The Attendance Evidence

The most consistent finding in the religion-and-health literature is not about belief — it is about showing up. Across decades of follow-up and millions of person-years, people who attend religious services regularly die at lower rates than people who never attend. This page walks the actual cohorts behind that claim, the shape of the dose-response curve, and the confounding stack that decides how seriously you should take any of it.

The Question These Studies Ask

The reason attendance dominates this literature is practical, not theological. Belief is hard to measure; behavior is not. Cohorts can ask one clean question — "How often do you attend religious services?" — with answer categories that mean roughly the same thing to a Baptist in 1965 and a Presbyterian in 2010. Attendance is also a repeated behavior, so the exposure accumulates in a way a single questionnaire item about "spirituality" cannot capture. The trade-off is that attendance measures the whole package at once: the belief, the practice, the community, the music, the morning structure, the dress code, the potluck. Untangling which thread carries the effect is the job of the next page — here we establish that the package, taken whole, predicts survival.

The Famous Cohorts, Side by Side

The finding did not emerge from one study. It emerged from the same pattern showing up in every cohort that bothered to ask. The table collects the landmarks; each row is a real study with real follow-up, and together they are the reason this topic gets taken seriously at all.

The Dose-Response Shape

The second reason to take the association seriously is its shape. It is not a binary — believers versus everyone else — but a gradient, and gradients survive confounding better than categories do. In the Nurses' Health Study analysis, each step up in attendance frequency carried a step down in risk, and the ordering held after adjustment for diet, smoking, alcohol, physical activity, and medical history. A graded dose-response is what you would expect if something about attendance itself were doing work — though it is also exactly what you would expect from self-selection, since people who manage weekly attendance tend to be people who manage other health-relevant routines. Hold both readings at once.

The Seven-Year Number, Read Honestly

The seven-year figure deserves the careful reading the headlines skip. It is a descriptive difference in life expectancy — the observed gap between two groups, not a causal estimate of what attendance would add to any given life. People who attend weekly and people who never attend differ from age twenty onward in hundreds of ways that are not measured in any survey, and the difference in their average lifespans is the sum of all of them. The same caution applies to the 33% figure from the Nurses' Health Study: a hazard ratio of 0.67 is a strong association by epidemiology's standards, but "associated with" carries every unmeasured difference between the women in those two rows of the data.

The Confounding Stack

What actually differs between attenders and non-attenders? The honest list is long, and every item on it predicts mortality on its own:

• 🚬 Health behaviors. Attenders smoke less, drink less, and are less likely to engage in risky habits — the single largest confounder, and the first thing every analysis adjusts for.
• 🩺 Health selection. Seriously ill people attend less — they cannot get there. This biases the association toward making attendance look protective even if it does nothing.
• 👥 Social integration. Attenders are embedded in networks that carry practical help, monitoring, and belonging — the channel the Community & Faith topic documents.
• 🧭 Personality and orderliness. People who keep a weekly appointment differ in conscientiousness and stability — traits with their own mortality signatures.
• 📊 Measurement noise. Attendance is self-reported, and the question is asked once, years before the deaths it is asked to explain.

When analyses adjust for the measurable items — smoking, alcohol, diet, exercise, medical history — the association shrinks meaningfully but survives. What no adjustment can do is remove health selection and unmeasured personality, which is why the residual effect remains an open question rather than a settled quantity.

Attendance Is a Community Variable First

The deepest thing these studies measure may not be religion at all. Weekly attendance is, structurally, a standing appointment with a community: fifty-odd gatherings a year, same room, overlapping faces, shared songs and shared obligations. That is precisely the recurring-contact infrastructure the Community & Faith topic finds in the longest-lived populations on record — Blue Zones residents belong to faith communities and show up. The deep-vs-weak-ties topic adds the mechanism: dozens of familiar-but-not-intimate faces, the exact network layer that predicts survival in large cohorts. None of this requires the doctrine to be true; it requires the doors to open on a schedule. That observation is the bridge to the next page, which asks the question this one deliberately postpones: what, precisely, is the active ingredient?

Questions, Answered Briefly

• 📅 How much attendance is enough? In the Nurses' Health Study the gradient started at less than weekly — but the largest step sat between weekly and more-than-weekly. The data describe a slope, not a ticket to punch; treat them as motivation to protect a recurring community rhythm, not as a dosing schedule.
• 🙋 Does it work for men too? The headline women-only analysis exists because the Nurses' Health Study is a women's cohort. The Alameda County Study, NHIS analysis, and meta-analyses include both sexes and find the same direction. The evidence is thinner for men than women, not absent.
• 🌍 Is this just a U.S. effect? Mostly, the strong data are American and Christian — a real limitation. European and Israeli cohorts generally find weaker or no associations, which is itself evidence that the cultural wrapper matters as much as the attendance.
• 🤔 Is it the belief or the showing up? The attendance studies cannot tell you. That is exactly the question the active-ingredient page takes apart.

Related Topics