I think it's worth discussing when such interventions need to begin - elementary school? high school? As an elementary school teacher, I sadly see many girls fall into, and actually embrace, the stereotype of the "ditzy, but pretty, girl" who shouldn't do better than or know more than the boys. Perhaps this is a naive comment, but at times, I think there is a cultural element to this thinking as well. Yes, I know it is an issue that crosses cultures and socio-economic groups, but there is that trend within my little bubble of the world that continues to bother me. I try to have the "sometimes-girls-need-to-work-harder-to-show-how-smart-we-are" speech, but then am left wondering how far that will go...

Thanks for keeping my head in the game over the summer!!!

Engineering/science calculus at a public university may not be sufficiently far out on the tail of mathematical ability for aptitude to trump a little extra diligence. I'd suggest re-running the experiment with an exam like the Putnam Competition (which is way out on the mathematic aptitude tail) to see if it boosts the number of women scoring in the top 200.

I've always found these studies disconcerting -- I don't like the idea that my mind can be toyed with is ways that can affect something like exam performance. I think KDeRosa may be right in suggesting that the mechanism is a boost in concentration that comes with confidence.

I also find it unnerving, from a policy perspective, that seemingly small details of how a test is framed can have measurable effects on the outcomes.

I read the study, and have to read it again. But I see a problem right away. In the abstract, the authors state:

"Test performance of women in a stereotype-nullifying presentation of the test in an experimental group was raised significantly to surpass that of the men in the course. In a control group, in which test-takers were given the test under normal test instructions, women and men performed equally."

Now, it is not quite true that the test-takers in the control group were given the test under "normal" test instructions. In fact, all test-takers, both in the stereotype-threat condition and the non-threat condition, read the following statement:

"For the next 20 min, you will be taking a math test aimed at measuring your mathematical abilities. Why? As you probably know, math skills are crucial to performance in many important subjects in college. Yet surprisingly little is known about the mental processes underlying math ability. This research is aimed at better understanding what makes some people better at math than others. After you finish the test, your teacher will score it. This will enable us to analyze your performance and compare it with other students taking this test."

The authors were deliberately invoking stereotype threat here. They say it clearly: "Our study, in contrast [to past studies], explicitly stated that the test was diagnostic of math ability. We believe that these instructions not only represent a stronger and more direct manipulation of stereotype threat, but also generalize more to real testing situations."

The subsequent statement about lack of gender differences on the test was given only to those in the non-threat condition:

"What about gender differences? This mathematics test has not shown any gender differences in performance or mathematics ability. The test has been piloted in many mathematics courses across the nation to determine how reliable and valid the test is for measuring mathematics ability. Analysis of thousands of students' test results has shown that males and females perform equally well on this test. In other words, this mathematics test shows no gender differences."

The authors believe that the first statement both invokes stereotype threat (as they intended) and generalizes to real situations. I'll give the authors the benefit of the doubt for the former point for now--but I wonder, does the statement really generalize to real situations? Does a statement like that accurately recreate a diagnostic test situation? I doubt most diagnostic tests would be worded in such a way. I wonder if the statement might not be setting up an artificially tense situation.

I wonder what the results would have been if the students had not been given the initial statement. The study is interesting, and I'm not negating the authors' findings offhand--but I do question the "normal" conditions of the control group.

Women and men have different bodies designed for different functions so why is it so implausible that their minds are also somewhat different?

Is there a different correct answer to an algebra equation depending on one's sex?

Sorry to disappoint but reality and actual results don't always conform precisely to a politically correct worldview and ideology.

George writes:
Women and men have different bodies designed for different functions so why is it so implausible that their minds are also somewhat different?

Why does it seem implausible? Perhaps because mathematical ability has no obvious connection with reproductive role...

Perhaps because mathematical ability has no obvious connection with reproductive role...

Maybe because of the reproductive role nature can take more chances with males and tolerates less veriability in females, hence the double x chromosome thing which tends to serve as an averging function and reduces outliers. So the female's decreased variability for reproductive reasons may spill over to other genetic traits like visiospatial ability.

