Last year, Princeton University recommended that the incoming students in the class of 2019, as well as any and all other members, read the book Whistling Vivaldi by Claude M. Steele. I learned at that time that it was a study about the biases we carry and the way society's collective biases can affect us in different situations (along with what corrective actions can be taken with respect to those biases), so it seemed pretty interesting at that time. My interest was further piqued by seeing that my advisor was reading it then too, so I figured I should read it at some point. Classes and (later) preparing for my general examination got in the way, so I didn't really have a chance to read through it until now. Having read through it, I'm writing a short review below (my first formal book review for this blog in a long time), consisting of a summary (that is probably going to be incomplete) as well as some longer questions after the jump. At this point, I should say as a reminder that I am a physicist, not a sociologist/psychologist/anthropologist, so a lot of the questions or remarks that I make will probably be easy to refute with current research in the field; given that it isn't my specialty (so I wouldn't necessarily know how to go about it), and in the interest of starting a conversation, I would like to invite you, the reader, to respond to these issues in the comments below.
The book is written by a psychologist, and focuses on a series of controlled experiments designed to understand the role that societal biases (regarding group identities, such as race and gender) have on people's performance in situations where their presence may or may not be socially expected (e.g. women versus men in STEM fields, or white versus black people in professional basketball). These experiments have been replicated in a variety of settings and seem to show fairly consistently that especially among people who would otherwise be expected (based on prior experiences, such as test scores or basketball performance in high school) to perform well in a field where their group is not well-represented, being aware (generally through life experience, or explicitly in the course of a test in that field) of their group identity can negatively impact their performance, because their minds are subconsciously consumed by the stress of trying to consciously break the stereotype of poor performance in that field by other people in their group, to the extent that this rumination hinders their ability to focus on the task at hand and causes them to perform worse than people in a more fully represented group; by contrast, people in an underrepresented group who are told that their performance isn't meant to be judged as diagnostic of their ability, or who are otherwise convinced that past stereotypes are irrelevant to the test at hand, do just as well as people in a more fully represented group (whose members do not generally see a performance boost from such a treatment). These experiments are fleshed out to a fuller extent over the course of the book, looking at different instances of stereotype threats in different situations, and looking at different metrics (including physiological ones, like blood pressure or heartbeat patterns) for measuring the impact of stereotype threats on performance. As the author offers several lessons throughout the book on what can be done to counter stereotype threat issues, the book closes with a summary of what works under different broad circumstances, and what further issues need to be faced.
There are a few issues that I have with the book overall. One is that the "introduction" of various people (mostly academic collaborators, students, and others in academia) that the author has met seems extraneous. The sense I got was that he wanted the reader to feel personally engaged with the details of the author's academic life, but by giving only cursory importance to these meetings (and then, only to set up descriptions of the experiments they ran), his writing on this matter came out rather stilted in the end. Thankfully, the book itself is fairly short (only 220 pages), so this is less of a problem. The two bigger problems that I have follow. The first of these problems is that in the middle of the book, the author introduces the story of a man named John Henry Martin, born in the early twentieth century, who eventually freed himself from sharecropping servitude and worked incredibly hard to better his lot, though he paid a steep price in doing so (in the form of much worse health and shortened lifespan). This anecdote is used to demonstrate how a great work ethic in conjunction with terrible struggles against difficult circumstances can generally adversely impact a person's life. However, though the chapter ends not too far after that, the remainder of the chapter essentially engages in idle speculation about whether previous examples of people struggling under stereotype threats would have suffered in deeper ways if the stereotype threats were chronic (rather than short-term/acute). Not only was I confused and lost in that short section, but that whole passage seemed rather disconnected from what came before and after. As far as going into the effects of chronic stereotype threats, what comes before is sufficient, and a smoother transition to what comes after would be all that is needed. The second of these problems is that throughout the book, the author tries to use a mix of anecdotes and general descriptions of studies and their results to engage a general audience. The issue I have is that either the anecdotes should be made more compelling to justify the qualitatively-described studies, or the studies should be described more quantitatively to justify the brief anecdotes that precede them; the author tries to be a too clever in splitting the difference, but the end result feels a bit less meaty than I'd like. (As a matter of personal preference, given my expectations of what a description of controlled studies should have, I would prefer that the book feature a level of quantitative description similar to One Nation, Underprivileged by Mark Robert Rank (which I read about a year ago).)
