Whistling Vivaldi: How Stereotypes Affect Us and What We Can Do (Issues of Our Time)

by Claude M. Steele

And mistakes were costly, as when, at thirteen, after arriving at six in the morning, I waited all day to be hired as a caddy at an area golf course, only to be told at the end of the day that they didn’t hire Negroes. This is how I became aware I was black. I didn’t know what being black meant, but I was getting the idea that it was a big deal.

What makes both of these contingencies identity contingencies is that the people involved had to deal with them because they had a particular social identity in the situation. Other people in the situation didn’t have to deal with them, just the people who had the same identity he had.

by imposing on us certain conditions of life, our social identities can strongly affect things as important as our performances in the classroom and on standardized tests, our memory capacity, our athletic performance, the pressure we feel to prove ourselves, even the comfort level we have with people of different groups—all things we typically think of as being determined by individual talents, motivations, and preferences.

We know that anything we do that fits the stereotype could be taken as confirming it. And we know that, for that reason, we could be judged and treated accordingly. That’s why I think it’s a standard human predicament. In one form or another—be it through the threat of a stereotype about having lost memory capacity or being cold in relations with others—it happens to us all, perhaps several times a day.

I’d been a fool. I’d been walking the streets grinning good evening at people who were frightened to death of me. I did violence to them by just being. How had I missed this… I tried to be innocuous but didn’t know how…. I began to avoid people. I turned out of my way into side streets to spare them the sense that they were being stalked…. Out of nervousness I began to whistle and discovered I was good at it. My whistle was pure and sweet—and also in tune. On the street at night I whistled popular tunes from the Beatles and Vivaldi’s Four Seasons. The tension drained from people’s bodies when they heard me. A few even smiled as they passed me in the dark. (Chapter 01)

I was wondering if Eric’s (from The Nod podcast) laugh was influenced by this same phenomenon. It’s a very goofy, threat-vaporizing laugh that seemed out of “character” to me, but would put anyone at ease—not that I know him well at all. Also: hence the book title: makes so much sense now! :)

What it did for sure was change the situation he was dealing with. And how it did this illustrates nicely the nature of stereotype threat. In a single stroke, he made the stereotype about violence-prone African American males less applicable to him personally. He displayed knowledge of white culture, even “high white culture.” People on the street may not have recognized the Vivaldi he was whistling, but they could tell he was whistling classical music. This caused him to be seen differently, as an educated, refined person, not as a violence-prone African American youth.

Like code-switching, he felt he had to put on some whiteness to separate himself from the stereotype and be less threatening to others. Ugh, so much work, but totally evident.

They found something very interesting: white students who were told the golf task measured natural athletic ability golfed a lot worse than white students who were told nothing about the task. They tried just as hard. But it took them, on average, three strokes more to get through the course.

What about the black participants, did they show the same behavior? Or is there a specific golf stereotype that threatens?

Jeff and his colleagues had put a group of black Princeton students through the same procedure they’d put the white students through. And, lo and behold, their golfing was unaffected. They golfed the same whether or not they’d been told the task measured natural athletic ability.

They tested this idea in a simple way. They told new groups of black and white Princeton students that the golf task they were about to begin was a measure of “sports strategic intelligence.” This simple change of phrase had a powerful effect. It now put black students at risk, through their golfing, of confirming or being seen to confirm the ancient and very bad stereotype of blacks as less intelligent.

White students told this vs not told this performed equally as well in this experiment, but the black participants had an even wider gap than the white students did in the “natural athletic ability” experiment.

The contingencies they faced were threats in the air—the threat that their golfing could confirm or be seen to confirm a bad group stereotype as a characterization of their group and of themselves. Still, it was a threat with a big effect. On a course that typically took between twenty-two and twenty-four strokes to complete, it led whites to take three more strokes to complete it, and blacks to take five more strokes to complete it.

the line for black students was consistently lower than the line for other students. At every level of entering SATs, even the highest level, black students got lower grades than other students. If we assume the SAT is a rough measure of preparation for college, this meant something dramatic: that among students with comparable academic skills, as measured by the SAT, black students got less of a return on those skills in college than other students.

I wondered on that flight home whether these two pieces of evidence—about their grades and about their sense of belonging—had anything to do with each other.

