Back in grad-school I was the TA for the upper level Social Psychology course. The teacher, Ed Hirt, would assign 2-page papers for the students to write, on some particular topic in Soc Psych, usually asking them to relate the particular phenomenon to a real experience they had. When we had the sections on stereotypes and then prejudice, he asked them to write about some time where it had turned out relying on a stereotype had been beneficial.
It was amazing how many of the students just could not grasp that notion, and returned a paper where they condemned stereotyping.
This inability to consider an upside to stereotypes extended to at least some of the grad students. Me, with my cognitive psychology background wanted to think about them like schemas, summaries that could be very useful, although at times quite negative, and tried to start discussing this with one of my pals, knowing full well that this was not quite the PC way to discuss the topic, but we were scientists, damnit, and as Mind Hacks say, science takes no sides, and I wanted to explore it. It was akin to holding a cigarette lighter to a gas leakage, because everything out of her mouth were about how bad stereotypes are, and how much she did not want to be stereotyped as a woman, and on and on and on. And, yes, I sympathize with some of this as neither one of us are particularly traditionally gender conformists, but it was, in some ways, odd. Stereotype = bad. And, here I wanted to wrest it back, in order to try to understand why we have them in the first place, and seem unable to ditch them, if they supposedly are so bad.
(And, yes, I go the schema way –we have to order and simplify our world, and frankly, there are plenty of times I’m happy if people stereotype me and treat me accordingly, for example in my role as a university teacher, and I do not want them to get to know me personally and intimately in my own unique snow-flaky way, because that would really interfere with teaching).
Although there were others of the social grad students who mentioned that there were some work on stereotype accuracy and the kernel of truth, it was kind of like hush hush. Even more so than talking about IQ. But, we were only grad students.
I think the person they talked about was Lee Jussim, who has written an absolutely excellent book called Social Perception and Social Reality – why accuracy dominates bias and self-fulfilling prophecy.
I was glued to this book. His writing is very engaging and clear. He is well aware that this is a controversial topic, and he lays out his thesis carefully and logically. And, he provides data. Essentially, this is a book long series of meta-analyses, either his or others, involving a whole series of seminal, influential studies, where he brings in and interprets the numbers. The effects. It is beautiful!
At the same time, I was teaching my advanced social cognition course. We use Taylor and Fiske’s social cognition book (newer edition, but now out of print), which is dense, wide-ranging, and fantastic in many ways in what it covers. (They even do not too shabby entries both on neural nets, and on emotion, although they are very clear about the fact that this is a bit outside their focal areas of research).
The amount of research they pull together is immense. It is also all together narrative. Effects are stated in directions:” participants who listened to an expert were more persuaded than participants who listened to a non-expert.” Sometimes the results are more complex: “those who were in a happy mood were more persuaded by the experts than those in a neutral mood, but over time this effect dissipated.” Etc. Anybody reading psychology are familiar with these statements. In papers they are accompanied by numbers, graphs, inferential statisics, the magic p value, and perhaps an estimated effect size dutifully copied from an SPSS output. In the text book the numbers are rarely (if ever) presented. (Sentences in quotes are not actual citations, but plausibel statements I made up. Those types of effects do exist, but it is more for an illustrative purpose).
It is quite difficult to assess how strong these effects are. Would I notice it or not? Does it have real consequences? Are they so strong that people living in incommensurable worlds, enveloped in biases, bounced around by circumstances, uncapable of controlling feelings, thoughts and behaviors that are problematic yet difficult to even get out of?
To that last, Jussim says, no. This is, in some cases, how findings are presented – and he gives plenty of examples. The effects are real, but they are small to moderate. Now, he only goes through research that is particularly interesting for his assessment of how accurate the stereotypes we use are, and there are loads of more very interesting effects in social psychology that need this kind of treatment. (Which is why I want to get into meta-analysis, because, why collect new data, when so many people have spent so much time doing it already, and now it needs to be pulled together).
But to go back to some of the results he discuss, that are frequently cited as large factors in behavior even outside social psychology, without really going in to how large they are, or are not.
He starts with Rosenthal and Jacobson’s (1968) Pygmalion effect. He has done quite a bit of research in schools and on learning, so this fits in with what he does. The Pygmalion effect is used as an argument for how malleable human performance is. I had actually heard a researcher mentioned that it doesn’t replicate, although he didn’t elaborate on it (just enough to sow a seed of doubt), but you see it regularly mentioned as evidence in all sorts of circumstances in a hopeful manner. The story is that teachers were told that some students were going to bloom, planting en expectation of excellence. And, this expectation caused the students to perform better. This sounds wonderful! Simply expect excellence from your students, and they will just magically fulfill your prophecy and do just wonderful!
