I run the annual State of JavaScript developer survey, and I was thrilled to see survey respondents more than double this year from 16,085 in 2021 to 39,471 in 2022.
Yet despite that big increase one thing that remains constant is that only 7% of respondents identify as women, non-binary, or gender-non-conforming.
A consequence of such a skewed ratio is that the opinion of people who identify as men risks drowning out all other voices. So to combat this, I've been focusing on creating new filtering tools to make these voices louder.
Today, I want to show you how I used these tools to dig deeper into the survey's data, and then broadened my research by asking other communities, platforms, and content creators about their own audience demographics.
Two Important Notes About Gender
This post will focus on men vs women, and I am well aware that reducing gender to a dichotomy is both factually inaccurate, and also very unfair to the many people who don't fit into said dichotomy.
Sadly, when it comes to non-binary or gender non-conforming audiences the State of JS survey doesn't provide us with a lot of data. For that reason, I have decided to focus primarily on women as a first step.
Additionally, I know that the vocabulary around gender is often the subject of contention and debate. So if I happen to mis-use a term while discussing these issues or mis-gender an individual, please let me know in the comments so that I can fix it.
Introducing Filters
Our internal API has long supported complex filters, and starting today you can now control them right from the survey's UI:
These new filters work on most of the State of JS 2022 charts, as well as on the State of CSS 2022 report. One thing you can do with them is take a chart, and create variations for different subpopulations in order to compare how values differ among them:
Alternatively, you can take a chart and break down each bar according to a variable such as gender:
You can learn more about these filtering features in our documentation.
An Outlier Appears
You might have noticed the high proportion of women having gone through coding bootcamp as a first step to learning JavaScript in the chart above, and it was while applying that same "faceting" feature to various charts that I stumbled on another outlier:
The e-learning platform Scrimba was apparently doing a much better job of appealing to women than its competitors!
This immediately brought many questions to mind:
- Why were some platforms doing better than others in terms of inclusivity?
- How did other programming-focused communities compare in terms of gender ratio?
- Why was the gender ratio so skewed in the first place?
Why This Article?
When sharing drafts of this article, the most common reaction I got was something along the line of "Why are you doing this? You should focus on improving your own survey's inclusivity instead of looking for excuses elsewhere!"
And I think that's quite understandable. But I want to make it clear that I am not writing this to abdicate responsibility for the survey's own demographics in any way. No matter how generalized the gender imbalance issue may be, I know I still have a duty to try and improve things at my own level.
Moreover, this article itself is actually part of these inclusivity efforts. My hope is that writing about the gender imbalance issue will help me reach the very audience that is currently under-represented in the survey, and if not encourage them to participate in future surveys or, at the very least, make them aware that the survey exists.
With this out of the way, let's get to the data!
Surveys
As I said earlier, from the start the State of JavaScript survey has struggled to attract women respondents.
And even though these numbers are quite bad, it does seem like they are in line with other comparable surveys.
The 2022 Stack Overflow survey reported a 91% ratio of respondents identifying as men) out of 70,853 responses.
And the JetBrains survey found a similar 93% ratio of male respondents.
Things are slightly better over in CSS land, where the 2022 State of CSS survey reported a 88% ratio of respondents identifying as men.
But it's quite possible that all of these surveys share common biases due to their similar methodologies, so I wanted to broaden my research beyond online surveys, which is how I ended up focusing on online education.
Why Educators
Among the users who completed the 2022 State of JS survey, we saw the highest percentage of gender and race diversity in the lower end of the "years of experience" spectrum.
If we can make sure this new generation of developers isn't turned away by entrenched biases, this could prove to be a great opportunity to make the industry more diverse:
So I decided to focus on educational content in the hopes of highlighting that more diverse segment since this content can often serve as an entryway into the community for people looking to learn JavaScript.
Moreover, today's educational content creators also operate at a larger scale than almost anybody else. Recent surveys have made it obvious just how large an audience YouTubers can command, with Kevin Powell for example being responsible for 10% of all 2022 State of CSS respondents!
And education is also a space that shows a lot of promise in terms of inclusion: many women have managed to create a sizeable following by creating awesome programming content, including Techworld with Nana, CodeBeauty, Mayuko, Coder Coder, Ania Kubow, Tina Huang, and many more.
YouTube
Fireship
With 1.8M subscribers, Fireship is by far the largest coding YouTuber around, with popular videos regularly crossing the 1M mark.
As a reminder, the 2022 edition of the State of JavaScript survey got around 40k respondents. In other words, a single Fireship video can offer an audience sample size 25 times larger than an entire survey!