Despite these reservations, I did enjoy reading the book, as it opened my eyes not necessarily to biases per se, but to the ways that biases can affect people's performance (as well as how people affected by these biases view and respond to them). It certainly got me thinking about some of the other ways that biases and stereotype threats can cut. With that in mind, follow the jump to see some questions that I have about the book, and please do respond in the comments with thoughts of your own (especially if you have citations for more recent research that could answer some of my naive questions).
Is there a limit to how broadly stereotype threats can be applied? An example in the first chapter discusses an experiment showing how white people and black people perform about equally well at golf when given no further instruction; from this baseline, the performance of essentially just the white people drops (compared to the baseline) when all participants are told that the experiment is to measure natural athletic ability, while the performance of essentially just the black people drops (compared to the baseline) when all participants are told that the experiment is to measure "sports strategic intelligence". This is supposed to play into the stereotypes of lesser intelligence among black people and lesser athletic ability among white people. Yet I have to ask: is the equal performance of white and black people at the baseline due to a much more equitable mix of people from different racial/ethnic backgrounds in professional golf now, compared to 30 years ago? From what I understand, golf wasn't considered a sport that required particular athleticism until Tiger Woods came on the scene; if Tiger Woods didn't arrive and golf hadn't changed in the last 30 years, would the issue of athleticism continue to be a stereotype threat for white people if they didn't credibly believe that athleticism was really a factor in golf? To put it another way: would the stereotype threats of athleticism come into play (lowering the performance of white people while not affecting the performance of black people) for a written exam if it were framed as a matter of "physical athletic endurance while writing" (or something like that), or is there a threshold of credibility that needs to be crossed for a stereotype threat to become effective? An experiment later in the book about made-up stereotypes of "overestimators" and "underestimators" seems to suggest that there is no such threshold, but these are stereotype threats that were made up on the spot, and for which the test subjects had no prior experience. I'm not sure that stereotype threats can be easily reversed (not countered, but fully reversed, such that a group that was the beneficiary of a stereotype threat is now a target, and vice versa) when they already exist in society; in particular, I'm not sure that telling a group of people taking a hard math test that "women are known to do better on this test than men" would cause men to do worse than average (while buoying the performance of women) if there is already a more pervasive stereotype that men do better than women in math.
Then again, that leads me to a further question: can positive stereotypes (like that of Asians being good at math) be detrimental under certain circumstances to their beneficiaries? I didn't see much in the book about this, but I wonder if by being the beneficiary of a positive stereotype, people in that group actually end up doing worse because of the added pressure of upholding (rather than breaking down) the stereotype. In particular, I wonder if, for a math test administered to white and Asian people in which test takers are individually given information about the test and then left to take the test separately (so that test takers don't know who else is taking the test, as knowledge of the other test takers may by itself invoke stereotype threats in different groups), a statement that "Asian test takers are known to do better than average on this test" would cause anxiety that would lead to worsened performance among those Asian test takers compared to those in a control group (with no statement one way or the other).
The author of the book describes an experiment determining the effects of stereotype threats on black and white high school students in Los Angeles, CA, specifically trying to break down the effects of stereotype threats on high- and low-performing students in each racial group; they found that stereotype threats have significant performance suppression effects for high-performing black students (and their elimination can eliminate the performance gap for that group in turn), while having no effect on high-performing white students or low-performing students of either racial group. The author concludes that low-performing black students may already be alienated from school and career pursuits by this point, which is why they don't care whether stereotype threats are present or countered. However, this leads me to the following question: assuming that the proportion of black students in the low-performing group is higher than those in the high-performing group (which may not be a valid assumption), could the alienation from scholarly pursuits seen in low-performing black students in high school be the result of stereotype threats discouraging more black students earlier in the educational system (as the stereotype threats accumulate and chronically persist, these students internalize them enough to depress performance over the long term, so only a smaller proportion of students can make it through high school remaining interested in academic activities)?
The author then describes the transient notion of race, giving the examples of Anatole Broyard, a black man who (thanks to his light skin color) managed to pass as white for essentially his entire adult life (and was able to access all of the opportunities that white people had that were denied to black people of his time), as well as several prominent black American artists (musicians, writers, and so on) who moved to Paris and found that the stereotypes that hung over them in the US had vanished in France, so their American blackness (or black Americanness) meant something completely different (and generally much more benign) in France than in the US. Yet it seems like most of the examples of stereotype threats given in the book are for a historically marginalized group in a particular field in relation to a historically normalized/dominant group in that same field (such as blacks versus whites in scholarly pursuits, whites versus blacks in athletic pursuits, women versus men in math, et cetera). How then might things look when considering two groups interacting in which both groups are marginalized in a broader sense? As a particular example, consider a working-class Latino family that moves into a predominantly black working-class neighborhood. Are the same stereotypes that hang over Latinos in white American culture still going to hang over this family in the black neighborhood? Will the Latino family try to assimilate more with their black neighbors, even though black culture is not dominant in broader American society? What sorts of stereotype threat situations might be observed here?
The author further describes an experiment testing the math performance of Asian women college students, looking at what happens when stereotypes of "women being bad at math" or "Asians being good at math" are raised (in comparison to no stereotypes being raised); the experimenters found a significant positive effect for the latter stereotype and a significant negative effect for the former stereotype. I'd be interested though to see what happens when both are combined: this gets at what many activists call "intersectionality", which is that people can see (sometimes simultaneously) different aspects of privilege and marginalization from different facets of their identity. Would being reminded of being both Asian and female have a net neutral effect when averaged over a large group, because different Asian women may value their identities as female versus Asian to different degrees?
This is further extended to young Asian girls given age-appropriate math tests, and being made to color a picture before the test; the picture could be a girl holding a doll, a landscape, or a person eating rice with chopsticks. The above results were qualitatively replicated even for these young girls. That said, I wonder if there are other issues (brought up in the book) that aren't being captured for these girls and women in these tests, especially regarding the context-dependent nature of stereotype threats. If many of these girls and women are coming from cultures where women are (perhaps stereotypically) expected to be just as good in STEM fields as men are, is the decrease in performance when stereotype threats related to femaleness are raised going to be of smaller magnitude compared to that observed for white girls and women given the same corresponding tests? Additionally, if these girls and women are coming from more communitarian cultures where issues of honor and shame for the family are (perhaps stereotypically) more emphasized, could those factors play a role in the performance of these test takers, or is that going to be too hard to measure? Moreover, if stereotype threats are context-dependent, would Asian American girls and women taking these tests respond differently when taking them in the presence of white American girls and women versus Asian girls and women (who were born and raised in the corresponding Asian countries), with regard to issues of assimilation into American culture versus retention of Asian culture and values? (I'm intentionally and sloppily using "Asian" as a catch-all here.)
