Researchers at the University of New South Wales and the Australian National University have performed an analysis based on 227 studies (a so called “meta-study”) covering well above 800,000 students of each gender. Results show that, overall, girls obtain significantly higher grades than boys and the grades of the former gender group present less variability than the latter. Both gender differences in mean and variance seem to be lower at the university stage (probably due to selection bias), although at the school stage they do not seem to decrease with age. Furthermore, these gender gaps have not changed in any significant way over the last eight decades. The surprise arrives when focusing on subject type: although girls’ superiority in average performance is relatively higher in non-STEM (as expected), the gender gap in the variability of grades was lower in STEM fields. The result is entirely driven by the lower standard deviation displayed by the grades of girls within non-STEM relative to STEM fields. As it turns out, within boys, no differences in variance are found between the two subject types.
The authors exclude the “variability hypothesis” as an explanation of the low female presence in STEM programs of higher education. They emphasize the importance of making their path in this direction more attractive. They speculate that a girl faces great barriers like the “collective beliefs of the society she is raised in” and also, that they may follow a non-STEM road because their relative advantage to boys is larger in that case (something that their very same results confirm).
Comment: Biologists since Darwin have noticed an intriguing pattern in many traits that is common to many species: males seem to be of a more variable kind. The pattern has proved to be quite robust and it covers both physical and psychological attributes. Many debates on gender differences have built arguments that gravitate around this observation and it may have important implications that underpin the empirical regularities found in the choices of boys and girls when it comes to their university studies and career paths. A meta-study like this is thus of great interest. It is also commendable the methodological sophistication that the authors employ in its implementation.
The only noticeable weakness can be found in the main conclusion that the authors seem to draw from their main result, namely, that the “variability hypothesis” cannot be used to explain the male over-representation in STEM fields. This is striking. First, the hypothesis has been mostly invoked as an explanation of potential gender differences in the tails of the distribution. For example, Steven Pinker, Professor of Psychology at Harvard University, has resorted to this regularity as one of many explanations that could be behind the lower female presence among Math Professors. As he eloquently puts it, “We are talking about extremes of achievement. Most women are not qualified to be math professors at Harvard because most men aren’t qualified to be math professors at Harvard.” A larger variability among men could produce more prodigies but also more idiots among their ranks. Second, the authors seem to be aware of this point when they write that their data “does not preclude a gender gap among the exceptionally talented” but that “it should be noted that STEM careers are not restricted to the exceptionally talented”. In such case, why invoke the variability hypothesis in first place as something to be debunked in this context?!
Text: Iñaki Rodríguez Longarela, Stockholm Business School