Do you have a particular study you could point me to? I do find twin studies useful.

I found

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3629371/

which addresses a variation of the question. The unpaired half of twins with discordant marital status show significantly (in the case of men) higher rates of depression; women also seem to too, but because there were so few twins of discordant marital status in general, the test is pretty underpowered and so the confidence intervals are huge. Given that, it's actually pretty impressive that it was able to find any significant differences at all.

Sure, here it is: https://journals.sagepub.com/doi/10.1111/j.1467-9280.1996.tb...

Thanks for sharing the findings of your quick research. Depression score is indeed not exactly the same as happiness, or life satisfaction, but for the purpose of our conversation I would say they are all close enough to what matter to us, so they are all sufficiently meaningful (and probably very correlated to each other).

Eyeballing the paper you cited, there appears to be evidence that widowed and divorced people are more depressed, but not single people (compared to the baseline, married people). At least when running the regression with all the other important variables controlled. Am I reading it wrong?

For the paper I found, I was looking at Table 4, which is of single never-married people comparing to the baseline of married people. Table 3 seems to correspond to what you're seeing. Skimming the discussion, they guess that the difference in depression and BMI scores between never married/married is due to health selection effects, although AFAICT there's not really a way to differentiate between health selection effects and marriage playing a causal role.

I'll take a look at your paper later.