Still, a pattern that mirrored the scenes in Jane Elliott’s classroom emerged. Women tended to underperform in advanced math classes, where evidence suggests they feel the collar of gender stigma, but not in advanced English classes, where evidence suggests the collar is less felt.

We’d gotten to a point where two plausible, but very different, ideas could explain our simple finding that, after we had carefully selected women and men who had strong and equal math skills, the women did worse on a difficult math test we gave them than the men—that is, classic underperformance. Our explanation was that frustration during a difficult math test made women worry about confirming, or being seen to confirm, the societal view about women’s poor math ability, and that this worry, in turn, interfered with their performance. This is how we saw the “collar” of stigma interfering with math.

Whereas Larry Summers glibly assigned weight to the genetic hypothesis. Without any data to distinguish between these two hypotheses examined here.

The challenge was to find something extra to real life that would lower the pressure women normally feel during such tests, that would somehow remove the “collar” of stigma during difficult math tests.

We stewed, feet on our desks; then we had a simple idea. We’d present the test in a way that made the cultural stereotype about women’s math ability irrelevant to their performance. We’d say something like this: “You may have heard that women don’t do as well as men on difficult standardized math tests, but that’s not true for the particular standardized math test; on this particular test, women always do as well as men.” (This is a close rendition of what was actually said in the real experiment.)

And the results were dramatic. They gave us a clear answer. Among participants who were told the test did show gender differences, where the women could still feel the threat of stigma confirmation, women did worse than equally skilled men, just as in the earlier experiment. But among participants who were told the test did not show gender differences, where the women were free of confirming anything about being a woman, woman performed at the same high level as equally skilled men. Their underperformance was gone.*

Wait. I thought the control was going to that they weren’t told anything about gender and test performance. The control should’ve been more similar to normal math test conditions (no mention of gender), right?

social science observers have been trying to explain poor outcomes—economic, social, educational, medical—experienced by blacks throughout the twentieth century. Like the sportswriters, Scott argues, they have tended to focus on deficiencies, one of which dominates all others—what he calls “psychic damage.”

We used the same test on which blacks had underperformed under ordinary testing conditions. But we told a different group of participants that the test was a “task” for studying problem solving in general, and emphasized that it did not measure a person’s intellectual ability. With this instruction, we made the stereotype about blacks’ intelligence irrelevant to interpreting their experience on this particular “task,” since it couldn’t measure intellectual ability. With this instruction we freed these black participants of the stigma threat they might otherwise have experienced on a difficult test of verbal reasoning.

These differences seem so small, I wonder if there are other effects going on. Are these trials blind to the people interacting with subjects, or do *they* know which instructions subjects are being given? There may be lots of other environmental influences in play if the experimenters are aware of the conditions while experiment is being conducted/test being taken. Must be blind studies! (Which can be done by writing the disarming statement on the exam without anything needing to be said). I don’t think this would invalidate the results necessarily, but could point to mechanisms that stereotype threat are imposed.

did people experiencing stigma pressure actually worried about confirming the negative group stereotype.

Grrrr typo count has escalated in recent pages. This just an example

Interestingly, when black students expected to take a test of ability, they tended to spurn things black, reporting less interest in, for instance, basketball, jazz, and hip-hop than white students. But when the test was presented as unrelated to ability, black students strongly preferred things black. They seemed to be eschewing these things when preferring them would have encouraged a stereotypical view of themselves. It was the spotlight of the negative group stereotype they were avoiding.

This could be *heavily* influenced by experiment or behavior if not blind! Careful on the conclusions here.

Finally, there was evidence that the threat of the stereotype pressured a search for excuses, a search for something other than oneself to blame poor performance on. We asked participants how much sleep they’d gotten the night before the experiment. Black students expecting to take the ability test, reported getting fewer hours of sleep than black students expecting to do a nonability task, and fewer hours of sleep than whites with either expectation.

Same experiment risk here too. Have to think they thought of this but why isn’t it called out in the text?

He gave a difficult thirty-minute test (a section of the SAT verbal exam) to groups of white and black students in spare classrooms. For those groups in which he wanted black students to experience the pressure of the racial stereotype, he did as Josh and I had done; he simply said the test was a test of verbal ability. Remember, this minimal statement reminds blacks that this is a test on which their performance could confirm the standing stereotype about their group’s intellectual ability.

This sounds like it wasn’t a blind study at all!

