In a recent post on the Rosenhan Experiment, I mentioned that some of the most interesting research comes from very simple ideas or premises. A few years ago, a team at the University of Virginia decided to try out a classroom game that saw university students arbitrarily labeled with a colour, green or purple. The results make for some uncomfortable reading for those of us who think we're above discriminatory behaviour.
"There is little disagreement that some economic inequities may have arisen from historical and cultural factors. Such inequities would be less worrisome if they tended to diminish as attitudes and laws changed, but economic theory offers the disturbing possibility that experience-based (statistical) discrimination may be self-reinforcing, even in the absence of continuing biases and asymmetric opportunities." It was this type of discrimination that the researchers - Roland G. Fryer, Jr., Jacob K. Goeree, and Charles A. Holt - wanted to expose in their paper, Experience-Based Discrimination:
Classroom Games [1].
The students were divided into 'employers' and 'workers', and the workers were randomly assigned a colour - green or purple. Working through a web-interface, the students went through three procedures. Firstly the workers were asked to decide whether to spend a specified amount of virtual money to get an 'education', which would then increase their chances on a 'test'. Then the test was conducted, which was based on a random number, with a bias added to load the odds in favour of those who had paid to be educated.
The third stage was the employers turn - each employer was presented with only two pieces about the candidates - the result of their test, and their arbitrary colour. On this basis, the employers had to decide whether to employ each worker. The experiment was run through 20 rounds, and each time the program made the average aggregate investment and hiring percentages for workers of each colour available to the participants, so that they could use this information in their decision-making.
The results varied, but were fairly ominous. In a number of the experiments, a disparity appeared and then grew as it reinforced itself. "In some of our experiments, we have seen that investment and hiring decisions differed sharply on the basis of the worker’s color. In other experiments, there has been very little separation. We have found that students tend to notice discrimination, when it exists, even if the patterns are fairly subtle in the data."
Suppose that by random chance, green workers happen to invest less in education on average than purple, or are slightly unluckier in their test results. Very quickly, the employers spot the difference, and consciously or otherwise this impacts their decision-making. In response, the green workers now find that they have a lesser incentive to invest in education, since there is a lower chance that the investment will pay off. This depresses education standards among green workers further, thus reinforcing the views of the employers, and so on and on in a vicious circle.
As part of the exercise, researchers held a discussion session with the students, prompting some interesting comments from them. So interesting in fact, that forgive me for quoting a large section verbatim:
"A young woman with this role exclaimed, "this game is fixed," assuming that we had fixed the parameters in
such a way as to induce purple workers not to invest. We assured her that there were no systematic differences between the two groups. We then addressed the same question to employers, and a young man said, "Purple workers just can’t be trusted…they won’t invest." A young woman then retorted, "I stopped investing because no one would hire me." To this, the young man stated, "I did not hire you because you did not invest." These comments captured the essential intuition behind statistical discrimination. Because of their personal involvement in the exercise, it seemed that the students were able to firmly grasp this intuition.
Several of the students with employer roles realized that purple workers were not investing, so they stopped "giving them the benefit of the doubt." Suspecting this, some workers reacted in different and interesting ways. One student chose to free ride; he indicated that, given he was a green worker, he knew he would be hired, so why invest all the time. A close look at the data gives a better depiction of this student’s strategy. He chose to invest enough to ensure that employers continued to give green workers the benefit of the doubt but used several occasions to shirk and avoid the investment cost. On the other hand, another student announced, "I invested every time, even when costs were high, because I felt confident that I would be hired—because I am green."'
The real impact of this paper comes from the fact that its results are apparent in society around us, in the persistent divide between the fortunes of Black and White workers in the U.S. and Britain. The research demonstrates that it is frighteningly easy for even a small initial difference to result in something like a bifurcation event, with two sections of the population veering off on different courses. How do we fix this?
The solutions that came out of the discussion groups are interesting. The students suggested that subsidies could help alleviate the training gap, and probationary hiring periods might allow employers to give the benefit of the doubt more often. The researchers were surprised to note that "nearly all students who openly expressed an opinion agreed that results-oriented policies that required equal representation among workers would not be an adequate solution. Employers indicated that they did not want to be forced to hire purple workers, and workers indicated that their incentives to invest would decrease."
