5/31/2011 17:19

Gender Differences in Physics Education

A research article from Science last November provides quantitative evidence that the educational gender gap (specifically in physics test scores) can be dramatically reduced by "values affirmation".

"Values affirmation" is a technique previously demonstrated in ethnic minority education which basically has the student spend a few minutes writing on whatever is most important to the student before proceeding with the coursework. It is so trivial an intervention that it is puzzling why it works at all, but in fact it provides measurable and lasting grade enhancement. I think I did something analogous when I was teaching, but not under that name and not as a grade-enhancing technique.

"In many science, technology, engineering, and mathematics disciplines, women are outperformed by men in test scores, jeopardizing their success in science-oriented courses and careers. The current study tested the effectiveness of a psychological intervention, called values affirmation, in reducing the gender achievement gap in a college-level introductory physics class. In this randomized double-blind study, 399 students either wrote about their most important values or not, twice at the beginning of the 15-week course. Values affirmation reduced the male-female performance and learning difference substantially and elevated women's modal grades from the C to B range. Benefits were strongest for women who tended to endorse the stereotype that men do better than women in physics. A brief psychological intervention may be a promising way to address the gender gap in science performance and learning." [abstract]

This article also shows a small negative effect among the men. In other words, asking men to expound on their most important values may have a negative effect on their test grades, although smaller and perhaps less durable than the positive effect on women.

The grades are shown only in the aggregate, so the gender effects are not clear from this report.

1. Perhaps gender differences are directly influenced by values affirmation. It isn't clear what the gender differences themselves would be.

2. There might be an interaction between values affirmation and socialized gender identity. This interpretation may be supported by the observation that the effect is "strongest for women who tended to endorse the stereotype". (This is also consistent with the large effects on minority middle school students previously reported.)

3. Gender differences may not be directly involved. Suppose for example that there are some people who are affected positively, others who are affected negatively, and perhaps still others who are not significantly affected. The distribution of positive and negative responders may be gender-biased, but there could be both types of responder among both men and women.

The difference in interpretation could be important ethically. If the positive responders change their scores upward more than the negative responders lose downward, there could be more students who are affected negatively by this intervention (even among the women) than are helped. That may be unlikely, but it certainly is not implausible; if true, this intervention could be ethically unsupportable, at least if applied unselectively.

What we know is that there is a trivially simple intervention which in the aggregate improves women's performance in physics (as well as that of minority middle-school students of both genders). The results are robust enough that we should start to consider applying this intervention broadly, wherever there is a persistent difference in performance between socially defined categories of students.

What the report leaves us unsure about is whether values affirmation may have significant negative effects on some students. The potential harm is large enough that, even if we feel this unlikely, it is ethically necessary to proceed cautiously and to study the effects more closely.

In this specific instance, it would be at least foolish to avoid values affirmation in light of its demonstrated success. But it would surely be dangerous to apply this intervention universally without consideration of the hints of possible harm.

It is sometimes said that "a little knowledge is a dangerous thing", but in actual reality all our knowledge is profoundly limited. It would be dangerous to presume more certainty than can be justified by the knowledge we actually possess; it would be foolhardy to ignore the knowledge that we have discovered.