Science: it's a role model thing

I recently gave a talk for the Huntsville, Alabama Chapter of the American Association of University Women (AAUW), and was asked to speak about what efforts there are to get girls interested in science. What I didn't realize is that the National Research Director of AAUW, Dr Catherine Hill, would be speaking right before me and covering a lot of the same territory, based on her co-authorship of an outstanding 2010 report on women in science, technology, engineering and mathematics (STEM) [1]. There was a slightly embarrassing moment when we said different things about the effectiveness of role models, which prompted me to look into the issue more deeply.In the great tradition of preparing talks the night before, I had looked up the 2007 report from the Institute of Education Studies entitled 'Encouraging girls in math and science' [2]. The team of authors examined existing studies and came up with five recommendations, and a label for the amount of evidence supporting each. Recommendation number 3 is 'Expose girls to female role models who have succeeded in math and science,' but the panel rates 'the level of evidence that supports this recommendation as low.' So, that's what I said in my talk, and this idea did not go over well with the room.It turns out that the panel made this rating because in 2007 the few studies they could find that documented role-model effectiveness focused on college students, and not on pre-college girls. These studies also focused more on immediate testing performance and belief in a girl's own math ability, for example, than longer-term consequences of role modeling. Since then, there has been an explosion of online efforts to involve girls and women in science, but peer-reviewed studies showing that these activities are actually effective remain scant.For test-based studies especially, a key concept is 'stereotype threat', or the effect of anxiety or negative emotion when a subject is put in a position where they might confirm a negativ