsy accent.
The same has been shown to be true in the US for resumes with typically African-American names vs the same resume with a typically Caucasian name. I don't really know where your skepticism is coming from, as it's not even surprising that these things happen.
This contradicts the evidence that: "When individuating information is clear and relevant, its effects are “massive” (Kunda & Thagard, 1996, yes, that is a direct quote, p. 292), and stereotype effects tend to be weak or nonexistent."
After the interview, you have individuating information. In fact, you probably have individuating information beforehand, such as the resume, maybe references etc.
So your claim contradicts this study. As I haven't seen any evidence for your claim, my tendency is to believe the study.
But certainly something that probably warrants further study.
The very same paper is littered with examples of cases in which stereotypes affect the judgement of people when your clause ("only applies to people that you are completely unconnected to and have never interacted with") doesn't apply. So I'm not sure if you're being intellectually honest, or cherry-picking the most extreme circumstances examined in the paper and generalizing them. Just one of a multitude of examples cited:
> a child from a high socioeconomic background who displayed a mixed performance on an ability test was rated as higher in ability than a child from a low socioeconomic background with the same performance (p. 293, citing two different studies)
So if you honestly believe that the stereotypes associated with gipsies cannot impact their lifetime opportunities in a harmful way, please do read in detail the Kunda & Thagard paper. And by all means believe the paper.
See, the whole idea of treating everyone as special is a fancy individualist invention that has currency in some countries. It also matters how much said person is special and how many of them are. Then you can call it a problem or not. Or by tracking outcomes of people subject to prejudice with strong statistical tools which do not involve cherry picking by hand.
Instead keep the fancy unexpected results published and cited, even if irrelevant or wrong.