Wednesday, September 7, 2011

Too Pretty to Write this Blog Post: On Math Performance, Stereotypes, and Bad T-Shirt Choices

I know, I haven't posted in a while. Mea culpa, etc. etc. I'll try to be better about this from now on :)

Recently, two stories came across my desk (well, desktop, but whatever) that I've been meaning to blog about. It seemed like a perfect opportunity to talk about some of my early research interests. I've always been interested in the media: how it shapes our perceptions, and how it influences our behavior, performance on various tasks, and even our health (something I've done some research on and should definitely post about one of these days). One of my first studies I worked on - for my master's thesis - was about how the media influences women's math performance.

The most recent pop culture reference to women and academic ability was on a t-shirt marketed toward girls 7 to 16 at the major retailer, JC Penney. Though JC Penney eventually pulled the shirt, many argued that this t-shirt perpetuated negative stereotypes about girls and women. This is not the only instance of stereotypes about women popping up in the media. Abercrombie and Fitch prompted a "girlcott" in 2005 when they released a t-shirt with the phrase "Who needs brain when you have these?" printed across the bustline. T-shirts aren't the only outlets for these messages. Perhaps you remember the "Math is hard" Barbie? Teen Talk Barbie was the cause of controversy when one of the many phrases she uttered is "Math class is tough!" And advertising is constantly filled with gender stereotypes (both of men and women).

But wait a minute? Do these messages really affect women's math performance? Research (and not just my own) finds that they can. To avoid looking biased toward my own research, I won't even discuss my thesis.

But first, to give you some background information. Much of this research is influenced by the work of Claude Steele and his colleagues, and on Steele's concept called "stereotype threat". Essentially, stereotype threat occurs when the negative stereotype about a group's performance on a task makes a member of that group underperform on that task. To give a concrete example, Steele first explored this concept in a study on standardized test performance by African-American students. He and his colleague Aronson found that making African-American students conscious of racial stereotypes resulted in underperformance on a standardized test. Even asking them report their race on the test lowered performance. When racial stereotypes were made irrelevant through test instructions, African-American students performed at the same level as White students. (Read the original article here).

Stereotype threat became an important area of study because it provided evidence to counter claims of genetic inferiority that have been used to explain past race differences in standardized test scores. Steele and his colleagues went on to apply this phenomenon to a variety of groups, and used it to explain another group difference: women’s scores in math testing. In a conceptual replication of their study on African-American students' test performance, they found that when some women were told a math test had shown gender differences in the past, they performed worse than men; when they were told the math test they were taking had not shown gender differences, they performed at the same levels as men. (Full paper available here).

Prior to Steele and his colleagues' research, biological explanations were applied to gender differences in math and the sciences. Many people believed that women were just naturally bad at math. These opinions continue even today; early in 2005, Harvard University President, Lawrence H. Summers, presented a variety of explanations for women’s under-representation in math and science, including that the difference may be due to genetic inferiority (he resigned not long after, though this speech was just one of many marks against him).

Steele established three conditions that must be satisfied in order for stereotype threat to occur: 1) there must be awareness of the stereotype by society at large, that is, the stereotype that women are bad at math must be well-known; 2) the individual must be identified with the domain of interest, that is, a woman must be math-identified in order for stereotype threat to take place; and 3) the negative stereotype must be relevant to the individual during the domain-specific situation, that is, the stereotype that women are bad at math must be relevant to the individual when she encounters a math test. Though some people have called proposition 2 into question, and have even shown that stereotype threat can operate among women who don't really care about their math ability/performance, the other two propositions are necessary.

What these propositions mean, however, is that instructions from an experimenter aren't necessary for stereotype threat to occur. Simply having a high awareness of the stereotype (abstract here), being strongly woman-identified (full paper), or believing that gender stereotypes are correct (full papers here and here) can lead women to underperform in math even in the absence of any stereotype cues from the experimenter. It's not even necessary that the information made salient is related to math. Being the only woman in a group of men is one way (abstract, full paper). Receiving unwanted homework help from your parents is another (abstract here).