But this didn’t happen for blacks in the academic rear guard of Mikel’s high school sample. The black test takers who cared less about school were unfazed by the stereotype. They performed the same regardless of whether the test was presented as an ability test or as a nondiagnostic laboratory task. And in both of these groups, they performed at the same level as white students who, like them, didn’t care much about achieving in school and didn’t have strong skills.

I am thus proposing something simple: the sense of having a given social identity arises from having to deal with important identity contingencies, usually threatening or restrictive contingencies like negative stereotypes about your group, group segregations of one sort or another, discrimination and prejudice, and so on, all because you have a given characteristic. What raises a characteristic we have to a social identity we have are the contingencies that go with the characteristic, most often, threatening contingencies.

Identities do have positive and neutral contingencies too—things one confronts in society because one has a given identity that are not threatening, but just neutral or even positive. Men have to go to men’s bathrooms and women to women’s bathrooms. This arrangement is indeed a contingency, of sexual identity. Yet it is so routine as to be essentially neutral. We don’t notice it.

Except when the lines to the two restrooms are of significantly different lengths? Or the restrooms’ sanitary conditions differ significantly? (He also addresses non-binary/trans issues)

And this is just what happened. The results were dramatic. White males taking the difficult eighteen-item test, represented as one on which “Asians tend to do better than whites” performed, on average, a full three items worse than white male participants who were told nothing about the test.

Artificially introducing stereotype threat where there was none before has the same effect.

No special self-doubting susceptibility seemed necessary.

Women whose background questionnaire reminded them of their gender identity—with questions about whether their dorm was coed and why they would prefer coed living—got 43 percent of the math test questions they attempted correct, whereas women whose background questionnaire asked questions that did not remind them of their gender identity—with questions about their telephone service—got 49 percent of the questions they attempted correct.

Importantly, though, when the background questionnaire reminded them of their ethnic identity—with questions about what languages they spoke at home and how many generations of their family had lived in America—this underperformance was eliminated entirely. They got 54 percent of the items they attempted correct. Simply varying which of their identities they were reminded of before taking the twelve-item math test produced an average difference of two points in their score—

Just before women math students took a difficult math test, we reminded them that they were Stanford students. This reminder greatly reduced stereotype threat’s effect on their performance. We later found that R. B. McIntyre, R. M. Paulson, and Charles Lord had independently found the same thing. They dramatically reduced stereotype threat’s impairment of women’s math performance by reminding them, just before the test, of positive women role models.

Carol said that when this advice is offered to white and Asian students, most of them readily take it, dropping the course for a grade and following one of the alternate strategies. To Carol’s surprise, though, when the advice is offered to black students having trouble, they more often rejected it, persisting in the course past the point when one can drop it without getting a grade, and thus often getting a low grade that jeopardized their medical school chances.

It was as if the black students she described were staying in this course to disprove the stereotype hanging over their heads—following their and my parents’ advice. They pushed on when a person not facing this “allegation” might have simply switched to a better strategy.

Was the syndrome of over-effort and self-sufficiency evidenced in Treisman’s research, and the observations of Carol Porter’s organic chemistry advisers, caused by stereotype and identity threat? Or was it perhaps a general characteristic of African Americans that stems from a socialization process—again, I hear my own father’s words in my ears—that stresses working twice as hard as others to succeed?

Another group of participants, however, went through exactly the same procedure, except that they were told the anagram task was a measure of cognitive abilities. For the black students in this group, this labeling made the stereotype about blacks’ cognitive abilities relevant to the anagram task. Their frustration on the task could now confirm the stereotype about their group’s abilities. Unlike the whites in this group, then, they were now under stereotype threat.

Black students tried 2x the optional anahrams as white/Asian. Was this researcher-blind study though? Or did they know they were giving instructions to specific groups (control vs exp, or black vs white)? Exp conditions can have as much of an effect as the stereotype threat conditions, perhaps acting through similar means, but perhaps not.

We thus had answers to both of our questions. Academic over-efforting among black students could be evoked in the laboratory—easily so. It’s a real phenomenon. Second, it seems to be caused by the identity pressure of stereotype threat. It didn’t happen without this threat, when the anagrams were presented as just anagrams, as puzzles unrelated to cognitive abilities.

On the difficult test, women did worse under stereotype threat than women not under stereotype threat and worse than men in either group. But the tables were turned on the easier test. Women under stereotype threat did better than women under no stereotype threat and better than men in either group.

