CHAPTER 9

It’s About Me, Not My Group

Closing the Achievement Gap

Muriel Wiggins was excited when her third-grade teacher, Miss Gill, asked the students to write a poem as a homework assignment. Muriel knew a lot of poems because her father, who was college-educated, often asked Muriel and her siblings to memorize verses and recite them at the dinner table. (When Muriel was telling me this story, nearly seventy years after it happened, she recited all ten stanzas of Edward Rowland Sill’s “The Fool’s Prayer,” a poem she had memorized for her father.) She was thus excited to try her hand at writing verses of her own, and decided to try to capture some of the religious sentiments that she, the granddaughter of a Presbyterian minister, often felt. Here is what she wrote:

I often dream of the star of love

The star that shines so far above

The star that came so long ago

That men of this earth His love might know.

Not bad for a third grader! Muriel handed in the poem the next day, proud of her work. One can only imagine the shock and humiliation she felt when, a few days later, Miss Gill reprimanded her in front of the entire class. The teacher held up Muriel’s poem with two fingers, as though it were a piece of paper she had taken out of the garbage, and proceeded to tear it to shreds. “Muriel must have copied this poem because there is no way she could have written it,” she said sternly, and as punishment made Muriel stand in the cloakroom for the rest of the morning.

Muriel, it so happens, was one of five African American students in a class of thirty-five kids. Miss Gill was an older white woman who had little time or affection for those five students. One time, Muriel and a black classmate ran into Miss Gill on the street after school. “Hello, Miss Gill,” they said timidly.

“Don’t ever speak to me outside of the classroom,” the teacher snapped.

Most of the other teachers in the school (all of whom were white) turned an equally cold shoulder to the African American students. Muriel’s fourth-grade teacher, Miss Rosenberger, held court on the playground for the white girls, who would cling to her dress and vie for her attention. A lucky few were allowed to hold Miss Rosenberger’s hand. No black girls were allowed anywhere near this inner circle.

Fortunately, such blatant acts of racism, which occurred decades ago, are less frequent in today’s schools. But they haven’t disappeared. Consider what happened to a very bright Hispanic girl whom I will call Maria, the daughter of migrant workers who had recently settled in the city where I live. Maria excelled in elementary school and by the sixth grade was in the most advanced reading group. She was an enthusiastic member of both the school band and orchestra (she played the clarinet and the violin), which was highly unusual—she was one of only a handful of students to play in both musical ensembles. But when she moved up to middle school, she was inexplicably placed in a lowtrack English class and told that she had to choose between the band and the orchestra because she was not academically advanced enough to handle both. There is no way to know for sure, but it certainly seems like her ethnicity and social class led to this oversight. Busy school administrators probably just assumed that she did not speak English very well and could not possibly do well in middle school if she played in both the band and the orchestra. Maria protested, but to no avail. She asked her parents to talk to the principal, but they spoke limited English and did not know how to help.

This story has a happy ending, but only because of a random set of circumstances. Maria was taking violin lessons as part of a scholarship program funded by a local musician, and she mentioned to her violin teacher that she was no longer in the orchestra. The teacher happened to know my wife, who at the time was a member of the school board. My wife made a few phone calls, and Maria was soon given a reading test and found to qualify for the most advanced English class—and to be eligible to rejoin the orchestra and band. Muriel Wiggins, too, was able to overcome her teachers’ racism and succeed in school, eventually graduating from college near the top of her class.

But many, many other students of color are not so fortunate and fall behind academically. Muriel reports that most of the other black kids in her school were unable to overcome the racism of the teachers. She remembers the names of the other four black kids in her class and has fond memories of playing with them on the playground and after school, but she has no memories of them in the classroom. The teachers did their best to make the black kids invisible, and to a large degree they succeeded. Even today, Hispanic and black students are more likely to drop out of school and are less likely to go to college. And, when test day arrives and standardized tests are given, Hispanics and blacks do worse, on average, than whites. Though the size of this achievement gap dropped dramatically in the 1980s, it has remained relatively stable since the 1990s and in some cases has even widened. On a 2007 reading test, for example, white eighth graders outperformed black eighth graders by an average of 27 points and Hispanic eighth graders by an average of 25 points (on a scale of 0–500). In math, white eighth graders outperformed blacks by 32 points and Hispanics by 26 points. These differences might not seem that large, but by the age of seventeen, a black student at the fiftieth percentile is performing only as well as a white student at the twentieth percentile.1

Why has the achievement gap persisted so stubbornly? Is the kind of prejudice Muriel and Maria encountered responsible, or are other factors at play? Many causes have been proposed, including higher poverty rates among blacks and Hispanics; the quality of the education that blacks and Hispanics are receiving; social and cultural factors, such as different parenting styles; biases in the tests; and, most controversial of all, differences in native intelligence between racial groups. In this chapter, we will consider the evidence for each of these factors and talk about ways of reducing the achievement gap, including a promising new technique based on the story-editing approach.

DEBUNKING CLAIMS ABOUT GENETIC DIFFERENCES

We can dismiss one claim about the achievement gap outright: there is no credible evidence for innate differences in intelligence between racial groups. I will not summarize all the evidence against the hereditarian view, because Richard Nisbett has done so cogently in his book Intelligence and How to Get It. Here are a couple of the most telling findings: first, if there were genetic differences in intelligence between people of African and European descent, then the more ancestors a person has from Africa, the lower his or her IQ should be, on average. Several studies have tested this hypothesis by, for example, measuring people’s racial heritage with blood tests and correlating that with their IQs. Overwhelmingly, these studies show no relationship between racial ancestry and intelligence. Second, the gap between blacks and whites on IQ tests and standardized tests narrowed substantially in the fifteen-year period from 1975 to 1990, which is much too rapid a change to be accounted for by changes in the gene pool. And finally, there is ample evidence that environmental factors are completely responsible for the differences between the races in IQ and standardized tests.2

What are these environmental factors? The conditions mentioned above probably contribute to the gap, namely, the lingering effects of racism, the quality of schooling that minorities are likely to receive, and cultural differences. As we will see, some of the efforts to address these problems have paid off. But one environmental cause of the achievement gap is not so obvious, and in fact has come to light only recently through research on story editing in social psychology. Fortunately, attempts to rectify it with small story-editing interventions have reaped big dividends. But let’s first consider some of the most commonly mentioned causes of the achievement gap and the interventions designed to address them.

