Religion & Well-Being · GSS 1972–2024
Religion is good for happiness — but it’s the showing up, not the believing, that does the work. A horse race inside half a century of American survey data.
For decades, surveys have found that religious people report being a little happier than everyone else. The usual reading is spiritual: faith brings comfort, prayer brings peace, belief in something larger brings meaning. That story is intuitive. It is also, the data suggest, mostly wrong about which part of religion matters.
The General Social Survey has asked Americans the same blunt happiness question since 1972 — taken all together, how would you say things are these days? — alongside how often they attend religious services, how often they pray, and how strongly they believe in God. Pull those apart and run them against each other, and a clean pattern falls out. The — physically attending, the congregation, the people in the next pew — carries almost all of the happiness link. The private channel — prayer, belief — carries strikingly little once attendance is in the room.
This is, more or less, the finding Chaeyoon Lim and Robert Putnam reached in 2010: the well-being dividend of religion runs through congregational social networks, not private devotion. “It is the friends you make at church.” Here is that argument rebuilt from the ground up in the GSS, with the seams left showing.
Start with the raw picture: average happiness at each level of frequency. Score happiness 1 (not too happy) to 3 (very happy), weight every respondent by the GSS survey weight, and trace the line.
Attendance climbs steeply. People who never attend services average 2.06; people who attend more than weekly average 2.35 — a span of 0.29 happiness points, monotone at almost every step. Now overlay prayer or belief in God on the same frame. The private lines are real but shallow: prayer spans about 0.18, belief in God only 0.15. Same outcome, same scale — one line rises like a staircase, the other barely tilts.
You can read the gap a second way, with correlations — and to keep it a fair fight, compute all three on the same set of respondents (the years where attendance, prayer and belief are all recorded). Happiness correlates with attendance at 0.13, with prayer at 0.08, and with belief in God at 0.08. Attendance is the strongest of the three by a comfortable margin — well over half again the size of either private measure.
Side-by-side gradients are suggestive, but attenders also pray more and believe more strongly — the three travel together. The cleaner test puts all three in one regression and asks which coefficient survives when they compete for the same variance. Standardize every measure, weight by wtssps, add the usual controls — age, income, sex, race, education, marital status, and a dummy for every survey year — and watch what happens to each bar when it goes from running alone to running together.
The picture is stark. Attendance enters alone at 0.09 and, with prayer and belief fighting it for the same respondents, barely flinches — it lands at 0.08. Prayer goes the other way: alone it pulls 0.07, but together it collapses to 0.01 — statistically a whisper. Belief in God shrinks from 0.07 to 0.03. The social channel keeps almost all of its weight; the private channels hand most of theirs over. The pew survives the prayer.
Here is the obvious objection. Maybe none of this is about God. Maybe people who go places — anywhere, regularly — are simply happier, and a church service is just one more place to go. If that’s the whole story, religious attendance should behave no differently from secular socializing.
The GSS lets us test it directly. It asks, in the same language, how often you spend an evening with friends and with neighbors. Those are “showing up” too. So run the same horse race with all three forms of gathering — church, friends, neighbors — in one model. Flip the toggle on the chart above to Showing up, and the result is its own kind of answer.
Religious attendance does not behave like just another evening out. Running together against secular socializing, attendance holds at 0.11 — more than twice the friends slope (0.05) and the neighbors slope (0.04). Gathering in general is good for you; gathering in a congregation is better. Whether that extra is the shared belief, the moral community, the weekly rhythm, or the depth of those particular ties, the data can’t say — but it is not only that religious people happen to leave the house.
The honest caveat comes first, because it is the strongest alternative explanation. None of this is causal. The arrow may run backwards: you cannot sit in a pew if you are sick, frail, depressed, or isolated. Healthy, connected, already-content people are exactly the ones able to show up every week — so part of the attendance premium could be selection, not benefit.
We can put a dent in that worry. Split the sample by self-rated health and re-draw the attendance gap inside each band — comparing weekly-plus attenders to rare attenders who report the same health. If the premium were purely a healthy-people artifact, it would vanish within bands. It doesn’t.
The gap is positive in every band — from 0.21 among those in good health down to a still-real 0.15 among those in poor health. And when self-rated health is added as a control to the main horse race, attendance shrinks but stays in front, at 0.06, even though health itself is a powerful predictor of happiness (standardized coefficient 0.25). Selection is real; it is not the whole story.
Other limits, stated plainly. The measures are ordinal — we treat the happiness scale as if a step from “not too happy” to “pretty happy” equals a step to “very happy,” which is a convenience, not a fact. Coverage differs by item: attendance runs the full 1972–2024, but prayer begins in 1983 and belief in God only in 1988, so the horse race lives in the overlap years. In 2021 the GSS moved to a push-to-web mode, which can shift levels; we pool across years and flag it rather than lean on any single recent reading. And “religious attendance” bundles a lot — belief, ritual, community, identity — so calling it “social” is a claim about which bundle does the work, defended here by the secular-socializing comparison, not a proof that no spirituality is involved.
What survives all of that is a single, reproducible shape. Across fifty years of American answers, the happiness that travels with religion travels mostly with the part you can see from the parking lot. It’s the pew, not the prayer.