Contact, conflict, or opportunity?

The question

Three leading theories make competing predictions about what happens when people encounter more out-group members. Contact theory holds that exposure reduces prejudice and fosters cross-group ties. Conflict theory predicts the opposite: a growing out-group triggers threat, intensifying in-group preference. Structural opportunity theory argues that observed patterns merely reflect the changing pool of available partners—preferences stay put. We test all three in Czech lower-secondary classrooms, where classroom composition varies substantially across schools and grades, creating a natural quasi-experiment in diversity.

Data

Seven independent datasets collected in Czech lower-secondary schools, covering 228 classrooms and nearly 5,000 students across grades 5–9. Each dataset records peer-nominated friendship and rejection ties alongside three demographic attributes: gender, ethnicity (Czech or non-Czech), and socio-economic status. The collection includes a nationally stratified sample of sixth-grade classrooms (PanelSYRI) and six regional studies from Prague, Brno, and the South Moravian region. Combining them maximises variation in classroom composition across datasets, approximating a natural experiment in demographic diversity.

Approach

We combine seven independent datasets covering nearly 5,000 students in 228 classrooms. Each student provided both friendship and rejection nominations, letting us examine positive and negative ties separately. To disentangle preference from opportunity, we model the nomination process using the Wallenius non-central hypergeometric distribution—an urn model that accounts for the finite, asymmetric pool of available classmates and the depletion of that pool with each successive choice. A multilevel model links each student's estimated preference parameter to classroom composition, yielding posterior model probabilities for each of the three theories across gender, ethnicity, and socio-economic status.

Findings

The results are sharply asymmetric across social dimensions.

For ethnicity and socio-economic status, preferences are largely invariant to classroom composition — consistent with structural opportunity theory. More diverse classrooms create more cross-group contact opportunities, but students' underlying willingness to nominate out-group members as friends stays put. What changes is who they can befriend, not who they want to.

For gender, the pattern flips. Same-gender preference in friendships intensifies as the other-gender share grows — supporting conflict theory. A larger out-group is experienced as threat rather than opportunity: the more other-gender students in the room, the stronger the pull toward same-gender ties. Contact theory finds no support across any social dimension or tie type.

What this means

Compositional diversity — mixing classrooms — reliably expands the structural opportunity for cross-group contact. But it does not, by itself, shift the preferences that determine whether students take that opportunity.

This has concrete implications for school integration policy. Ethnicity and socio-economic status seem amenable to structural interventions: give students more cross-group contact opportunities and some ties will form. Gender segregation is more stubborn — and can actually intensify under naive mixing. Interventions that aim to shift gender norms or reduce threat perception may be necessary before compositional change has the hoped-for effect.

Preprint