LOW EXPECTATIONS

As we saw in the previous chapter, racism has far from disappeared from contemporary American life. Remember the incident at the swim club, where white suburban parents objected to black and Hispanic kids using the pool? It happened in 2009 in suburban Philadelphia, a mere thirteen miles from the elementary school that Muriel Wiggins attended seventy years ago. Fortunately, blatant acts of discrimination in the schools, such as Miss Gill’s treatment of Muriel, have become less common. But even so, teachers’ low expectations of minority students can hurt in more insidious ways, as illustrated in one of the most famous studies in social psychology.

In the 1960s, Robert Rosenthal and Lenore Jacobson administered a test to all the students in an elementary school and gave the results to the teachers. They told the teachers that, based on the results of the test, some students were particularly likely to bloom academically in the upcoming year, whereas others were not. The catch was that, unbeknownst to the teachers, the “bloomers” were chosen by drawing names out of a hat, not by how well they actually did on the test. In fact, the test was bogus and didn’t really measure anything. Instead, the researchers were interested in what would happen if they instilled expectations in the minds of the teachers and told them that some of their students were especially gifted. They sat back and waited, and by the end of the year the results were clear: the “bloomers” had scored significantly higher on an actual IQ test than the nonbloomers did!

The only way this could have happened is via a self-fulfilling prophecy in the minds of the teachers. The students themselves didn’t know that they had been designated as bloomers, and neither did their parents. Only the teachers knew, and their expectations caused them to act in ways that actually made the kids smarter. But how, exactly, did this happen? Perhaps the teachers made callous, conscious decisions to treat the bloomers more favorably than their less promising classmates. But this was not the case; the teachers did their best to treat everyone equally. Instead, despite their best efforts, the teachers acted more warmly toward the bloomers, challenged them more with difficult material, and helped them with their answers when they were wrong. In short, they were better teachers toward the bloomers, without knowing it.

In real life, of course, psychologists do not give teachers false expectations about how well their students will do. But teachers are only human. On the first day of class, they know their students’ race, gender, and social class, and maybe something about their family history (for example, whether they have an older sibling who did well or poorly in school). Any one of these factors could instill expectations in the minds of the teachers and lead to self-fulfilling prophecies, just as in the Rosenthal and Jacobson study.

To the extent that teachers develop low expectations about black children, then, we have identified a contributor to the achievement gap—teachers treat black children in ways that make the teachers’ expectations come true. But is there evidence that this actually happens in real classrooms? Researchers have looked for such evidence and concluded that it does exist. One study found that teachers’ negative expectations are especially likely to influence students who are black or from poor socioeconomic backgrounds.3

Further, in everyday life, teachers’ expectations can easily be transferred to the minds of their students. Muriel Wiggins reports that after a while she gave up trying very hard in school. What’s the use, she thought, when Miss Gill won’t even believe I’m capable of writing a four-line poem? Nor did it escape Muriel’s notice that the teachers rarely called on her or the other black children, or let them perform the coveted classroom duties that are signs of status at that age, such as hall monitor. Who wouldn’t give up under such circumstances?

POVERTY AND SOCIAL CLASS

In 2009, 12.3 percent of whites in the United States fell below the poverty line, according to the U.S. Census Bureau. That might not seem like a large number, but poverty levels are defined very conservatively by the United States government. For a family of two people with no children, for example, only those earning a combined income of less than $14,366 were classified as poor. For a single parent with one child, the cutoff was $14,787. Thus there are a large number of families who are struggling but are not officially designated as poor. These figures are all the starker when we consider black and Hispanic families. Even with the conservative cutoffs, 25.8 percent of blacks and 25.3 percent of Hispanics live below the poverty line.4

The fact that black and Hispanic children are more likely to live in poverty than white children contributes to the achievement gap in multiple ways. First, children living in poverty are more likely to experience biological deficits that influence intelligence, including prenatal exposure to drugs and alcohol, low birth weight, little or no breast-feeding, exposure to lead, and malnutrition. Second, poor children have less access to good schools, particularly those with experienced teachers and small class sizes. And they obviously have fewer of the educational advantages that money can buy, such as books, summer enrichment camps, private tutors, and SAT preparation courses. Children living in poverty are also more likely to move frequently, which can mean a disruptive change of schools in the middle of the year.

These disadvantages could be addressed by raising the income of people living in poverty. Though this is easier said than done, surely we can do better than we’re doing now, given that the poverty level in the United States is higher than in many other developed nations. But there is another set of issues associated with poverty and social class that is more controversial, namely, differences in child-rearing practices between lower-class and middle-class parents. As mentioned in chapter 4, middle-class parents talk more to their children, read to them more, and are more likely to encourage them to question the world and analyze why it works the way it does, all of which likely contribute to the higher academic achievement of middle-class children.5

One reason that Muriel Wiggins was able to overcome the racism of her teachers and graduate from college is that she did not have these disadvantages. She came from a supportive family that expected her to do well in school. Her father was college-educated, and two of her uncles were doctors. There were lots of books in her house. The other black kids in her class, she points out, didn’t have these advantages and didn’t do nearly as well. One way to reduce the achievement gap, then, would be to provide enriched experiences for low-income children in preschool and beyond, essentially giving them the advantages they are not able to get at home. There have been many attempts to do this, as we will see shortly.