Overall, the tone of the book is an optimistic one, and I am likewise optimistic that it could lead to more interesting research and developments with regard to bringing marginalized groups in any field out of those margins. Already, I have seen programs like the Meyerhoff Program at UMBC (and similar programs, at other universities, modeled after it) counter the detrimental study habits of high-achieving black and other minority students that arise due to stereotype threats. I have also recently seen articles that further develop the ideas of having minority college students read about upperclassmen in those universities discuss their struggles and eventual academic and social successes in order to boost the sense of belonging and reduce the feelings of alienation and impostor syndrome in these students. Additionally, several months ago, I went to a talk in Princeton University about further experiments discussing how achievement gaps can be further diminished between normalized and marginalized groups when the idea of a person's knowledge being expandable through a process of learning, rather than being rigid and based on innate ability, is emphasized. Clearly, even an outside observer like myself can see that these ideas are taking hold and are being further developed, so I'm hopeful to see not just more research but also more implementation of this in the near future.
The book is written by a psychologist, and focuses on a series of controlled experiments designed to understand the role that societal biases (regarding group identities, such as race and gender) have on people's performance in situations where their presence may or may not be socially expected (e.g. women versus men in STEM fields, or white versus black people in professional basketball). These experiments have been replicated in a variety of settings and seem to show fairly consistently that especially among people who would otherwise be expected (based on prior experiences, such as test scores or basketball performance in high school) to perform well in a field where their group is not well-represented, being aware (generally through life experience, or explicitly in the course of a test in that field) of their group identity can negatively impact their performance, because their minds are subconsciously consumed by the stress of trying to consciously break the stereotype of poor performance in that field by other people in their group, to the extent that this rumination hinders their ability to focus on the task at hand and causes them to perform worse than people in a more fully represented group; by contrast, people in an underrepresented group who are told that their performance isn't meant to be judged as diagnostic of their ability, or who are otherwise convinced that past stereotypes are irrelevant to the test at hand, do just as well as people in a more fully represented group (whose members do not generally see a performance boost from such a treatment). These experiments are fleshed out to a fuller extent over the course of the book, looking at different instances of stereotype threats in different situations, and looking at different metrics (including physiological ones, like blood pressure or heartbeat patterns) for measuring the impact of stereotype threats on performance. As the author offers several lessons throughout the book on what can be done to counter stereotype threat issues, the book closes with a summary of what works under different broad circumstances, and what further issues need to be faced.
There are a few issues that I have with the book overall. One is that the "introduction" of various people (mostly academic collaborators, students, and others in academia) that the author has met seems extraneous. The sense I got was that he wanted the reader to feel personally engaged with the details of the author's academic life, but by giving only cursory importance to these meetings (and then, only to set up descriptions of the experiments they ran), his writing on this matter came out rather stilted in the end. Thankfully, the book itself is fairly short (only 220 pages), so this is less of a problem. The two bigger problems that I have follow. The first of these problems is that in the middle of the book, the author introduces the story of a man named John Henry Martin, born in the early twentieth century, who eventually freed himself from sharecropping servitude and worked incredibly hard to better his lot, though he paid a steep price in doing so (in the form of much worse health and shortened lifespan). This anecdote is used to demonstrate how a great work ethic in conjunction with terrible struggles against difficult circumstances can generally adversely impact a person's life. However, though the chapter ends not too far after that, the remainder of the chapter essentially engages in idle speculation about whether previous examples of people struggling under stereotype threats would have suffered in deeper ways if the stereotype threats were chronic (rather than short-term/acute). Not only was I confused and lost in that short section, but that whole passage seemed rather disconnected from what came before and after. As far as going into the effects of chronic stereotype threats, what comes before is sufficient, and a smoother transition to what comes after would be all that is needed. The second of these problems is that throughout the book, the author tries to use a mix of anecdotes and general descriptions of studies and their results to engage a general audience. The issue I have is that either the anecdotes should be made more compelling to justify the qualitatively-described studies, or the studies should be described more quantitatively to justify the brief anecdotes that precede them; the author tries to be a too clever in splitting the difference, but the end result feels a bit less meaty than I'd like. (As a matter of personal preference, given my expectations of what a description of controlled studies should have, I would prefer that the book feature a level of quantitative description similar to One Nation, Underprivileged by Mark Robert Rank (which I read about a year ago).)
Despite these reservations, I did enjoy reading the book, as it opened my eyes not necessarily to biases per se, but to the ways that biases can affect people's performance (as well as how people affected by these biases view and respond to them). It certainly got me thinking about some of the other ways that biases and stereotype threats can cut. With that in mind, follow the jump to see some questions that I have about the book, and please do respond in the comments with thoughts of your own (especially if you have citations for more recent research that could answer some of my naive questions).
Is there a limit to how broadly stereotype threats can be applied? An example in the first chapter discusses an experiment showing how white people and black people perform about equally well at golf when given no further instruction; from this baseline, the performance of essentially just the white people drops (compared to the baseline) when all participants are told that the experiment is to measure natural athletic ability, while the performance of essentially just the black people drops (compared to the baseline) when all participants are told that the experiment is to measure "sports strategic intelligence". This is supposed to play into the stereotypes of lesser intelligence among black people and lesser athletic ability among white people. Yet I have to ask: is the equal performance of white and black people at the baseline due to a much more equitable mix of people from different racial/ethnic backgrounds in professional golf now, compared to 30 years ago? From what I understand, golf wasn't considered a sport that required particular athleticism until Tiger Woods came on the scene; if Tiger Woods didn't arrive and golf hadn't changed in the last 30 years, would the issue of athleticism continue to be a stereotype threat for white people if they didn't credibly believe that athleticism was really a factor in golf? To put it another way: would the stereotype threats of athleticism come into play (lowering the performance of white people while not affecting the performance of black people) for a written exam if it were framed as a matter of "physical athletic endurance while writing" (or something like that), or is there a threshold of credibility that needs to be crossed for a stereotype threat to become effective? An experiment later in the book about made-up stereotypes of "overestimators" and "underestimators" seems to suggest that there is no such threshold, but these are stereotype threats that were made up on the spot, and for which the test subjects had no prior experience. I'm not sure that stereotype threats can be easily reversed (not countered, but fully reversed, such that a group that was the beneficiary of a stereotype threat is now a target, and vice versa) when they already exist in society; in particular, I'm not sure that telling a group of people taking a hard math test that "women are known to do better on this test than men" would cause men to do worse than average (while buoying the performance of women) if there is already a more pervasive stereotype that men do better than women in math.
Then again, that leads me to a further question: can positive stereotypes (like that of Asians being good at math) be detrimental under certain circumstances to their beneficiaries? I didn't see much in the book about this, but I wonder if by being the beneficiary of a positive stereotype, people in that group actually end up doing worse because of the added pressure of upholding (rather than breaking down) the stereotype. In particular, I wonder if, for a math test administered to white and Asian people in which test takers are individually given information about the test and then left to take the test separately (so that test takers don't know who else is taking the test, as knowledge of the other test takers may by itself invoke stereotype threats in different groups), a statement that "Asian test takers are known to do better than average on this test" would cause anxiety that would lead to worsened performance among those Asian test takers compared to those in a control group (with no statement one way or the other).
The author of the book describes an experiment determining the effects of stereotype threats on black and white high school students in Los Angeles, CA, specifically trying to break down the effects of stereotype threats on high- and low-performing students in each racial group; they found that stereotype threats have significant performance suppression effects for high-performing black students (and their elimination can eliminate the performance gap for that group in turn), while having no effect on high-performing white students or low-performing students of either racial group. The author concludes that low-performing black students may already be alienated from school and career pursuits by this point, which is why they don't care whether stereotype threats are present or countered. However, this leads me to the following question: assuming that the proportion of black students in the low-performing group is higher than those in the high-performing group (which may not be a valid assumption), could the alienation from scholarly pursuits seen in low-performing black students in high school be the result of stereotype threats discouraging more black students earlier in the educational system (as the stereotype threats accumulate and chronically persist, these students internalize them enough to depress performance over the long term, so only a smaller proportion of students can make it through high school remaining interested in academic activities)?
