The Rules of Sociological Method

The Rules of Sociological Method (1895) is a short, combative book with an unusual character: it is a manifesto written before the demonstration. Durkheim set out how sociology should be done, and only two years later, in Suicide (4.2.4), showed that it could be.

This lesson closes Topic 4.2 by taking the rules one at a time — what each demands, what it was for, how it has held up, and what it still tells anyone doing research.

Imagine founding a science that everybody already thinks they understand.

A physicist has an advantage: nobody comes to a lecture on thermodynamics with strongly held prior views about entropy. The subject is unfamiliar, so the audience arrives empty.

A sociologist has no such luck. Every person on earth has lived in a family, been to school or not, held a job or not, obeyed and broken rules, been included and excluded. Everyone has a theory of why people are poor, why crime happens, why marriages fail, why some children succeed. These theories are held with conviction, drawn from experience, and often mutually contradictory.

And they get in the way before the investigation starts — because they supply the categories. Ask "why do people commit crime?" and you have already accepted that "crime" names a single kind of thing, that people who do it are a distinguishable type, and that the interesting question is about them rather than about the process that made certain acts criminal.

Durkheim's first and most important instruction is therefore not about data collection at all. It is: get your own ideas out of the way first. He calls the ideas you must remove prénotions — pre-notions — and he thought clearing them was the precondition of everything else.

The book has two jobs and they pull in different directions.

Job one: intellectual. Specify a method that produces genuine knowledge about social life. What counts as a social fact, how do you observe one, how do you establish a cause?

Job two: political. Win sociology a place in the French university system. This meant demonstrating that it had an object no other discipline claimed and a method as rigorous as any other science's — that it was not warmed-over philosophy or amateur history.

The second job explains the book's tone, which is unusually assertive, and some of its excesses. Durkheim overstates the analogy with natural science, insists on rules that are too strict, and pushes the exclusion of psychology further than his own practice warranted.

Read it as a founding charter under competitive pressure and both its force and its overreach make sense.

The rules

The most famous, treated fully in 4.2.1. Its practical content:

How it has held up. The core instruction is permanent, and it is one of the discipline's most valuable habits (1.4.6). But the demand for wholly external definitions is too strict, and Durkheim's own later work breaks it: The Elementary Forms (4.2.6) is a book about meaning. The interpretive tradition (4.3.1, 5.4) built itself on the objection that you cannot study human conduct while ignoring what it means to those doing it. The workable modern position: be explicit about your definitions and their consequences, and never let common-sense categories operate unexamined — which is Durkheim's point, without the prohibition.

Rule 2: Distinguish the normal from the pathological

Durkheim's most vulnerable rule. A social fact is normal for a society of a given type at a given stage of development if it is general — widespread — in societies of that type; otherwise it is pathological.

The purpose was practical: he wanted sociology to be able to advise, and advice requires a criterion for what needs remedy. Without one, sociology can describe but not diagnose.

The famous application: crime is normal. It occurs in all societies; a society without crime is impossible, since if serious offences vanished, minor deviations would attract the same outrage; and crime performs functions — it provokes the collective reaction that reaffirms shared standards (4.2.2), and it is the necessary precondition of moral change, since every reformer begins as a deviant. "According to Athenian law, Socrates was a criminal."

The objection, and it is fatal in the form Durkheim states it.

If normal = general, then whatever is widespread is healthy. Slavery was general for millennia. Systematic exclusion of women from public life was general. Judged by generality, both are "normal" and objecting to them is an error of method.

Durkheim adds a qualification — a fact may be general and yet pathological if conditions have changed — but this makes the criterion depend on a judgement about which conditions matter, which is the value judgement he claimed to be avoiding.

The rule is generally abandoned. What survives is the insight underneath: that certain forms of deviance are structurally inevitable in any social order, and that the reaction to deviance does real work. That insight founded the sociology of deviance (10.1, 10.2) and is entirely detachable from the normal/pathological framework.