BIG INTERVENTIONS TO REDUCE THE ACHIEVEMENT GAP

The picture we’ve seen so far is bleak. A host of societal and cultural factors contribute to the achievement gap, some with deep roots. It seems like it will take major interventions to make up for the many disadvantages faced by students of color, and, indeed, this is the approach most experts and policy makers have taken. Some argue that school environments and high-quality teachers are the key and that we must throw everything we have into changing our educational system to eliminate the achievement gap. Others suggest that the gap is far too big a problem for schools to solve alone, and that we must focus on issues of poverty and class and culture outside of the classroom. We’ll get to those approaches, but first let’s evaluate a simpler one: giving disadvantaged students more of an incentive to do well in school.

Paying Kids

According to some experts, disadvantaged kids have little incentive for doing well in school. They have few role models showing them that education is a way out of poverty, and, in the case of students of color, academic success might reduce their popularity with their peers. Perhaps we can get kids to overcome these disincentives by giving them financial rewards. Or maybe it would work to penalize families (such as those receiving welfare) if their kids fail to meet educational goals.

There have been a number of attempts to use incentives and punishments, in the United States and elsewhere, and the results are mixed. Consider, for example, an ambitious project by the economist Roland Fryer, who helped initiate incentive programs in four American public school systems. In two, New York and Chicago, kids were rewarded for doing well on standardized tests or in their courses, that is, for their academic “outputs.” In the other two, Dallas and Washington, D.C., kids were rewarded for behaviors that lead to academic success, that is, for their academic “inputs.” In Dallas, for example, second graders received two dollars for every book they read. In Washington, D.C., students were rewarded for such inputs as attending school, good conduct in the classroom, and turning in homework. All told, more than 38,000 students in 261 schools were paid a total of $6.3 million for these inputs or outputs.

To Fryer’s credit, he did not just implement these incentive programs, he tested them experimentally. In each city, some schools were randomly assigned to participate in the program, whereas others were randomly assigned to a control group that did not participate. By comparing students in the two sets of schools, Fryer assessed the causal impact of the incentive programs.

The results of the programs were, for the most part, disappointing. In fact, paying kids had a statistically significant positive effect in only one city, Dallas, where kids were paid to read books. Second graders who took part in that program achieved significantly higher scores on a reading comprehension test and received better grades than did second graders in the control schools. But in the other three cities, paying kids had little or no effect on test scores or grades.

There are many reasons why paying kids did not work as well as expected. As we saw in chapter 4, there is a danger to providing kids with strong incentives: doing so can convince them that they are “doing it for the money” and undermine any intrinsic interest in school they had at the outset. On the other hand, if kids have no interest in an activity to begin with, providing them with rewards can’t hurt. Unfortunately, there is no way of knowing what kids’ initial level of interest in school was when they started the incentive programs, so we can’t tell whether the incentives ultimately helped some kids but hurt others. The bottom line is that giving kids monetary incentives does not, as of yet, appear to be a very effective way of closing the achievement gap.6

Early Childhood Interventions

Given that some of the causes of the achievement gap affect children very early in life, perhaps the best approach is to intervene as early as possible. A number of interventions have attempted to do just that by offering free preschool and day care programs that include educational and enrichment activities. Head Start, the best-known program, began in 1965 and targets disadvantaged children ages three through five. In 1995, it was extended to children from birth to three years old. Both versions of Head Start provide a wide range of educational and health services to its clients. Both are expensive, costing billions of dollars each year.

Do they work? Perhaps the most shocking answer to this question is that we are only now beginning to find out. When Congress reauthorized the Head Start program in 1998, it did so with the proviso that the U.S. Department of Health and Human Services determine its impact—thirty-three years and billions of dollars after the program was implemented. The result was the Head Start Impact Study, which began in 2002. A sample of three-year-olds, who had applied to Head Start programs at 383 centers in twenty-three states, were randomly accepted into Head Start or assigned to a control group, and their progress was monitored carefully until the end of first grade.

For at least two reasons, the effects of Head Start would have to be very strong to be detected by this experiment. First, more than 40 percent of the children in the control group ended up in another preschool program, which means that the study wasn’t testing whether Head Start kids did better than kids who didn’t participate in any preschool program, but whether they did better than a heterogeneous control group that included kids who participated in other preschool programs (in fact, 18 percent of the kids in the control group managed to get into Head Start centers that weren’t part of the study). Second, not all the kids assigned to Head Start stayed in the program when they turned four, and kids in the control condition were allowed to enroll in Head Start at age four if their parents chose to do so. The net result of this was that in year two of the program, only 63 percent of the three-year-old cohort remained in Head Start, whereas 50 percent of the kids in the control group entered Head Start at that point. The study was thus a test of whether participating in Head Start at age three has any benefits, compared to a control group of kids who (for the most part) did not participate in Head Start at age three, but many of whom did participate in other preschool programs and began Head Start at age four.

It turns out that beginning Head Start at age three did yield positive results. After the first year of the study, the Head Start children showed significant gains (relative to children in the control group) on an overall measure of academic proficiency that included tests of early writing and math skills. In addition, they exhibited fewer behavior problems than did kids in the control group, and their parents showed positive changes as well—they were more likely to read to their kids and less likely to spank them. Finally, the Head Start kids received higher scores on a measure of overall physical health and were more likely to have received dental care.

Did these positive effects persist? Some did: by the end of the first grade, there was evidence of a small advantage in oral comprehension skills in the kids who began Head Start at age three, plus these kids had better relationships with their teachers than the control kids did. Further, some of the positive parenting styles persisted: the parents of the Head Start kids reported that they had less of an authoritarian parenting style. But, perhaps not surprisingly, many of the positive effects did not persist beyond the first year of the program. Recall that, after year one, 50 percent of the children in the control group enrolled in Head Start, which may have brought some of them up to the level of the kids who began at age three. In fact, there was another group of kids who were randomly assigned to begin Head Start at age four, and they, too, showed increases in several measures of academic proficiency by the end of that year.

