A partner is worth about half a rung of the Cantril ladder, and the premium holds wherever you live. But a civic-rich county does not rescue the isolated: the gap between the partnered and the alone is the same in high-social-capital places and low. Connection is personal — and it does not trade in for community.
The chapter this plate belongs to is called Alone Together, and the phrase is an argument: that Americans can be surrounded by institutions — churches, associations, a dense civic fabric — and still be alone in the way that matters most. The Gallup data lets us test the two halves of that idea against each other. Does personal connection move well-being? And can the community kind of connection stand in for the personal kind?
The first answer is an emphatic yes. People with a partner — married, in this coding — rate their lives at 7.23 on the eleven-point ladder; the single and the separated, divorced or widowed sit together near 6.73. That half-a-rung gap is one of the largest demographic differences in the whole file, and unlike the income gradient in the companion plate, it is not about where the partnered live: it survives almost untouched when we compare people only to their own county neighbors. A partner is worth the same wherever you stand.
The second answer is the interesting one — a clean null. If a county's stock of social capital could substitute for a personal tie, the penalty for being alone should shrink in the most civic places. It doesn't. The gap between the partnered and the unpartnered is essentially flat across the entire range of county social capital, from the most atomized counties to the most richly associational. A standard deviation of a county's social fabric does not buy back a tenth of a partner. Four steps follow: the size of the premium; its survival within places; the community-compensation null; and a within-county horse race for what actually moves the ladder — where the deepest wound turns out to be not solitude but unemployment.
Start with the raw gap. Married Americans rate their lives at 7.23; those who are single and never married sit at 6.73, and the separated, divorced and widowed almost exactly alongside them at 6.73. The partnered are also markedly less sad — 13% report sadness yesterday against 18–23% for the unpartnered — though, tellingly, not much less stressed. Marriage in this data is less a shield against the hard day than a lift to the verdict on the life: the partnered rate their lives higher and grieve less, while the daily grind of stress is shared about equally.
Is the partner premium really about partnership, or about the kinds of places the partnered tend to live — leafier, safer, richer counties? Fit the gap twice: once across the whole country, and once within counties, comparing each person only to their own neighbors. The two come back nearly identical — 0.497 pooled, 0.495 within county. Holding the place fixed changes nothing. Unlike a county's average income or its life expectancy, the value of a partner does not travel with geography; it is a fact about two people, and it weighs the same in Manhattan and in rural Mississippi.
Here is the test the chapter is named for. Sort counties into ten rungs of social capital — the density of associations, the strength of family and institutional ties, the project's JEC index — and draw the partnered and unpartnered ladders separately in each. If a rich civic fabric could rescue the isolated, the two lines would converge in the most associational counties: the gap would close. It doesn't. The lines run nearly parallel the whole way; the partner premium is 0.45 of a rung in the least-civic counties and 0.50 in the most. A full sweep from the most atomized America to the most richly connected America barely dents the penalty for being alone. The community kind of connection is real and valuable — but it is not the personal kind, and it does not substitute for it.
Put the personal forces in one model — within counties, so geography is netted out — and let them compete. Income still matters most among the positives: a standard deviation of household income lifts the ladder 0.35 of a rung. But a partner (0.21, net of income, education and work) rivals a college degree (0.24), and beats anything else in the social column. The deepest mark, though, is negative and economic: being unemployed costs three-quarters of a rung (−0.75), more than a partner or a degree can give back. The picture that emerges is the chapter's thesis in miniature — connection and money both move the good life, neither substitutes for the other, and the sharpest blow to how an American rates their life is to be out of work.
The largest caution is selection, and it cuts at the heart of the marriage finding: happier, healthier, more financially stable people are more likely to marry and to stay married, so part of the partner premium is the kind of person who partners, not the partnering itself. This is a repeated cross-section, not a panel — no one is followed into or out of a marriage — so "a partner is worth half a rung" is a comparison between people, not a before-and-after, and the true causal effect is surely smaller than the raw gap. The within-county and within-income controls narrow the gap (to about 0.2 rungs) but cannot close the selection problem. What is more robust is the comparative claim the chapter rests on: whatever marriage is worth, a county's social capital does not substitute for it — and that comparison holds the selected kind of person roughly fixed on both sides.
The usual Gallup caveats apply: the extract is unweighted (associational only); "married" is the available partnership proxy and misses cohabiting and other committed partnerships, which would attenuate the gap; the county social-capital index is a single ~2018 cross-section; and the ladder is a coarse self-report. The horse-race coefficients are descriptive partials among correlated predictors, not causal effects. What survives is the shape the chapter was built to show: Americans can be richly surrounded and still alone, because the fabric of a place and the tie to a person are different things, and only one of them is what a good life seems to turn on.
cty_social_capital context column (~2018). Off the project's
tract spine (CLAUDE.md §13).ladder ~ married, estimated
pooled and with county fixed effects (pyfixest, SEs clustered on county). The
community-compensation panel splits respondents into deciles of county social capital and
takes the married/unmarried ladder mean in each. The horse race adds standardized income,
a college-degree indicator and an unemployment indicator, all within county.