Researcher develops tools to remedy race, gender gaps in standardized test performance
March 24, 2008
After finishing a lecture about using euphemisms to discuss embarrassing or upsetting topics—death, bodily functions, sex—Dr. Matthew McGlone, assistant professor of communication studies, was approached by an African American woman in the audience whose daughter was about to enter kindergarten.
"Do you have any suggestions on how I can talk to my daughter about African American stereotypes she might have to confront in school?" she asked.
A heartbreaking, yet relevant, question considering that social stereotyping emerges so early in children's thinking. Gender typing begins around age 2. By the age of 5, most children endorse the prevailing gender and ethnic stereotypes in their environment. By their middle elementary years, children become aware of intelligence-oriented stereotypes, such as the belief that white people are smarter than black people and boys are better at math than girls.
Dr. Matthew McGlone teaches courses on cognition, persuasion and prejudice in interpersonal communication. He is writing a book with Dr. Joshua Aronson at New York University on stereotype threat, due out in 2009.
McGlone first became interested in the influence of stereotypes on academic performance as an undergraduate statistics instructor in the 1990s. He was perplexed to observe many talented women and ethnic minority students stumble on standardized tests, despite appearing to master concepts in homework, class discussions and one-on-one interaction.
According to the College Board, which administers the Scholastic Aptitude Test (SAT), women consistently achieve lower SAT math scores than their male counterparts, while African Americans achieve lower scores than whites. Some researchers have attributed these gaps to hormonal differences between men and women or genetic differences between different ethnic groups.
McGlone was intrigued by what he'd seen with his students and felt strongly that biological factors were not responsible for the differences. He was delighted to learn that an acquaintance, Joshua Aronson, associate professor of applied psychology at New York University, was investigating the psychology of stigma. Since then, the two have collaborated in investigating ways to remediate race and gender gaps in educational achievement and standardized test performance.
Cuing Social Identity
Survey researchers have known for years that identity issues influence the way people answer opinion questions, especially in the context of political research. Women respond differently in political opinion surveys depending on the gender of the survey administrator, with a tendency to report more liberal attitudes when asked by a woman.
"What's surprising is that identity issues can come into play into what is ostensibly a test of your knowledge," said McGlone. "Heightened awareness about your identity as a man or woman or member of a certain group could influence your performance on a standardized math test."
This phenomenon is called stereotype threat—the fear that one's behavior will confirm an existing stereotype of a group with which that person identifies, leading to impaired performance. It was first articulated by Aronson and social psychologist Claude Steele of Stanford University.
"Stereotype threat can be induced by a variety of subtle cues in the testing environment," McGlone said, "such as the gender composition of a class or being asked to indicate one's ethnicity or gender on a test demographics question. These cues heightened awareness of people's 'ascribed identities'—for example, identities based on things about themselves that they can't easily change."
McGlone acknowledged that many aspects of personal identity are achieved—membership in social categories based on individual achievements—rather than ascribed. He contended that deficits in test performance caused by stereotype threat could be mitigated by reminding test takers of the achieved identities they possess for which there are positive performance expectations.
"In other words, by putting women in a situation where they're not preoccupied with negative gender stereotypes, you can significantly reduce the gender gap in standardized testing performance," he said.
McGlone tested his hypothesis by priming different social identities among undergraduates prior to administering the Vandenberg Mental Rotation Test (VMRT), a standardized spatial reasoning test linked to math performance. The VMRT typically produces the largest documented gender difference in any cognitive ability, a difference some academics have attributed to genetic differences in intelligence favoring men.
McGlone and his colleagues asked male and female students at Lafayette College in Easton, Pa., to take the VMRT. Prior to the test, the participants completed one of three short questionnaires composed of six questions designed to cue a particular social identity: their residence in the northeastern U.S., their gender, or their status as students in a selective private college.
He found that women who were primed to contemplate their identity as students at a selective private college performed at a significantly higher level on the VMRT than those primed to contemplate their gender or a test-irrelevant identity. In contrast, priming selective private college status among the male participants did not improve their performance. However, priming their gender status (men are better at math) did improve their performance.
"These results suggest that priming a positive achieved identity (selective private college student) can alleviate women's anxiety about confirming the negative stereotype that 'women can't do math,'" said McGlone. "When we primed this positive identity in men—for whom there is no negative stereotype regarding their math acumen—their performance was no better than when their gender was primed.
"We were able to significantly reduce an allegedly large gender difference with a pretty simple manipulation," said McGlone. "Regardless of whether the documented gender gap is due to biology or socialization, we can narrow it by psychological means."
Applications for these findings might include eliminating subtle cues from standardized math testing environments that might make gender identity issues salient to women and impair their performance.
"We're pushing for the College Board and other standardized testing organizations to move demographic questions to the end of the test," said McGlone. "Testers think they're just collecting data in asking for gender, ethnic and geographic information, but there's a subtle—and consequential—communication going on here. It says, 'Your gender matters."'
"By simply manipulating when questions are asked we can appreciably improve SAT scores," he said. "Ideally, cues that heighten awareness of any negative stereotypes—ascribed or not—should be eliminated from testing environments."
The Vandenberg Mental Rotation Test typically produces the largest documented gender difference in any cognitive ability, a difference some academics have attributed to genetic differences in intelligence favoring men. Illustration: Nicholas Bright.
Until that happens, students, especially women and those in ethnic minority groups, should consider focusing on attaining additional identities—those associated with positive academic expectations—as a means of improving academic performance.
It's important to note that while stereotype threat accounts for some of the disparity in standardized test performance among men, women and ethnic minorities, there also are accumulated socialization factors that contribute to these differences.