Except, it is not that simple, and that was not quite what the study actually did, although there seemed to be some effect of expectation. the original study, and several versions of follow-ups, he concludes there is some effect of expectancy. If I recall, about the level of r = .200. What he also brings up (and summarizes research of), is that teachers are good at accurately assess their students. Their expectancies are not overinflated, or overly malleable, but relatively spot on. Also, he shows that the students themselves are not passive beings that will simply respond to whatever someone expects of them. No they are themselves interested in signaling clearly who they are and what they are capable of. Does not this seem reasonable? Don’t you want your friends and loved ones to know you (and you ensure that they have the opportunity to), so that the expectations are somewhat aligned with who you are? There is slosh– but limited slosh. Both perceiver and perceived (or both interactant) are interested in the relationship being somewhat accurate. I think this is kind of what Robert Kurzban is on about here also (although there is more to it).
He discusses Hastorf and Cantril’s (1954) famous study where students from Princeton and Dartmouth watched a football game between their respective teams, and then assess the judgments of the play. They differ. This is interesting. But, Jussim goes to the numbers and reasons about how to assess agreement and disagreement, and places the numbers in the larger context of a football game (this starts on page 22, and goes on for a couple of papers). Particularly interesting exercise for someone like me who likes the numbers, but is not interested enough in American football to have a prior feel for how many fouls or plays or what have you we are talking about. There is actually an awful lot of agreement, although there clearly is some disagreement also. This is an interesting bias, of course, and sometimes very small biases can have big consequences. But it is not as if the two groups saw different games.
A third eye-opener was of a classic study on the halo effect by Snyder, Tanke, & Berscheid, (1977). The premise is that there is a belief – a stereotype – that attractive people are warmer friendlier smarter what have you. And, the study is elegant. Male students were given a picture of a woman they were going to speak with via the telephone. In some she is plain, in others beautiful. The person the participant talks to is randomly assigned to condition, so there is no relationship between the attractiveness of the person on the photo, and the attractiveness of the woman in the other room. The conversation is recorded. Independent raters are then rate the conversation on warmth and friendliness. Those men that believe the woman is attractive are rated as behaving warmer and friendlier. So is the woman responding! This is taken as evidence that the activation of a stereotype (beautiful is good), can result in a self-fulfilling prophecy (the responding woman becomes warmer and friendlier). It is a nice study, but Jussim proceeds to analyze it suggesting that it is not necessarily strong evidence for the biasing effect of stereotypes (pages 95-97). (Also, the presented analyses does not allow for a retrieval of the effect sizes). For one, later research has shown that the attractiveness stereotype perhaps is not biased as there is some evidence that attractive people are more socially skilled. So, perhaps the men were simply acting in accordance with more or less correct expectations (although now the cue of attractiveness was uncorrelated with the target woman). The men had a false expectation. It is also not quite clear that there was an activated stereotype, but, as he says, why would 19-20 year old men possibly want to impress on a pretty 19-20 year old woman? In my quick re-reading, I don’t see him addressing that the women (of arbitrary attractiveness) became warmer when the men behaved more warm. What he is after is, what does it say about stereotypes and their biasing effect? And the answer is, it is unclear. It is an elegant, inventive study, but it leaves questions open. And, interestingly, for such an influential study, there has only been one attempt at direct replication By Anderson & Bem (1981) (and as far as direct replication, I don’t think it would be considered that in the current climate of replication interest). Incidentally, the effects did not replicate! For such an influential study, this is kind of remarkable.
He reviews this research, because they have been used as evidence against his work on stereotype accuracy. Stereotypes cannot be accurate because of self-fulfilling prophecies, for biased observations, for halo effects that will render them inaccurate and misleading. But, although the effects are there, they are not particularly strong.
He proceeds to summarize the work on stereotype accuracy. (These are chapters after chapters, and very interesting). Of course, he is entirely in the clear about that stereotypes can be pernicious, negative, and used for horrible purposes, and that they can be wildly inaccurate and smearing. But, is this all they are? Why do we have them, if they are so misleading? Because having a wildly inaccurate view of the surroundings cannot possibly be beneficial. He then walks through evidence where stereotype accuracy is assessed. That is, does peoples beliefs about some group (women, men, African Americans, etc) on some dimension (e.g. school performance) somewhat match how that group actually behaves and performs? Although people are not perfect, they are more accurate than inaccurate in many instances. It is like the stereotype is a summary of knowledge from which one can make a reasonable bet. It is a bet, so it may not pay off, of course. It is only as good as the accuracy of that stereotype is, and, at least when you interact with people from another group, it may be of interest to keep that stereotype updated so as not to make too many mistakes.
It doesn’t mean that stereotypes can’t be wildly off, or that they can be highly negative. It doesn’t mean that there are those motivated to pursue negative biased stereotypes about others (as in propaganda). Doesn’t mean that stereotyping can lead to very bad consequences, or that, as an individual it can become tiresome if not outright painful to frequently be assessed as something you are not, and he is very well aware of this.
