Discussion
RE: https://mastodon.social/@grimalkina/116581190573338371
I am keeping my promise and downloading this data right now. We will see if I make it through anything interesting.....we're having some turbulence on this flight :)
You can download the jetbrains survey raw data here! I will be downloading it for my flight to Toronto!!!
https://devecosystem-2025.jetbrains.com/#methodology-and-data
I should read this against the methodology page actually. I wish there were a script to quickly replicate their cleaning and weighting here, maybe that'll become clearer, because there's a lot of data cleaning choices listed at a high level. But that's ok we can make our own choices too!
not me staring at this dataset of survey weights trying to remember the last time I weighted responses
hmmmmmmmmmm not exactly a construct map in this folder, but we're going to guess that parent groups are the construct
A huge number of these items are characteristics: psychology measures are the most FUN though so let's try to find those
There are some items about emotional exhaustion, stress, and being overwhelmed, really curious what we could do with those
they highlight some national diffs like this that I'm really curious about
I feel like I'm going to have to download a vapoRwave theme though 😂
y'all if you are waiting on this thread to be something interesting CHECK BACK LATER, clearly this flight is just going to be me updating my packages and trying to figure out weird conflicts 😭
R, holding an extremely long cigarette in a cigarette holder, gazing out the window as I stand in the doorway with my hat in my hands: oh you're BACK? from your NEWSLETTER and KEYNOTES or whatever?
while R is being all "I'm not mad I'm just disappointed", I want to look back at the psychology measures in this survey. Something complicated here is some of these groups contain a more global mood ("I experienced feelings of low mood or sadness") but then there are specific attributions ("I feel overwhelmed by the constant need to learn and adapt to new technologies in my field"). You could feel overwhelmed by other things.
Think we need to think about collinearity here. This is where constructs are useful! Could be nice to use the global affect measure to group people and then compare some outcomes for subgroups. Some imposter syndrome-type measures here too which is potentially just interesting to map
OH. A RATING OF ROLE-RELATED IDENTITY CLOSENESS/SALIENCE. INTERESTING. That is a really cool potential grouping variable. ("How much do you align your identity to your job")
the social identity theory psychologist in me jumped OUT when I saw this
in my headphones Espresso started
Magic moment
this & imposter syndrome* made me remember to check what % of this survey is men and it's uh, 92%
*often has gender diffs
*something interesting about gender diffs; in my studies, professional developer women are much younger samples than men. This is something you want to think about to not accidentally conflate "younger" with a gender effect
Which you wouldn't do in more complex statistics but you know, this is a reason simple descriptive stats on one variable can mislead
My ability to really estimate this is complicated by the fact that I always make my demographic items optional though; so it greatly reduces the sample size. Something I debate a lot but I don't want to make inability to feel safe answering a demo question a reason to not be counted in my studies.
oh you thought you were going to get to see DATA in a thread like this, jokes on you social scientists spend 99% of their time thinking about the FORMAT of data not the data
oh I found my problem
I forgot what I named the data and tried to call it by a different name
*looks directly at the camera*
I've given a lot of talks this week
hang on I need to go leave a github star on the vapoRwave repo now that I KNOW ABOUT GITHUB STARSSSSS
So uh, mental health familiarity by country.......what's going on*
*I know what's going on. Really big culture*gender differences on this probably (don't forget this is a 92% male sample)
somehow we're mucking around with fonts now because that's important to The Aesthetic
R, standing in the middle of the room inside of the vaporwear I've dressed her in, looking the way my dog looks when we give him a bath, staring into the middle distance
OH MY GOD WE'RE LANDING SOON????????? BUT MY GRAPHS
Ok, exploratory graph time. This was not what I expected to see!!
This could be unstable given the relatively small number of people in the "just a means to an end group" (we lost a lot of responses because of the survey logic probably, a relatively small subset answered the questions in combo I wanted to look at).
@grimalkina Could a point plot with CI whiskers give this more context? I don’t know what the R/ggplot2 equivalent would be to seaborn’s point plot function. I agree with the subsample size concern, and this is how I’d address it if I had the data open in front of me.
@grimalkina Or even just standard error whiskers would be helpful, honestly
Install bonfire.mavnn.eu
Get the full app experience