The author then describes the transient notion of race, giving the examples of Anatole Broyard, a black man who (thanks to his light skin color) managed to pass as white for essentially his entire adult life (and was able to access all of the opportunities that white people had that were denied to black people of his time), as well as several prominent black American artists (musicians, writers, and so on) who moved to Paris and found that the stereotypes that hung over them in the US had vanished in France, so their American blackness (or black Americanness) meant something completely different (and generally much more benign) in France than in the US. Yet it seems like most of the examples of stereotype threats given in the book are for a historically marginalized group in a particular field in relation to a historically normalized/dominant group in that same field (such as blacks versus whites in scholarly pursuits, whites versus blacks in athletic pursuits, women versus men in math, et cetera). How then might things look when considering two groups interacting in which both groups are marginalized in a broader sense? As a particular example, consider a working-class Latino family that moves into a predominantly black working-class neighborhood. Are the same stereotypes that hang over Latinos in white American culture still going to hang over this family in the black neighborhood? Will the Latino family try to assimilate more with their black neighbors, even though black culture is not dominant in broader American society? What sorts of stereotype threat situations might be observed here?
The author further describes an experiment testing the math performance of Asian women college students, looking at what happens when stereotypes of "women being bad at math" or "Asians being good at math" are raised (in comparison to no stereotypes being raised); the experimenters found a significant positive effect for the latter stereotype and a significant negative effect for the former stereotype. I'd be interested though to see what happens when both are combined: this gets at what many activists call "intersectionality", which is that people can see (sometimes simultaneously) different aspects of privilege and marginalization from different facets of their identity. Would being reminded of being both Asian and female have a net neutral effect when averaged over a large group, because different Asian women may value their identities as female versus Asian to different degrees?
This is further extended to young Asian girls given age-appropriate math tests, and being made to color a picture before the test; the picture could be a girl holding a doll, a landscape, or a person eating rice with chopsticks. The above results were qualitatively replicated even for these young girls. That said, I wonder if there are other issues (brought up in the book) that aren't being captured for these girls and women in these tests, especially regarding the context-dependent nature of stereotype threats. If many of these girls and women are coming from cultures where women are (perhaps stereotypically) expected to be just as good in STEM fields as men are, is the decrease in performance when stereotype threats related to femaleness are raised going to be of smaller magnitude compared to that observed for white girls and women given the same corresponding tests? Additionally, if these girls and women are coming from more communitarian cultures where issues of honor and shame for the family are (perhaps stereotypically) more emphasized, could those factors play a role in the performance of these test takers, or is that going to be too hard to measure? Moreover, if stereotype threats are context-dependent, would Asian American girls and women taking these tests respond differently when taking them in the presence of white American girls and women versus Asian girls and women (who were born and raised in the corresponding Asian countries), with regard to issues of assimilation into American culture versus retention of Asian culture and values? (I'm intentionally and sloppily using "Asian" as a catch-all here.)
Overall, the tone of the book is an optimistic one, and I am likewise optimistic that it could lead to more interesting research and developments with regard to bringing marginalized groups in any field out of those margins. Already, I have seen programs like the Meyerhoff Program at UMBC (and similar programs, at other universities, modeled after it) counter the detrimental study habits of high-achieving black and other minority students that arise due to stereotype threats. I have also recently seen articles that further develop the ideas of having minority college students read about upperclassmen in those universities discuss their struggles and eventual academic and social successes in order to boost the sense of belonging and reduce the feelings of alienation and impostor syndrome in these students. Additionally, several months ago, I went to a talk in Princeton University about further experiments discussing how achievement gaps can be further diminished between normalized and marginalized groups when the idea of a person's knowledge being expandable through a process of learning, rather than being rigid and based on innate ability, is emphasized. Clearly, even an outside observer like myself can see that these ideas are taking hold and are being further developed, so I'm hopeful to see not just more research but also more implementation of this in the near future.