But it was work by Paul Davies and his colleagues that found stereotypical television commercials could lower math performance in women and impact their interest in math. In one study, some participants saw two stereotypical commercials: a teenage girl jumping for joy about a new acne medication and another of a woman drooling over a new brownie mix. Others saw counter-stereotypical ads: one in which a woman impressed a man with her knowledge of automotive engineering and another of a woman discussing health care concerns. Men and women who saw the counter-stereotypical ads did not differ on math performance. Men and women who saw the stereotypical ads did, with women receiving significantly lower scores. In a second study, women exposed to stereotypical ads avoided math questions in favor of verbal questions more frequently than women who saw neutral ads or men in either condition. A third study found that exposure to stereotypical television ads lowered women's interest in quantitative careers.

None of these television commercials said anything about women's math ability. Why then would these influence math performance? It has been argued that these messages create an "atmosphere of stereotype threat", where the messages are ubiquitous and inescapable. No special instructions are needed to make women question their math ability; messages present in their environment are enough. Some have argued that the controversy of JC Penney's t-shirt is misplaced; after all, it's just a t-shirt. But is it? Based on what we know about stereotype threat and how similar messages have influenced women's math performance, can we really shrug this off as "just a t-shirt"?

So what was the second story that came across my desktop? A cultural exploration of not math performance, but a related concept: spatial reasoning. In this really creative natural experiment, researchers examined two tribes in India, the Karbi and the Khasi. These two tribes are very similar in lifestyle, diet, even DNA (they likely share some common ancestors). The main difference is in gender roles. In the Karbi society, land is passed to male children when the parents die. Women rarely own land. In the Khasi society, women control the land and goods; men can't own land and their earnings are turned over to their nearest female relative. The researchers went to different members of these two tribes (in total, 1300 people) and found some striking results: in the Karbi society, men outperformed women, but in the Khasi society, no differences were found.

There are still people who argue that stereotype threat doesn't explain these differences, or if it does, is only one variable among many that explain these differences. There is definitely some validity to that argument. The effect sizes in these studies are small to moderate at best, meaning there is more pushing math scores around than stereotype threat. Furthermore, the original article that prompted this blog post is a t-shirt about "doing homework" in general - not just the math work. There aren't many stereotypes that women are bad in school - in fact, in some cases, the opposite stereotype exists - so would this t-shirt really have the effects I discuss above? What do you think?

Thoughtfully yours,


  1. I just read about the Karbi/Khasi study a few days ago. I'm probably misunderstanding the study, but if the Khasi women perform equally to the men, where the men are at a disadvantage, doesn't that actually reinforce the idea that the men have some innate spatial abilities? For example, if I can hit as many home runs as Babe Ruth if you tie Babe's arms behind his back (meaning none!), that does not mean my skills are equal to the Bambino.

    I would think the Khasi women should have outperformed the men given the role-reversals, but instead they tied.

  2. An excellent point, and I love the Babe Ruth analogy. If the effect is entirely cultural, why don't we see a reversal in abilities in the same way we see a reversal in roles in this society? There's actually an interesting phenomenon I'm not as knowledgeable about called 'stereotype lift', where making a stereotype salient gives the un-stereotyped group a boost. In studies of stereotype threat & math performance, this manifests as an increase in men's performance in the stereotype threat condition compared to men in the neutral condition. This effect is also very small - so small, in fact, that we only saw a significant stereotype lift effect in this research through meta-analysis, a procedure that allows researchers to combine the results of multiple studies on a topic - but still often visible when results were graphed.

    I'm not sure how pervasive gender stereotypes are in these societies, which may explain some of the effects. I'm also not aware whether men in the Khasi society are denied other opportunities, like education, that could have influenced spatial ability.

    :) I love intellectual discussions. Anyone else have any thoughts?