loading...

Some Things You May Have Missed in the Stack Overflow Developer Survey

remotesynth profile image Brian Rinaldi Originally published at remotesynthesis.com ・6 min read

Yeah, I know what you are thinking, "The Stack Overflow Survey was last month's news!" True. But there was so much data to digest in there that, honestly, it takes a lot of time to process. This meant that many of the early posts focused on headline info, like the most loved/dreaded/wanted languages. Now that I've had a little bit of time to sift through all the information, I wanted to share some things that struck me as important, even if they didn't necessarily catch the biggest headlines.

I should note that in all cases below, when I refer to developers, I mean respondents. Obviously, even though over 100,000 people took the survey, it probably has some bias due to the fact that it is heavily tilted towards at least somewhat regular StackOverflow users.

We are not diverse - like, at all!

The demographics section overall was disappointing overall. This isn't helped by the fact that a substantial number of developers don't think diversity is a priority (including both men and, surprisingly, women. The net result is that the prototypical developer is a white, male under the age of 35 with no children.

We are almost all men

Developers are, sadly, a very homogenous group - almost 93% male, over 74% white, and over 93% straight. Compare that to the general population that is obviously slightly over 50% female.

And all the diversity efforts to bring more women into the industry doesn't appear to be changing things in any real noticeable fashion. Looking at just students, the numbers shift by less than a percent. The only potential bright spot is that women represent a growing share of developers with less than 5 years experience.

We are mostly white

The racial breakdown is especially disappointing since 63% of the respondents are from North America and Europe. Nonetheless, over 74% of all developers identify as white or of European descent. The geographic breakdown leads one the conclusion that, in North America and Europe, the breakdown would be significantly more white than the overall percentage. This bears out in the commentary, which notes that "7.4% of professional developers in the United States identified as black, Hispanic or Latino/Latina, or Native American" compared to over 30% of all respondents. On a positive note, this number improves slightly when reducing to only students - so change is happening in this area, but very slowly.

We are mostly young, without children

Almost 73% of developers are under the age of 35. Personally, I am on the cusp of falling into the 45 or older category, which represents slightly less than 7% of developers - and I can attest to my age already coming up at most every developer event I have attended over the past 5 years. Slightly more than 71% have no kids, which matches pretty closely with the under 35 category.

Web technologies rule everything...

..but the paycheck

JavaScript, HTML and CSS rule the top languages used by developers making up more than two-thirds of respondents, with JavaScript being the most prevalent of the three (the numbers even go up slightly for all three when filtered for just professional developers). Also, a combined 64% are using either Angular or React (with Angular showing up as modestly bigger) and almost 50% say they are using Node.js.

However, popularity doesn't translate into pay. Web technologies are at the bottom of the pay scale (both globally and in the US where only JavaScript makes the list at second to last). If you want to make money, it appears to be in the less widely used back-end languages like Erlang, Scala, Ocaml and F#. Front-end developer as a title sits somewhere in the middle of the salary stack overall.

Mobile is still a (large) niche

The native mobile contingent wasn't quite as large as I expected. Swift is at 8.3% and Objective-C is still at 7.3%, which, when keeping in mind that this category allows overlap, means that very likely less than 10% overall are doing iOS development (which would fit with the XCode numbers which sit around only 10% and the correlated technologies section shows there is indeed some overlap), although 15% did say that they are developing for iOS as a platform overall. Android isn't broken out of Java in the languages section, but the IDE section shows Android Studio at 19.3% and platforms show it at 29%, which is a bit unusual given the dearth of native iOS developers. Since well over 50% of respondents doing mobile were doing Android, I wonder if there might be a bit of a Java/Android bias to the StackOverflow audience.

Other mobile-focused development technologies were also relatively small, with Xamarin and Cordova showing a combined 15% (since I'd assume minimal overlap between the two). Tools like React Native or NativeScript did not make the list. Unfortunately for Xamarin and Cordova is that they also both top the most dreaded tools/frameworks list.

Back-end development is all over the place

Java is still the king when it comes to back-end development, but Python and C# are not that far behind. A lot of people (31%) still do PHP, and it isn't just Wordpress (less than 16% use WordPress). Significant numbers of developers do C and/or C++ while small but not insignificant percentages do Ruby and Go. The numbers would indicate that many developers work with multiple back-end languages.

Rust and Kotlin developers really love their languages

Rust and Kotlin developers are clearly a small but passionate group. Rust didn't even make the language list and Kotlin came in at only 4.5%, but they significantly outpace most ever language in terms of the developers who use it wanting to continue working with it. Rust knowledge is fairly high up on the global pay scale, so perhaps that is a factor.

Overall, developers generally want to learn Python or JavaScript/TypeScript and want to get away from some older .NET technologies in particular. However, both Python and JavaScript are among the lowest paying languages.

Developers generally like their jobs

This is a bright spot. Over 70% of developers are at least slightly satisfied with their career with a majority (55%) moderately or extremely satisfied. Those numbers are essentially unchanged in terms of satisfaction with their current job. However, a significant number of developers changed their job in the last year and a majority (56%) within the last two years. So, perhaps that job satisfaction number is ephemeral.

Perhaps part of the reason developers are happy is that the job pays well overall. Even the lowest paid category in salary by developer type in the U.S. is paid about $20k more than the average salary of a person with a four-year college degree even though just about 75% of professional developers have a college degree.

Posted on by:

remotesynth profile

Brian Rinaldi

@remotesynth

Brian Rinaldi is a Developer Advocate at Stackbit. Brian has worked for over 9 years focused on developer community and developer relations at companies like Progress Software and Adobe.

Discussion

markdown guide
 

// , The diversity question may have more to do with income brackets and related behavior than (direct) racism as a proximate cause, especially since StackOverflow makes pseudonymous participation without a photo easy. I used a squirrel photo for the first couple of years. Maybe they could do an experiment?

As to your note about the student demographics, I have seen some of the weirder causes of it in person.
And it calls to mind the Scandinavian paradox.

This may be part of two broader trends:

One trend that could be part of the cause is that young USA citizens don't see much benefit from pursuing careers in technology, especially where money and social status are concerned:

economist.com/node/21586624/all-co...

Another trend that could be part of the cause is the proportion of people who enter computer science through Video Games, which make much of their money by exploiting the competitive inclinations of young men.

These (admittedly limited) studies and their cited studies examine that:

researchgate.net/publication/22862... (No Control Group)

researchgate.net/publication/27471... (Observational study, correlation only)

From what I've seen in person, best and brightest of my CS classes were the laid back guys who had done all the homework 2 months ago, and spent the rest of the semester in the back of the class hacking their NVIDA cards to optimize their computer games, trying to hack each other's cell phones, or doing something mysterious looking on a couple of xterm windows, if they even bothered to show up for class at all.

The lowered social status of STEM among young people,and the demographics of competitive computer games, as combined filters of interest and motivation, might go some way toward explaining the phenomenon.

 

Being a lot on Twitter lately, I am wondering if these stats are accurate when it comes to the dev world in general, or demonstrate what happens particularly in Stack Overflow.

I do agree that we do need more diversity, but I think a lot of the groups you mentioned are avoiding posting in Stack Overflow due to the attitude there. I remember I tried to post a bit in my early career days but after a few "Well actually" answers, I gave up on it, and only used it for finding information rather than being part of the community, as I didn't want to risk my (already kinda low) confidence to try to take up space there. I didn't respond to the survey either, I didn't even realise when it was on. I m still a dev though, a female one, with a bit more than 4 y of experience.

I believe it could be the case that more people are like me, they just don't see Stack Overflow as a community they could contribute in. Wondering if that raises the question if we believe it is an inclusive community and how much we want it to be.

 

This is a very good point. As I mention in the intro, when I say developers here it means respondents - so any bias in the population of responses will impact the results. As you mention, due to the nature of the community, there is likely some bias in terms of the demographics. Some have argued that they think that Stack Overflow audience skews younger and less experienced while, as you mention, the audience could also skew more towards white males.

We can't really know how close to the truth the data is since, unlike political polls, there is no election or other final result that proves the data right or wrong. That being said, it is one of many data points (though an important one given its size and scope) that reflect a lack of diversity in our community in general. Perhaps it is off by some percentage points here and there, but I'd be shocked if the actual numbers are hugely different. For example, might the overall developer population be more like 10-15% female? That certainly doesn't seem outrageous, but nonetheless I think the data here on that subject just simply reflects that, even if we take potential bias into account, we are nowhere near where we want to be or should be.