Sadly, the channel's demographics reveal the same bias as the survey, with a 94% male audience.
But maybe Fireship is an outlier? Let's keep digging to find out!
Web Dev Simplified
Web Dev Simplified has 1.2M subscribers, and their demographics stats confirm the trend with a similarly skewed 91% men ratio:
Jack Herrington
Medium-sized channels are no different. Jack Herrington is a popular YouTuber with over 100k subscribers who covers mostly front-end and web development topics and whose videos routinely get over 50k views.
According to YouTube statistics, his audience is 96% male. From the number alone, you might think Jack spends his videos talking about professional football, beard grooming, Joe Rogan, and other stereotypically male preoccupations. But nothing could be further from the truth.
I asked Jack if he was conscious of this issue, and here's what he had to say:
100% yes, and from the beginning I've made specific content choices to address that.
I use they/them pronouns in regards to all generic role types in an organization, including engineers. I very strongly skew against the "brogrammer" culture (which I despise anyway).
In general I just try to visualize that the viewer is by default a woman.
Jack also informed me he intends to do more collaboration with women on his channel going forward and it'll be interesting to see if that moves the needle.
This is heartwarming since it shows that there are at least some YouTubers putting efforts towards inclusivity, but at the same time the fact that those efforts don't seem to have a strong impact is a bit depressing…
Other Channels
As additional datapoints, Wes Bos revealed that his YouTube audience is 96% male, while Kent C. Dodds has a 93% male ratio.
Coder Coder
You might have noticed that the YouTubers featured so far present as men. So what about creators who identify as women? It turns out that Jess Chan's Coder Coder channel has a much higher ratio of women in its audience, with a comparatively low 83.3% proportion of men:
As Jess points out, this is both due to focusing on introductory content, and a consequence of who they are:
When it comes to creating my coding tutorials, the only factors about my audience that I consider is that they are mostly beginner level, self-taught web developers. I don't cater the content to one or another demographic, I just stick to making educational content.
However, as someone who is visibly different from the majority white, male-presenting creators, I feel like just existing on YouTube is my way of representing minority and marginalized groups. In some anecdotal demographic comparisons, I have significantly more viewers from Asian countries and viewers who identify as female. This is no coincidence, because we will naturally gravitate to people who look like us.
TechWorld With Nana
With 742k subscribers, TechWorld With Nana ranks among the largest programming-focused YouTube channels out there, and popular videos can even cross the 1M views mark.
The channel features demographic stats that are very similar to Coder Coder's, with 86.8% of men in its audience:
At this stage it's hard to say whether the two women-hosted channels' more diverse audiences are the result of gender, a focus on introductory content, or both. But it's certainly interesting to notice such a clear separation with channels hosted by men in terms of demographics.
The YouTube Factor
At this point there's something else that is important to highlight: it's hard to know how much influence YouTube itself is having on these stats.
Although YouTube claims that “>50% of [their] audience is female”, the YouTube algorithm can potentially bias video recommendations (for example, it might have decided not to show programming-related videos to women). Also, we don't really know what logic YouTube uses to determine a viewer's gender.
So while YouTube stats can be interesting (especially since few people proactively track their audience's gender stats otherwise), it's also good to keep these caveats in mind.
Note that one way of getting more reliable data is actually through the surveys themselves. For example, since Fireship used a special ?source=fireship tag when linking to the survey, we are able to segment the sub-group of people who followed that link and filter it by gender, which in this case confirms Fireship's own YouTube stats:
More Women Voices
I want to take a moment to address the scarcity of women voices in this very article. I have reached out to more women creators, and I am hoping to include their data in Part 2.
This is also a good place to talk about the lack of women and non-binary video creators in a recent State of JavaScript 2022 question, which raised some concern about the survey's inclusivity.
This was the first year this new question was introduced, and I completely agree that I didn't put in enough work to ensure that the list was accurate and representative.
This was careless on my part and sent the wrong message to an already-minoritized demographic, and I appreciate those who took the time to rightfully call out this oversight. Going forward I will improve the list, and make an effort to clarify and spell out criteria for inclusion for survey items.
Stay Tuned
Stay tuned for part 2 where I:
- Follow-up with Scrimba to know more about their audience.
- Get data for one of the largest e-learning communities around.
- Talk with an independent teacher and creator about how they their personal brand and voice relates to being inclusive.
- Try to identify which "funnel" is responsible for the lack of women in developer surveys.
- Share my overall takeaways.
