What Religion Does Sociologically

The prediction failed. The question the founders were actually arguing about did not.

Durkheim and Weber gave different answers to what religion does in a society, and the difference is not a matter of opinion — they are two research programmes, generating different predictions, and both have been tested (see 7.2.3).

This lesson sets out what each predicts, what the evidence supports, and the single finding that has done most to settle the argument.

Two congregations, one city, and a list.

The first is a large suburban church. On a Tuesday it runs a toddler group. On Wednesday a lunch club for the isolated elderly, with volunteer drivers. Thursday is the food bank. The noticeboard carries three job vacancies, two rooms to let, and a hand-written card offering help with a form.

Its members lend each other money without paperwork. Newly arrived families are told, within a fortnight, which school to apply to, which doctor's surgery takes new patients, and which of the local secondary schools is worth the commute (see 8.3.3). A substantial number of its members met their spouse there.

When someone dies, it produces a funeral, a rota of meals for three weeks, someone to sit with the widow, and — a year later — someone who remembers the date.

The second is a storefront congregation of recent migrants, meeting in a converted shop unit.

It does everything on the first list, at higher intensity, plus: interpreting at appointments, a queue of people needing help with immigration paperwork, coordination of money sent home, temporary accommodation on the floor of the back room, and a network that will find someone work within a week of arrival.

It is, functionally, a welfare state, an employment agency, a credit union, a legal advice centre and a burial society — run by volunteers, out of a shop, for people who have access to none of those things.

Now count what on either list requires anybody to believe anything.

Almost none of it. The childcare, the food, the lending, the introductions, the information, the funeral, the memory of the date — these are produced by people meeting weekly, knowing each other, and having obligations to each other.

And the last item is the one that matters most for this lesson. When someone dies, a religious congregation has a script, a set of actions, and a division of labour, ready. Most secular institutions do not, which is why the moment of death is where religion's social function becomes visible even to people who believe none of it.

Everything in that story is Durkheim. The question is whether it is all of the story — and Weber's answer is that it is not.

Two programmes, two objects, two sets of predictions.

Durkheim's object is solidarity. On his account religion is society representing itself to itself; the division of sacred and profane marks the collective off from the ordinary; and the mechanism is ritual — assembling, doing the same thing simultaneously, generating the heightened collective state that renews attachment to the group (see 4.2.6).

Its predictions are testable. Participation should matter more than doctrine. Religion should track group boundaries and intensify when the group is threatened. The effects should come from assembling, so a religion whose members never meet should not produce them.

Weber's object is meaning and conduct. Religion supplies answers to the problem of suffering and injustice — theodicies — and different answers license different conduct in the world (see 4.3.10).

A doctrine of predestination, a doctrine of karma and rebirth, and a doctrine of a coming redemption produce different orientations to work, to wealth, to suffering and to political action.

Its predictions are also testable. Content should matter, not merely participation; groups with different theodicies should behave differently under similar conditions; and the carriers of a doctrine — clergy, virtuosi, intellectuals — should shape which version prevails.

And Marx's third answer belongs here too, and it is more sympathetic than its slogan.

In the passage that produced the famous phrase, the sentence before it reads that religion is the sigh of the oppressed creature, the heart of a heartless world, the spirit of spiritless conditions — and the metaphor that follows was, in its own century, the standard reference for a painkiller. The claim is that religion is a real response to real suffering, which both consoles and, in consoling, makes the suffering endurable rather than intolerable.

Its prediction: religiosity should be highest where deprivation and insecurity are greatest — which is 9.4.1's existential security finding, arriving from a different tradition.

Three programmes. Now the evidence.

What the evidence supports

One — networks and social capital, which is the best-established finding and is decisively Durkheimian.

Religious participation is among the largest single sources of associational life in most societies: of volunteering, of charitable giving (including to secular causes), of civic participation, and of the dense acquaintance networks that 8.3.3 identified as an opportunity channel.

And the finding that settles the argument comes from a large study designed to disentangle the components.

The civic effects of religion are produced by the social networks inside congregations, not by the beliefs.

People with more friends in their congregation are markedly more likely to volunteer, give, and participate civically. People who hold the same beliefs but have no religious friends are not. And once the network measure is included, theological variables add essentially nothing.

Read what that establishes. It is not that belief is irrelevant to believers. It is that the mechanism producing the civic outcome is assembling with people and acquiring obligations to them — which is precisely what Durkheim's theory predicts and Weber's does not.