Before explaining, classify. Durkheim proposes ordering societies by their composition — by the number and arrangement of their constituent segments, from the simple horde upwards.

How it has held up. The specific taxonomy is dead; the principle is not. Comparative sociology cannot proceed without some way of grouping cases, and every modern typology — Esping-Andersen's three worlds of welfare capitalism, varieties of capitalism, regime types, Hall and Soskice's coordinated versus liberal market economies — does what Durkheim said must be done. And the requirement to say what kind of case you are studying before generalising from it is now standard methodological hygiene.

The purpose is disciplinary autonomy (4.2.1). If social facts could be explained psychologically, sociology would be a branch of psychology.

How it has held up. As a claim about disciplinary boundaries, it did its institutional job. As a rule of explanation, it is too strong, and its status is one of the discipline's permanent arguments.

The case against it: methodological individualism (5.3) holds that social phenomena must ultimately be explicable through the actions of individuals, and that explanations lacking microfoundations are incomplete. Coleman's Foundations of Social Theory (1990) and the analytical sociology of Hedström and others have built substantial programmes on this. Even sympathetic critics note that Durkheim's own explanations covertly rely on psychological mechanisms — the whole account of collective effervescence (4.2.6) is a psychological mechanism, and Suicide's argument requires assumptions about how unbounded aspiration feels.

The case for it: emergent properties are real, aggregating individual accounts does not reconstruct them, and explanations that stop at individual motives systematically miss structural causes. Nobody's individual intentions explain a suicide rate, a market crash or a language.

The reasonable settlement, which most working sociologists hold implicitly: social facts constrain and enable, individuals act, and good explanation moves between the levels. Coleman's "boat" diagram — macro conditions shaping individual situations, individuals acting, actions aggregating to macro outcomes — is the standard picture, and it is neither Durkheim's nor the individualists', but a synthesis both can live with.

Rule 5: Use the comparative method

Sociology cannot run experiments on societies. Comparison is its substitute, and Durkheim adopts John Stuart Mill's canons — while rejecting the ones he thinks unusable in social science.

His preferred instrument is concomitant variation: rather than seeking cases where a cause is simply present or absent, look for cases where cause and effect vary together in degree. If integration falls and suicide rises across many independent comparisons, you have evidence of a relationship.

He is also emphatic about the requirement to eliminate rival explanations, and adds his rule that one effect has one cause — if a phenomenon has different causes in different cases, it is not one phenomenon but several.

Suicide is The Rules executed, and reading them together shows exactly what a method is for.

Rule 1 (things, not preconceptions). Durkheim opens by defining suicide externally — "all cases of death resulting directly or indirectly from a positive or negative act of the victim himself, which he knows will produce this result" — rather than by any assumption about despair or madness. He then takes the most objective indicator available: official rates.

Rule 4 (social facts by social facts). He explains the rate by religion, marital status, family size, military service and economic conditions — never by individual psychology, which he devotes a long opening section to eliminating.

Rule 5 (concomitant variation). He does not simply compare Protestants with Catholics. He shows that the rate varies by degree with the degree of integration, across religion, marriage, parenthood, political crisis and military rank — several independent gradients, all running the same way.

And the elimination. Mental illness, heredity, race, climate and imitation are each taken up and shown not to fit the pattern of the rates.

Where he breaks his own rules. He commits the ecological fallacy (7.2) by inferring about individuals from area data. He treats official statistics as transparent, which Douglas and Atkinson later showed they are not. And the one-cause-one-effect rule is what forces him into a fourfold typology when a multi-causal account would fit the evidence better.

That combination — a method genuinely applied, genuinely productive, and genuinely violated by its author — is why the two books are taught together.

