The State of Devs 2025 survey results are now available, and they contain quite a few interesting insights! I encourage you to check out the whole thing for yourself, but in the meantime I thought we could explore some of the data together—and maybe learn a little about statistics in the process.
Zelda Players Earn $30,843 More on Average Compared to Minecraft Players??
Let’s start with an “insight” that well… isn’t really one! You’ll see what I mean.
When talking about surveys or scientific research, you often hear that “correlation is not causation”. But what does that mean exactly?
Here’s a concrete example. It turns out Minecraft players earn way less on average than developers who play Tears of the Kingdom:
In other words, a developer’s income is correlated with their favorite game.
But does that mean that switching from Tears of the Kingdom to Minecraft will magically results in a salary increase? Of course not!
What’s going on here is that the median age for Minecraft players is 27, vs 35 for Zelda players. And naturally, older developers with more professional experience earn more.
Now we could draw a causality link between both age and income on one hand, and age and video game preference on the other. But even that would purely be a hypothesis informed by our pre-existing knowledge about the world, and not something the data can prove one way or another.
So whenever you’re exposed to any kind of statistical data, keep in mind that:
- Correlation is not causation.
- Statistics can only show correlation.
Engineers Earn $44,939 More on Average Compared to Developers??
The previous example was easy to debunk, but let’s look at something a bit trickier. It turns out, job titles containing “engineer” in them carry quite a premium!
So what’s going on here? Do engineer positions really pay that much better, even though the e.g. “frontend engineer” and “frontend developer” are virtually synonymous?
U.S. vs The World
Before we can advance a hypothesis, we need to consider an important variable that has a huge impact on income: respondent country.
It turns out, U.S. respondents earn a lot more than any other country:
And while positions containing the word “engineer” only make up 30% of responses worldwide, they represent 56% of responses in the U.S.:
In other words, the fact the engineers earn more than developers could be due at least in some part due to the fact that a larger proportion of engineers live in the U.S.–and all programmers earn more in the U.S., no matter their job title.
Looking at the U.S.
But if respondent country is the only reason for this income gap, we would expect it to disappear when controlling for the respondent’s country.
And when excluding the U.S. from the dataset, the gap does shrink quite a bit:
Yet somehow it’s still very much present when limiting the data to U.S. respondents:
And just to eliminate one more variable, the gap also exists even when comparing the exact same position (“Frontend Developer” vs ”Frontend Engineer”).
But Why?
At this point you’re probably waiting for me to reveal the big reason why engineers are valued so much more, at least in the U.S. Is it because of the certification? Differing job descriptions? The fact that “engineer” just sounds cooler?
Sadly, this is where this kind of surface-level statistical analysis shows its limits, and where real, on-the-ground research would be needed–a.k.a. what actual researchers do.
Because even though I might play one on TV, at the end of the day I’m not a data scientist. I’m just a regular frontend developer–I mean, engineer–with an affinity for charts and graphs.
The Query Builder
But I’ve always believed data scientists shouldn’t be the only ones that can have all the fun. This is why all the charts I’ve shown today were created with the survey’s own built-in query builder:
The query builder makes it super easy for anybody to dig deeper into the data to find interesting correlations without having to learn data processing tools or import the whole dataset, and I encourage you to try it out!
Discover the State of Devs Results
This whole article was just one big long preamble to encouraging you–now that you have a solid understanding of what you should or shouldn’t conclude from survey data–to explore the survey results by yourself.
So go ahead and get lost in the data, then let me know on Bluesky what you found!