Other researchers have tried to estimate the long-term impact of Head Start using nonexperimental designs and have found encouraging results. For example, one study compared children who participated in the program with siblings who did not, and another compared children who lived in counties that had Head Start (because they met poverty guidelines for the program) with children who lived in adjacent counties that did not (because they just missed the poverty-level cutoffs). Both studies found that the Head Start program had long-term benefits. In one, children who participated were more likely to graduate from high school and attend college and less likely to commit crimes and became teenage parents. Though these results are encouraging, we need to be cautious about them, given the built-in problems with the research designs (e.g., the children living in neighboring counties might have differed in lots of ways besides their participation in Head Start).7

We have a better idea about the long-term effects of some other childhood interventions because they been tested with rigorous experimental designs and the children have been followed for several years. One of the best-known programs is the Carolina Abecedarian Project, founded by Craig Ramey and his colleagues. Disadvantaged African American children took part in an intensive preschool program, beginning when they were about four months of age and continuing until they started school. The program had just about everything one would want—a very low teacher-to-student ratio and a multitude of activities that targeted the children’s educational, social, emotional, and cognitive development. There was a particular focus on language skills. The program took up the whole school day, five days a week, twelve months a year. And this was one of the few childhood interventions that included a randomly assigned control group of children who did not take part.

To see what effects the intervention had, the researchers followed the children in both groups for several years. At twenty-one years of age, those who had attended the preschool program, compared to those who had not, had higher IQ scores; were less likely to have been assigned to special education classes or to have repeated a grade; scored higher on tests of reading, math, and cognitive ability; were more likely to have graduated from high school and attended college; and were less likely to have become teenage parents. The magnitude of these effects was impressive—for example, at age twenty-one, the percentage of kids who were attending a four-year college was nearly three times higher in the group who had gone to the preschool (35 percent) than in the control group (13 percent).

Other early childhood intervention programs have also had positive effects, such as the Perry Preschool Project, conducted in Ypsilanti, Michigan, in the 1960s. Thus early interventions can work; and by “early,” I mean soon after birth. But they must be intensive and of very high quality. And though they have lasting benefits, they do not completely eliminate the achievement gap.8

Redesigning Schools

A problem with early childhood interventions is that they are, well, early, and stop by the time children reach school. Although these programs can have positive long-term effects, the fact is that many children continue to live in poverty and attend poor schools. Maybe we need to develop programs for kids in elementary school and beyond.

Virtually every school district in the nation has given this a try, implementing myriad programs that target reading, math, and other subjects. Undoubtedly, some of these programs are beneficial. My beef is that we don’t really know, because the programs are seldom tested sufficiently using experimental designs. Things are improving in this regard: in 2002, the United States Department of Education launched a website called the What Works Clearinghouse (WWC), which presents a set of standards for evaluating educational interventions and reviews studies that meet those standards (http://ies.ed.gov/ncee/wwc/). This is a big step, because, if nothing else, it will call attention to the need for more rigorous research.

Other educators have decided that individual programs are not enough and that we need to restructure schools from top to bottom. Hundreds of charter schools have opened throughout the country, for example, with the claim that they do a better job of educating disadvantaged children than public schools do. Again, rigorous experimental studies evaluating these schools are hard to come by, and what evidence there is suggests that most charter schools do no better than public schools at closing the achievement gap.9

One exception may be the Knowledge Is Power Program (KIPP), founded in 1994 by two teachers, Michael Feinberg and David Levin, who were fresh out of the Teach For America program and had innovative ideas about how to structure a school in order to help disadvantaged kids. Today there are more than sixty KIPP schools in nineteen states and the District of Columbia. The students they serve are largely from low-income families, and more than 90 percent are African American or Hispanic. One key feature of KIPP schools is that students receive about 60 percent more instructional time than do students in public schools. At a typical KIPP school, students arrive at 7:30 a.m. and stay until 5:00 p.m., attend half-day Saturday sessions twice a month, and attend a three-week summer session. KIPP schools also attempt to create an environment of high expectations, commitment, and discipline.

Although an experimental test of the effectiveness of KIPP schools is under way, for now we need to rely on nonexperimental tests that have compared KIPP students to similar, non-KIPP students. These studies have yielded promising results: KIPP students scored better on standardized tests of reading and math than did the non-Kipp students. But for now we cannot be certain that this difference is due to the schools or whether it exists because the children who attend KIPP schools are different in some way to begin with (for example, they may have parents who are especially motivated to see their children do well).10

We can be more confident about the success of the charter schools called Promise Academies, which are operated by the Harlem Children’s Zone (HCZ). HCZ, which began in the 1970s, is an ambitious organization that attempts to mitigate the effects of poverty in the Harlem section of New York City. It targets a ninety-seven-block area of Harlem, where it has developed more than twenty programs and charter schools. As with the KIPP program, there is no experimental evaluation built into the program. But HCZ was evaluated by clever social scientists who took advantage of a key fact: by law in New York State, if a charter school receives more applications from students than it can accept, it must choose students randomly. This essentially creates an experimental design whereby the researchers can compare the academic achievement of the students who won the lottery (and thus attended the charter schools) with those who lost (and thus did not). The results? By the eighth grade, those who won and enrolled in the Promise Academies for at least two years were doing better in math and reading—by large margins—than students who lost the lottery. In fact, eighth-grade Promise Academy students, who are all minorities (about 90 percent black and 10 percent Hispanic), outperformed white students in the rest of the New York City public school system in math to a statistically significant degree. Plus, there is evidence that the schools, and not the other HCZ community programs, are responsible for these impressive gains, because kids who attended the schools, but did not take advantage of the other programs, show the same academic gains as those who did take part in the programs.11

It thus seems that massive attempts to redesign schools can work to close the achievement gap. We don’t know for sure why the KIPP and HCZ schools have been so successful, though one key factor may be the sheer volume of instruction that the students receive. Like KIPP students, those who attend HCZ schools have an extended school day and year, and students who need additional instruction attend classes on Saturdays. Another likely factor is that the schools attract dedicated and talented teachers, and only the best of these are retained.