Stereotyping Among Children
"It's important to get kids to think about defining themselves in ways that transcend their gender and ethnicity early on," said McGlone. "But talking to a 5-year-old about coping with gender and ethnic stereotypes is fraught with problems."
There are many tools for measuring stereotype ideation in adults, but they typically require sophisticated reading and reasoning skills that make them inappropriate for the under-10 set.
In 2002, while working as a fellow at the Center for Research on Culture Development and Education (CRCDE) at New York University, McGlone and his colleagues assessed the social factors that influence elementary and middle school children's academic achievement in New York City public schools. Among the factors being assessed were children's access to and use of popular media and communication technologies, which contribute to the disruptive influence of self-relevant stereotypes.
"We conducted in-home visits to assess the parents' relationships with their children and how these relationships influenced the children's cognitive and social development," said McGlone. "Familiarity with gender and ethnic stereotypes was one of the many aspects of social development we investigated. We were particularly interested in how the parent-child relationship influenced the age at which children exhibited familiarity with stereotypes.
"In conducting these assessments, we needed a tool that allowed us to measure and quantify the kids' stereotype beliefs in a subtle manner."
McGlone created an age-appropriate measurement tool based on the classic children's game "concentration" or "memory." In the traditional game, players examine a set of cards placed face down in a grid formation. Each card has an image on its face that is identical to one other card in the grid. On each turn, the player turns over two cards on the grid. If two cards match, they are removed from the grid. The object of the game is to remove all cards from the grid by identifying all of the matches.
The twist to McGlone's game was the gender and ethnic composition of the faces and the stereotypes presented.
During their in-home assessments, researchers gave children one of three decks of cards with the face of a boy or girl on one side and pictures of objects (cooking utensils, trucks, etc.) on the other. The decks consisted of the stereotype deck, which re-affirmed stereotypes such as cooking/girl or doctor/boy, the counter-stereotype deck, which went against convention with cards such as computer/girl, doll/boy and the non-stereotype deck, which included a combination of both stereotype and counter-stereotype cards.
McGlone and his team found that children performed the game much faster with the face-object relationships conformed to gender stereotypes (for example, two boys who like trucks or two girls who like cooking) than when they did not (for example, a boy and a girl who like trucks).
"We measured the time it took children to finish stereotype-consistent or inconsistent grids with a stopwatch to create an index of their endorsement of gender stereotypes," he said. "And we found that the children were quite familiar with gender-based activity stereotypes (for example, boys like fishing, girls like cooking) by the age of 5.
"Five- and 6-year-olds love this game and can fly through the stereotyped cards in two to three minutes," said McGlone. "But it takes kids much longer to go through the counter stereotype deck. Even kids from highly liberal households performed much better on the stereotyped deck than the counter- and non-stereotype deck."
This demonstrates that children's intuitions are in line with stereotypes, which is why the counter- and non-stereotype decks—where you can't rely on your assumptions—were harder for children.
Armed with the findings from his research with the CRCDE, McGlone and his colleagues created workshops to teach middle schoolers about stereotype threat and how to handle it when taking their entrance exams for the New York City magnet schools.
"To be forewarned is to be forearmed," said McGlone. "We talked to the kids about the things they're good at and their multiple identities: girl, soccer player, science enthusiast, etc."
By instructing the students to think of their achievements and things in which they excel, McGlone and his colleagues were able to take the weight off the stereotype—girl, African American, etc., and dramatically cut the performance gap on magnet school entrance exams.
"By telling kids 'what makes you YOU is what you DO,' we got them to focus on their achievements," he said. "It sounds like a very simple manipulation—and it is—but it has significant effects on students in a standardized testing situation."
The game proved to be a valuable tool for measuring stereotypes among children and—equally important—the children were enthralled with it. It got McGlone thinking about the mother's question, "How do you talk to children about stereotypes?" And "How can you prevent stereotypes from interfering with children's academic success?"
Enter Trait Mate, an online game he is developing for measuring and modifying children's social stereotype beliefs. A Web-based version of the "memory" card game, Trait Mate serves as both a tool for researchers to measure children's knowledge and beliefs about stereotypes.
In addition to being used as diagnostic tool for researchers and educators, Trait Mate can be a stimulus to negate stereotype ideations.
"For example," McGlone said, "after completing several Trait Mate grids in which no stereotyped trait mates are present, and gender and ethnicity cues have no value in determining trait mates, kids' endorsement of social stereotypes may be reduced by virtue of this exposure.
"I foresee teachers incorporating Trait Mate into their social studies curriculum and using the results as a springboard to discuss prejudice and gender stereotyping," McGlone said. "A teacher could have the kids play the online game and then discuss why one version of the game was easier than the other."
Trait Mate is being developed—under McGlone's guidance—by the members of Girlstart, a nonprofit organization that empowers girls in math, science, engineering and technology. McGlone hopes to make it available online to researchers and educators in the coming year.
The Mind as a Muscle
McGlone is interested in further exploring interventions and is conducting preliminary research on getting children to think about intelligence as something that can change.
"Many people think about intelligence as something that is fixed or something you're born with," he said. "I'm intrigued with the idea of teaching kids to think of their mind as a muscle, which can get kids excited about learning."
McGlone and a colleague are developing a prototype for an online math game targeting fifth- and sixth-graders featuring an avatar that gets bigger as players build their skills, allowing players to do more things, go further in the game and ultimately score more points.
"If I can get these kids, who often think of intelligence as something that can't change, to think about the fact that how they're doing in school has more do to with what they're doing, not what they're born with, that can make them less vulnerable to stereotype threat," McGlone said.