Support Us

Learning about Outgroups: Comparing Deep versus Broad Connections

Work & Education India

Photo Credit: Gareth Nellis

Policy Context

Ethno-religious divisions have been shown to cause conflict and impede development in low-income countries. Intergroup contact has been proposed as an important tool. First-generation studies of the contact hypothesis demonstrate the promise of collaborative intergroup contact for reducing bigotry and bias. Yet, recent landmark studies find limited or no generalized effects of contact. That is, improvements in prejudicial attitudes and behaviors remain confined to the experimental context and do not extend to the outgroup as a whole.

Professor Gareth Nellis and team hypothesize a reason why. Research on the contact hypothesis largely uses randomized controlled trials with fixed groups, where individuals interact repeatedly with the same set of outgroup members during an intervention (e.g., Lowe 2021, Ghosh 2021). But when people have contact with only a limited number of outgroup members, they may consider those individuals to be exceptional, and not prototypical, of the outgroup (Paluck and Clark 2020). Consequently, what we call Deep contact – long interactions with only a few outgroup members – may lead to less generalization than Broad contact – shorter
interactions with many outgroup members.

Study Design

CEGA funding will be used to complete an innovative, ambitious RCT that is already in the field. Nellis and team have set up a factory from scratch in Barasat, West Bengal. At this factory, the researchers are recruiting 1,000 unemployed or underemployed Hindu and Muslim adults to make bags in pairs (a collaborative production task that requires minimal training).

Individuals complete baseline surveys and then are randomly assigned to one of 4 groups for a 6-day joint production task. The 4 groups are: 1) control (no 6-day joint production task) 2) a group with rotating outgroup partners (a new partner every day), 3) a group with the same outgroup partner for all 6 days, or 4) the same ingroup partner for all 6 days (Hindus only).

The experiment is underway and CEGA funding will be used to bring 200 additional individuals into the study.



Results and Policy Lessons


  • Gareth Nellis (UCSD); co-authors: Arkadev Ghosh (Duke)
  • Matt Lowe (UBC)
  • Anujit Chakraborty (UCD)

2024 — 2025

Share Now

Copyright 2024. All Rights Reserved

Design & Dev by Wonderland Collective