But, if you are a scientist, you have to actually try to look at your topic area and try to understand what is really going on, rather than being influenced by other agendas. It is perfectly fine, of course, to want to end pernicious stereotyping, but it is also a good idea to try to accurately understand the phenomenon. He discusses this extensively (really, read the book. It is great). There is a confusion about meaning when it comes to stereotypes, and it is inconsistent. One could consider stereotypes those types of summary information that are wildly biased and plain wrong (for whatever reasons, lack of knowledge, motivation to derogate), but then there needs to be a term – like perhaps schema – for the summaries we have that aren’t perfect maps of everyone, but are rough priors for orienting among people – but, as he points out, we don’t.
It is not unreasonable that we do have some clue about how to assess others, using some kind of rough summary knowledge acquired from multiple sources that may or may not be accurate, of course.
It squares with work on snap-judgments where, supposedly, you can assess better than chance a number of traits from rapidly presented snapshots of faces (Todorov). Or non-verbal assessment. Having a degree of accuracy in our perceptions of others (or, at least not really wildly wrong) would be in our interest.
As I also have some, as yet written up for publishing work where I assess attitudes via IAT, and where I find a link between this and the judgment I’m interested in, but it is incredibly weak, and the more impressive “ground” information is how top of mind it is about the problem with stereotyping, I really need to know this type of calibration. Should you really write about it as pernicious hidden prejudice? Or is it the echo of both our frequency of encounters with “others”, and our need to actually categorize, and a perfect equipotential is just not something that can ever be reasonably expected among most people within a fairly wide swath around the mean.
I have the research interest here, but my students, especially those that are mainly interested in getting an overview, have a practical interest also, in that stereotyping and prejudice and that type of relations are a very active problem. And, I honestly am not sure if what we know from research can help resolve this in a “once and for all” manner – but points more to that this is a bit like cleaning house, or looking over the infrastructure. Just a couple of minor anecdotes dealing with the idea of priors and wrong bets.
I recently went to what was called an inspiration seminar at my university, which was inspirational. One of the topics were the work on recruiting and retaining people from underrepresented groups, and to work on making sure that people from these groups feel welcome.
During the panel talk, one of the members recounted anecdotes from people – students, personnel, not sure – at that department who, well, don’t look like me. I really look stereotypically Swedish (blonde, blue-eyed, rather tall) so a Swedish assumption is correct. But, of course, there are plenty of swedes who look nothing like me, who nevertheless are as Swedish as I am (more so, since I’m actually now American). These were people I gather who looked Asian or middle-eastern (because their ancestry were from there), but who were from here. But, people would make the bet they were not, and start speaking English to them. I can see that that is quite annoying. I would think that anybody that has lived in non-homogeneous places have heard those stories, and even on occasion being one of those that did the wrong bet.
But, I also have this alternate anecdote. This year, I have met two women with very Swedish first and last names. In fact, both blonde. As I don’t care whether I use Swedish or English I spoke English to them because they spoke English to me first, although I wondered whether I really should speak Swedish. (There have been people who start speaking English to me on account of my last name.) One of them is a California girl with a Swedish parent, and the other a Canadian woman with a Swedish husband. I would have bet wrong here too, if I had started speaking Swedish with them.
This then becomes a real conundrum. On the one hand, the annoyance of people in minority status is completely understandable, but the mistaken bets of individuals are also understandable and are likely not due to serious bias or racism, but more due to sensitivities to frequencies, and having made a wrong bet. The experience is asymmetric.
Now, I can analyze that from a science point of view, but, practically, what do you do? Be timid? Like the prof at my old uni who kind of hesitantly went up to me when I was a rather large 8-months pregnant and asked, aheming, if, well, perhaps, was I possibly, maybe, pregnant? I know where that comes from, but it was kind of funny.
Perhaps with time the missed bets about background will also fade, as what you look like will become less of a reliable marker, as was first suggested by Kurzban’s study, and that Tom Stafford recently covered. (And, anecdotally, after 23 years in the US, my view of “white” contains many different ancestries from all of Europe and the middle-east). I don’t think the mistakes are signatures of hidden prejuice. I think they are honest mistakes from bets/heuristics that went wrong.
I’m beginning now to think about the frequencies, the distributions out there, and how our psychology picks those up – but I think I have to leave that to a later post.
Some refs – for those I didn’t link.
Anderson, S. M., & Bem, S. L. (1981). Sex typing and androgyny in dyadic interaction: Individual differences in responsiveness to physical attractiveness. Journal of Personality and Social Psychology, 41, 74–86.
Hastorf, A. H., & Cantril, H. (1954). They saw a game: A case study. Journal of Abnormal and Social Psychology, 47, 129–143.
Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom: Teacher expectations and student intellectual development. New York: Holt, Rinehart, and Winston.
Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception and interpersonal behavior: On the self–fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656–666.