STEM-readiness and the gender gap in STEM at university

Consider the proportion of university enrolments that are in STEM (Science, Technology, Engineering, and Mathematics). In contrast with enrolment in the social sciences (for example), the pre-requisite learning at high school is higher for STEM subjects in university, since there are minimum levels of prior mathematics and (often) basic sciences. So, we can think about the proportion of university of enrolments in STEM as represented by the following equation:

That is, the proportion of all enrolled university students (ENR) that are STEM students is equal to  the proportion of students who meet the pre-requisites (READY) who enrol in STEM, multiplied by the proportion of all enrolled students who meet the pre-requisites. This is an over-simplification, but it helps us to understand the difference in the gender enrolment rate in STEM, where typically a higher proportion of male university enrolments are in STEM. This gender difference might arise because more male students who meet the pre-requisites enrol in STEM (compared with female students who meet the pre-requisites), or because more male students who enrol in university meet the pre-requisites (compared with female students who enrol). Knowing the source of this gender gap has important policy implications if you want a similar proportion of enrolments in STEM for each gender, since it tells you whether you need to increase female participation in the high school pre-requisite courses (the second term in the equation), or if you need to encourage female enrolment in STEM for those that meet the pre-requisites (the first term in the equation).

A recent NBER working paper (alternative ungated version here) by David Card (University of California, Berkeley) and Abigail Payne (University of Melbourne) performs a similar disaggregation to the one above, using data on 413,656 university entrants in Ontario over the period 2004 to 2012. First, there is a clear gender gap in the data in terms of STEM enrolment:

Overall, 30.3% of females (a total of 72,033 women) and 42.5% of males (a total of 74,763 men) enrolled in a STEM program. Note that the gap in the proportion of students within each gender group who register in STEM is large (12 percentage points) despite the fact that nearly half (49%) of STEM registrants are female. This reflects the much larger fraction of females than males who enter university in the province.

They find that the difference in the second term in the equation above dominates. Their preferred decomposition results show a:

...13.2 percentage point gender gap in the fraction of newly entering university students who enroll in a STEM program. Overall, 2.1 percentage points are attributable to a lower rate of entering a STEM major by STEM ready females than males...; 1.7 percentage points are attributable to the slightly lower fraction of females than males who are STEM ready at the end of high school and the slightly lower fraction of STEM ready females who enter university...; and 9.4 percentage points are attributable to the higher fraction of non‐STEM ready females who finish high school with enough qualifying classes to be university ready.

That last bit is not surprising in one sense, that female university students are less likely to be STEM-ready (that is, to have the pre-requisites for enrolling in STEM). However, it is surprising in the sense that the reason for that difference is the much greater numbers of female students enrolling who are not STEM-ready, compared with male students who are not STEM-ready. So, the gender gap could be reduced by encouraging more female students to take the pre-requisite courses in high school, or by encouraging more male students who are not STEM-ready into university courses in non-STEM disciplines.

The one potential negative about the research is that the STEM subjects include:

engineering, physical sciences, natural sciences, math, computer science, nursing, environmental science, architecture, and agriculture.

They do make a good case for why nursing is included, but I also wonder about agriculture, and whether excluding those two subjects would make much of a difference to the results.

However, overall the results are interesting and help us to better understand the gender gap. I would be really interested to see if something similar holds for New Zealand.

[HT: Alex Tabarrok at Marginal Revolution, back in September]