The impact of a good teacher cannot be overestimated. As mentioned, Muriel Wiggins had pretty much given up in elementary school after her horrific experiences with racist teachers, but then she was assigned to Miss Simons for the sixth grade. Miss Simons saw something in Muriel that the other teachers had not, and treated her with kindness. She gave Muriel coveted duties, such as cloakroom monitor (whose job it was to open the cloakroom door when it was time for the students to get their coats), and small gifts, such as a box of leftover crayons. She was also tough on Muriel when she didn’t live up to her potential. One day, Muriel realized that she had forgotten to do her homework, which was to read a newspaper article and summarize it. When she went home for lunch, she grabbed the newspaper and scribbled a few words on a piece of paper. But wouldn’t you know it—that afternoon, Muriel was the first person Miss Simons called on to tell the class about the article she had read. All Muriel could remember was the headline, and after reciting it she stood there in mortified silence. As punishment, Muriel lost her cloakroom-monitor job. With both encouragement and discipline, Miss Simons got Muriel back on track academically.

STEREOTYPE THREAT AND ACADEMIC ACHIEVEMENT

We’ve just seen that early childhood interventions and attempts to redesign schools can help reduce the achievement gap. But what about the story-editing approach, in which we try to redirect people’s narratives about why they are having difficulties? Consider first the results of a study conducted by researchers at the University of Oklahoma. African American and white college students were given a standard intelligence test called the Raven’s Advanced Progressive Matrices test (Raven’s APM), on which people see a series of geometric figures and have to guess which figure goes next in the sequence. The only thing the researchers varied was the instructions they gave to participants. Some were randomly assigned to receive the instructions in the manual for the test, which uses the word “test” in several places and describes it as a “measure of observation and clear thinking.” Another group was randomly assigned to receive instructions that described the Raven’s APM as an IQ test. “Like the SAT and ACT,” these participants learned, “this test is frequently used to measure individuals’ intelligence and ability.” A final group was randomly assigned to receive instructions that said, “The task you will be working on is a series of puzzles.” The word “puzzles” was used repeatedly, and neither the word “test” nor “intelligence” was ever used. After receiving one of these three sets of instructions, the participants spent forty minutes working on the test.

Now, keep in mind that all students took the exact same test—the only thing that varied was which set of instructions they received. It might seem like this would not influence people’s performance; after all, the Raven’s APM is a widely used intelligence test that has been developed and validated for decades as a “pure” measure of reasoning ability. Because it involves only geometric figures, with no verbal content, it supposedly tests basic reasoning abilities devoid of any cultural bias. Surely it doesn’t matter whether the test is described as a measure of IQ or a set of puzzles. Well, it turns out it does matter, and it matters a lot, at least to one group of people: African Americans. As seen in the figure opposite, when the test was described as a measure of IQ, the standard achievement gap was evident: whites outperformed blacks, to about the same extent as is typically found on other measures of intelligence and achievement. When participants received the standard instructions, which described the test as a measure of observation and clear thinking, the achievement gap was still present, but to a lesser degree. But look what happened when the test was described as a set of puzzles: as seen in the two bars on the far right of the figure, the achievement gap completely disappeared! In fact, in this condition, African American students did slightly better than white students. What was going on here? Clearly, something about the way in which the students interpreted the test made a big difference in their performance—something that isn’t supposed to happen on tests of intelligence or achievement.12

image

What was going on is an important phenomenon called stereotype threat, discovered by the social psychologist Claude Steele (and detailed in his riveting book, Whistling Vivaldi). Stereotype threat is a psychological state that people experience when they feel they are at risk of confirming a negative stereotype about a group to which they belong. African American students, for example, are well aware of the stereotype that blacks are less intelligent than whites. When they are given a test and told it measures their IQ, they thus might worry that they will confirm the stereotype—a concern that a white student does not have when taking the same test. This worry about confirming a negative stereotype uses up mental resources and triggers anxiety, which makes it harder to concentrate on the test. We know this from dozens of studies that have given the same tests to different groups of people with slightly different instructions about what the test measures. As seen in the study just described, when blacks are told that the test measures intelligence, they do worse than whites. But when they are not told that the test is a measure of intelligence, they do just as well as whites on the very same test.

You may have noticed in the figure that whites did best when the Raven’s APM was described as an IQ test and worse when it was described a set of puzzles. Although these differences are not statistically significant, they reflect a trend that has been found in other studies—people’s performance improves when there is a stereotype that their group does better on that task than other groups. Because of the stereotype that whites are more intelligent than blacks, whites did better when told that they were taking an IQ test than when they were told they were doing puzzles—a phenomenon that is called stereotype lift. Our focus here, however, will be on the conditions under which people do worse because they are at risk of confirming a negative stereotype about their group.13

Stereotype threat is not just experienced by minorities. It can make anyone do worse on a task, as long as doing so would confirm a negative stereotype about his or her group. Women do worse than men on math tests when gender differences in math are emphasized, for example, but do just as well when gender differences are downplayed or said not to apply on the specific math test they are taking. Women who were told that a simulated driving test was designed to investigate why men are better drivers than women were more than twice as likely to run over a pedestrian (with the simulated car) than women who took the same test with no mention of gender differences.14

Stereotype threat wields its insidious effects by triggering a vicious cycle of thinking, similar to the one we discussed in chapter 1. People start thinking about the stereotype and whether they will confirm it, which makes them anxious and aroused. This leads to further rumination and self-doubt, as people monitor their performance (e.g., “Oh no, I’m having trouble with this math problem; maybe the stereotype is true!”). Next, people use mental energy to try to suppress these disrupting thoughts. The problem is that all these steps—becoming anxious and aroused, monitoring one’s performance for signs of stereotype confirmation, and attempting to suppress thoughts about the stereotype—use up mental resources, leaving less room to concentrate on the test. And the harder it is to concentrate, the worse people do, triggering the whole process over again. With all this stress, worry, rumination, and thought suppression going on, it is no wonder that people under stereotype threat do worse than people free to concentrate on the test without anything else to worry about.

