We all know that the impostor syndrome is really common between developers. We've all been there. But what if the data don't show us the same results?
Are we really under-confident?
While reading the 2019 Stack Overflow survey, I came across this seemingly simple question:
For the specific work you do, and the years of experience you have, how do you rate your own level of competence?
The question featured these 5 possible answers:
- Far below average
- A little below average
- Average
- A little above average
- Far above average
If you haven't completed the survey, think of your response for a couple seconds before reading further.
I have to be at least as good as the average. Why not?
That was probably what I thought the first time I saw the question. And there's something funny about this: it seems a lot of people made the same assumption as me.
Almost 70% located themselves above the average, but just 10% considered themselves to be below it. And even 1 out of 5 people thinks they are far above.
Stack Overflow's ironic analysis made me laugh the first time I read it:
This is statistically unlikely with a sample of over 70,000 developers who answered this question, to put it mildly.
What's happening here?
The above-average effect
What we see in these results is a common consequence of the above-average effect: a cognitive bias that makes us overestimate our own competence in relation to other people.
It is closely related to the Dunning-Kruger effect. I don't wanna dive deep into this right now, but you can always read about it here.
In a nutshell: we tend to forget that statistics also apply to us.
As much as we all may want to be above the average person, we just can't. If we are all above the mean, the mean wouldn't be the mean. See what I'm saying?
There are a lot of other examples of this bias, E.g. how we all tend to think that we are better drivers that anyone else (and I'd have to plead guilty on that one), or how we think that our startup will succeed even though the records show that almost 9 out of 10 fail.
Addendum
I don't wanna forget to mention that men are more likely to consider themselves to be above the average than women and non-binary, genderqueer, or gender non-conforming people.
I can't say I didn't see that one coming. In next surveys, it would be awesome to observe that people from gender minorities started to feel more and more confident about themselves.
What do you think? Do developers tend to be overconfident? Are you the above average?
Let's discuss it in the comments.
Top comments (11)
You caught me. I think part of the problem is that we compare ourselves against the limited scope of our experience. I think well I'm good at that or well x people come to me for help, so therefore I'm a little above the average of the people I talk to day to day. In reality, there are lots of people we don't interact with so we don't see the actual average. It becomes relative, based on time in the chair or titles that had more to do with HR than us.
I had some similar thoughts and questions today too...
Am I (not) a mid-level dev yet?
Austin Standing ・ 1 min read
Well that's really interesting: how do we define "the average" anyways? I think it's better to keep growing personally and professionally without looking that much to what the rest does.
Definitely.
That's not exactly true. The median competency would have to go up because there's by definition 50% of the people above the median and 50% below it. Now it's entirely possible that only 10% of people are below the mean competency if they're really really bad coders.
Now that begs a more relevant question in my opinion : How the hell do you determine mean competency and median competency ? And what does that even mean ?
Hi Nico! Of course that it's entirely possible; it does not represent a statistical impossibility. However, as I quoted above,
The 70% who considered themselves to be above the average would have to be really close to it, whereas the 10% below should be extremely below. That's why I disregarded that option.
I totally agree with you in that you cannot determine competence just like that. And I don't think that it would mean anything, in any case.
However, I just wanted to point out the bias and to contrast this results with the well known impostor syndrome.
Funny this was something that caught my eye as well. I believe our industry is in an interesting spot right now where the work is in high demand and it leaves a lot of us feeling very valuable with higher than average salaries in an industry that you don't exactly have to have a formal education to succeed in.
Maybe the demand is so high that if they don't value your work in Company X, you know that there's some Company Y that will definitely hire you. Given that, is easy to feel proud of yourself and the work you do (and sometimes that's good too!)
Since skills are widly spread in tech, I don't think an all-in-all average makes any sense. How would you compare let's say a dev at a start-up with widespread but shallower skills (vulg. Full-Stack, don't @ me), with a dev at a larger company who is an expert in one field. Both may deliver similar value and earn the same.
Even with certain technologies. Do you count knowledge of syntax, or the produced result?
Do you look at the artist's neat brushwork, or their cultural impact?
In the end it's a gut-feel. At least we know when we really suck at something and if we usually help people out with things, we might be a bit above average.
My wild guess is that I'm well above average at CSS and average, maybe slightly above on JS and HTML. Definitely far below average in Elixir. And the worst person on earth when it comes to any testing.
Nice analysis! It's really hard indeed to measure competence (and competence in what exactly, anyways?).
I think the most interesting side of it is the cognitive bias we all experience, regardless of the way of measuring competence.
Stackoverflow surveys are the worst indicator of any genuine data.
Could you please elaborate on that? I read your post about how you don't think stack overflow surveys really represent technological choices.
But how is that related to this?