Five honest questions about faith and flourishing — answered at the real strength of the research, and what it does and does not touch.
Religion is not a wellness accessory that recently acquired a research literature. It has been the organising centre of human societies for as long as there have been societies, and it remains the most consistent correlate of meaning and purpose that social science has found. The question worth asking is not whether it does any good, but what kind of good, how much, and by what mechanism.
That is what this page sets out, one question at a time. Every figure was checked against the actual publication — not against a press release, and not against another page that cited it. Where the evidence is strong we say so plainly. Where it is weak, absent, or points away from what we would have liked, we say that too, in the same voice and the same size type.
Yes — consistently, across the largest study ever run on the question. But the headline figures are unadjusted for age, health, income and education; the one carefully adjusted estimate is small and fragile; and in South Africa specifically it runs slightly the wrong way. All of that belongs on the same page as the good news.
Flourishing here is a 12-item composite — happiness and life satisfaction, mental and physical health, meaning and purpose, character, relationships, and financial security — each scored out of 10. Pooled across all 22 countries, and unadjusted for age, health, income, education or marital status, mean flourishing rose steadily with attendance:
| Attends religious services | Mean flourishing (95% CI) |
|---|---|
| More than once a week | 7.67 (7.44–7.89) |
| Once a week | 7.42 (7.27–7.57) |
| One to three times a month | 7.21 (7.03–7.39) |
| A few times a year | 7.08 (6.89–7.28) |
| Never | 6.86 (6.65–7.07) |
The gradient is orderly, and it held in almost every country — including highly secular ones. It varied enormously in size, though: from 2.33 points in Hong Kong down to 0.15 in India, where the confidence interval crosses zero.
These are unadjusted means. Older, sicker, poorer and more settled people attend at different rates, and none of that is controlled for here. You will see the figure 0.81 quoted as the gap between weekly attenders and never-attenders — it is not. It is the gap between more than weekly and never, the weekly gap is 0.56, and because it is a subtraction of two pooled means there is no confidence interval for it at all. We do not use it.
The study also asked what predicts adult flourishing from childhood. Attending weekly at age 12, adjusted simultaneously for relationship with each parent, parents' marital status, family finances, abuse, feeling like an outsider, childhood health, immigration status, birth cohort and gender, produced a much smaller estimate:
0.27 points (95% CI 0.17–0.36), E-value 1.61. That E-value is the honest part: an unmeasured factor associated with both childhood attendance and adult flourishing at a risk ratio of 1.61 would account for the whole thing. This is a real finding, and it is a fragile one.
This is retrospective recall — adults remembering their own childhood — not a study that followed children forward. And the estimate was largest in Poland (1.02) but slightly negative in Kenya (−0.22), South Africa (−0.15) and Tanzania (−0.05). Those confidence intervals all include zero, so they are null results rather than evidence of harm — but quoting the global figure while leaving out the countries where it ran the other way would be choosing our data, and we are not going to do that.
South Africa has its own dedicated paper. Half the sample attends religious services weekly or more; 38% were unemployed and job-seeking. Across 69 separate health and wellbeing outcomes, those attending more than once a week held the most favourable result on 26 of them — more than any other attendance group. Those who never attend held the least favourable result on 20.
The authors go on to name never-attenders, alongside women, divorced people and those with nine to fifteen years of schooling, as groups that “may be especially vulnerable subpopulations.”
“This could be because attending religious services more regularly promotes flourishing or because those who are flourishing to a greater degree are more likely to attend religious services with greater regularity; this association could also be confounded by other factors.”
One dataset, not four. The global paper, the South African paper, the religious-centrality paper and the European paper are all Wave 1 of the same survey — the South African 2,651 and the European 84,399 are subsets of the same 202,898 people. They are one line of evidence reported in several places, and we count them once.
And these particular figures are a snapshot. The attendance gradient above is cross-sectional — everyone measured at one moment. Further waves of the study have been collected and longitudinal work from them is now appearing, so this is a fast-moving literature and our reading of it will need updating. What a snapshot cannot tell you is which way the arrow points: whether attending lifts flourishing, or people who are already flourishing attend more. The study’s own authors say exactly that.
Yes, and substantially — while you are in it. What the research finds underneath Sabbath-keeping is detachment, social contact and a week with less chaos in it. That is an account of how the practice works, not a reason to abandon it. And the benefit fades once you return to normal — which is the best argument there is for keeping a rest day every week.