There are some interesting side issues the paper doesn't cover that I'd like to see addressed. One obvious extension would be to see what effect might appear if the employers themselves identified with a particular colour. Another obvious idea would be to implement policies such as affirmative action, subsidies or probations and see what the effect would be. In fact, if I had one criticism of this paper it is that the authors don't pursue this avenue, although to be fair it was probably beyond the scope of their work, given that they were presenting in an education journal.
It would be unfair to call these participants "racist", or to suggest that all people are innately prone to racism but it does highlight how easy it is for us to trap ourselves with rationality. At every point in this simulation, individuals made decisions that were perfectly rational, yet the end result in many cases was endemic discrimination. It is a lesson we should remember.
[1]Fryer, R., Goeree, J., Holt, C. (2005). Experience-Based Discrimination: Classroom Games. The Journal of Economic Education, 36(2), 160-170. DOI: 10.3200/JECE.36.2.160-170
http://layscience.net/trackback/152
The potential for this experiment to royally mess with the mental states of both extreme right- and left-wingers as they argue about hiring policies and racial/gender/ethnic discrimination is wonderful and should be pursued.
I find the most interesting thing, however, to be the possibility that group membership, once identified in the mind of the individual, can have a serious and re-inforcing effect upon self-esteem: Purple realises that they're discriminated against, gives up and therefore fails ability tests; green realises they're privileged, gets confident and passes interviews by impressing the interviewers or bounces back from failures by expecting to get the next break.
This seems to be intuitive of course, and I've been away from study of this and related issues for too long to remember what's been done or to know what's been done recently. Having done my Honours in the effects of self-confidence on self-handicapping, and self-handicapping on self-esteem, however, I'd love to know if anyone's looked at the link between [race/gender/ethnicity/social background], self esteem and what could be termed "get-ahead" behaviours such as further education, career planning and better/more job applications.
You aren't aware of anythning similar, are you?
Completely agree about the potential to mess. I'm often accused of being right-wing or left-wing, but I have little patience with either - all I care about are the facts. The beauty of this game, in my view, is that it gives a platform on which to test out theories in practice.
My area of research is ecology and complexity, so I'm somewhat outside my "comfort zone" research wise, but I've not found much on the esteem question. There's strong evidence that your group has a strong psychological effect - see for example my article on Conspiracy Theorists a while back. On esteem I've heard plenty of anecdotal suggestions that, for example, women lack confidence in their careers compared to men, but I've not yet found a good study on it (which says more about how little time I've spent looking, probably).
I suppose (and this is purely my brainstorming now) you could argue that there might be two components to the esteem issue. Discrimination might give someone a jaded view of government/society (as we seem to see here and in the Conspiracy Theorist paper). Then there might be influences on how people rationalize that view - i.e. do you think that society is stupid and keep your self-confidence, or do you think that maybe you really are less capable. Is it just discrimination that leads to low self-esteem, or is it discrimination plus 'x'.
Either way, I'd love to see this taken further.
Martin is the editor of layscience.net.
Follow Me!
RSS | Twitter
We once did an exercise similar to this at school.
My drama class were all given little red badges, about a centimetre square. We were told to behave in exactly the same way as we normally would and monitor the responses of people in a one week period. Only the class, our teacher and the head-teacher knew about our experiment.
In other classes we would stand out because some had these badges on and others didn't.
The exercise had to be shut down after about 3 days following a member of staff accusing a "badge-wearer" of being akin to a prostitue (presumably because of the colour of her badge). One other incident involved a "non-badger" manufacturing badges and selling them on!
Is this the result of some environmental development or is it genetic?
It does not do to dwell on dreams and forget to live
"Purple realises that they're discriminated against, gives up and therefore fails ability tests."
