House of Castes: A School-Level Experiment on Intergroup Relations in India

Policy Context

Caste-related violence among youths in Tamil Nadu has prompted government efforts to reshape caste relations. Beyond violence, a more general phenomenon is the wearing of caste-colored wristbands – a reflection of strong caste pride. Pilot data (N=636) shows that 54% of students report wearing such wristbands. In addition, researchers have found that children as young as seven show strong same caste bias in friendship formation.

In response to these issues, one Indian Administrative Service (IAS) bureaucrat, Johny Tom Varghese, landed on the idea of introducing the House System to schools, with (i) students and teachers randomly assigned to one of four houses, represented by four different caste-neutral colors, (ii) houses scoring points for the good behavior – in academics, sports, and otherwise – of individual students, (iii) team activities conducted in houses, with houses earning points based on team performance, and (iv) prizes for the house with the most points at the end of each academic year. Varghese successfully introduced the House System to all 293 secondary schools in the district of Ramanathapuram. In collaboration with Varghese, Professor Matt Lowe and team seek to evaluate the effects of the House System through a school-level RCT with primary and secondary schools in three new districts.

Study Design

Working with NGO implementing partner, Vidhya Vidhai, researchers are in the process of selecting 300 secondary schools in Tirunelveli and Thoothukudi districts and 300 primary schools in Ramanathapuram with strong caste tensions. Amongst these schools, the research team will randomize half to have the House System introduced in summer 2024, at the start of the next academic year. Control schools will be without the House System for at least three years permitting the team to explore longer-term effects.

The team will measure caste and house relations using surveys at the end of each school year, beginning in summer 2025. Measures will include (i) social networks, (ii) reports of bullying and violence (37% report being bullied in our pilot data), (iii) social preferences as measured with dictator games played with ingroup and outgroup partners, (iv) the prevalence of wearing caste-colored wristbands, and (v) measures of well-being and depression.

Results and Policy Lessons

Forthcoming.