Until then, let me know what you think of this whole enterprise by leaving a comment here or pinging me on Mastodon – especially if you've got stats of your own you'd like to add to the set!
Update: Part 2 is now available!
Cover image created using Chromata by Michael Bromley.
Thanks to Jeff Delaney, Kyle Cook, Jack Herrington, Jess Chan, Nicole Hiller, Per Harald Borgen, Wes Bos, Kent C. Dodds, Quincy Larson and Josh Comeau for being willing to share their data and feedback.
And huge thanks to Martin Heidegger, Stephanie Tuerk, Jess Chan, Sameera Kapila, and Sophonie Bazile for reviewing drafts of this posts.
Top comments (115)
I wonder, if there was a concerted effort to do outreach to communities of women who program, such as Women Who Code, Girls Who Code, and the many other communities who focus on the demographic, whether more women would fill out the survey. It's asking for a little commitment of time - and many women have many claims on their time, of which this might rank lower than for other groups. In my experience, outreach makes a world of difference
I always try to set aside time for outreach, but sadly –speaking about my own personal experience here– I tend to get a very low response rate to these emails. It makes sense I guess, cold emailing organizations to ask for favors is always a long shot.
Can't say much about cis women, but in trans circles I notice a lot of "thanks for the link but I'm not knowledgeable enough to take that survey" type responses, and it's really annoying sometimes because more often than not that's not true and even if it was, hearing from beginners is valuable information as well (and it's not like cishet male beginners aren't a big part of the respondents anyway)
Maybe a bit more emphasis on the fact that it's not a survey of the leading experts, but anyone who uses the technology on a semi-regular basis could somewhat help in overcoming this, but again, I can't say whether this applies widely or is just a statistical anomaly in my friend group.
I think you already might know this, but you don't have to go this alone. There's a lot of work that can be done just by including others in your process.
I think we've talked before about the inclusion of a11y in the survey, and my offer to help with some of these logistics or just even provide early feedback still applies.
That would be awesome! Would you be able to join our Discord maybe?
This is, at the end of the day, a matter of linguistics; a field that has long ago formed the consensus that words just mean what people understand them to mean, so "gender" can really be anything.
The dogmatic reluctance to accept any change in definition is almost always politically motivated. An attempt to control discourse by controlling vocabulary, and to lend credibility to the non-existence of a concept by refusing to put a label on it.
It has never been a scientifically sound argument, and shouldn't be dwelt on for too long. It's a red herring.
There's many models for defining gender, really.
I like the simplicity of self-identification; and it's probably the most viable on a societal level, as it is by far the easiest to evaluate and has so far proven to have an incredibly small error rate (No, there isn't any noticeable number of people "pretending" to be another gender, that's just not a thing).
But that isn't to say we can't form more specific models of what gender ultimately is, at its core. That's just not the same discussion, because we can't look into people's heads and transcribe what we see onto their ID cards. We also can't look into people's heads to figure out their "biological" pronouns.
So the whole "someone who identifies as x" is really just the practical side of it. It's a rule of thumb to guide our behaviours, based on the idea that other people are both sentient and reasonable, and can probably be trusted to tell us who they are. We can' give better explanations of what gender is. Those are just not practical in everday life.
It's funny because I've probably had this user-name for longer than you've been allowed on the internet. And I'm not just talking about your 1-day-old sockpuppet account with a suspiciously similar naming scheme to the other 1-day-old sockpuppet account in the conversation.
Either way, setting your ad-hom cope aside, a strict binary of biological sex has no basis in reality and none of the "arguments" for it are even consistent, nor conclude in an actual binary.
Chromosomal sex isn't a binary.
Gamete types aren't a binary.
Genitalia aren't a binary.
Secondary characteristics aren't a binary.
Gender isn't a binary.
And worst of all, none of these necessarily align and if you bundle them all up you just end up with more and more degrees of freedom that allow for countless permutations. At that point, best you can do is to identify clusters of data-points. Gender isn't a binary.
Yet you insist that sex is a binary against all evidence. That is dogma.
It's very telling how you're entirely dismissing any possible violence (including sexual assault) between men in prison. This isn't about violence in prison, this is about hating and demonising trans people.
I'm not going to read all of that word salad, and I'm not going to attempt to fix your arrogance. You clearly think you're smarter than all the experts in many fields of science, ranging from biology to linguistics, and without even understanding the foundations of the current consensus in any of those fields. You're beyond help.
It’s been proven that women tend not to apply to a job advert where they don’t meet 100% of the criteria. Men, on the other hand will almost always apply for a job they meet less than 100% of the criteria for.
