[epwr]

I haven’t read the underlying paper, but from the article it seems the paper looks at if a person gets healthier when they get married.

This seems flawed because you get older as you get married, which is generally something that makes you less healthy.

This does not seem to compare health between a cohort that stayed single and a cohort that got married, so I don’t place a lot of value on the conclusions.

[solardev]

Good point. The article wasn't clear on that, but I think the full study did do that (my emphasis):

> Marital Status: The main independent variable is a time‐varying measure of marital status at each interview, calculated from marriage dates in the 1985–2011 PSID marital history file. Observations were censored after the end of a first marriage to focus on health difference between married and never‐married adults. To account for cumulative health effects of marriage (Dupre and Meadows, [ 12] ), this variable distinguishes among being never married; being in a first marriage formed 0–4 years ago; being in a first marriage formed 5–9 years ago; and being in a first marriage formed 10 or more years ago. An alternative specification of this variable as continuous duration in the first marriage (equaling 0 in never‐married observations) led to the same findings.

> (Partial results): In OLS models for both men and women, any duration of marriage is significantly associated with a better (i.e., lower) score on the health scale, compared to remaining never married. The random‐effects models assess whether this association persists after controlling for respondents’ heterogeneity in underlying health status. With the addition of random intercepts, both the men's and women's models continue to show statistically significant associations between all marital durations and better health, relative to remaining unmarried, but the magnitude of the marriage coefficients is one‐third to one‐half smaller than the OLS estimates. Finally, the fixed‐effects models test whether within‐person change in marital status or marital duration is associated with improvement in general health. Among men, no such association is evident, while among women, only a weak protective effect of being married for 10 or more years is supported by the fixed‐effects model. While the latter finding is consistent with the accumulation of a protective marital effect at longer marital durations, the fixed‐effects coefficient (−0.06) is approximately half as large as the coefficients in the random effects or OLS models. Therefore, considering all birth cohorts together, within‐person improvements in health are weakly attributable to long marriages among women, but are not caused by accumulation of time in marriage among men.