At the frontier of their skills, stereotype threat and the motive to disprove the stereotype hurt performance, presumably through the set of interfering reactions described earlier. But back a ways from that frontier, where the task was easier and the frustration less intense, trying to disprove the stereotype boosted performance beyond that of all other groups.

When men drop out of quantitative majors in college, it is usually because they have bad grades. But when women drop out of quantitative majors in college it usually has nothing to do with their grades. The culprit, in their case, is not their quantitative skills but, more likely, the prospect of living a significant portion of their lives in a domain where they may forever have to prove themselves—and with the chronic stress that goes with that.

He conceived a program that, as far as studying calculus is concerned, and put a bit crudely, tried to get black students to study more like Asian students—in particular, to work in groups, groups that spent a lot of time (at least six hours per week) together outside of class talking about calculus, among other things.

Treisman’s workshops teach the skills of group learning that directly make learning calculus easier. But what were these skills correcting for? Here his anthropological research was revealing. They were correcting a tendency among these black students to protectively isolate themselves and to over commit to self-sufficiency—strategies that might help them avoid people who they worried might stereotype them, but that would also isolate them from help they needed.

Want to know more about this group formation and forced interaction: practically speaking, how did they go about doing it? Was there any resistance and what to do in the face of it? (Especially before they had determined that it would lift performance?)

But the mean arterial blood pressure of their black counterparts rose dramatically while they took the test. People under stereotype threat might not be able to report that they were anxious or even whether their feelings were anxiety or love, but it didn’t mean that they weren’t anxious. Their physiological responses told us plainly they were.

Talking to a black stranger, compared with talking to a white stranger, should put white participants at greater risk of being seen stereotypically, as perhaps racially insensitive. And if this stereotype threat causes anxiety, these participants should have higher blood pressure. They did, substantially higher.

Wow, and the white person trying hard not to be seen as racist is more anxious, and then might misinterpret the anxiety as dislike for the black stranger they’re talking to? Aw, man.

When we’re at risk of confirming a stereotype that we don’t like, and it’s about something we care about, our minds race. They’re probably doing all sorts of things: arguing against the stereotype; denying its applicability to us; disparaging anyone who could ever think that of us; feeling sorry for ourselves; trying to buck ourselves up to disprove the stereotype. We are defending ourselves and coping with the threat of being stereotyped. We’re probably aware of some of this defending and coping. But much of the time we may miss it, unless we try very hard to listen.

“[N]early a third of black men (34%) and black women (31%) are considered hypertensive, compared to 25% and 21% of white men and women, respectively.” One might think these disparities are due, in part, to black-white differences in income, education level, body mass index, smoking, and the like, all factors that cause hypertension. But these disparities persist even when they are adjusted for the effect of these factors.

The formidable conditions faced by low-income blacks in this rural area were not enough, by themselves, to elevate blood pressure. For that to happen, people had to be high in John Henryism beliefs; they had to care about succeeding enough to endure a struggle against difficult conditions. Race was a factor, too. Whites who lived under these conditions and were high in John Henryism did not show elevated blood pressure. It was high John Henryism pitched against the conditions of being poor and black in these rural, southern areas that raised blood pressure.

even when black, Latino, and Native American students overcome other disadvantages in trying to gain parity with white and Asian classmates, they face the further pressure of stereotype and identity threats. Even privileged students from these groups have an extra, identity-related pressure working against their achievement.

The Massey team, however, did find something that alleviated this effect—black professors. Black and Latino students in these schools experienced virtually no stereotype threat in classrooms where the professor and probably more of the other students were black and Latino.

But one form of feedback did work, for both black and white students. I will call it the Tom Ostrom strategy. The feedback giver explained that he “used high standards” in evaluating the essays for publication in the teaching magazine. Still, he said, having read the student’s essay, he believed the student could meet those standards. His criticism, this form of feedback implies, was offered to help the student meet the publication’s high standards. Black students trusted this feedback as much as white students, and trusting it powerfully motivated them to improve their essay.

Why was it so effective? It resolved their interpretative quandary. It told them they weren’t being seen in terms of the bad stereotype about their group’s intellectual abilities, since the feedback giver used high intellectual standards and believed they could meet them. They could feel less jeopardy. The motivation they had always had was released.