As science gets tough, the girls get going … to some other career?

January 21, 2013 at 12:09 pm | Posted in Uncategorized | 1 Comment
Tags: , , , , , ,

The graph shows that courses with fewer women students rank more highly on a rating of scientific difficulty

The x-axis of this boxplot shows the percentage of female students accepted onto 73 higher education science courses in the UK in 2011 (latest available data, from the official UK source, UCAS).

The y-axis shows how the course subjects rank on an index of scientific rigour (Fanelli, 2011), with ‘harder’ sciences like physics having a higher ranking (nearer 1) and ‘softer’ sciences like psychology a lower ranking (the lowest being 20).

The boxes show where most of the courses cluster, and the top-and-tailing lines show how widely the data are spread.

(I’ll say more below about those three little outlying numbers in the top right-hand corner: they don’t fit the pattern. For more detail on the making of this graph, see this background information PDF, Fanelli_background.)

The graph seems to suggest that tougher courses have fewer women students, while softer sciences attract more girls.

This is not a new idea. In 2005, the President of Harvard University caused a ruckus by suggesting that ‘one reason there are relatively few women in top positions in science may be “issues of intrinsic aptitude”’.  (The quote is from Inside Higher Ed; the link to the transcript of Larry Summers’ speech is, alas, broken).

So, the UCAS data support this controversial claim, yes?


Here are some reasons why not.

We don’t know how much of the difference in career choices made by (mostly) teenage boys and girls is down to ‘innate factors’, like genes, and how much comes from cultural, family and peer pressures.

We have observed that scores on the internationally-standardised PISA maths and science tests show a smaller gender gap in countries which treat their women better, according to a paper (written by a man) in PLoS ONE.

This suggests either that:

a) having more innate factors for male brilliance goes with being a nation of sexist so-and-sos (in which case some nations should be getting much better PISA scores than they are),


b) socialisation may have more of a role than Prof Summers initially thought (as indeed he later acknowledged). For an interesting take on this, see Ceci and Williams, 2010.

Thus there are many potentially confounding factors contributing to the apparent link between softer science and female interest. This is why getting firm answers about gender differences  – let alone sex differences – in intellectual ability is a problem with so many sharp and poisonous thorns.

Looking a little further

For example, what if we look at the UCAS data in more detail, and ask how many women are accepted onto courses in each of the Fanelli subject groupings? If men are keener on tough science, then the highest-ranking subjects should have the lowest number of women, and vice versa.

Thus a plot of the Fanelli ranking of a course against the number of women taking it should be highly-correlated (i.e. nearly a straight line).

Actually the correlation’s about 0.32, dropping to 0.25 if you control for the total number of students on each course. Not that numbers should be taken too seriously, with such a tiny dataset; the point is that the correlation coefficient is nowhere near 1.00. And the plot looks like this:

Women taking a course, by scientific 'hardness' of the subject

A straight line this is not. The only such line is the solid one marking is where gender parity would be if we had it. The graph shows the percentage of women accepted onto courses for each of the Fanelli subject groups – arranged in rank order from the highest (‘toughest’) on the left to the lowest (‘softest’) on the right.

Unsurprisingly, computer science and engineering are well below gender parity, and psychology is well above (Dorothy Bishop has blogged about this).

However, not all high-ranking subjects lack women (e.g. plant & animal sciences), and some lower-ranking ones do (e.g. materials science).

What’s going on?

Who knows? As so often, we have insufficient data and a surfeit of hypotheses. More specifically, we do not have the data to decide between the hypotheses. They are, to use the technical term, ‘underdetermined’: we can’t determine which are incorrect.

Thoughts and speculations

One possibility often cited is that, overall, women may tend to be more interested in people and animals, and men in things and abstractions. Women may, in general, have better theory of mind, or be more empathetic, than men (as researchers such as Simon Baron-Cohen have suggested). This may give them more interest in – and a greater tendency to study – other beings, especially other beings with minds.

If so, one prediction about the very simple dataset I’m discussing is that fewer women should choose the ‘plant’ subjects (Forestry and Botany) in the plant & animal sciences, and more should opt for the ‘animal’ subjects (Animal Science, Pre-clinical Veterinary Medicine, Zoology).

Here’s the table:

% Women Applying

% Women Accepted










Pre-clinical Veterinary Medicine



Animal Science



It seems that, the more animalcules, the more the ladies love it. Looking back at the first graph, you can see three case numbers which are outliers: high-ranking subjects taken by lots of women. These are, of course, Zoology, Veterinary Medicine and Animal Sciences.

Evidence for the ‘girls like living things’ claim? Barely. This is a blog, not a research paper. Even in a paper, one little table with five data points would need considerable backup. Also, the claim needs both fleshing out and refining to become an adequately testable hypothesis. Even were it correct, it would say nothing about whether the group-level tendencies were ‘innate’.

What we can say is that my first graph, suggesting that women tend to avoid the tougher scientific subjects, glosses over a more complicated story. Some tough sciences draw plenty of girls.

To make a sweeping generalisation of my own, this is why academics are still useful members of society. They’re trained to look beyond initial impressions, to dig deeper, to criticise. That may make them come across as carping nit-pickers, but nit-picking is a necessary service in a world replete with nits. Generalising is easy, quick, and misleading. (Not all academics do a good job.)

Talking of complicated …

Finally, it’s worth thinking about why the life and social sciences are deemed softer and less rigorous than the hard physical sciences. For one thing, the latter tend to be older and better-established. Physics had the Newtonian framework well before evolution, molecular biology or even modern psychology took shape. It’s had longer to polish its methods and brush up its mathematics.

But there’s also the well-known human trait of efficiency – or laziness – to be considered. We do like our low-hanging fruit, in science as elsewhere. We started with the easy stuff: the inorganic world of physics and chemistry. Organisms, especially human organisms, are far more complex, far harder to study, and far more difficult to understand. Furthermore, they may involve such intricate interactions that traditional reductionist science struggles to get a grip.

Looked at this way, someone still wanting to make such generalisations might say that overall, men like more rule-bound, less complex sciences with lots of bits and pieces, while women prefer the harder, more holistic problems of understanding living things. In other words, when the science gets tough, the girls get going.

Perhaps. But that’s another idea still needing refinement, still hugely underdetermined by the available data, and with lots of competitors, from hypotheses about math anxiety to sociological ideas about how science relates to power.

It might be better to avoid making inadequately-supported generalisations about gender differences. That, however, is really hard work.


1 Comment »

RSS feed for comments on this post. TrackBack URI

  1. Weird that the bar graph considers “business & economics” not only a science but a “harder” science than biology. I beg to differ…

    I love your alternative to the classic grouping of “harder” and “softer” sciences. I will definitely – and gleefully – relay that to my physicist boyfriend. Birds are absolutely harder to study than photons!

If you'd like to leave a comment ...

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Blog at
Entries and comments feeds.

%d bloggers like this: