Using interdisciplinary research to combat bias and discrimination
2020 marked a global shift in the public discourse around bias and discrimination. While the Covid-19 pandemic led to greater disparities in healthcare access and financial instability, protests erupted around the world to denounce systemic racism and police brutality. Momentum from these events and decades of work by grassroots and social movements has motivated governments, NGOs, donors, and members of the private sector to redouble efforts to fight discrimination and increase diversity and inclusion.
What is the role of research in helping to address this persistent challenge? Across disciplines such as psychology, sociology, economics, and more, researchers have been documenting the nature and consequences of discrimination around the world. As support for anti-discrimination policies and programs continues to grow worldwide, another area in which researchers could play an instrumental role is in helping us to understand what policies to combat bias and discrimination are the most effective, and how best to tailor them to individual contexts.
With these questions in mind, J-PAL hosted a webinar on October 25, 2021, on the need for more rigorous, interdisciplinary research in this space. In particular, randomized evaluations can help fill this gap by assessing the impact of different interventions in real-world settings. Because effective strategies should account for human psychology, as well as historical and sociological context, interdisciplinary research is another critical piece towards countering discrimination.
We were joined by Betsy Levy Paluck (Princeton University), Salma Mousa (Yale University), Sendhil Mullainathan (University of Chicago; J-PAL affiliate), and Mario Small (Harvard University) for a conversation moderated by Marianne Bertrand (University of Chicago; Co-Chair, J-PAL’s Labor sector). Read on for a recap of key insights from the researchers.
The state of the field
Over the course of the session, the panelists highlighted the ways in which rigorous interdisciplinary research could help identify promising anti-discrimination strategies. Salma Mousa, an assistant professor of political science, presented findings from a randomized evaluation that tested whether positive and cooperative interactions through mixed religion soccer leagues can improve relations across groups in post-conflict communities. The positive results of this study highlight how randomized evaluations can be leveraged to combine methods from economics and theory from psychology to measure the impact of anti-bias interventions in real-world settings.
Betsy Levy Paluck, a professor of psychology and public affairs, summarized findings from a meta-analysis of prejudice reduction interventions. She shared that despite a big uptick in prejudice reduction research, many studies still face methodological challenges, such as small sample sizes and reliance on light-touch interventions, leading to modest overall impacts.
Mario Small, a professor of sociology, shared insights from his study around sociological perspectives on racial discrimination. He argued that discrimination models in economics are valuable but insufficient to account for the role organizational and institutional practices play in reinforcing discrimination.
Sendhil Mullainathan, a professor of computation and behavioral science, discussed the importance of algorithms in shaping beliefs and interactions on social media, and in decision-making in health care and labor markets. He argued that if deployed correctly, algorithms could be powerful tools for curbing individual and institutional-level discrimination, and randomized evaluations could help us to understand the impacts of different algorithms.
Looking forward
These varying fields bring their own unique perspective and valuable insights on potential means of combating discrimination. Combining insights from across disciplines can help policymakers develop effective solutions that consider historical context, local norms, human psychology, and more. It could also enable the development of programs that counter discrimination in both formal institutions and social settings, where norms often develop.
However, there is not yet enough rigorous, field-based research in real-world settings on the effectiveness of interventions to combat discrimination. Developing this evidence base is crucial for scaling and implementing effective anti-discrimination policies and programs. For instance, panelists highlighted that many well-intentioned organizations have been investing in diversity training programs, despite the fact that their impact is not clear. To truly improve diversity in the workplace, it is critical to ensure such programs are achieving their stated goals.
In addition to reducing bias and discrimination in workplaces and other organizations, other themes requiring more real-world research include identifying what tools effectively counter innate bias and prejudice, addressing hate speech and discrimination in online platforms (including via algorithms), and reducing discrimination in services such as healthcare, criminal justice, and more.
Programs may also have varying impacts across different contexts. Implementing similar programs in different regions around the world could help shed light on what contextual factors are critical for impact. For example, based on the findings of the Iraq evaluation, Salma Mousa and co-authors are currently replicating the soccer leagues in Lebanon with native Lebanese youth, Syrian refugees, and those descended of Palestinian refugees.
Despite widespread support for anti-discrimination policies and programs, the impact of many common interventions is not clear. Combining interdisciplinary insights on ways to counter discrimination with rigorous, real-world impact evaluation are critical steps towards fostering diversity, equity, and inclusion.
Interested in learning more about J-PAL’s efforts in this space? Contact Anupama Dathan, Policy Manager, at [email protected].
Missed the webinar? You can watch a recording of it here: