In-Groups, Out-Groups, Reference Groups, Peer Groups

Belonging to a group does something immediate. It draws a line — and everyone on the other side of it becomes, in some measure, them.

This lesson takes four concepts that describe how membership organises perception: the group you are in, the group you are not, the group you measure yourself against, and the group of your equals.

In the early 1970s the social psychologist Henri Tajfel ran a series of experiments designed to find the minimum conditions under which people would begin favouring their own group.

He expected to have to build up: shared interests, competition for a prize, some history of interaction. So he began with a deliberately absurd baseline, intending it as a control.

Schoolboys were shown pairs of abstract paintings and told they had been sorted into a "Klee group" and a "Kandinsky group" on the basis of which they preferred. In some versions the sorting was openly random — they were told it was decided by a coin. The boys never met the others in their group. They did not know who was in it. There was no contact, no history, no shared interest, nothing to compete over, and nothing to gain personally.

They were then asked to allocate small sums of money between two other participants, identified only by a code number and a group label.

They favoured their own group.

More striking still: given a choice between an option that gave their own group more money in absolute terms, and one that gave their group less money but more than the other group, a substantial number chose the second. Maximising difference mattered more than maximising benefit.

Groups that had existed for four minutes, based on nothing, produced discrimination.

That result should be handled carefully rather than dramatically, and this lesson will do both.

Taken naively, it says human beings are hard-wired to divide the world into us and them, and that everything from school rivalry to communal violence is an expression of an inbuilt tendency. That reading is popular, and it is exactly the reductionism 1.4.1 and 1.5.3 warned you about — it explains a variable outcome (some societies are far more riven than others; the same two groups are hostile in one decade and intermarrying the next) by an unvarying constant.

Taken carefully, it says something more useful and more interesting: the categorisation itself does work. Simply being told there is a line is enough to start producing differential treatment — which means the sociologically important question is not whether people form in-groups, but which lines a society makes salient, who draws them, and what is attached to them.

That is a question about power and history. Nothing is hard-wired about caste, nation, religion or race. The capacity to treat a line as meaningful is general; which line, and what follows from it, is entirely social.

In-group and out-group

The terms come from William Graham Sumner (1906) — the same book as folkways (2.1.3) and ethnocentrism (2.1.8), and the connection is not accidental. For Sumner, ethnocentrism is the in-group's view of itself as the centre and standard.

Four well-documented consequences of the in-group/out-group division, each with sociological rather than merely psychological significance:

In-group favouritism. Members are given the benefit of the doubt, preferential treatment, and more generous interpretations of their behaviour. Note that in most real settings, discrimination operates far more through favouring one's own than through overt hostility — which is why it is so hard to detect and so easy to deny. The referral network in 1.3.2's job interview was in-group favouritism, and nobody in it was hostile to anybody.

Out-group homogeneity. "They are all the same"; "we" are individuals with differences. This is why members of a dominant group experience themselves as persons and members of a marked group are experienced as representatives (2.1.12, 2.2.2's tokens).

Attribution asymmetry. An in-group member's good act reflects character and their bad act reflects circumstances; for an out-group member the attributions reverse. This is the fundamental attribution error (1.4.1) with a boundary added, and it is the engine of a great deal of stable prejudice.

Boundary maintenance. Effort is spent policing who counts as a member — 2.1.7's boundary work, and 2.1.5's symbolic boundaries. Note that the boundary usually matters more than the content: what defines "us" is frequently just not being them.

Reference groups: the people you measure yourself against

A distinct and enormously useful concept, introduced by Herbert Hyman and developed by Merton (5.1).

The crucial move is that last clause. Your reference group need not be your membership group. People routinely take their standards from groups they aspire to join, and evaluate their lives against the situation of people they do not know.

Two mechanisms follow, and both are among the most useful ideas in Part 2.

Anticipatory socialisation. Adopting the values, manners, speech and outlook of a group you hope to enter, before entering it. The student who begins dressing and speaking like the profession she intends to join; the migrant learning the destination's codes before departure; the junior officer who starts thinking like a senior one. Merton's finding was that anticipatory socialisation predicts advancement in open systems — and produces isolation in closed ones, because if you cannot in fact enter, you have distanced yourself from your own group for nothing (2.1.9's cleft habitus).

Relative deprivation. Dissatisfaction depends not on absolute conditions but on comparison with a reference group. This came out of The American Soldier — the same study behind Lazarsfeld's reversal in 1.4.6 — and produced a genuinely counterintuitive finding: military police, whose promotion prospects were poor, reported greater satisfaction with promotion than air corpsmen, whose prospects were excellent. Because in a unit where few are promoted, not being promoted is normal; where many are, it is a personal failure.

That single finding explains an enormous amount that puzzles people:

Why rising prosperity can be accompanied by rising discontent.
Why revolutions frequently occur during periods of improvement rather than at the depths of misery (11.2).
Why people compare themselves with neighbours and colleagues rather than with the very rich, and why segregation therefore dampens resentment (1.2.2).
Why a pay rise that everyone receives feels different from the same rise given to you alone.

Practical consequence: if you want to understand someone's grievance, do not measure their conditions. Find out whom they are comparing themselves with.

Peer groups

Its distinguishing feature is horizontality. Family, school and workplace are all vertical: someone is in charge. The peer group is where a person deals with equals, and that makes it a distinctive socialising environment (2.3).

Three things worth knowing:

It is the first group most people choose. Family is ascribed; peers are, in some measure, selected. This makes the peer group the first arena in which a person practises being someone in particular.

Its influence peaks in adolescence — not because of the psychology of teenagers alone, but because of a structural fact: modern societies place large numbers of same-age people together in institutions for years, with limited adult supervision of their interactions, at exactly the age when adult roles are being anticipated. Age-graded schooling manufactures peer groups (9.2). Societies without it produce a different adolescence.

And it is where subcultural formation happens (2.1.7): a peer group with a status hierarchy of its own is precisely the alternative scoreboard that lesson described.

The best-known study of in-group hostility is Muzafer Sherif's Robbers Cave experiment (1954). Twenty-two boys at a summer camp were divided into two groups, which quickly developed names, norms and flags; competitive games were introduced; hostility escalated to raids and name-calling; and the conflict was then reduced by superordinate goals — tasks requiring both groups to cooperate, such as fixing the camp's water supply.

The findings are frequently taught as established fact. Two things should be said.

What has held up: superordinate goals are a genuinely well-supported mechanism for reducing intergroup hostility, replicated in many settings, and consistent with Allport's contact hypothesis — the finding that contact between groups reduces prejudice only under conditions: equal status in the situation, common goals, cooperation rather than competition, and institutional support. Contact without those conditions can increase hostility, which is why desegregation efforts that ignore the conditions have sometimes backfired.

What should be qualified: archival work on Sherif's records — most prominently by the journalist Gina Perry — has shown that an earlier attempt at the same study failed when boys refused to turn on each other, that this failure went unreported, and that experimenters actively encouraged hostility in the successful run. That does not make Robbers Cave worthless; it makes it a demonstration that hostility can be manufactured under certain conditions, rather than evidence that it emerges spontaneously.

Why this matters for you as a student: 1.4.6 warned that a finding sounding obvious is not evidence. This is the companion warning — a study being famous is not evidence either. The mature use of any classic study is to know what it showed, what it did not, and what happened when people looked again. You will impress no one by citing Robbers Cave, Zimbardo or the Hawthorne effect without the qualifications; you will be doing sociology if you supply them.

Go deeper

Which lines become salient, and why

Return to the puzzle. If any line can produce in-group favouritism, the sociological question is which lines a society activates.

Four observations from the research on ethnic and communal conflict (8.5, 9.5, 11.2):

Salient boundaries are made, not found. Categories that were once locally fluid can be hardened by administration, politics and violence. The colonial census in India, by requiring one caste and one religion per person, contributed to converting flexible local identities into countable political blocs (1.3.3, 6.2.1). Nations are, in Anderson's phrase, imagined communities — real, powerful, and constructed within recent history (9.5).

Entrepreneurs activate them. Boundaries do not become politically salient by themselves. Specific people and organisations benefit from mobilising them, and the timing of communal or ethnic polarisation usually tracks political competition better than it tracks any underlying cultural difference.

Structural conditions matter. Competition over scarce resources, segregation that reduces cross-cutting ties, and institutions that allocate by category all raise salience.

Cross-cutting ties reduce it. Where people's group memberships overlap in complex ways — a trade union containing several religions, a neighbourhood mixing castes, intermarriage — a single line cannot cleanly divide the population, and mobilisation becomes harder. Ashutosh Varshney's comparative study of Indian cities found that towns with dense inter-communal civic associations — business groups, unions, clubs, festivals organised jointly — had substantially fewer communal riots than otherwise similar towns without them. Everyday cross-cutting association, not merely goodwill, did the work.

That last finding is worth carrying, because it converts an apparently intractable problem into an institutional one.

Tajfel and Turner's social identity theory proposes that people derive part of their self-concept from group memberships, and are therefore motivated to see their groups as positively distinct — which explains the minimal group results.

Sociology's response should be neither dismissal nor adoption. The mechanism is real and well-evidenced at the level of individual cognition. What it cannot do is explain the content: which groups exist, which are ranked, which are enforced by law and violence, and why the same two groups coexist for a century and then do not. Those require history, institutions and power.

