Results of Stack Overflow survey of 20,000 developers

Seemed like an awfully large amount of developers make less than 20 grand.<p>I did some analysis that I think is a little more useful.<p>Compensation &amp; Job =&gt; filtered by &#x27;US only&#x27;, &#x27;not looking for a job&#x27;, and &#x27;provided a job&#x27;. <a href="https://www.statwing.com/demos/dev-survey-2#workspaces/18770" rel="nofollow">https:&#x2F;&#x2F;www.statwing.com&#x2F;demos&#x2F;dev-survey-2#workspaces&#x2F;18770</a>

The data itself is really interesting! But the UI, gee, lots of frustration with tiny scrollbars, things not fitting on my large monitor, mal-aligned tables if scrolled right, gigantic help screen appearing everytime you close something, &quot;&gt;&quot; and &quot;$&quot; of the values used in alphabetical sorting, ... I&#x27;d prefer oldschool HTML tables over this...

Some things I noticed:<p>- 55% of programmers surveyed are making less than $20k&#x2F;yr, which is ridiculous.<p>- 4.9% of programmers surveyed are women<p>- 2.7% of programmers surveyed are 51+<p>- 17.5% of programmers earning less than $20k&#x2F;yr live in California.<p>That last fact caught my eye, so I did some math:<p>This article (<a href="http://www.wired.com/2014/04/no-exit/" rel="nofollow">http:&#x2F;&#x2F;www.wired.com&#x2F;2014&#x2F;04&#x2F;no-exit&#x2F;</a>) says that the author pays $1250&#x2F;mo &quot;for a mattress on the floor, behind a panel of imbricated torn shower curtains, in an unheated rabbit warren of 20 bunk beds under a low converted-­warehouse ceiling.&quot; Assuming they all live in the same space that the author lived in, they spend $15000&#x2F;yr on rent alone. Assuming that they have health insurance, a cell phone plan, and pay $3&#x2F;day for transportation, they have (and I swear I&#x27;m not making this up) about $3.50 per day to spend on food.

Aaaargh. This really frustrates me. I would rather have a bunch of text files and a link to the R manual.<p>Really, this is Stack Overflows annual &quot;OK Cupid&quot; moment and instead of a really linkable blog post with two gee-whiz insights I get &quot;here you go, find an insight by spending two hours learning a UI you will never need again&quot;<p>Did I miss the blog where they had already looked at the data themselves?

After 6-10 years of experience, some bunch of people wind up self-reporting as &#x27;full stack&#x27;.<p>After 10 years, this declines and they become &#x27;Manager &#x2F; Team Leader&#x27; or &#x27;Other&#x27;, categories which previously did not self-report very much.<p>Not super fascinating, but it confirms my own path (15 years commercially and counting).

This is really cool - I found it interesting relating the &quot;job satisfaction&quot; statistic with various other things like &quot;hours per week: fixing bugs&quot; or &quot;hours per week: meetings.&quot; It goes pretty much as you would expect except that the two extremes seem to have the highest satisfaction. The only one that was exactly like I would have expected was &quot;hours per week: technical support&quot;

Did they do some filtering of their users or just raw data? In that case they asked 20000 <i>stackoverflow users</i>, and their results will not be representative of developers as a whole.

I wanted to make fun of it too but started playing with the data and found it to be really easy and actually a lot of fun.<p>How dare they ruin a good cynical take.

Full time remote workers have the highest average salary. That&#x27;s an interesting data point. Correlation is not causation obviously, but it&#x27;s interesting that the more time you spend out of the office, the more money you make. (Compensation plus bonus)

Compare OS and compensation. Apparently all the highest paid developers use Windows 7.

I enjoyed &#x27;hours surfing the internet&#x27; for those &#x27;currently looking for a job.&#x27; If you have enough time to stay employed and surf the internet: you probably just feel like you&#x27;re employed to receive a paycheck. Not challenged at all and seeking adventure in a new shiny grass field. Let&#x27;s read about those greener fields (or burn time) instead of acting on it.

Would love to see some analysis of the data. Too lazy to do it myself.

Only ~13k of the 20k reported a StackOverflow reputation, and none of them had more than 10k rep? That seems odd.

It&#x27;s a fun data set to explore, but without the ability to compare interaction effects the UI doesn&#x27;t allow you to get very deep into the data. Also, it&#x27;d be nice to be able to filter out the highly correlated tables.<p>Otherwise a cool presentation, I particularly like the little arrows for significance p-vals and CI tooltips.

A bit off-topic but the page looks really heavy and responds slowly on a rather powerful machine that I have. Any idea if we can get a csv or raw data of this?

I was surprised more developers used Linux than OS X.

I find it interesting that developers at small companies (1-25) are apparently the happiest, but also spend the most time surfing the web.

Can someone give interesting stats out of this mess ?

statwing rocks...