The contact hypothesis, seventy years of testing

Gordon Allport published in 1954 a book setting out four conditions for contact capable of reducing prejudice: equal status in the situation, a common goal, cooperation rather than competition, and institutional support. The book, The Nature of Prejudice, appeared in segregated America, the very year of the Brown v. Board of Education ruling.

The publicly retained formula, that mixing is enough, is therefore a weakened version of the initial proposition, which insisted on the conditions. Allport himself warned against the idea that mere co-presence could suffice, since superficial contact can on the contrary reinforce stereotypes.

Thomas Pettigrew and Linda Tropp published in 2006 a meta-analysis of several hundred studies. They find a negative association between contact and prejudice, of moderate size, found in varied contexts.

They also find that Allport's conditions strengthen the effect without being strictly necessary, a result that fed a lasting debate about the theory's reach.

An extension proposed in 1997 by Stephen Wright and colleagues, known as extended contact, holds that knowing that someone close has a friend in the other group would be enough to reduce prejudice, without direct contact. It links this field to that of friendship networks.

People who have contact with a group are also those with the least prejudice to begin with. The measured correlation therefore mixes the effect of contact with the selection of contacts. This is the classic objection to any correlation between mixing and attitudes.

Longitudinal analyses and the rare designs exploiting unchosen allocation, notably of university roommates, find real but weaker effects than cross-sectional correlations. The random allocation of roommates, practised by some universities, does indeed provide a situation in which contact was not chosen.

Several recent studies, including field trials, found no lasting effect, and some measured reverse effects in contexts of competition for resources. A review published in 2019 by Elizabeth Levy Paluck, Seth Green and Donald Green noted the small number of randomised studies measuring effects beyond the intervention itself.

Those results have led to specifying the theory's domain of validity rather than abandoning it: it describes what contact produces under certain conditions, not what mere co-presence produces in general.

This literature meets that on induced homophily: contact depends on the composition of schools, neighbourhoods and workplaces, which determines possible encounters.

That is why the most recent work is less interested in attitudes than in allocation policies, which alter exposure without intervening on people. The reasoning changes scale: the aim is no longer to persuade people but to alter the opportunities open to them.