Use social identity theory to explain the readiness. Use sociology to explain the arrangement. This is 1.4.1's division of labour, applied to a real case.

"The minimal group experiments prove tribalism is human nature." They show that categorisation alone can produce differential allocation. Which lines matter, how much, and with what consequences is social and variable — and that variation is the thing to be explained.

"In-group bias means hatred." Most discrimination operates through favouring one's own, not through hostility. This is what makes it deniable and durable.

"Contact reduces prejudice." Only under conditions — equal status, common goals, cooperation, institutional support. Contact without them can worsen matters.

"A reference group is a group you belong to." It need not be. That is the whole point of the concept.

"Relative deprivation means people are ungrateful." It means satisfaction is comparative. This is a finding about how evaluation works, not a moral judgement.

"Peer influence is a teenage problem." Peer groups operate at every age — professions, neighbourhoods, colleagues. Adolescence is distinctive because schooling manufactures dense same-age groups.

"Robbers Cave showed hostility arises naturally." The archival record complicates it considerably. Cite it with the qualification.

Why it matters

It explains grievance better than any measure of conditions. Relative deprivation is the single most useful idea in this lesson: to understand dissatisfaction, find the comparison. It applies to a workplace, a marriage, a caste, a region and a nation.

It shows where prejudice actually lives. Not primarily in hostility but in favouritism, homogeneity perception, and asymmetric attribution — mechanisms that operate in decent people who would sincerely deny prejudice, and which therefore require structural rather than attitudinal remedies.

It identifies a workable lever. Cross-cutting ties and superordinate goals are among the best-evidenced instruments for reducing intergroup hostility, and both are institutional: they can be built.

And it teaches you how to use a famous study. Which is a skill Part 7 will demand of you continuously, and which separates a reader from a repeater.

In-group and out-group (Sumner): the "we" and the "they." Consequences: in-group favouritism (the main engine of real-world discrimination), out-group homogeneity, asymmetric attribution, and boundary maintenance.

Tajfel's minimal group experiments show that categorisation alone, even when random, produces differential allocation — which makes the sociological question not whether people form groups but which lines a society makes salient, who draws them, and what is attached.

A reference group supplies the standards by which people evaluate themselves, whether or not they belong to it. Hence anticipatory socialisation (adopting a group's ways before joining — predictive of advancement in open systems, isolating in closed ones) and relative deprivation (satisfaction depends on comparison, not conditions — from The American Soldier).

Peer groups are horizontal groups of equals; modern age-graded schooling manufactures them, which is why adolescence is distinctive.

Robbers Cave supports superordinate goals and, with Allport's contact hypothesis, shows contact reduces prejudice only under specific conditions — but its archival record requires honest qualification.

Salience is made: by administration and census, by political entrepreneurs, by competition and segregation — and reduced by cross-cutting ties (Varshney's finding on Indian cities).

In-group — a group one belongs to and identifies with.

Out-group — a group one does not belong to and may regard with indifference or hostility.

In-group favouritism — preferential treatment of one's own group members, often without hostility to others.

Out-group homogeneity — the perception that members of another group are more alike than members of one's own.

Boundary maintenance — the ongoing work of policing who counts as a member.

Minimal group paradigm — Tajfel's experimental design showing discrimination following arbitrary categorisation.

Social identity theory — Tajfel and Turner's account of self-concept derived from group membership.

Reference group — a group whose standards one uses for self-evaluation, membership or not.

Anticipatory socialisation — adopting the norms of a group one hopes to join.

Relative deprivation — dissatisfaction produced by comparison with a reference group rather than by absolute conditions.

Peer group — a horizontal group of people of roughly equal standing and age.

Superordinate goal — a shared objective requiring cooperation between conflicting groups.

Contact hypothesis — Allport's conditional finding that intergroup contact reduces prejudice given equal status, common goals, cooperation and institutional support.

Cross-cutting ties — overlapping memberships that prevent a single boundary from cleanly dividing a population.

Imagined community — Anderson's account of the nation as a constructed community of strangers (9.5).

One — find the comparison. Think of something you are dissatisfied with. Identify precisely whom you are comparing yourself with. Then imagine a different comparison group and notice whether the dissatisfaction changes. That is relative deprivation, from the inside.

Two — audit favouritism. Recall the last time you helped someone get an opportunity — a referral, an introduction, a recommendation. How did you know them? No hostility was involved, and a boundary was operating.

Three — map your cross-cutting ties. List your significant group memberships. Do any of them contain people who differ from you on a line your society treats as salient? Varshney's finding suggests those associations are doing more work than they appear to.

Four — spot anticipatory socialisation. Identify someone adopting the manner, speech or dress of a group they have not yet joined. Then ask the harder question: is the system open enough for it to pay off?