Stack Overflow’s developer survey analysis hurts women

Note: Graciously, Stack Overflow has corrected many of the issues discussed in this piece. Many of the quotations you see in this article are no longer a part of the live survey findings. Read their response.

The findings of Stack Overflow’s 2016 developer survey say a lot more about the internal culture and biases of the company than about the current state of developers.

Let’s start with some numbers. Only 5.8% of survey respondents identified as women (or, more specifically, “female”, because apparently Stack Overflow wants to know what kind of genitals you have). Industry representation for 2016 according to NCWIT is more like 25%. While Stack Overflow does hedge a bit by speculating that their registered user base is closer to 10% (ETA: since this writing, they have updated the report to say 12%) women, I’d be willing to bet that women who use the site without an account form a much greater portion.

This raises an interesting question: why aren’t there more women registered on Stack Overflow? A couple of reasons I’d stipulate would be that women (and other minorities), on average, have less time outside of work to participate on forums such as these, and are less willing to spend productive work time responding to questions on a public forum. The culture of the site also encourages hostile corrections in the form of mansplaining and “well actually”-ing from other users, which many women and other underrepresented groups find unwelcoming. I personally remain an unregistered and passive Stack Overflow user for these reasons.

This is all to say that the woman survey respondents make up a fraction of a fraction of a fraction of Stack Overflow’s actual woman user base, and yet Stack Overflow generalizes about women in tech at large.

Sweeping Conclusions

There’s no way to sugarcoat it. Lots more men are writing code than women. (ETA: this quote has been removed from the survey. It originally appeared in the section linked above.)

When I read this, I hear that Stack Overflow doesn’t believe that I exist.

Sure, lots more men responded to your survey than did women. There is no doubt that fewer women are programmers than men, but as I shared above, the situation isn’t nearly as dire as Stack Overflow suggests (and seems resigned to).

women are on average less experienced than their male counterparts.

What I really hear when I read this is that Stack Overflow believes that women are not as good at engineering as men.

As my friend Jessica Paoli points out, couldn’t this also mean that women are entering tech more rapidly than men? The data also suggests that women of a certain age or experience level leave tech—why might that be? This question isn’t even raised.

Women are more likely to be Designers than any other developer type

It is a well-documented phenomenon that men assume that women at technical conferences and meetups have little experience or are not a part of the field at all. Data like this just reinforces the stereotype. Because design and even front-end development are not considered to be as technical as server-side engineering, it is assumed that women fall into these occupations, and they are often undervalued. (Fun fact — at a meetup I co-organized, I was presumed to be someone’s girlfriend. On other occasions I have been presumed to be a recruiter, marketer, or designer; all fields which I have never worked in.)

Surprisingly, there was no mention that 6.1% of the woman respondents were machine learning developers—a highly technical and rigorous field—versus 0.1% of developers who responded on the whole.

As another treat, “Female Developer Age” has its own section, while men are not similarly scrutinized. From that section:

We can’t claim to know the answer to bridging the gender gap in tech, but we think sharing data with the public is a productive step in advancing more dialogue across the industry.

Notice that the tone here is noble and lofty, as though Stack Overflow is doing us a favor by sharing this. The biases inherent in this survey do actual harm under the guise of progressiveness and good intentions. I shudder to think what further sexist and inaccurate conclusions will be drawn about me and other women [by people who have no business analyzing data] from this expression of generosity.

Questionable Questions

Stack Overflow reinforces monoculture and stereotypes by asking respondents whether they prefer Star Wars or Star Trek, or if they self-identify as ninjas, rockstars, or surprisingly—gurus. The design of these questions excludes anyone who does not know or care about science fiction and condones the use of culturally appropriative terms for engineering prowess, a subtle signifier that developers who do not share the same interests and cultural backgrounds are not “real” developers.

My Conclusions

I don’t see the women I know and work with reflected in this survey. We are a highly competent, experienced, and well-rounded bunch. Some of us like gaming and science fiction, but just as many of us don’t. We come from many different cultures and backgrounds.

All I could think about when I read these findings is how many men in tech will use this data as a reason to stop caring about diversity—or even to believe that achieving it is impossible and not worth striving for. Their misperceptions are only confirmed when unrepresentative data with inaccurate and biased conclusions are published.

It’s quite simple. If you deny my existence, there’s no way you can imagine me being your peer, let alone your superior. There’s no way you can respect my contributions or my perspective. And this is one of many reasons why women aren’t more represented in the field.

Stack Overflow, you have given men in tech ammunition for trivializing diversity outreach in their organizations. You have given them a reason to continue believing all of the sexist and untrue things they believe about women, and use against them to deny jobs, promotions, raises, speaking gigs, and leadership roles. I cannot overstate how irresponsible publishing this data and these words—especially with your reach and influence—was.

If you really want to present factual data and accurate conclusions, might I suggest you read up on confirmation bias and survey design?

Paradoxically, you would have done us more of a favor by saying nothing about women.