What of the Jews, who have long endured bigoty, but are nevertheless "overrepresented" (as is the term) in the professions and large, successful business enterprises? To tentatively answer my own question, I think one reason is that Jewish culture puts a tremendous emphasis on success, perhaps primarily to prove themselves better than their detractors. That said, I think that's far from the whole story:
http://home.comcast.net/~neoeugenics/IQgenes.htm
One thing I find amusing is that dog breeders and and fanciers speak quite comfortably about the different levels of intelligence and personality traits (viciousness, courage, loyalty, gentility, etc.) characteristic of different breeds (as you know, all dogs, like all humans, belong to a single species). Likewise, biologists don't inherently preclude differing behaviors or intelligence between different races of wild animals. Indeed, that is what evolutionary theory would predict. Yet if one suggests such things apply to human races, in particular different average levels of intelligence, one is immediately branded a "racist" and despite the scientific evidence to the contrary, is told that such a concept is untenable.
What scientists and people in general have to wrap their minds around, is that there is nothing inherently "evil" or regressive in recognizing that on average, certain races likely have greater intelligence than others. As it is, Ashkenazi Jews (though not a race, they do constitute a distinct group)do better on IQ tests, on average, than American blacks (also not a race, but race is a largely artificial construct anyway). That said, there are many individual American blacks who have higher IQ's than many Ashkenazim, so each person has to be judged AS AN INDIVIDUAL. Indeed, I've yet to meet "the average person"! Furthermore, there are frankly some truly terrible people with high IQ's and fine individuals with lower IQ's. There are differences in behavior, too. Mediterranean peoples generally express their emotions more readily than northern peoples. This is sometimes called "the Latin temperament", though it applies equally, I believe, to Greeks, Albanians, etc.
One can become obsessed with either proving these things (the racists, who are usually white gentiles, which puts them on tenuous ground indeed!) or disproving them (most social scientists, who maintain that average IQ differences, while true, are wholly societal).
I believe that the social scientists need to drop their bias that because differing IQ averages are perceived as negative, they can't (mustn't) be true. And the racists need to recognize that we are all individuals, each with his or her strengths and weaknesses. I think that different average IQ's need to be recognized only to develop sound social policy (and NOBODY should ever be denied the opportunity to get ahead based upon his/her race!). Beyond that, I think we should all stop obsessing about "race", respect each other and celebrate our individual differences and similarities. (By the way, American blacks have their own particular genius. I can enjoy Beethoven and Mozart, but blues and jazz speak to me like no other music. And I really dig that so many blues and jazz bands are "color blind". That's the way we should be all the time!)
Carl said: "Yet if one suggests such things apply to human races, in particular different average levels of intelligence, one is immediately branded a "racist" and despite the scientific evidence to the contrary, is told that such a concept is untenable."
As the Human Genome Project noted: "two random individuals from any one group are almost as different [genetically] as any two random individuals from the entire world." A Jew could easily have more in common with an African genetically than with other Jews.
There are slight "clines" where we see genetic drift over large distances, but there is no such thing as a genetically distinct "race", and there is nothing to support the idea of genetic differences in intelligence between races. Cultures vary in their ability to perform IQ tests, but then it's universally acknowledged that IQ tests are culturally biased to begin with, and on top of that we have to consider different education standards in different parts of the world.
By the way, Dog breeds are the worst possible example to use. Domesticated dogs are widely regarded by science to be an unusual case study - an exception to the rule if you like. Virtually no other mammal displays the same "plasticity" that dogs do.
Martin is the editor of layscience.net.
Follow Me!
RSS | Twitter
That's a very cool example actually, and it makes you realize the power of labels like the Star of David that had to be worn by Jews in WWII.
I would tend to say genetic, but I don't know if there's any specific research backing it up. I would imagine it comes down to the methods the brain uses in spotting patterns in the world around it, that might be appropriate for some tasks but less so for others...
Martin is the editor of layscience.net.
Follow Me!
RSS | Twitter
Martin, you said, “As the Human Genome Project noted: "two random individuals from any one group are almost as different [genetically] as any two random individuals from the entire world." A Jew could easily have more in common with an African genetically than with other Jews.”