One remarkable thing about these deficits in performance is how easily they are corrected. A simple reinterpretation of the meaning of a test can eliminate the achievement gap, as seen in the examples above. So can attempts to reduce the salience of the negative stereotype—by, for example, emphasizing positive aspects of one’s group or introducing people to a positive role model from the stereotyped group (for example, a female math whiz). Teaching people about the nature of stereotype threat can also help. All these approaches can succeed in bumping people out of the vicious cycle of thinking that stereotype threat causes, allowing them to concentrate better on the task at hand.15

This could be taken to mean that the societal and cultural variables we discussed earlier—racism, poverty, child-rearing practices, and schooling—are not as important as we thought in creating the achievement gap. If we can eliminate the gap by removing stereotype threat from testing situations, maybe we’ve been barking up the wrong tree. According to this argument, the achievement gap is due largely to an uneven playing field in testing situations and not to hard-to-change societal or educational factors. Look again, for example, at the figure that showed the results of the Raven’s Advanced Progressive Matrices test: when stereotype threat was removed from the testing situation, the achievement gap was completely eliminated. Black college students did just as well as white college students when the playing field was evened, psychologically speaking.

It would be a mistake, however, to conclude that the entire achievement gap is due to testing bias and that we no longer need to worry about poverty and racism. Yes, some of the difference is likely due to threats raised by the tests themselves. But we should keep in mind that most of the studies on stereotype threat have been conducted with college students, who are obviously a highly selective sample. They have already overcome numerous hurdles to get where they are, particularly if they are poor and black. Many other disadvantaged students don’t make it over those hurdles. Think of Muriel Wiggins’s black classmates, who didn’t have the family support she had. Or the Hispanic students unfairly placed in lower reading groups who never convince school officials to retest them. In the year 2000, only 14 percent of Hispanics and 27 percent of blacks attended college, compared to 38 percent of whites. We must continue to look for ways of removing barriers for minority groups, including implementing the kinds of big interventions we discussed earlier that have shown signs of success.16

It is also important to note that stereotype threat can act as a barrier to early academic success, in addition to influencing how college students react in testing situations. The fear of confirming a negative stereotype can have insidious effects, not only while taking a test but in everyday life from childhood onward. The costs of such spiraling anxiety can be high, both psychologically and physically. Encountering stereotype threat, for example, has been shown to raise blood pressure among African Americans. Many of us, when constantly faced with this kind of threat, would bail out of the whole pursuit. After all, one of the best ways of coping with something that is hard to get is to devalue it, like the proverbial fox and the grapes—the grapes must be sour if we can’t reach them, and school must not be very important if it is such a source of stress. Unfortunately, minorities are more likely to bail out of school than whites. In the year 2008, for example, the high school dropout rate for African Americans was double what it was for whites (9.9 percent versus 4.8 percent). And for Hispanics, it was nearly double what it was for African Americans (18.3 percent). Minority kids who stay in school generally value academics as much as nonminority kids, but the stress of stereotype threat takes its toll, lowering their performance. As Claude Steele puts it, “Disproving a stereotype is a Sisyphean task; something you have to do over and over again as long as you are in the domain where the stereotype applies.”17

But what if we catch kids at a point where they are particularly vulnerable to stereotype threat and the downward spiral to which it can lead? Maybe a dose of story editing could redirect people’s thinking in a way that reduces stereotype threat, thereby preventing the self-defeating cycle of thinking. There is reason to believe that a good time to try this is when kids start middle school. Most kids switch schools at the beginning of sixth or seventh grade and are faced with new academic pressures, and it is at this point that grades often drop and behavioral problems increase, especially for minority children. Could this be a crucial point at which stereotype threat leads to stress and poor performance? And, if so, could a story-editing intervention prevent this from happening?18

STORY-EDITING INTERVENTIONS TO REDUCE THE ACHIEVEMENT GAP

One of the first story-editing interventions designed to reduce the achievement gap targeted seventh graders at a rural school in Texas. About two-thirds of the students were Hispanic, 20 percent were white, and 13 percent were black, and a majority of the students (about 70 percent) qualified for a free or reduced-price lunch, which is a standard measure of poverty in schools. The researchers reasoned that this population was especially vulnerable to stereotype threat and designed two story-editing interventions to prevent this from happening. For the first intervention, the students were assigned a college-student mentor who kept in touch with them regularly throughout the school year and emphasized that many students encounter academic difficulties when they begin middle school, but that they typically do better when they learn how to navigate their new environments. This intervention, based on the study with college students discussed in chapter 1, was designed to prevent a self-defeating cycle of negative thinking, in which people blame themselves for poor academic performance and give up.

For the second intervention, the students were also assigned mentors, but instead of talking to the kids about the transition to middle school, the mentors emphasized that intelligence is a skill that is learned and is not a fixed quality that people are born with. The researchers reasoned that this approach, based on Carol Dweck’s research on “mindsets,” is another way to prevent a self-defeating thinking cycle. As we saw in chapter 4, many kids believe that intelligence is a fixed quality; thus, when they encounter academic difficulties, they take that as a sign that they are not very smart. This can lead to giving up: why keep knocking your head against the wall if you don’t have what it takes? But if kids can be convinced that intelligence is a skill that develops with practice, then they might take academic difficulties as a sign that they simply need to work harder.

Other kids were randomly assigned to a control group. These kids also received mentors, but the mentors taught them about the perils of drug use and did not discuss academics. As it turned out, both of the story-editing interventions improved academic performance. At the end of the school year, students who got either of the two interventions achieved significantly higher scores on a standardized reading test than kids in the control group. On a standardized math test, girls who received the interventions did substantially better than girls in the control condition, and just as well as boys. The interventions did not boost boys’ performance on the math test, possibly because they were not experiencing stereotype threat in this domain as much as girls were. (The boys were, in fact, already performing at a higher level than the girls in math.) Another study that used a similar “mindset” intervention—teaching kids that intelligence is malleable instead of fixed—had similar success with minority seventh graders at a New York City school that primarily serves black and Hispanic students.19

Is it possible that an even simpler story-editing intervention might work just as well? Geoffrey Cohen, Julio Garcia, Valerie Purdie-Vaughns, and their colleagues decided to find out. They, too, targeted seventh graders—in this case, at a middle school in New England that was about evenly divided between African American and white students. The intervention was striking in how minimal it was, at least on the surface. At the beginning of the academic year, students completed a fifteen-minute writing exercise in one of their classes. They were given a list of things they might value, such as relationships with friends and family, religion, music, and politics, and asked to pick the two or three that were most important to them. Then they wrote about why these values were important to them. Depending on the version of the study, the students completed this writing exercise three to five times during the school year, usually at times of high stress, such as test days. Students randomly assigned to the control group either wrote about how a value that was unimportant to them might be important to others, or about a neutral topic, such as their morning routines.

That’s it—there were no attempts to improve the kids’ study habits or to teach them math or reading skills. The entire intervention was the writing task. Why, you might ask, did the researchers expect such a simple exercise to close the achievement gap? Because of a central principle of social psychology, namely, that people are motivated to perceive themselves as good, competent, moral people, and that when that view is threatened, they do what they can, psychologically, to repair their self-image. It is worth tracing the path that led from social psychological theory to the middle-school intervention, because it is a fascinating illustration of how theory can lead to simple yet powerful story-editing interventions—interventions that never would have occurred to people if it weren’t for the prior basic research.

Literally hundreds of experiments, conducted mostly in the laboratory with college-student participants, have documented how important it is for people to maintain a positive self-image and the lengths to which they will go to accomplish this. Typically, people try to deal with threats to their self-esteem directly. For example, being a competent research psychologist is a central part of my identity, and if something threatens that view, such as receiving negative feedback on an article I have submitted to a journal, I will do my best to remove the threat. I’ll rewrite the article and send it to another journal, say, or try to improve the research on which it is based. But what if that doesn’t work? Human beings are excellent rationalizers, and I might well make myself feel better by finding an explanation that deflects blame away from me. The journal editors don’t know what they’re talking about; the expert reviewers they consulted must have been too busy to read the article carefully; my research findings are so novel and advanced that people just can’t comprehend them (“Yeah, that’s the ticket!” as a character on Saturday Night Live used to say). I don’t do this consciously, of course—rationalization works best if it occurs behind the scenes, so that I don’t know that I’m coming up with these ideas in order to make myself feel better. It feels like I am simply telling it like it is.

But even rationalization has its limits. I can only go so far with the “misunderstood genius” defense before it crumbles in the face of reality (further journal rejections, say). What do I do now? Throw in the towel and admit that I’m a failure in a central domain in my life? Not so fast! Claude Steele had the insight, with his self-affirmation theory, that there is another way out: reminding ourselves how competent we are in some other domain in our lives. “Maybe I’m not doing well as a research psychologist,” I might think, “but I am a dedicated family man, which, after all, is much more important than some silly journal article.” Steele and his colleagues demonstrated this process of self-affirmation in several laboratory experiments. Giving people the opportunity to affirm themselves, even with a task as simple as writing about the values that are important to them, turns out to be a powerful prophylactic against threats to their self-esteem. If music is important to us, for example, then thinking about what great musicians we are can take the sting out of doing poorly at school.20

What does all this have to do with closing the achievement gap in middle school? You can probably see the connection. Imagine that you are an African American seventh grader who has experienced repeated stereotype threat throughout your school career. Unlike the white kids in your class, you have a lot to worry about when taking tests or writing papers, namely, that you will confirm a negative stereotype about your group. As we’ve seen, stereotype threat takes a serious toll, both psychologically and physically, and lowers the performance of people who experience it. But what if we remind you that there are other parts of your life that are important to you and that you value? Areas that have nothing to do with your race and are thus “identity-safe”? Cohen and colleagues’ writing exercise was designed to do just that. It allowed the students to self-affirm treasured nonacademic values, such as their love for their families or their religion. The researchers had the hunch that this would take the heat off of black students the next time they took a test or completed a difficult assignment. Academics would no longer seem like a do-or-die situation that might confirm all their worst fears about themselves as well as other people’s stereotypes about them, because, after all, there are other important things in life. And to the extent that this helped students concentrate better and do well on a test, their confidence would be boosted, making it easier for them to study for the next test.

The study was a spectacular success. African American students who completed the self-affirmation writing exercises, compared to African American students randomly assigned to the control group, achieved significantly higher grades over the next two years of school. The intervention was particularly powerful among African American students who had been doing poorly academically. Among these students, those who did the writing exercise, compared to those in the control group, achieved markedly higher grades and were significantly less likely to be placed in a remediation track or to be held back a grade. The writing exercise had no effect on white students’ grades, presumably because this group did not have excessive stress in school due to stereotype threat.

