Software developers are not depressed
(but everyone else is)
I often see stories associating software development with depression and other mental health issues. A lot of people saw a 2013 talk by Greg Baugues in which he talks about his bipolar disorder and the possible suicide of one of his co-workers. There are a lot of posts talking specifically about Aaron Swartz. There’s a website and forum devoted to the topic of developers with depression. The perception amongst developers is that we are a high-risk group for depression and related disorders.
The problem is, the data disagrees: all formal research says precisely the opposite. Concerning mental health in general, a 2003 study for the UK’s Health and Safety Executive concluded that “the prevalence of psychological disorder was lower [than average] in natural scientists, engineers and technologists, … electrical and electronic trades … computer analysts, [and] programmers.” Concerning depression specifically, a 2007 study for the US Department of Health & Human Services found that only 6.2% of “mathematical and computer scientists” experienced such an episode, putting them below the average of 7%.
It seems like the circumstantial and anecdotal evidence in blog posts is the victim of a systematic bias in favor of depression; a bias which is only revealed by formal research. Every post I’ve seen on software development sites takes the high prevalence of depression amongst developers as granted, and then rationalizes this belief by giving hypotheses that explain the apparent phenomenon. For instance, J. Timothy King attributes it to three aspects of the software development career: (1) stress, (2) perfectionism, and (3) isolation. Most other posts make the same hypotheses.
Concerning stress, another HSE paper from 2000 tabulates ‘work stress by occupation’, showing the ‘science and engineering’ occupation as rated ‘high stress’ by 13% of people — below the average of 15%. Concerning perfectionism, I could find no data to support or dismiss the idea that software developers are more perfectionist than most — but not many people would describe themselves as less perfectionist. Concerning isolation, are developers — with their open-plan offices, stand-ups, conferences, company holidays, and so on — really more isolated than, say, textile workers? According to the studies mentioned above, the highest-risk occupations are not ones I would associate with isolation — “arts, design, entertainment, sports and media”, for instance.
So what’s going on here? Why do developers think they are more depressed than everyone else, while research says the opposite? My theory is that everyone thinks they are more depressed than everyone else because they only know about their own problems — a kind of occupational myopia, or a kind of inverse Dunning-Kruger effect. This theory is supported by the fact that other the same phenomenon occurs in other occupations. For instance, there are lots of posts from the legal profession saying that legal professionals are high-risk for depression — but the data doesn’t support that, either. And here’s a post from an architect’s website dismissing the idea that architects are more prone to depression — so apparently it’s a myth amongst architects, too.
So, a depressing conclusion: software developers might be depressed, but everyone else is more depressed.