Ever since I looked up the data a couple of years ago and found it to be far more fascinating than I expected, it’s been a bit of a hobby for me, to follow the trajectory of the percentage of women in computer science.
The U.S. data is very easy to get. You can find it yourself right here. It’s published at “government speed,” meaning here in 2020 we finally have the data from 2018. So what does it say?
If you were unaware of the peak of 37% women in 1984, or the fact that we’ve gone through two distinct periods of decline – I wrote about that last time. There’s some speculation in that post about reasons behind those declines, though I haven’t found any further information on the subject since then. Suffice it to say, our gender balance was better in the past. We went through a 10-15 year plateau at around 28%, and after the decline correlating with the dot-com bust stabilized, we’ve been at a 10-ish year plateau at around 18%. I’m happy to say that the numbers are slightly better now that we’ve added the last two years: we’ve risen from 18% to 20% women!
If nothing else, keep that number in your head: don’t settle for goals below 20% in your admissions, hiring and other activities. Actually, since this is 2-year-old data, let’s not settle for less than 22%, mm’kay?
Another good thing in this data: we’ve finally, *finally* started graduating more women with CS degrees than we did in 1986 and 2003. Think about that – in just total numbers, more women got CS degrees in 1986 than in 2016.
One observation that I don’t know whether to feel hopeful about or afraid of. A little of both. The total number of CS grads per year is skyrocketing! While that says great things about us finally being able to fill the work force we need – because we need a lot more people across all industries who have tech skills – I’ve seen graphs like this before. I may be too much of a downer here, but it looks like a bubble to me. It can’t go on like that forever, can it?
A pessimistic observation would be – so far women don’t seem to come out too well whenever bubbles burst in this industry. We can’t afford to let the percentage drop again. I’m still P.O.’d that it went from 28% to 18% on my watch. No way, no how, we can’t drop again.
So: let’s choose to be optimistic that even if the tech industry starts getting over-populated, it could spur on a renaissance in other industries who’ll get an influx of people with tech skills.
The other bellwether I like to watch is the gender balance of Advanced Placement CS exams. That’s published here.
What you see here is the data from three different courses. The main AP CS class that has been available for a while is Computer Science A. In earlier years there was a more advanced class, Computer Science AB, which was discontinued about a decade ago. And in 2017 the College Board started offering a new class, Computer Science Principles. Principles is a more introductory class aimed at familiarizing students with the basic concepts.
Of the three classes, the most advanced one, AB, had the worst gender balance. The “standard” class, A, has varied over the years, but you could roughly say it hovered around 16-18% until 2014 and then started rising (hooray!) — it’s now up to 25% female. And the Principles class has done great so far, starting at 30% and rising to 33%. What I’m *really* watching is I want to see if Principles tends to increase the number of young women going on to take AP CS A and then computer science in college. So far I am still hopeful!
And comparing the numbers between AP CS and CS degrees:
- In 2018 almost 16K women received CS degrees.
- Looking back 4 years earlier to when those women were in high school, in 2014 about 7.4K young women took AP CS. Those weren’t all seniors, but it gives us a sense of scale between the two.
- About 30K high school girls took AP CS Principles in 2019. Pretty good considering that was only 13K in 2017!!
- Between Principles and A, a whopping 46K young women took AP CS in 2019!
So I guess what I am saying is – I have a lot of hope. 🙂
Note: cover image courtesy of #WOCinTech, wocintechchat.com
P.S. it’s harder to get data about racial makeup and LGBT folks. Once in a while I go looking. NCES has a bit; it’s also more complicated and I’m not quite trained to do the breakdowns well. But I’m interested in that data too.