But why did writing about an important value a few times have such lasting effects? It did so, the researchers argue, in the same way that the small story prompts discussed in chapter 1 helped college students get better grades: by interrupting a self-defeating psychological cycle that causes students to spiral downward. African Americans in the control group were at particular risk for getting caught in a self-defeating cycle, in which poor grades signaled to them that they didn’t fit into the school environment and might not have what it takes to succeed. African Americans who received the intervention, however, were more likely to avoid this cycle—if they did poorly initially, the researchers found, they were less likely to conclude that they could never make it or that they were a bad fit at the school.21

Similar story-editing interventions have been found to help minority college students. Black students at a predominantly white university, for example, might be particularly prone to feel that they don’t fit in or belong at that university, especially if they experience an academic setback, as many students do in their first semester. If so, then an intervention designed to redirect their narratives from “I don’t fit in here” to “Everyone experiences bumps in the road” might increase their sense of belonging and improve their academic performance. To find out, researchers conducted a study with black and white first-year students at a predominantly white university. In the treatment condition, the students received statistics and read interviews with upper-class students indicating that most students worry that they don’t belong when they begin college, but that these worries lessen over time. To reinforce this message, the students wrote a speech illustrating how this lesson applied to them; that is, how their own worries about belonging were likely to be temporary. They delivered this speech in front a video camera, ostensibly so that it could be shown to future students at their school. Participants in the control group underwent the same procedure, except that they learned that social and political attitudes change over the course of one’s college career—they heard nothing about changes in one’s sense of belonging.

The entire session lasted only an hour. Yet, as with other story-editing interventions, it had dramatic long-term effects on the black students’ performance and well-being. Those who got the message about belonging, relative to those in the control group, believed they fit in better at college, became more engaged in college academically (by studying more, attending more review sessions, and asking more questions in class), and achieved better grades in the rest of their college careers. Not only that, but on a questionnaire they completed right before they graduated, black students who had received the “belonging” intervention reported that they were in better health, had visited a doctor fewer times, and were happier than did black students in the control group. The “belonging” message had no effect on the white students, because most of them already felt that they fit in at their university.22

The researchers who conducted these story-editing studies of academic achievement emphasize that their interventions are not meant to replace other ways of closing the achievement gap, such as providing better preschool education and redesigning schools. It would be absurd to say that getting kids to write for a short time about their values is all we need to do to help minority kids do better. The intervention obviously won’t teach a child who can’t spell to spell. Having kids affirm themselves is likely to work best in supportive environments where children have the resources they need to learn and excel. But it would be equally absurd for schools to ignore this simple, inexpensive way of helping minority kids do better academically by avoiding stereotype threat and increasing their sense of belonging.

USING IT

Research on stereotype threat is relevant to everyone, because we all belong to groups that are the target of negative stereotypes. As we have seen, blacks do worse on achievement and intelligence tests, and women do worse on math tests, when they feel they are at risk for confirming the negative stereotypes about their groups. Some groups are less stigmatized than others, such as white males, but even this privileged group can experience stereotype threat. One study, for example, found that white men did worse on a math test when they thought their results would be compared to those of Asian men, a group thought to be superior at math. And—there is no escaping it—all of us are going to get old (if we are lucky), and old people have been found to do worse on memory tests when they think the tests are examining age differences in memory, because that triggers stereotypes about the forgetfulness of the elderly.

Given that all of us are at risk for stereotype threat at one time or another, it is important to remember the ways of preventing it. Reminding oneself that performance is more about hard work than innate talent can help. So can the kind of self-affirmation task used by Geoffrey Cohen, Julio Garcia, Valerie Purdie-Vaughns, and their colleagues, in which students write about values that are important to them. As we’ve seen, reminding ourselves of our treasured values can lower the heat in anxiety-provoking situations. And simply knowing about stereotype threat can help by reminding us, “Wait a minute, this memory test is about me, not my group.”

We can also create a sense of belonging in those who are at risk of confirming a stereotype about their group by, for example, portraying women and minorities in photographs included in company newsletters. Although it might seem like rounding up minorities and women for a company photo shows an excessive concern with political correctness, it matters—because conveying the impression that the company has a critical mass of “people like me” is one way to reduce stereotype threat. To use Claude Steele’s words, anything that makes a setting feel “identity-safe” will help, because people will then not have to dwell on their group membership.

We should not, of course, just convey the impression that minorities and women are well represented in a corporate or academic setting. We should do everything we can to make sure they actually are well represented, not only to achieve basic goals of fairness and social justice, but also because the composition of a group communicates important messages to its members. A woman who is the only female in an advanced math course, or an African American who is the only black in his graduate program, is likely to feel the spotlight of attention and be on guard about how he or she is viewed. The best way to prevent this is to increase the representation of minorities and women so that the spotlight isn’t on just one or two people. Consider a study of college women who took a difficult math test along with two other students. In one condition, all three of the students were women. In another condition, two were women and one was a man. In a third condition, one was a woman and two were men. This seemingly arbitrary gender composition of the group had a dramatic effect on the women’s math performance: they got 70 percent of the answers right when in an all-female group, 64 percent when one of the other group members was male, and 58 percent when two of the other group members were male. The gender composition of the group had no effect on women’s performance on a verbal test, suggesting that the presence of men in the math groups triggered the stereotype that women are not as good at math.23

Stereotype threat is one of the most striking examples of the power of self-narratives. It is a process that occurs beneath the surface: research shows that people are not always aware that they are thinking about their group membership or that this is handicapping their performance. But with relatively simple story-editing interventions, people can be redirected out of the vicious cycle of anxiety and rumination with remarkable long-term benefits. Again, this doesn’t mean we should abandon larger structural approaches to reducing the achievement gap. Early childhood interventions are important, as are ways of redesigning schools to better serve disadvantaged students. But we should also be mindful of the messages social environments send to people about who they are and how they fit in, and seek to make these environments identity-safe.