Modern Class Schemas

A theory of class can afford to be subtle. A classification cannot.

A statistical agency has to assign every person in a country to a category, using answers that fit on a form, in a way that is consistent across decades and comparable across countries. The theory has to survive contact with a tick box — which is 7.2.2's operationalisation problem at national scale, and it is where most of what you will ever read about class actually comes from.

Two questions, thirty million people.

Here is what a national statistical agency has to work with. On the census or the labour force survey, it asks:

What is your job? — recorded as a free-text description, later coded into an occupational classification with several hundred categories.

And are you an employee, self-employed with employees, or self-employed without employees? Plus, usually: do you supervise others, and how many people work at your workplace?

That is essentially it. From those answers, everyone must be placed.

Here is how the standard modern scheme does it, and the logic is Goldthorpe's from 8.2.1 made operational.

Step one: is the person an employer? If self-employed with employees, they are in the small-employer category — and the number of employees separates small proprietors from larger ones.

Step two: self-employed without employees? Own-account workers — the petty bourgeoisie, in the older vocabulary.

Step three: for employees, what kind of employment relationship does this occupation typically involve? And this is the crucial move: the scheme does not ask the individual about their contract. It uses the occupation as an indicator of the typical relationship, because the theory says the relationship is a property of the position rather than of the person.

Occupations where work is hard to monitor and expertise is specific — professionals, senior managers, specialists — are assigned to the higher categories, on the ground that employers there must secure commitment with salary, progression, security and prospective rewards.

Occupations where work is easily monitored and skills are general are assigned to the lower categories, where the relationship is payment for effort or time.

And the categories in between — technicians, supervisors, lower-grade administrative work — are assigned to intermediate positions where the relationship mixes elements of both.

Now our five people from 8.2.1.

Ravi, self-employed with twelve employees: small employer. Sunita the nurse: a professional occupation with a service relationship — the lower professional and managerial category. Anil the builder: own-account worker. Meena the shopkeeper: small employer, despite earning less than Sunita. The scheme does not care what she earns, because it is classifying her position, not her income (see 8.2.1). And Farooq: higher managerial and professional, the top category.

Notice what happened to the Marx–Weber argument. It has been settled in practice, by an agency that needed an answer, in favour of the employment relationship — and the settlement is now the source of nearly every official statistic about class in Britain and much of Europe.

A classification is a theory, and it produces the picture it was built to see.

This is not a criticism. It is the point. But it has consequences that are easy to miss.

Before 2001, British official statistics used a scheme dating from 1913 which sorted occupations into five groups according to their "general standing in the community" — a judgement about social prestige, made by officials.

That is a status scheme, not a class scheme, in Weber's precise sense (see 8.2.1). It measured honour, and it was used for eighty years to produce evidence about mortality, health and life chances — evidence that was then interpreted as being about class.

It was replaced because it had no theory. Nobody could say what the five categories were categories of, which meant nobody could say what a finding about them meant.

The replacement has a theory, and it was validated against it: the categories were shown to differ, as predicted, on measurable features of the employment relationship — how people are paid, how secure they are, whether there is a career structure, how much autonomy they have (see 7.2.2 on criterion validity).

That is what a properly built classification looks like, and it is worth knowing that the one before it was in use for most of a century.

The three families of scheme

One — employment-relations schemes, which dominate official statistics.

The European standard schema distinguishes around eleven positions, collapsible to seven, five or three depending on what you need: higher and lower professional and managerial; routine non-manual; small proprietors and farmers; technicians and supervisors of manual work; skilled manual; semi- and unskilled manual; agricultural labour.

Its purpose was comparative, and it was built to support the international study of social mobility (see 8.3). The British official classification implements the same logic in eight, five or three categories, and a harmonised European version extends it across member states.

Why it works as well as it does. It is derivable from two survey questions. It is stable over decades. Its categories differ steeply and consistently on outcomes — mortality, morbidity, children's educational transitions, security in a downturn — and they do so after controlling for income, which shows it is capturing something income does not.

Two — asset-based schemes, in the Marxist tradition.

Wright's operationalisation asks direct questions rather than inferring from occupation: do you own a business, how many employees, do you supervise others, do you participate in policy decisions, what credentials does your job require? From these, positions are assigned across the capital, organisation and skill dimensions (see 8.2.1).

Its distinctive contribution was comparative. Applying the same instrument across countries showed that class structures differ substantially in composition — the size of the managerial and supervisory apparatus varies between economies at similar levels of development — which is a finding an occupational scheme cannot produce, because it is a finding about how many managers a society employs.

Three — capital-based schemes, in the Bourdieusian tradition.

The argument: occupation captures economic position and misses the cultural and social resources that Bourdieu showed are convertible into advantage (see 5.8.1). A scheme should therefore measure economic capital, cultural capital and social capital directly and let the classes emerge from how they combine.

The best-known attempt measured household income, savings and house value; participation in "highbrow" and "emerging" cultural activities; and the number and social range of people known — then used a clustering method to identify groups. It produced seven classes with memorable names, including an elite at the top and a precariat at the bottom, and it reached the public in a way academic classifications almost never do.

And that last scheme attracted serious methodological criticism, which is 7.3.2 arriving in a headline.

The main dataset was a self-selected web survey promoted by a broadcaster. It attracted a very large number of responses — and the respondents were heavily skewed towards the well-educated, the professional and the culturally engaged, exactly as 7.3.2 predicts for a voluntary online instrument. Some occupational groups were represented by a handful of people; others by tens of thousands.

A smaller, properly drawn national sample was collected alongside it, and the analysis used both. The critics' objection was that the class structure itself was derived substantially from the self-selected data, so the categories reflected who chose to participate.

Three further criticisms were made, and they are about method rather than politics.

The number of classes was chosen by the analysts. The clustering technique used requires the researcher to decide how many groups to extract, guided by fit statistics that rarely give a single clear answer. Seven was a defensible choice among several defensible choices, and it was presented as a discovery.

The categories are not relational. They describe combinations of resources, not positions defined by their relation to each other — so the scheme abandons the feature that made class analysis explanatory rather than descriptive (see 8.2.1). It is, in this sense, a sophisticated gradational scheme.

And the elite and precariat categories are small and defined by the method, which makes them vulnerable to exactly the sampling problem above — the very top and the very bottom being where voluntary surveys are least reliable.

The defence is real and should be stated too. Occupational schemes do miss cultural resources that plainly matter, and the scheme's central substantive finding — that cultural capital has split into a "highbrow" form and an "emerging" form that is not simply lowbrow, and that the second is concentrated among the young and educated — is interesting and has held up in other data. And it made class a public conversation in a way no occupational schema had for decades.

The proportionate reading, which is 7.9.1 in operation: an important substantive argument about cultural capital, delivered through a classification whose particular categories rest on a sampling design that cannot support them. The argument survives the criticism; the seven classes do not.

The problems every scheme has to solve

Five practical difficulties, none of which has a clean answer.

One — whose class is a household's class?

Occupational schemes were built when the "head of household" convention seemed unproblematic, and classifying married women by their husband's occupation was standard. The feminist critique of this was decisive: it made half the population's own position invisible by definition, treated women's employment as economically irrelevant, and guaranteed that no analysis could detect the effect of women's own class position on anything.

The alternatives all have costs. Classify each individual by their own occupation — which loses the fact that living standards are shared. Use a dominance approach, taking the higher of the two positions in a household — which is defensible for living standards and makes the lower earner's position vanish. Or use both, for different purposes, which is the current practice and is the honest answer.

Two — the self-employed. A category containing a builder with a van, a shopkeeper, a consultant on a day rate, and a driver working through an app under conditions indistinguishable from employment. Own-account work has grown, and much of it is employment that has been contractually reclassified — which is 8.1.1's Meena and Farida, and it means the category is measuring a legal form rather than a position.

Three — people not in work. The retired, the unemployed, the long-term sick, those doing unpaid care, students. The convention is to classify by last occupation, which works reasonably for the recently retired and badly for those who have never worked or left long ago. And a scheme built on employment relations has, by construction, least to say about people outside employment — who are disproportionately those in the worst positions.

Four — comparability across countries. The schemes were designed for it, and the difficulty is real: the same occupational title carries a different employment relationship in different labour markets, which is 7.2.2's measurement invariance problem. A nurse's contract, security and autonomy differ enough between countries that the category is not identical.

Five — and the largest. Occupational class schemes were built for economies where most work is formal, waged and recorded. Where a large share of employment is informal, unregistered, casual or in household enterprises, the classification has no purchase — and that describes the majority of the world's workers. Applying a European schema to such an economy classifies the minority who fit and treats everyone else as residual, which is exactly 6.8.1's structure appearing as a technical limitation.

Because when you read a statistic about class, you are reading the output of one of these schemes, and they are not interchangeable.

Four habits.