Indeed! As I noted in my previous comment, race is a largely artificial construct anyway. That said, one can unerringly distinguish an East Asian from a sub-Saharan African from a European. Among Europeans, few would have trouble separating Swedes from Spaniards. And of course among black Africans, Pygmies are radically different than Tutsis, so clearly there is a significant positive correlation between genotype and phenotype. Races and sub-races, like tribes and families, do exist, interbreeding notwithstanding. They're just not sharply defined, fixed and absolute, as many lay people imagine them to be.
As to the example you gave, it’s hardly revelatory that “A Jew could easily have more in common with an African genetically than with other Jews.” My understanding is that it’s the Ashkenazim, who are arguably closer genetically to Kazars (as well as Central Europeans) than to the ancient Hebrews, that particularly shine in IQ tests. The modern Jewish subdivision most closely allied genetically with the ancient Hebrews are the Mizrahim from the Middle East and North Africa. Their middle-eastern/African heritage could well mean that they have more in common genetically than say an Ethiopian than an Ashkenazi Jew.
All of this notwithstanding, it’s still true that in nearly every case, the more two organisms have in common genetically, the stronger their resemblance and similarities in observed characteristics. Hence our closest relative, the bonobo, more closely resembles a human than does a baboon in appearance, behaviour and intelligence.
As for cultural bias, educational standards, etc., various studies have taken all of these things into account, but the differences remain. Cultural bias also doesn’t account for the genetic illnesses frequently found among Ashkenazim and the observed correlation between people with these illnesses and markedly high IQ. Indeed, the theory put forward to explain this association is that there was tremendous social pressure due to “anti-Semitism” that led the Ashkenazim to develop a high IQ at the expense of a cluster of genetic diseases. This is homologous to the malaria/sickle-cell anemia dynamic amongst (primarily) Africans.
By the way, I disagree that dogs are a poor example. Obviously the differences between dog breeds can be enormous; hence it’s difficult to imagine that the Great Dane, Chihuahua and pug are all members of the same species. Similarly, the border collie shows great intelligence and loves to herd. The greyhound, not so much. I used the dog as an example because it’s an exemplar we’re all familiar with. The fact that we are not as “plastic” in our heritability as the dog doesn’t mean that the traits observed in dogs don’t apply to humans, as you seem to imply by your dismissive statement. A Swede may resemble a Ugandan more closely than a Chihuahua does a Great Dane, nevertheless, nobody would have trouble distinguishing a Swede from a Ugandan! The same principle applies to average intelligence. For all our cultural differences, intelligence is primarily inherited (so we are told), and hence subject to all the evolutionary pressures as any other characteristic. When social scientists refuse to acknowledge a genetic basis for intelligence, they show a shaky grasp of evolution. If intelligence is inherited, then how, if not via our genes? And if people pass on their intelligence to their offspring, then where does this heritability end? As a biologist, I know you can appreciate what I’m saying.
The fact that races, subraces, tribes, etc. blend into each other and “cross-pollinate” doesn’t mean that there aren’t observed different characteristics of different peoples. I believe, based upon my understanding of evolution and what I understand has been observed from countless IQ tests (including “culturally adjusted” ones), that like hair, skin and eye colour, like physical stature, like general dispositions, different groups of related peoples we generally term races and sub-races, have different average IQ’s.
That said, it would be very wrong for “Dubya” to feel smug about his “superior IQ” to the clearly more intelligent Barack Obama. Individuals must always be judged on their own merits and failings.
Third paragraph, last sentence should read, "Their middle-eastern/African heritage could well mean that they have more in common genetically WITH [not than], say, an Ethiopian than an Ashkenazi Jew."
Carl, I think you're labouring under a number of misconceptions that need clearing up.
1) There is no scientific consensus that Ashkenazim Jews perform better in intelligence tests - the study that was done was heavily criticized and didn't actually present more than a hypothesis anyway. Even if there was, there is no accepted definition of intelligence, and the authors of the paper didn't rule out any number of non-selective causes like culture or founder effects.
2) There is not necessarily a linear correlation between genetic changes and phenotype - the true story is much more complex.