Ninety-one independent samples, pooled. Psychological detachment — genuinely coming off duty, not merely being away from the desk — tracked with less emotional exhaustion (r = −0.36), less fatigue (r = −0.42), better sleep, better mood and higher life satisfaction.
It also, revealingly, did nothing for work engagement (r = 0.04, not significant) and almost nothing for task performance (r = 0.09). Rest is worth having as rest. It is not a productivity strategy, and we will not sell it as one.
This is the largest body of evidence on this page, but not the best-designed. Most of the underlying studies measured both the exposure and the outcome by self-report at the same sitting, which inflates correlations.
The best-designed rest study we found. Faculty were measured before, during and after their sabbatical, against matched colleagues carrying on as normal. The comparison group did not move on anything. Among those on sabbatical, measured before → during → after:
| Measure | Before | During | After |
|---|---|---|---|
| Burnout | 2.36 | 1.28 | 2.17 |
| Faculty stress | 2.78 | 1.90 | 2.60 |
| Resource gain | 4.33 | 13.15 | 4.70 |
Burnout more than halved. Resource gain tripled. Those are large changes in a real life, set against colleagues who did not budge. After the sabbatical most measures moved back toward where they started — resource gain returned to its pre-sabbatical level — though the paper reports some effects still significantly better than baseline afterwards.
Most of it does not last. Positive affect returned to its pre-sabbatical level and life satisfaction had fallen back by the third measurement. Which specific measures remained better than baseline, and for whom, depends on moderators we have not fully verified, so we make no claim about that here. The fading is not a weakness of rest — it is the nature of it.
Among 1,309 Christian-school teachers across four countries, Sabbath-keepers reported burnout 40–49% of a standard deviation lower — which the authors call “sizeable”, though by the usual conventions it is small-to-medium. Among 1,316 United Methodist clergy, Sabbath-keepers looked healthier on every mental-health measure too: less depression, less anxiety, less burnout, more flourishing. Physical health was never significantly related to Sabbath-keeping at any stage.
Then the clergy study adjusted for sleep, vacation, relaxing activities, social support, financial stress, life chaos and time demands — and every one of those health associations disappeared. What remained significant was spiritual wellbeing and quality of life. What predicted health instead was social support, significant on all twelve outcomes, and life chaos, where a single unit raised the odds of depression by 24%.
The difference between the two studies is the depth of adjustment, and it is decisive. The teacher study controlled for age, gender, marital status, years of experience and country. The clergy study controlled for all of that plus sleep, vacation, relaxing activities, social support, financial stress, life chaos and time demands — and the associations did not survive it. The two are not comparably adjusted, and should not be read as a disagreement between equals.
Read carefully, though. Several of the things the clergy study adjusted for — rest, support, less chaos — are plausibly how a Sabbath works rather than rival explanations for it. Controlling for the mechanism will erase a real effect. The honest reading is that the Sabbath adds little on top of detachment, support and a calmer week — not that it does nothing.
So why a seventh day, and not one long break a year? Because the benefit decays. A sabbatical halves burnout while you are in it and most of that is gone within months of returning. Something that wears off does not need to be bigger — it needs to come round again. Weekly is roughly what the decay curve asks for, and the tradition arrived at it about three thousand years before anyone measured one. Weekends and holidays are the same principle at different intervals.
It depends entirely on what you are asking prayer to do. Praying at a distance for a stranger's recovery has been tested about as thoroughly as anything in this field and does not change medical outcomes. Praying with someone, in the room, is a different intervention — and the evidence there points the other way.
Intercessory prayer — praying for the health of someone who is usually unaware of it — has been through ten randomised controlled trials. Pooled, there was no significant effect on death (RR 0.77, 95% CI 0.51–1.16), on general clinical state (RR 0.98), or on readmission to intensive care (RR 1.00). A separate meta-analysis of 14 studies found an effect size of 0.10, not distinguishable from zero — and 0.012 once one discredited study was removed.
The Cochrane authors concluded they were not convinced further trials of this intervention should be undertaken. We think a page like this has to lead with that, not bury it.
These are two reviews containing the large trials — not four separate findings. The famous individual studies sit inside these pools, and we do not count them twice.
