Ask a European how satisfied they are with their life, on a scale from one to ten, and one fact about them predicts the answer with surprising reliability: whether they are married. Across 37 countries and five waves of the European Values Study, married people report a higher life satisfaction than the unmarried in nearly every country measured — and once you hold age, sex and education constant, the premium is positive in all 37 of them. Ask instead how many children they have, and the needle barely moves.
That asymmetry is the story. The two facts most people would file together under "family" pull apart sharply in the data. Marriage carries a premium; children carry a non-premium. The contrast is not subtle, and it survives the obvious adjustments.
In every country, the married report higher life satisfaction
Each line connects the mean life satisfaction (1–10) of unmarried and married adults within a country. The gap — the family premium — is shown sorted from widest to narrowest. Toggle to the age-adjusted estimate, which nets out age, sex and education within each country.
Pool everyone together and the married sit at 7.25 on the ten-point scale; never-married singles at 7.17. That raw pooled gap is small — barely a tenth of a point — but it is the wrong number to lead with, because singles are on average younger, and the young are buoyant for reasons that have nothing to do with a wedding. Adjust for age, sex and education within each country and the picture sharpens: the typical married premium is 0.61 of a satisfaction point (the median country, 0.60), and it is positive everywhere.
The premium runs widest in Cyprus, where married adults outscore the unmarried by 0.92 points, and in Sweden at 0.85 — an unlikely pairing of a Mediterranean society and a Nordic one, which is itself a hint that this is not simply about traditional family norms. It runs narrowest in the South Caucasus: in Armenia the raw gap actually tips slightly negative, married adults sitting 0.16 of a point below the unmarried, and in Azerbaijan by 0.07 — though even there, once age is accounted for, the adjusted premium turns positive.
The other split is regional. In Western Europe the average age-adjusted married-versus-unmarried premium is 0.70 of a point; across the post-communist countries it is 0.56. The edge of marriage is real on both sides of the old Iron Curtain, but it is consistently larger in the West. (One caveat on labels: "Western" here is simply the residual — every country not in the post-communist list — so it sweeps in Turkey and Cyprus along with the usual north-western cases. Read it as "non-post-communist," not "EU-West"; the membership is spelled out in the notes.)
Now watch the children line go flat
Set the marriage premium beside the relationship between life satisfaction and the number of children a person has, and the contrast becomes visceral. The children line does not climb. It does not really fall. It wanders inside a band a fifth of a point wide and ends almost exactly where it started.
Marriage moves it. Children don't.
Left: mean life satisfaction by marital status. Right: mean life satisfaction by number of children, 0 to 4 or more. Same vertical scale on both, so the steps you see are directly comparable. Bands are 95% intervals.
Childless adults average 7.17. People with one child dip to 6.96, those with two recover to 7.05, three-child parents reach 7.15, and the largest families, four children or more, sit at 7.11. The entire range, from the most to the least satisfied child-count group, spans just 0.21 of a point — and there is no clean upward march to it. Adjust within each country for age and the rest, and the typical extra satisfaction per additional child is 0.08 of a point: positive in most places, but a rounding error next to marriage's 0.61. The slope is positive in fewer than nine countries in ten — about 86.5 percent of them — which is to say it is not reliably positive at all.
The marital-status panel, by contrast, has real structure, and most of it lives below the married line rather than above the single one. The genuinely large gaps are not between the married and the never-married; they are between the married and those whose marriages ended. Divorced and separated adults average 6.51, the widowed 6.31 — fully 0.74 and 0.94 of a point under the married. The same ranking shows up in the four-point happiness item as a secondary check. Whatever marriage tracks, its absence through loss tracks the steepest deficits in the entire portrait.
Two ways to read a premium
A cross-section like this cannot tell you that marrying someone would raise your satisfaction by six tenths of a point. The arrow may run the other way. Stutzer and Frey, following the same people over time, found that happier people are measurably more likely to marry in the first place and less likely to divorce — selection, not treatment, does part of the work. Their longitudinal evidence suggests selection explains part but not all of the premium, which is roughly the most honest thing the cross-section can say in agreement.
The divorce and widowhood deficits carry their own selection traps. The widowed are old, and age drags on satisfaction independently of marital loss; the divorced include a disproportionate share of people who were less satisfied to begin with. The gap you see conflates the genuine sting of an ending with the sorting of the less content into those endings. It is a real association. It is not a clean causal estimate of what divorce does to a person.
The full marital ladder — and where children sit on it
Mean life satisfaction by marital status (married vs not) crossed with parenthood (has children vs none). The rows separate cleanly — marriage moves the score. The columns separate too, but read that gap carefully: it is mostly a question of who among the unmarried has children. Hold marital status fixed, and parenthood almost vanishes.
None of this means children leave well-being untouched. The research that follows the same parents across the transition tells a more textured story than any single survey snapshot. Nelson, Kushlev and Lyubomirsky's review found a small overall negative tilt to parenthood with wide variation by circumstance; Margolis and Myrskylä showed the effect bends with the number of children already at home and depends heavily on a country's welfare and development context. A flat cross-sectional line is exactly what you would expect when a small drag in early parenthood is offset by a small lift later, averaged across societies that support parents very differently. The flatness is informative, not empty.
And the marriage premium is not the steady glow it looks like in a pooled snapshot either. Following couples over the years, VanLaningham and colleagues traced marital satisfaction declining with duration — the contented long-married and the recently wed are blended together in a cross-section, so the level overstates the durability. What the EVS can say cleanly is narrow and worth saying: across forty years of a secularizing continent, the married consistently evaluate their lives a little higher than the unmarried, everywhere; the number of children a person has does almost nothing of the kind.
What this does, and does not, show
Association, not a lever. Every figure here is cross-sectional. The married-unmarried gap reflects who marries and stays married as much as anything marriage might do, and the divorce gap blends the pain of an ending with the sorting of the less satisfied into it. Read the premiums as differences between groups, not as the effect of changing one person's status.
Evaluative only. The EVS measures how people judge their lives overall — life satisfaction on a 1–10 scale, with a four-point happiness item as a check. It carries no "yesterday" measures of daily mood, so nothing here speaks to the texture of a parent's actual day, only to the standing evaluation they report.
An unbalanced patchwork. Countries enter and leave across the five waves, so this is a structural portrait pooled over 1981–2017, not a trend. Each country counts once in the cross-country figures — an average national experience, not a population-weighted one — and any country-cell with an effective sample below 50 is dropped before it can add noise.