Find out which scheme. Official statistics almost always say. The answer tells you whether you are reading about employment relations, assets, capitals or — in older data — perceived social standing.

Never compare across schemes. "The working class has shrunk from 60 per cent to 25 per cent" is meaningless if the two figures come from different classifications, and older figures very often do (see 7.4.4 on trend breaks).

Check what "working class" means in the sentence. In political commentary it frequently means neither position nor identity but education level, or region, or ethnicity, or voting behaviour — none of which is class, and the substitution is usually doing argumentative work.

And ask whether the scheme could have found what is being claimed. A scheme built on employment relations will not detect cultural exclusion. A scheme built on capitals will not detect exploitation. Each finds what it was constructed to see (see 7.1.3), and that is a reason to know which one is in front of you, not a reason to distrust all of them.

One closing point that matters for the next lesson. These schemes exist because people needed to answer a question: do the categories predict anything? They were validated on that basis, and they pass — the outcomes differ steeply and consistently across them.

That is the fact that 8.2.3 has to reckon with, because the case that class no longer matters has to explain why the classifications still work so well.

A theory of class becomes a classification when it must place everyone from what a form can ask — occupation, employment status, supervisory responsibility, workplace size.

The standard modern scheme implements Goldthorpe's mechanism: occupation indicates the typical employment relationship, so occupations where work is hard to monitor and expertise is specific are placed in the higher categories, where employers must buy commitment with salary, security and progression. The Marx–Weber argument was settled in practice by a statistical agency that needed an answer.

Britain's previous official scheme, in use from 1913 to 2001, sorted occupations by their "general standing in the community" — a status scheme, with no theory of what its categories were categories of. Its replacement was validated against measurable features of the employment relationship.

Three families. Employment-relations schemes dominate official statistics, are derivable from two questions, stable over decades, and differ steeply on mortality, health, children's chances and security net of income. Asset-based schemes ask directly about ownership, supervision, decision-making and credentials, and their distinctive contribution is comparative — class structures differ in composition between similar economies. Capital-based schemes measure economic, cultural and social capital and let classes emerge.

The best-known capital-based scheme attracted serious methodological criticism: its structure was derived substantially from a large self-selected web survey skewed towards the educated and culturally engaged; the number of classes was an analyst's choice presented as a discovery; and the categories are not relational, which abandons what made class analysis explanatory. Its substantive argument about the splitting of cultural capital is interesting and survives; the seven classes do not.

Five unsolved practical problems: whose class a household's is — where classifying women by their husbands was decisively criticised and every alternative has costs; the self-employed, a category now containing reclassified employment; people not in work, whom an employment-based scheme is least equipped for; cross-national comparability, where the same job title carries different relationships; and the fact that these schemes assume formal waged work, which describes a minority of the world's workers.

And when reading: find the scheme, never compare across schemes, check what "working class" is standing in for, and ask whether the scheme could have found what is claimed.

Occupational coding — assigning free-text job descriptions to a standard classification.

Employment relationship — the basis of the standard schemes: labour contract against service relationship.

EGP schema / national socio-economic classification — the comparative European schema and its official national implementations.

Criterion validation of a schema — showing the categories differ as predicted on pay form, security, progression and autonomy.

"General standing in the community" — the pre-2001 British basis; a status scheme used as a class scheme.

Asset-based scheme — classification by ownership, organisational control and credentials.

Capital-based scheme — classification by economic, cultural and social capital.

Latent class / cluster analysis — methods that group cases by patterns, with the number of groups chosen by the analyst.

Relational vs gradational classification — categories defined by relations between them, against combinations of resources.

Dominance approach — assigning a household the higher of its members' class positions.

Own-account work — self-employment without employees; increasingly a contractual form rather than a position.

One — classify yourself twice. Find your country's official class classification and place yourself. Then place yourself on economic, cultural and social capital. Note whether the two agree.

Two — find the scheme behind a statistic. Take any published figure broken down by class and identify the classification used. Then find the year it was introduced, and check whether the series you are reading crosses that date.

Three — test the substitution. Find three uses of "working class" in political commentary this week. For each, work out whether the writer means position, education, region, or voting behaviour.

Four — apply the household problem. Take a household you know with two earners in different positions. Work out its class three ways — each individually, and by dominance — and note what each answer would be useful for.

Five — try the scheme on informal work. Describe a job you know of that is casual, unregistered or platform-mediated, and try to place it in the official classification. Note exactly where it fails, and what that says about who the classification was built for.