Findings to date

Update 4    spring 2026

...is long overdue. Sorry. There's a ton of data now I just haven't had a chance to dig into it - too many other things to fix before summer.

I'm specifically not trying to link emails to accounts here, but if you want to go on the mailing list for updates, just email hwaicom24@gmail.com and I'll add you. I've managed one update in the last 10 months so safe to say I won't be spamming you :)

Update 3 - 1000 people!    7/11/25

We've now collected data from over 1,000 people. Amazing. If you want to see who all has actually done this thing, have a look at Some Histograms. Or, if you want to poke around the data yourself - check out the data explorer. (It's a bit clunky but it works).

Update 2    5/22/25

Just crossed 500 people. Here's a slightly interesting finding: anxiety looks different as we get older. These plots show everyone so far, projected onto a 4-factor non-negative matrix factorization. (You know what that is, right?) Basically four "underlying components" of anxiety. The average for each group is plotted as the larger circles. Overall, younger people tend to report anxiety characterized by "fidgetyness" or "racing thoughts". Older people tend to have less of that, and a little more fatigue / lack of motivation. BUT, in a more general sense, we see a small shift toward better mental health as people get older and/or coupled/married.

Nice.

Update     4/14/25

Closing in on 300 people and that seems to be enough to start doing some (basic) machine learning to test a few things...

Here's what we did: We took each of the baseline 20 questions and trained models to guess a person's answer to that question, based on their answers to the other 19 questions. Some of them were easier to predict than others, but - and this is huge - the weights the models assigned to the input features all seem to make sense. In the example below, you can see that predicting how successful someone feels in life primarily takes into account how mentally healthy they identify as, how much they like their job, and how optimistic they are. No, we're not exactly cracking the code mental health just yet, but it's encouraging to see that en masse, the data makes sense and gives sensible results. (You can explore the other model weights here).

First read-out    3/22/25

Here's a result from what we've collected so far (n=200 people). These are the correlations between people's answers to the first 15 questions. For a given pair of questions, a lighter square means the answers tend to positively correlate, a darker square means the answers tend to negatively correlate.

So far we see correlations where we'd expect them (people who are older are more likely to be coupled / have kids, people who self-describe as autistic tend to disagree with the statement "I regularly make new friends"). Is this an amazing result? Well - yeah, kind of. People seem to be willing to come to this random ass website and answer questions about themselves honestly. Also apparently people who like horror movies don't eat healthy. So yeah, eat healthier, horror movie enthusiasts.