Go deeper

The criticisms, gathered

Positivism. The book is the founding statement of sociological positivism, and the whole interpretive tradition is a reply. Weber's Verstehen (4.3.1) insists that social action must be understood through the subjective meanings actors attach to it — not because meanings are nice to know but because an action is not identifiable without them. Two men cutting wood: one for wages, one for exercise, one in a rage. Externally identical, sociologically different. Durkheim's rules cannot tell them apart.

The exclusion of psychology is untenable and was never observed. See above.

Circularity in functional explanation. Treated in 4.2.7.

The normal/pathological distinction smuggles values into a supposedly value-free method. Treated above.

And the reflexivity problem. If the sociologist is a member of the society being studied, how do they achieve the external position the method requires? Durkheim's answer is essentially discipline and training. Later thinkers found this insufficient: Mannheim's sociology of knowledge (4.1.7, 5.15), Gouldner's demand for reflexive sociology, and Bourdieu's insistence on participant objectivation — turning the analytical tools on one's own position in the academic field — are all attempts to solve a problem The Rules did not see.

What the rules gave the discipline

Set the criticisms beside what actually persists in research practice, and the balance is favourable.

Operational definition. Defining your concept by observable criteria before measuring is now the first step of any empirical study, quantitative or qualitative.

Distrust of common-sense categories. Every good study asks where its categories came from and whose they are.

Comparison as the substitute for experiment. The whole of comparative-historical sociology (7.6) — Skocpol, Moore, Tilly — rests on this, with far more sophisticated logic (Mill's methods, qualitative comparative analysis, process tracing) but the same premise.

Elimination of rival explanations. This is what "controlling for" means in regression, and what a good qualitative study does by seeking disconfirming cases.

Levels of analysis. The insistence that group-level phenomena require group-level explanation is why multilevel modelling exists, and the ecological fallacy is taught in every methods course.

And the demand that sociology be more than commentary. Durkheim's real achievement was insisting that a sociological claim must be checkable. Whatever else is wrong with the book, that demand is the reason sociology became a research discipline rather than a genre of essay-writing.

Durkheim's own career, briefly, because it is the argument's other half

He was born in 1858 at Épinal, in Lorraine, into a family of rabbis stretching back generations; he was expected to become one and did not. He took the first course in social science ever taught in a French university, at Bordeaux from 1887. He founded L'Année Sociologique in 1898 — a journal that was also a research programme, with a school around it: Mauss, Halbwachs, Hubert, Simiand, Hertz, Fauconnet. He moved to the Sorbonne in 1902, where his course on the history of education was compulsory for all prospective teachers, which meant that for two decades every secondary schoolteacher in France was trained in Durkheimian sociology.

He lost his son André at the front in 1915, and never recovered; he died in 1917, aged fifty-nine. Half his school died in the war with him — a generation of French sociologists lost in four years, which is one reason the discipline's centre of gravity moved to America.

The point for this lesson: the rules were not an abstract exercise. They were the constitution of an institution he built deliberately, and the institution outlived both the man and most of his specific claims.

"The Rules is a research manual." It is a manifesto written before the research. Read it with Suicide to see the rules applied — and violated.

"'Treat social facts as things' means only use quantitative data." Durkheim's greatest work is a study of religious meaning. The rule is about avoiding introspection and preconception.

"Normal means good." Durkheim's criterion is generality for a social type. This is the book's weakest rule and it is generally abandoned.

"Sociology must never use psychology." Durkheim said so, and did not observe it. The modern position is that levels interact and good explanation moves between them.

"Durkheim was anti-theory." He was against speculation without evidence. He theorised constantly, and The Rules is itself a theoretical treatise.

"The comparative method proves causation." It provides evidence about relationships and is vulnerable to confounding, selection and small-N problems. Modern comparative work is far more careful about exactly what it can establish (7.6).

Why it matters

It is the discipline's constitution. Whatever any sociologist now thinks of positivism, the argument that sociology has a distinct object, a distinct level of explanation and a checkable method is what secured its existence. Every department is downstream of this book.

