Faith & well-being · General Social Survey, 1972–2022

Losing My Religion

Americans leave the faith they were raised in at four times the rate they did fifty years ago. What that costs in happiness depends less on where you started than on where you stand now.

An analysis of 68,615 General Social Survey respondents with both a childhood and a current religion · weighted estimates · association, not causation

Run the country’s religious life backward and you find a great migration. In the early 1970s, fewer than one American in ten claimed no religion; by the early 2020s it was more than one in four. That headline hides the actual movement underneath it — a churn between four life stories, sorted by what you were raised with and what you hold now. Some kept the faith of their childhood. Some left it. A smaller number, raised with nothing, found one. And a growing share were raised with no religion and have stayed that way.

Tag these four trajectories by religion at sixteen against religion now, both collapsed to faith or none, and a clean ordering of happiness falls out — one that says something blunt about belonging. Your current affiliation matters more than your origin. Converts who moved in are happier than people who never left; leavers who moved out sit right alongside the lifelong unaffiliated. Leaving, in the averages, carries no extra penalty beyond simply being a none. Finding faith looks like a small gain.

Four ways through a religious life

The centerpiece below is an origin-and-destination map. On the left, the two starting points: raised in a faith, or raised with none. On the right, the two endings. Each of the four ribbons is one trajectory; its color and the number riding on it encode that group’s average happiness on the General Social Survey’s three-point scale, and its label gives the share of all classified respondents who travel it. Hover — or tab to — any path for its mean, its very-happy share, and its sample size.

The stable-religious are the happiest of the four, averaging 2.21; about 33% call themselves very happy. They are also, at five in six of the sample, by far the largest group — for most of this period, keeping your childhood faith was the unremarkable, default path. Converts come next at 2.12. Then a near-tie at the bottom: leavers at 2.07 and the lifelong unaffiliated at 2.05, with the same 23% very-happy share. The arrow you traveled barely registers; the side you ended on is what shows.

Does the ordering survive the people in it?

Leavers and converts are not random draws on the population. The people who walk away from religion are younger, more educated, more urban, and more recent than those who stay; the unaffiliated as a whole skew that way too. Some of the happiness gap could be those differences wearing a religious costume. So fit the obvious controls — age, sex, race, education, family income, and survey year — in a weighted regression and watch what happens to the gaps.

The ordering holds. The convert gap stays the smallest — about 0.04 points after controls. The leaver deficit barely moves, from about 0.14 to 0.12: who leavers are explains almost none of why they are less happy. The lifelong-none deficit shrinks the most, from 0.15 to about 0.10, as their youth and schooling get netted out — but it stays the largest of the three. Composition rearranges nothing essential. What it cannot reach is the part of the religious advantage that runs through congregational life itself — the friends in the next pew, the casseroles, the standing invitation to show up. A companion study, The Pew, Not the Prayer, takes that mechanism apart; here it is enough to say the gap is mostly not about who these people are on paper.

Leaving doesn’t cost you happiness beyond what being unaffiliated already does. The penalty is the destination, not the journey.

The rise of leaving — and a penalty that thinned

Half a century ago, walking away from your childhood faith was a deviant act. Among Americans raised in a religion, only about 7% had become nones in the early 1970s; by 2021–22 that leave-rate had climbed to roughly 22%. The left panel below tracks that climb. The right panel asks the harder question: as leaving went mainstream, did its happiness penalty fade? It draws the leaver line and the stable-religious line together, so you can see not just the gap but where the gap went.

The penalty did narrow. When leavers were rare, in the early 1970s, they trailed the stable-religious by about 0.23 points; by 2021–22 the gap was roughly 0.12. That is the shape a social-norm story predicts: deviance costs less as it becomes common. But read the two lines honestly and the comfortable version collapses. The gap closed because the top came down. The stable-religious fell from about 2.24 in the early 1970s to about 2.04 in 2021–22; the leavers did not rise to meet them — they drifted slightly lower too. Every bit of the narrowing comes from the religious themselves becoming less happy, not from leaving becoming painless. A penalty that shrinks because the comparison group is sinking is not the same as a penalty that has healed.

Two cautions sit on top of that. The last point spans 2021 and 2022, when the survey moved largely to the web after COVID; some of the late drop in both lines may be that mode change rather than mood. And the whole secularization is mostly generational, not a matter of individuals changing their minds in adulthood.

A generational tide

Sort the same respondents by the decade they were born and the engine of the change shows itself. Among Americans born in the 1920s, only about 5% end up unaffiliated. Among those born in the 1990s, it is roughly 38% — a near-eightfold rise across the cohorts. The nones are not, for the most part, older believers losing faith late; they are successive generations who arrive at adulthood already holding less.

What this shows — and what it doesn’t

This is a cross-section, and the arrow of cause runs in directions it cannot separate. Religion may lend happiness through community and meaning; equally, unhappy people may drift out of congregations, or be pushed out, and leaving may itself cost the belonging that buoyed them. The data record the association, not its direction. The four-group ordering is robust to the obvious controls, but the controls are coarse and cannot capture the congregational social channel that other work places at the center of the religious advantage. The narrowing penalty is real but confounded: it closed because the religious declined, so do not sell it as pure normalization. The categories are blunt — any faith against none — and the “none” bucket holds the spiritual-but-not-religious alongside the indifferent and the convinced. Childhood and current religion are reported together in adulthood, so memory and the late-2010s web-mode shift both leave fingerprints. And these are weighted GSS estimates of evaluative happiness, the trajectory itself a marker of eudaimonic and social standing — not a controlled trial of what faith does to a life.