I wonder if the same effect is at play for survey respondents? I have to admit from my own experience that the State of JS survey is demoralising for me because of how many technologies I’ve never heard of or used before.
It’s ridiculous because I understand the survey is gathering data on up and coming trends, but I can see that demoralising feeling being enough to make a more junior me feel like the survey isn’t targeted at me.
I am of course a sample size of one.
Yes the overall conclusion from all the feedback I'm seeing is that this might be the main problem we need to address. Not quite sure how yet but it's encouraging to at least know in what direction to look!
Yep, this sounds familiar. I sent the link to the state of JS survey to a friend who has done some JS stuff and her response was something along the lines of "I don't know enough to take that survey". So that could be a fairly common problem.
Here's my theory - studies have shown women have higher workloads than men, both professionally and at home. They don't have time to be sitting around watching all these YouTubers and whatnot because they've got things to do. I only know about these surveys because a coworker shared them in a group chat years ago and now I get alerts when new surveys come out, so I take them. The dev teams i've been on have been about 60% male/40% female so every time I see ~7% of respondents are female, I chuckle a little and wonder why.
That's a good point! I think having children especially puts more stress on women compared to men. I know that studies have shown that kids impact women's career more than men's overall, and I wouldn't be surprised if that same kind of pattern also affected the demographics studied in the article.
A similar point came up in a discussion about why there was a majority of male applicants wanting to speak at conferences and meetups. The organizer said they observed that men were more likely to abandon their chores and care work in favor of preparing a talk, prioritizing their ego over family duties.
That's just a cognitive bias. The environment you are in does not represent whole industry.
It would be a cognitive bias if I claimed the whole industry had a 60%/40% split (assuming that is what you are referring to). I was simply stating I have always found the difference between the results and my experience interesting. A quick Google search shows the actual breakdown is more like ~75%/25%, which is still far off from the survey results.
Yes according to the data gathered by Sacha, a perfect balance is probably not achievable, as the distribution at industry-level is itself skewed. In the short run, I think we should aim for a minimum 10% of non-male respondents to produce more representative results, and ideally start a virtuous circle.
This may be a bit long, but just giving my experience as a woman in the swe industry.
As a woman who's been a swe for over 4 years, I got in as a career transition from academics. There could be a combination of reasons for why there's less women in the swe industry, and I think one part of it is potentially that the field of swe did not seemed to be as inviting, like math wasn't as inviting to me growing up. Most of the guys in my classes were the top performers in maths, and I always thought I'm like Amy Tan and my skill was writing. But later on I found programming and I really enjoyed it. Most of the people I worked with early on was mostly males, and because I play a male dominant sports, I was used to be one of the very few females in a group. But I would say that in terms of mentorship I got a bit of it, but it did feel that during my time getting into the field, it was both a battle of filling in gaps of knowledge, adjusting to corporate culture, adopting a new identity of being a "technical" person, and also trying to push through even though there was little representation of my gender and/or race. But I still haven't given up, and if it weren't for those early years 5-6 years ago when online learning took off, and focus on teaching folks how to learn became a thing, I wouldn't be where I am now. And I hope that more people with empathetic mindsets and technical inclination do step up into the field because it opens the doors of swe industry from what I earlier on perceived as a simply cold calculated field to one where it's about innovation and creativity.
Thanks for sharing this. I like your point about online learning, it's something that has really taken off in the last decade or so and it might take a little bit more time before we see its effects.
But why would the natural conclusion of all this be "we need separate prisons for biological men and women"?
If there are opportunities for sexual assault because inmates do not have their own individual safe spaces and have to share a room for example, isn't that the actual problem we should address, i.e. prison overcrowding? Men can sexually assault each other, and so can women. Can you see why the fact that you specifically chose to stigmatize trans women for it can come across as bigoted?
I see you're more interested in asserting your dogma than finding the truth, so I see no reason to continue this discussion
Thank you for taking the time to do an in-depth analysis of the gender ratio of your surveys. I've been taking the surveys for about five years now, and every year I am painfully disappointed by the ratios, but not surprised given that the ratio accurately reflects my current working environment (sigh, AWS).
In regards to the YouTubers in particular, the sample size of me and a few of my past female colleagues says that we don't generally gravitate towards video content. For me personally, I just don't have time to watch anything during the workday, and if I want to learn something new, I'll choose the blog post with coding examples I can copy/paste and try out over any other medium. It's the same for podcasts - I can't just listen to something about JavaScript to learn, I need to have my VS Code open while I'm absorbing new information so I can put it to use and solidify the knowledge.