Is there a pre-selection bias in the sample? Women enrolled in an engineering/math/science track have been hearing and ignoring "girls shouldn't do math" propaganda for at least ten years.

Lots of comments to respond to – sorry to be late to the discussion.

Mimi – I’d like to hear more about your cultural hypothesis here – I *think* you’re saying that this is an issue for some racial/ethnic groups and not others?

Ken – Aren’t these precisely the women who would go on to math and science careers? While I think it’s interesting to look at whether the most mathematically talented women respond differently, I’m more concerned with the big pipeline for math and science careers.

Rachel – The strong role that testing conditions play really freaks me out as well.

Diana – Very good points. The question is whether these results should be seen as more or less normal than laboratory studies, from which there is a large base of evidence about stereotype threat. Btw, did you see that you got props for your comment on the National Review website?

George, Rachel, and Ken on biology – I’m not willing to totally discount some biological differences, and I think the extreme achievement question of Larry Summers infamy (i.e. that boys appear to be more likely than girls to have very high or very low achievement) should be investigated. I’ve been told that a good review of this literature can be found here:
Spelke ES. 2005. “Sex Differences in Intrinsic Aptitude for Mathematics and Science? A Critical Review.” American Psychologist. 60(9):950-58

Heather – The study intentionally sampled students in high-level college math courses to see if stereotype threat could be an issue across the entire spectrum.

Did the gender nullification statement boost female scores or did it retard male scores? Since males and females scored equally well when the "stereotype threat" only was present, is it possible that males interpreted the nullifying statement as a "political correctness threat?"

I always told my kids things like, "You've had great preparation for this test and worked very hard, you've learned what you need to know to do well, last year's students did well on this and you've worked just as hard as they did, this test was designed for kids like you to do well, etc." to try to at least trigger the idea that they were all in a group that was prepared and could be successful... of course, I have no studies to back up my "intervention."

Aren’t these precisely the women who would go on to math and science careers?

Yes, but I'm thinking that if you want to prove that stereotype threat rather than cognitive abilive is the cause for the disparity you'd want to use a testing instrument that is sensitive to cognitive ability and not other confounding factors such as extra diligence/studying. If the theory holds, women should be able to perform as well as the men once the stereotype threat was removed. In contrast, if your testing instrument is not sensitive to cognitive ability, other confounding factors may affect the achievement results.

According to Project Talent, men have a mean (male-female) difference of 0.12 standard deviations and a 1.20 (male/female) variance ratio advantage over women in math ability. If the testing instrument were sensitive to cognitive ability, this advantage would, in theory, show up in the results and if the stereotype-threat theory was valid removing the threat would equalize the results.

Art hit it: how do the control and the nullification group compare to each other? (A paranoid soul would note that people treat men as the normal and therefore assume that it is women who vary.)

And it doesn't matter whether the difference is plausible. What matters is if it exists.

– I’m not willing to totally discount some biological differences, and I think the extreme achievement question of Larry Summers infamy (i.e. that boys appear to be more likely than girls to have very high or very low achievement) should be investigated.

My gut feeling (based on a professional life spent partly among high achieving scientists) is that there may be something in this, and that it may significantly affect the gender distribution of Nobel prize winners and Fields medalists, and possibly even university math and physics department faculty.

But if it's used to justify trends seen in high school or middle school, I worry...

I had a (female) high school math teacher who complained that there were "too many" girls in the accelerated math class, and that "everybody knew" we weren't really interested in math, and we just got selected for the class because we worked hard. Fortunately, that's more than 30 years ago now...

Great, very interesting, post! I think the bias and messages for women start very young regarding ability and acceptable behavior in terms of gender stereotypes, and in fact, there is a body of work indicating that the way fremales are socialized to communicate may exacerbate the problem. I've been blogging about this at my own site (http://PowerfulMindCoaching.com/blog) and recently did a piece at Suite101: (http://americanuniversities.suite101.com/article.cfm/equality_for_female_professors). The conclusion I've reached is that although the sterotypes are still there and start very early, as Moms, professors, and women leaders, we have to jsut keep plugging away to model what is possible for women accross their academic careers. We have to keep tooting our horns and showing what is possible for all of us.