Its central habit is permanent. Clearing prénotions — refusing to let everyday categories do your analytical work unexamined — is the single most useful thing the book teaches, and it applies far beyond sociology.

It defines one pole of the discipline's biggest argument. Durkheim's insistence on external observation, social causation and the reality of the collective; Weber's insistence on meaning, interpretation and the individual actor. Almost everything in Part 5 and Part 7 sits somewhere on the line between them — and the next topic takes up the other end.

And it shows a founder being right about the enterprise and wrong about the details. The specific rules are variously abandoned, softened and superseded. The project they defined is alive, and the man who defined it also demonstrated that it could be carried out. That is the standard for a founding text, and this one meets it.

The Rules of Sociological Method (1895) is a manifesto written before the demonstration — Suicide followed two years later.

Rule 1: Consider social facts as things. Discard prénotions; define phenomena by external, observable characteristics before investigating; take objective indicators. Permanent as a habit; too strict as a prohibition on meaning.

Rule 2: Distinguish normal from pathological. Normal = general for a social type. Applied to show that crime is normal, inevitable and functional ("According to Athenian law, Socrates was a criminal"). The criterion is fatally flawed — slavery was general — and is generally abandoned; the insight about the inevitability and function of deviance survives.

Rule 3: Classify social types before explaining. The taxonomy is dead; the requirement is standard in comparative work.

Rule 4: Explain social facts by antecedent social facts, not by individual consciousness. It won disciplinary autonomy and is too strong as an explanatory rule; the modern settlement moves between levels (Coleman's boat).

Rule 5: Use the comparative method, especially concomitant variation, and systematically eliminate rival explanations. This is Suicide's architecture — including where Durkheim breaks his own rules, through the ecological fallacy and uncritical use of official statistics.

Criticisms: positivism versus Weber's Verstehen; the untenable exclusion of psychology; circular functional explanation; values smuggled in through normal/pathological; and reflexivity, which Bourdieu later addressed through participant objectivation.

What persists: operational definition, distrust of common-sense categories, comparison as the substitute for experiment, elimination of rivals, attention to levels of analysis — and above all the demand that sociological claims be checkable.

Prénotions — the everyday preconceptions that must be cleared before investigation.

Operational definition — defining a concept by observable criteria before measuring it.

Normal and pathological — Durkheim's criterion of generality for a social type; his weakest rule.

Social types — the classification of societies by their composition, as a precondition of comparison.

Concomitant variation — the comparative method of examining whether cause and effect vary together in degree.

Elimination of rival explanations — the systematic testing and dismissal of alternative causes.

Methodological individualism — the position that social phenomena require explanation through individual actions.

Microfoundations — the individual-level mechanisms underlying a macro-level regularity.

Coleman's boat — the diagram of macro-to-micro, micro-to-micro and micro-to-macro links in social explanation.

Participant objectivation — Bourdieu's requirement that the analyst objectify their own position in the field.

L'Année Sociologique — the journal Durkheim founded in 1898, and the school around it.

One — clear a prénotion. Take a category you use daily — "unemployed," "middle class," "immigrant," "successful" — and write down the official definition. Compare it with the one in your head. The gap is what Durkheim was worried about.

Two — define something externally. Define "family" using only observable criteria, without reference to what you think a family should be. Then test it against five real households. Notice what your definition includes and excludes that you did not intend.

Three — test the normal/pathological rule. Name three things that were general in a past society and that we now consider wrong. Then name something general today that a future society might judge the same way. The exercise shows exactly why the rule fails.

Four — build a concomitant variation. Pick any outcome that varies across countries — life expectancy, incarceration, fertility, trust. Find a candidate cause and see whether they vary together in degree across many cases. Then think hard about what else varies with both.

Five — read the two together. The first chapter of The Rules and the first chapter of Suicide, in sequence. About sixty pages, both freely available. It is the clearest demonstration in the discipline of a method being stated and then used.