And the corollary is a prediction worth watching: a religiosity that is individual, private and unassembled — believing without belonging (see 9.4.1) — should not produce these effects. The available evidence supports that.

Two — health and wellbeing, where the association is real and the causal reading requires care.

Religious attendance is associated with better health outcomes, including lower mortality, in a large literature.

Four things must be said together.

Attendance predicts better than belief does, again — which points to social support and behavioural regulation rather than to doctrine.

The mechanisms are identifiable: social support and reduced isolation; health behaviours, where some traditions substantially reduce smoking and drinking; and meaning-making resources for coping.

The causal problem is serious. Healthy people can attend and unhealthy people cannot, which produces the association in the wrong direction (see 7.6.2). The best longitudinal work, following healthy people forward over decades and adjusting extensively, still finds a real association — substantial in some cohorts — and its authors are explicit that residual confounding cannot be excluded.

And the literature has the profile of one with publication bias (see 7.8.2): many small studies, heterogeneous measures, and effect sizes that shrink as designs improve.

The proportionate reading: a real, moderate association operating mainly through social and behavioural mechanisms, with a causal component that is probably non-zero and smaller than the cross-sectional literature suggests.

Three — mobilisation, where religion supplies infrastructure.

Social movements need buildings, communication networks, pre-existing leadership, a shared culture of collective action, and leaders with incomes independent of the opponent. Religious organisations supply all five.

The American civil rights movement is the documented case. The Black church provided meeting places outside white control, a communication network reaching most of the population weekly, an existing organisational structure, a repertoire of collective ritual, and — critically — a leadership whose salaries came from the congregation rather than from white employers, so that they could not be dismissed for participating.

And the same infrastructure has served movements of every political character. A mass opposition movement in a communist state; liberation theology; a revolution; and religious nationalisms that have targeted minorities in several countries.

The course applies its own symmetry rule here (see 6.8.1). Religion is an organisational resource, and organisational resources are not politically directional. An account that treats religion as inherently conservative or inherently emancipatory is contradicted by the record in both directions.

Four — boundary marking, where religion functions as identity rather than belief.

In several of the world's most persistent conflicts, religious identity marks the boundary and religious belief is thin. Populations divide along confessional lines whose members attend rarely and believe inconsistently, and the division nonetheless organises residence, schooling, marriage, employment and violence.

Which is why measures of religious identity predict conflict participation better than measures of religious belief — and it is Durkheim rather than Weber, because what is operating is the boundary, not the doctrine.

Five — and the place where Weber's programme has the clearest support: the content of coping.

When people face bereavement, serious illness or disaster, religious frameworks are used to interpret what has happened — and which framework is used predicts the outcome.

Research distinguishing forms of religious coping finds that the content matters substantially. Framings involving a benevolent presence, collaboration, and meaning are associated with better adjustment. Framings involving punishment, abandonment or demonic attribution are associated with worse adjustment — worse than no religious framing at all.

This is a doctrinal effect, measured, with the direction depending on the doctrine — which is exactly what the Weberian programme predicts and what a purely functional account cannot generate.

And the causal arrow between religion and politics runs the other way more than expected.

The intuitive model is that religious convictions shape political positions.

Research tracking individuals over time finds a substantial flow in the opposite direction. For many people, political identity is formed earlier and more firmly, and religious affiliation adjusts to it — people whose politics do not fit their tradition's public alignment reduce their involvement or disaffiliate, and people whose politics fit are drawn in.

This is the mechanism behind 9.4.1's finding. The rise of religious non-affiliation was concentrated among people whose politics did not match the political identity that organised religion had acquired — so a political sorting produced a religious statistic.

Two consequences.

Cross-sectional correlations between religiosity and political attitudes overstate religion's causal role, because part of the association was produced by politics selecting who is religious.

And the process is self-reinforcing. As the loosely attached with mismatched politics leave, the remaining population is more politically homogeneous, which strengthens the alignment, which produces further sorting. A religious tradition can become more politically distinctive without any individual changing their mind.

Because the two programmes make different predictions and the evidence has largely decided between them — for some outcomes.

For solidarity, civic participation, health and mobilisation, the operative mechanism is assembling and belonging, not doctrine. The network finding is decisive, the attendance-over-belief pattern recurs across three separate literatures, and the mobilisation cases are about buildings, communications and independent incomes.

For meaning, coping and conduct under adversity, content matters and is measurable — and the coping literature is the clearest demonstration that a doctrine's specific character has consequences.

So the honest answer is that Durkheim describes what religion does to a group and Weber describes what it does for a person facing a problem — and treating them as rivals rather than as accounts of different objects wasted a great deal of the twentieth century.

Four things to carry.

When a religious effect is reported, check whether it survives controlling for social participation. Frequently it does not, and where it does not, the finding is about assembling.

When religion is described as a cause of conflict, ask whether identity or belief is doing the work. The measures are different and they behave differently.

When religion is described as politically conservative or progressive, apply the symmetry test. It has supplied the infrastructure for both, and an account that predicts only one is not an account.

And be alert to the reversed arrow. Politics shapes religious affiliation substantially, which means correlations between the two overstate religion's causal role.

One closing observation, which is what the storefront congregation is for. Everything on that list — the childcare, the loan, the lift to hospital, the room, the funeral, the person who remembers the date — is a function that a well-resourced society would provide through institutions, and does not.

Which is why religious participation is highest where those institutions are weakest, and why the decline of religious participation in societies with strong ones is not an accident (see 9.4.1). The question a secularising society has to answer is not what people will believe. It is who will bring the meals.

Two congregations produce childcare, food, loans, job information, school advice, introductions, interpreting, accommodation and funerals — and almost none of it requires anyone to believe anything. The moment of death is where the function becomes visible even to non-believers.

Three programmes. Durkheim's object is solidarity, produced by ritual and assembly, predicting that participation matters more than doctrine and that religion tracks group boundaries. Weber's object is meaning and conduct, through theodicies that license different orientations to work, wealth and suffering, predicting that content matters. And Marx's account — more sympathetic than its slogan, since the surrounding sentences describe religion as the heart of a heartless world and the metaphor referred to a painkiller — predicts that religiosity tracks deprivation, which is 9.4.1's security finding from another tradition.

Five evidential findings.

Networks are the best-established, and they are decisively Durkheimian: the civic effects of religion are produced by friends in the congregation, not by beliefs, and theological variables add essentially nothing once the network is included. Which predicts that unassembled private religiosity should not produce these effects, and it does not.

Health associations are real, operate mainly through social support and behaviour, and attendance predicts better than belief. The causal problem is serious, the best longitudinal work still finds an association while acknowledging residual confounding, and the literature has the profile of one with publication bias.

Mobilisation works through infrastructure — buildings outside the opponent's control, weekly communication reaching a whole population, existing leadership, a repertoire of collective ritual, and leaders with incomes independent of the opponent. The same infrastructure has served movements of every political character, which the symmetry rule requires stating.

Boundary marking explains conflicts where confessional identity is thick and belief is thin — and identity measures predict conflict better than belief measures.

And coping is where Weber's programme is clearest: framings involving benevolence and collaboration predict better adjustment, and framings involving punishment or abandonment predict worse adjustment than no religious framing at all. A doctrinal effect, measured, with direction depending on content.

And the causal arrow between religion and politics substantially reverses: political identity forms earlier and religious affiliation adjusts, which produced 9.4.1's rise in non-affiliation, overstates religion's role in cross-sectional correlations, and is self-reinforcing — a tradition becomes more politically distinctive without anyone changing their mind.

Sacred and profane — the division that constitutes a collective's self-representation.

Collective effervescence — the heightened state produced by assembly and simultaneous action, which renews attachment.

Theodicy — a religious solution to the problem of suffering and injustice; different solutions license different conduct.

Elective affinity — the mutual fit between a set of ideas and a group's material situation.

Religious social capital — the networks, obligations and information generated by congregational participation.

Attendance-over-belief pattern — the recurring finding that participation predicts outcomes better than doctrine.

Movement infrastructure — buildings, communication, leadership and independent incomes supplied by religious organisation.

Confessional boundary marking — religious identity organising a division where belief is thin.

Positive and negative religious coping — framings of adversity as collaborative or as punitive, with opposite associations to adjustment.

Political sorting into and out of religion — political identity shaping religious affiliation rather than the reverse.

One — make the list. For any religious congregation you know of, write down every practical function it performs. Then mark which require belief.

Two — test the network claim. Ask two religious people how many of their close friends attend the same place. Then ask two non-attending believers the same question about anyone.

Three — apply the symmetry test. Find one emancipatory and one repressive movement organised through religious infrastructure. Then ask what the infrastructure supplied in each case.

Four — separate identity from belief. Take a conflict described as religious and find out what proportion of each side attends or believes. The answer is frequently low on both sides.

Five — check the arrow. Ask someone who has changed their religious involvement whether their politics changed first. The order is informative and is usually not the one assumed.