Religion: measure belief, participation and institutional authority separately

Start with a mechanism

An institution is a durable pattern of roles, rules and relationships. Identify a process before treating an observed group difference as its explanation.

Sociology studies religious beliefs, practices, organizations and social consequences without deciding theological truth. Secularization claims need a specified dimension: affiliation, participation or institutional authority can change differently.

Predict and inspect

Predict which observation changes under the proposed rule or measurement. State a competing explanation before reading the worked analysis.

Worked case

Synthetic repeated surveys ask identical questions: affiliation falls from 70 of 100 to 60 of 100; weekly participation stays 30 of 100. The ten-point affiliation decline is not a participation decline. A separate legal change removing a religious organization’s control of public appointments concerns authority, not directly personal belief.

The case is deliberately synthetic. Its purpose is to make one process inspectable, not to establish how frequently it occurs in a country or institution.

Affiliation shares are not individual journeys

Predict whether a12-point increase in one response identifies how many people changed their affiliation. First check the denominator; then test that inference with a separate fixed-cohort example. No personal religious information is requested.

Inspect the observed census source and definition

ONS, Religion, England and Wales: Census2021, released29 November2022, corrected5 December2022; accessed29 October2026. Contains public sector information licensed under the Open Government Licence v3.0.

The voluntary question concerns affiliation, not belief or active practice. Published percentages use all usual residents: in2011/2021, Christian59.3/46.2%, No religion25.2/37.2%, and question answered92.9/94.0%. These rounded shares describe England and Wales, not the whole UK. The2021 reference date is21 March2021.

The bulletin corrects2011 nonresponse processing for three local authorities. Its December wording correction distinguishes the most common religious group from the most common answer. We use the corrected bulletin, not the superseded version. Overall census response and answering this voluntary item are different measures.

Initial claim: identify the denominator, measured construct and an unsupported individual inferenceCapture claim and settings Published response categoryNo religionChristianDenominatorAll usual residents: published sharesThose answering the religion question: reconstructed

To reconstruct the answered-only share, divide the published population percentage by the answered population percentage and multiply by100. For2021 No religion:37.2÷94.0×100≈39.5745%. This is arithmetic from rounded inputs, not an additional official estimate or imputation for nonrespondents.

Selected rounded published shares and optional reconstruction

Year	Published population share	Answered question	Selected denominator share
2011	25.2%	92.9%	25.2000%
2021	37.2%	94%	37.2000%

Selected change: 12.0000 percentage points. Switching denominators changes the question; it does not alter the underlying published observations. Nonresponse is not classified as No religion.

Independent fictional transfer: the same100 people at two times

Invented category X has25 members initially and37 later. Everyone is observed twice; nobody enters or leaves the population, every answer is complete, and there are only X and not-X. These assumptions do not describe the two censuses. X is an arbitrary label, not a religious group.

People leaving X: 0Try maximum offsetting flows

Rows: earlier category. Columns: later category. Counts of fictional people.

Earlier → later	X later	Not-X later	Earlier total
X earlier	25	0	25
Not-X earlier	12	63	75
Later total	37	63	100

Entries 12 − exits 0 = net increase12. All settings preserve both margins. Entries may range12–37 and exits0–25 under these invented assumptions; the margins alone do not choose one setting. These bounds are not estimates of religious switching in England and Wales.

Revised explanation: separate affiliation, participation and authority; explain the changed-case counterexample

Compare reasoning with feedback

The observed12-point population-share change and the fictional net12 people are different quantities. In the closed fictional cohort,12 entering/0 leaving and37 entering/25 leaving give the same margins. Real repeated censuses have changing populations and response patterns, so these fixed-cohort bounds cannot be applied to them. Affiliation data alone cannot establish individual transitions, practice, sincerity or a causal theory of secularization. A supported institutional explanation needs evidence about a specified process, not merely matching trends.

Copy your analysis report

Readonly JSON report{ "source": "https://www.ons.gov.uk/peoplepopulationandcommunity/culturalidentity/religion/bulletins/religionenglandandwales/census2021", "observed": { "category": "none", "denominator": "population", "rows": [ { "year": 2011, "answeredPercent": 92.9, "publishedPopulationShare": 25.2, "displayedShare": 25.2 }, { "year": 2021, "answeredPercent": 94, "publishedPopulationShare": 37.2, "displayedShare": 37.2 } ], "changePoints": 12.000000000000004, "derivedFromRoundedShares": false, "individualTransitions": null, "belief": null, "attendance": null }, "transfer": { "fictional": true, "total": 100, "beforeX": 25, "afterX": 37, "remainX": 25, "exitX": 0, "enterX": 12, "remainOther": 63, "netChange": 12, "entryRange": [ 12, 37 ], "exitRange": [ 0, 25 ], "appliesToCensus": false }, "worksheet": { "draft": "", "captured": null, "revision": "" }, "feedbackSeen": false, "temporary": true, "masteryAssessed": false }

Reset worksheet

Writing is temporary. Reset retains feedback exposure during this visit; reload clears it. Captured work is not independently assessed mastery.

Produce the evidence

Build a three-column measurement dashboard. Use Durkheim’s cohesion question, Weber’s social-change question and a conflict perspective on authority to propose different evidence. Explain why one measure cannot substitute for the others.

Record each intermediate step. For a network, label the direction and kind of each relationship; for a table, label units and denominators; for a process diagram, mark the decision point that changes the outcome. Include the original prediction beside the revised result so another reader can identify what changed.

Explain through a sociological perspective

Distinguish three claims: a pattern exists in this record; a specified institutional process could produce it; that process actually caused an observed real-world outcome. The first can be supported by a bounded record while the last needs further evidence.

A theory is useful when it directs attention to a process and makes a claim vulnerable to evidence. A perspective focused on coordination might examine how roles fit together; one focused on unequal power might ask who sets a rule and who bears its costs; an interactionist approach might examine how participants interpret a situation. Choose the perspective that fits your question and identify what it leaves unexplained.

Transfer before feedback

A festival grows while formal membership falls. What should a researcher examine?

Answer with a specific comparison or observation you would obtain. Explain which part of the case transfers and which assumption must be checked again. Do not substitute a group label for a causal mechanism.

Limits

Participation does not disclose sincerity. No tradition is internally uniform, and attendance trends do not justify claims about the character or truthfulness of believers or nonbelievers.

Source and extension

OpenStax sociology: Religion, Conerly, Holmes and Tamang; OpenStax, Introduction to Sociology 3e, 2021. Authored conceptual summary of the linked educational source. It provides a framework, not observations from the teaching case.

High school: Complete the case diagram or table and write a claim using two specific features of the packet. Keep the invented case distinct from the linked explanatory source.

College: Compare two theories using different predicted observations. Add an authentic bounded institutional record and a documented alternative interpretation; assess measurement and selection limits.