3) Social scientists don't "refuse to acknowledge" a genetic basis for intelligence. The fact is that no genetic basis for a racial difference has been found, nor yet is there any real hint of one, and so they have to operate on what they know until genetics shows otherwise.
4) There's no universally accepted measure of intelligence, and so the kind of black-and-white picture (no pun intended) you describe is at best not particularly useful, and at worse could be completely misleading for social research.
5) "Various studies have taken all of these things into account, but the differences remain". No researcher would publish this claim - there are an infinite number of possible correlates yet to be tested.
It's not the case that science refuses to address this question - there's a large community of scientists out there engaged in very strong debate about it. The fact remains though that there is no evidence that race has a genetic impact on IQ (and the possibility of finding it is diminishing), and that there is a vast body of research demonstrating considerable impacts from culture / education / upbringing. As a life-scientist I can't rule out the possibility, but my suspicion is that any correlation would be so tiny as to be swamped under random noise.
Martin is the editor of layscience.net.
Follow Me!
RSS | Twitter
"One other incident involved a "non-badger" manufacturing badges and selling them on!"
The badge manufacturer was no doubt influenced by Dr. Seuss' brilliant story "The Sneetches", in which the "fix-it-up chappie" adds and removes the sneetches' belly stars at will, as each group of sneetches -- star-bellied and non-star-bellied -- strive to become or maintain their status as the "in" group.
Taking your statements in order:
1) I don't think that anybody is taking issue with one group performing better on intelligence tests than another. Rather, I believe that the argument is whether the differences have a genetic component or if they represent a cultural bias (etc.) in the tests themselves.
2) I stated, "clearly there is a significant positive correlation between genotype and phenotype."
You responded, "There is not necessarily a linear correlation between genetic changes and phenotype - the true story is much more complex."
Your statement's not in opposition to mine. I neither stated nor implied a linear correlation. Indeed, "significant positive correlation" implies that it is less than total or "linear".
3) If you mean by "no genetic basis for a racial difference has been found," you mean there isn't a set of genes as yet clearly linked to intelligence, I agree. If, on the other hand, you mean that it wouldn't be expected that evolution of relatively isolated groups would produce such differences (as it clearly has for other racial attributes), and that intelligence tests don't show racial differences, and that these results do not also correspond to racial differences in brain size (which is heritable and held to be one indicator of intelligence), then I most strongly disagree.
4) Agreed. Nevertheless, the ability to do well on an IQ test is a generally fair measure of reasoning ability. Some tests are of course better than others, but dismissal of IQ tests as totally incapable of measuring intelligence I think is far from justified. As for my presenting a "black and white picture", I hardly think so. As I strove to make clear, some people from each and every group will be very intelligent, and some others, downright stupid. I said that this is about averages, and that I've yet to meet "the average person". I also mentioned that the very concept of race is considerably vaguer than most people realize. At any rate, whether such conclusions are particularly useful in social research is not an issue I took a position on, although of course it's always better to found policies on truth (whatever it may be) rather than delusions. I suspect the fact that there's no universally accepted measure of intelligence is largely fueled by the racial differences noted in testing that you dismiss because there's no universally accepted measure of intelligence. Talk about Catch-22!
5) Sorry, I don't know what you mean by this "infinite number of correlates yet to be tested."
As for the scientific debate on the issue, you are correct. Not ALL scientists refuse to address the question. An increasing number are supporting my position, but I well remember when about 20 years ago J Philippe Rushton at the University of Western Ontario first published a study suggesting that of the races, as generally defined, orientals had the highest IQ's, followed by Caucasians, etc. He was absolutely slammed by his peers, journalists, and others, not so much because his methodology was obviously flawed, but because "such racist views have no place in our society". Fortunately, that attitude is slowly beginning to change. We can't dismiss something merely on the grounds that we find it distasteful, and we can't permit such an oft-times unconscious bias to guide our conclusions.
BTW, I'm going to be traveling for the next few day, so I may not be able to continue our conversation. Try not to miss me too much. :)