Sexism in computer science

:: computer, politics, science

Anyone who says that the facts show that men are innately better than women in computing either does not know the facts, does not understand them, or is lying.

The facts

In 1971, about 14% of US computer science and information science graduates were women. By 1984, about 38% were. But by 2011 the proportion had fallen to under 18%1. Here is a graph of the proportions by year from 1971 to 2011:

CS & IS graduate ratio, US, 1971-2011

CS & IS graduate ratio, US, 1971–2011

What the facts show

This entire process happened in about two generations: the proportion of women more than doubled in less than one generation, and then about halved in a generation: some of the women studying CS in 2011 could be the daughters of the cohort of 1984, and the granddaughters of the 1970 cohort.

No genetic change in a human population can happen this fast: evolution operates on timescales of thousands to millions of years, not over a small number of decades. This means that whatever caused these changes was not a change in innate ability. There simply can be no question about that: there must be some other explanation, since the innate ability of women to do computer science, or any other innate ability, cannot have changed significantly over this period.

This means that the changes were caused by something environmental. Perhaps in 1984 there was enormous positive discrimination, or in 1970 and 2011 there was enormous negative discrimination, or some combination of the two2.

This data is also perfectly compatible with the conclusion that women may be innately as good at computing as men: 38% is not very far from 50%, and if we assume some level of sexism in 19843 it is easily possible that the underlying figure was 50%.

What this data tells us, unambiguously, that whatever has caused these changes is environmental, and is not due to any differences in innate ability as such changes simply cannot happen over this timescale. It also tells us that things have got a lot more skewed since 1984: progress in this area has not only stopped, it is being reversed and has been so since the mid 1980s: the situation now is only about 28% less skewed than it was in 1970.

What the facts don’t show

What the data does not say is why this has happened, except that it is not due to changes in innate ability.

While it is almost certain that there was strong institutional discrimination against women in 1970, it seems unlikely that, in 2011, there was any kind of institutional discrimination, as this would be illegal and institutions are pretty good targets for legal action. So it seems unlikely that the decline is due to institutional discrimination. However all the data says is that there has been a decline: not why.

If we assume that most of the change is not due to institutional discrimination then it’s tempting to speculate on what did cause it. Well, I’m not going to do that: I have theories but they are based either on no evidence or on anecdotal evidence. Perhaps someone has done proper research into the causes, but I don’t know. There is a vast surfeit of theories based on little or no data, and outright made-up stuff on the internet — wild speculation, outright lies and ‘alternative facts’4 — and people are dying of this surfeit: I won’t add any more to it.

One possible inference is that women who, today, succeed at computing degrees, have done so against significant odds. It’s very likely that this means that they are better than men who achieve the same grades. So companies, if they are legally able to, might consider actively selecting female candidates for jobs, on the grounds that they are, probably, better.

In any area where people make a claim that some group is innately better than some other group based on some metric, and where the scores of one or both of those groups has changed radically over time, then it is immediately safe to conclude that those claims are either lies, confusions or both, because either the metric is junk, or it is not measuring innate ability. The obvious example of this is racial ‘science’.

  1. Source. Later figures may be available, but I couldn’t find them. I also don’t have the figures for other countries but I expect they are broadly similar. 

  2. I worked in academic computing from shortly after 1984 to the late 1990s and although I am not female I can say with some certainty that there was not enormous positive discrimination. 

  3. Again, in my experience there was some level of sexism in academia in this period. 

  4. Which are, of course, lies.