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 correspon

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.