I'll confess that I don't really watch coding YouTubers either, even though a lot of them put out great content, probably just for lack of time.
But anecdotally, looking at the State of JS data, while some video creators have low women ratios, others have about the same 5% ratio as the survey overall – possibly indicating that it boils down more to individual YouTuber style more than a gender-based video-watcher vs non-video-watcher divide?
For sure! I guess my aim in pointing that out would be to encourage looking into the gender ratios of other mediums, like those who follow tech bloggers or documentation, to get a more holistic picture.
I'm just observing that there is a severe under-representation of women in tech, and the only reason that might be is that there's something keeping them out. It's that keeping out part that I have a problem with, not so much the actual demographic.
As for bending things to my will, you seem to have it all wrong. Most people don't want to exclude women or non-binary people from tech. The ones trying to bend things to their will are people who want to prevent inclusion from happening.
What even drives you to an article about a topic that you supposedly don't have any interest in, just to point out how much it annoys you? I'd get being indifferent to women in tech, but why are you being actively offended by the idea?
Privileged people (like me) often (used to) take their situation for granted but still subconsciously act to sustain it. As a white male tech expert, it might feel natural and productive to hire for "culture fit", but that means hiring more white male dudes and bros instead of a more open minded outlook for diversity and equality with the benefit of bringing new perspectives and inspiration as well.
As I have pointed out, for the purpose of society and law, self-ID is the easiest, most accurate and overall most viable way of defining gender. So no, any other definition is irrelevant and purely a matter of philosophy, which is not a discussion you seem equipped for.
You blatantly deny any progress science is making in this field and pretend none of it exists, as long as it doesn't confirm your dogmatic faith in the binary nature of sex and its 100% overlap with gender. Just because you don't want to understand, doesn't mean the rest of society has to wait for you. Catch up or get left behind.
No and this is a non-issue. Everyone agrees on this. You're just trying to frame your actual point as more reasonable, which is that you disagree whether sex or gender is a better metric in some specific circumstances. But that makes it sound like both sides might have valid points, so you'd rather pretend there's some gender conspiracy that wants to abolish the concept of sex.
Yes. In medical contexts, for example, it is important to distinguish people's biological realities more so than their gender. This does not mean that you shouldn't do so while respecting their identity, though, so going out of ones way to, say, call a trans man a biological woman is still morally wrong, just like calling someone an insignificant sac of meat is biologically correct, but very clearly an attempt at being offensive.
Yes. There is absolutely no ground to stand on to attack this. No people don't believe they're literally cats, no matter how much people insist on this weird fantasy they made up to attack trans people.
In 95% of cases it means she's a woman, end of story. Is there any reason for you to know what's between her legs? No? Then it's none of your business. The burden of proof is on you if you think it is somehow relevant what their chromosomes say.
Yes. That's the point of gendered dressing rooms, in so far that there even is one outside of a heteronormative and sexist idea that women need to be protected from male sexuality and men need to be protected from the temptation of female bodies.
Yes. This isn't as much of a problem as people make it out to be. Actual age will still be real as much as biology is still real despite accepting people's genders. And just the act of identifying as a different age, or hell, even as an attack helicopter doesn't impact you whatsoever. Nobody is going to sue you over not exploding after they made helicopter noises and gestured launching rockets at you.
If gender is entirely subjective, then there is no error rate. If gender is some sort of unmeasurable but real phenomenon in the human mind, then there can be an error rate (and the error rate will not be measurable either, so we'll only ever have a rough idea based on our intuition on individual cases).
Gender is not unrelated from one's sex. They are two distinct but adjacent phenomena. Nobody outside of tumblr is denying this. This also does not conflict with the existence of gender identities not based in biological sex.
The same way you know they're not all plotting to murder you: By acknowledging that you matter way less to people than they themselves and if they're telling you something about themselves it's vastly more likely that it's actually true than an intricate lie to hurt you in some way, specially if that lie isn't even necessary.
Because most people aren't criminals. As for your article, sorry but articles aren't science. And I'm not going to do your work of finding the actual study and reading it only for it to be the 100th time some "data" turns out to either a) contradict the vast majority of other studies or b) turn out to be so fundamentally flawed that only a transphobic cretin could ever take it seriously. This is how you people strategically waste our time.
So anyway, closing statement:
You're wrong, and this sea lion wants its job back.
Some comments may only be visible to logged-in visitors. Sign in to view all comments.