The largest single trial randomised bypass patients three ways: prayed for and uncertain, not prayed for and uncertain, or prayed for and told so. Prayer itself made no difference — 52% versus 51% complications.
But the patients who were told they were being prayed for did worse: 59% versus 52% (RR 1.14, 95% CI 1.02–1.28). Whether that reflects performance anxiety, or the alarm of being singled out as a serious case, nobody knows. It is a secondary finding from one trial and should not be over-read — but it should not be hidden either.
Set against those nulls is a randomised crossover trial of direct, in-person prayer — six weekly hour-long sessions, face to face. Depression and anxiety improved significantly against controls, along with optimism and daily spiritual experience, and the gains held a month later. In the MANTRA II trial, the signals that favoured anything at all favoured the bedside component — human presence — rather than prayer at a distance.
The most defensible reading of the whole randomised literature is the simplest one: presence works; distance does not.
Six hours of undivided compassionate attention from another person is a powerful intervention in its own right, and this trial had no comparison group receiving the attention without the prayer. It was also 95% female, its sample size is not stated in the published abstract, and it was null on cortisol — the one objective measure it took. It supports presence and sustained attention. It does not show that prayer heals.
Spiritual experience changes, measurably and durably. Mood is far less reliable. And across three independent studies the same pattern appears: what happens inside a practice matters more than how often you do it.
The best-designed study on this page. Nine hundred and eight people were randomised to twenty minutes a day for twenty-eight days of Centering Prayer with explicitly religious framing, the identical practice with the religious language stripped out (neutral words — ball, ticket, coffee), or a comparison group asked simply to notice and record their day. Seven hundred and two completed follow-up and were analysed.
On flourishing — the pre-registered primary outcome — the three arms did not differ. Religious versus secular: d = 0.03. Religious versus the comparison group: d = 0.18, not significant. Everyone improved from baseline, which is what expectation alone will do.
On what the practice is actually for, it did something: mystical experience d = 0.37 and daily spiritual experience d = 0.39, both against the comparison group. The spiritual-experience gain was still there a month later; the mystical-experience measure was only taken the day after, so nothing is known about how long that one lasts.
Those two spiritual-experience results are exploratory — the authors say so plainly, because the pre-registered primary outcome came out null. And religious framing did not beat the secular version on them either; the difference was against the comparison group. Twenty minutes a day for four weeks, in a general population, of one practice in isolation. Whether a full practice among people who already believe would move flourishing over months is an open question nobody has answered.
Five hundred and sixty adults, four kinds of prayer. The findings are not the ones people expect:
| What was measured | Association with wellbeing |
|---|---|
| Petitionary prayer — asking for things | nothing at all |
| Ritual prayer — set forms | more negative affect |
| Sheer frequency of prayer | less happiness (β −.14) |
| Meditative prayer | existential wellbeing |
| Prayer experiences | 4 of 5 outcomes |
Praying more often was associated with being slightly less happy — though the authors are careful here: the raw relationship was positive and non-significant, and only turned negative once controls were added, which they call “some ambiguity in its effect on happiness.” Frequency did predict greater religious satisfaction. And meditative prayer was the strongest predictor in the whole table for that outcome.
Having something actually happen in prayer was the most consistent predictor of wellbeing. The authors' own reading is that distress drives people to pray, and that saying prayers without encountering anything does not relieve it.
A 1985 sample in one American city, with modest coefficients and one weak scale. We use the pattern here, not the numbers.
Three sources, one message. The UCLA study of 112,232 students found prayer to be one of the strongest influences in the whole project — on religious commitment. The same study's authors state that growth in academic and personal outcomes accompanied growth in qualities like equanimity and care, and that “similar effects were not associated with growth in religious qualities.” Rohde found practice deepened spiritual experience without lifting flourishing. Poloma and Pendleton found experience mattered and frequency did not. Practice reliably deepens faith. It does not reliably lift mood. Anyone who tells you otherwise is selling something.
Deepening a practice destabilises some people, and the research is clear enough about it that we would be negligent to leave it out. There is also a pattern running through several datasets that changes how the whole page should be read.
Around one student in ten arrived high in religious struggle — feeling distant from God, angry at God, unsettled, at odds with family over belief. By the third year it was one in seven. Freshman struggle correlated negatively with psychological wellbeing (r = −0.27), and growth in struggle showed negative effects on leadership and on satisfaction with college.
