New experimental evidence suggests that intergroup contact does reduce prejudice, but the effects are smaller and less transformative than earlier research suggested, highlighting the need for more rigorous evidence and better-designed interventions to understand what actually works.
Editor's note: This episode of VoxDevTalks is available on Spotify and Apple Podcasts.
For decades, the idea that bringing members of different social groups into contact can reduce prejudice has shaped policy and research. Known as the intergroup contact hypothesis, the theory has influenced everything from school integration programmes to peacebuilding initiatives around the world.
In this episode of VoxDevTalks, Matt Lowe discusses new evidence that questions how large these effects really are. His recent review revisits decades of research and suggests that the benefits of intergroup contact may have been significantly overstated.
The conversation explores where the idea came from, how it shaped policy, and why more rigorous research methods are now producing more modest estimates of its impact. While the results may challenge long-held assumptions, Lowe argues that contact still has value – and that the field is entering a promising phase of more credible and ambitious research.
From theory to policy: how contact shaped real-world interventions
The idea that prejudice could be reduced through structured contact quickly gained influence following the publication of Gordon Allport’s The Nature of Prejudice, particularly during the civil rights era in the US. Policymakers and researchers saw it as a practical framework for improving relations between social groups. If the right conditions could be created, then intergroup contact might help reduce racism and discrimination. One of the most prominent examples was school desegregation, including the controversial policy of bussing students between neighbourhoods to integrate schools.
Contact theory also influenced programmes designed to reduce segregation within institutions. For example, schools sometimes introduce 'house systems', which randomly assign students to groups that cut across existing social divisions.
Beyond schools, contact-based approaches appear in many development and peacebuilding programmes. Governments and NGOs frequently design initiatives that bring people from different communities together to collaborate.
Examples include youth exchange programmes, conflict-resolution projects, and national service schemes that place young adults in different regions. As Lowe notes, many of these initiatives are motivated by the hope that shared experiences and cooperation will foster mutual understanding.
Why early evidence was weaker than it appeared
Despite its widespread influence, the empirical foundations of contact theory were relatively weak for much of its history. Early studies relied heavily on observational correlations, rather than controlled experiments. Researchers often measured how much contact people reported having and then compared it with their attitudes toward other groups. The problem with this approach is that it cannot clearly establish cause and effect.
“We would really worry that if you see correlation between contact and prejudice, it could be because the less prejudiced people are more likely to select into having more contact with the out-group.”
In other words, people who already have tolerant attitudes may simply be more willing to interact with members of other groups.
Another challenge was publication bias. Studies showing positive effects were more likely to be published than those showing no effect or negative results. Over time, this can create the impression that a theory is more strongly supported than it really is. These issues became clearer during the broader “credibility revolution” in social science, which emphasised rigorous experimental methods and transparency in research design.
What new experimental research reveals
In the past decade, researchers have begun running large-scale field experiments that directly test the effects of intergroup contact. These studies randomly assign participants to interact with members of other groups and compare the outcomes with control groups. Because participants are randomly assigned, the results provide much stronger evidence about causal effects. However, the findings have often been more modest than earlier research suggested.
For instance, experiments bringing together Christians and Muslims in Nigeria through vocational training programmes produced mixed results. Another study in Iraq created mixed football teams, encouraging cooperation on the field. While players continued to train together, the effects did not extend much beyond the sporting context.
As Lowe notes, these studies were well executed, but they produced “quite mixed, limited effects.” Meta-analyses of earlier research had estimated very large impacts from contact – around 0.4 standard deviations, suggesting substantial changes in attitudes and behaviour.
But Lowe’s new analysis focuses specifically on pre-registered experiments, where researchers publicly specify their methods and outcomes before collecting data. This approach reduces the risk of selective reporting. The result is a much smaller estimated effect – roughly 0.1 standard deviations.
Can we design better contact interventions?
Although the average effect of contact appears smaller than previously believed, researchers are now exploring ways to make such interventions more effective. One important question is how contact is structured. For example, Lowe and his collaborators have studied the difference between broad contact – interacting briefly with many people from another group – and deep contact, which involves longer interaction with fewer individuals.
Their findings suggest that different types of contact influence different outcomes. Broader contact can improve general beliefs about other groups, while deeper interactions are more likely to produce lasting social and economic relationships.
Researchers are also testing variations in the conditions originally proposed by Allport, such as equal status and shared goals. Experiments show that these conditions tend to improve the effectiveness of contact, even if they do not guarantee large effects.
Another key factor may be duration and intensity. Many experiments involve relatively short interactions, sometimes lasting only a few hours. Longer and more sustained engagement may produce stronger outcomes.
Why prejudice reduction research matters today
Understanding how to reduce prejudice remains a pressing issue in many societies. Despite economic development and social change, discrimination and intergroup tensions continue to persist. Lowe emphasises that this makes the search for effective policies particularly important.
“We might have thought that prejudice would be something that would just go away with development, or with new generations replacing old generations, but that's not really what we see.”
The good news is that the field of prejudice reduction is evolving rapidly. Researchers are increasingly running larger, more rigorous experiments and testing more ambitious interventions. While the evidence suggests that contact alone may not be a silver bullet, it still appears to play a useful role – particularly when combined with other approaches. As Lowe notes, contact is often relatively inexpensive to incorporate into programmes, and even small improvements in intergroup attitudes can still be worthwhile.
References
Lowe, M (2025), "Has intergroup contact delivered?" Annual Review of Economics 17: 321–344.