What raised it, in the study's own list: arriving at a college where many peers already struggle, frequent self-reflection, alcohol, heavy television, serious illness or injury, and bereavement. In an earlier pilot of 3,680 students, 22 per cent of the most spiritual reported high spiritual distress against 8 per cent of the least — a different sample and a different scale from the struggle figures above, so the two are not interchangeable.
Meanwhile the same cohort's mental health declined across the board while their spiritual qualities grew — those two things happened simultaneously, in the same people.
American undergraduates entering university in 2004, followed for three years, with a 40% response rate at follow-up. Whether any of it transfers to adults elsewhere in the world is an assumption, not a finding.
No numbers in this section. If you read nothing else on this page, read this.
None of this is a test of whether your faith is true. No study here asked whether God exists, whether prayer is heard, or whether you are right. Science cannot get at those questions and never claimed it could. What all these researchers measured is a much smaller thing: whether certain habits shift certain numbers.
That matters, because some famous results in this field sound devastating and are not. When researchers found that praying for a stranger in another hospital did not change their recovery, they had not discovered anything about God. They had discovered that a strange experiment did not work. Nobody prays in order to improve someone's surgical statistics. Most of the disappointing findings here are like that — the wrong question, asked of the wrong thing, because it was the easy thing to measure.
What we would rather you took from this page is what stood up. And a great deal stood up.
Of everything on this page, one of the best-evidenced findings anywhere here is that a strong sense of purpose goes with living longer. Not happier necessarily — longer. And purpose is exactly what a faith gives a person.
Comfort makes people satisfied. It does not make life mean anything. Across 132 nations, people in poorer countries report more meaning in life than people in wealthy ones — and the researchers found this was in part because people in those countries were more religious. When things fall apart, meaning is the part that holds.
Four weeks of contemplative prayer measurably deepened people's sense of the sacred, and it was still there a month later. It did not turn them into happier people. It made them people to whom something had happened.
A protected break more than halved burnout while people were in it, against colleagues who did not budge. Most of it faded once they went back — which is the argument for a rest day every week rather than one long break a year. Rest is a meal, not a monument.
Sitting in a pew among strangers does almost nothing — weekly attenders with no close friends in the congregation looked no different from people who never attend at all. Having even a handful of people there who know something true about your life is where the benefit turns out to live, for those whose faith matters to their sense of who they are.
And here is the honest part. Faith will not reliably cure your anxiety. It will not make you richer, thinner, or more productive. Being a believer does not appear to make a person happier than their neighbour in a comfortable country. We could sell you those promises — a great many people do — but they are not the tradition's promises, and they do not survive checking.
What the tradition actually says is that a life needs a purpose, that people are made for one another, and that attention given to God changes what a person notices. Those hold up. Every one of them. Refusing to claim the rest is how we keep the right to say so.
Some of these we believed ourselves until we checked. Each one is here because we found it circulating, traced it, and could not stand it up.
This matters more than the list above, because it is the list that applies directly to the course we sell.
We would rather publish that list than pretend it does not exist. It is also, frankly, the research agenda we would like to be part of.
Every figure was read at its source — the publisher, PubMed or PMC — not from a secondary citation, a press release or another website. Several claims died at this step.
Where a study is cited by the wrong year or the wrong statistic, we use the correct one. A widely quoted volunteering figure is a risk ratio, not a hazard ratio, and the two are not interchangeable. We also removed a claim of our own from this page after checking it: we had said a well-known prayer study was routinely miscited by year. It is not — the two dates are two genuinely different papers on the same data, and we were the ones who had it wrong.
Overlapping studies are counted once. Four prominent papers here are the same survey; two famous prayer trials sit inside the reviews we cite. Listing them separately would manufacture agreement out of one piece of evidence.
Nulls and negatives get the same prominence as positives, including where they undercut what we sell. A study with a persuasive title and ten participants is not evidence, however much we like its conclusion.
We name what we cannot verify rather than rounding it into a claim. One influential study's effect sizes exist only inside a book we have not yet obtained, so we make no claim that depends on them.
We discard compromised sources entirely. One frequently cited prayer study had a co-author later convicted of fraud, a senior author who withdrew his name, and participants never told they were in a study. It has never been retracted. We do not use it.
Found something wrong? Tell us and we will check it and change it. That is the whole point of publishing the sources.