The Impact of Group Discussion on Hiring Discrimination against Transgender Workers in India

Researchers:
Duncan Webb
Location:
India
Sample:
3,397 individuals
Timeline:
2023 - 2023
Initiative(s):
Crime and Violence Initiative (CVI)
AEA RCT registration number:
AEARCTR-0010953
Research papers:
Partners:

Discrimination imposes heavy social and economic costs. While legal protections are crucial, they often fail to change behavior. Researchers evaluated the impact of group discussions and rights messaging through videos on hiring discrimination against transgender workers in India. Individuals who engaged in group discussions and individuals who were shown a video that affirmed the legal rights of transgender individuals were more likely to hire transgender individuals. Group discussions were most effective.  

Policy issue

Discrimination carries high economic costs. Biased managers can make workers from minority groups less productive, while prejudice contributes to significant wage gaps that harm both individuals and the broader economy.1 Though individual prejudice is often seen as deep-seated and difficult to change, social norms can shift rapidly; in recent decades, many societies have seen dramatic increases in acceptance for gender equality, gay rights, and interethnic marriage.2 This suggests that interventions targeting social dynamics, not just individual beliefs, may be effective.

A well-studied approach is to foster contact between majority and minority groups, which has been shown to reduce prejudice.3 For instance, mixing Christian and Muslim youth in Nigerian classrooms successfully reduced discrimination.4 However, the reach of such interventions is limited by a simple reality: people mostly interact with others from their own identity group.5 This makes scalable intergroup contact difficult. It also raises an underexplored question for policymakers: can communication within the majority group itself be harnessed to reduce discrimination against an outside minority?

Context of the evaluation

This study focuses on the thirunangai, a community of transgender women in the state of Tamil Nadu, India, with a long-standing cultural and religious role in society. Their visually recognizable identity, however, leaves them particularly susceptible to economic discrimination and violence. Transgender people are often excluded from traditional forms of paid employment, pushing many into poverty and sex work. There are at least 0.5-1.25 million transgender people in India, implying that such discrimination results in large welfare and efficiency costs.

The study took place in urban Chennai with 3,397 non-transgender participants. This context is marked by a stark disconnect between norms and behavior. While discriminatory actions are common, there is widespread agreement that discrimination is wrong; 93 percent of comparison group participants described an act of discrimination as wrong. Recent legal changes, such as a 2014 Supreme Court ruling affirming the constitutional rights of transgender people, have had relatively limited impact on behavior, partly due to low awareness (36 percent of the participants were unaware of these rights). This gap between what people do (descriptive norms) and what they say is right (prescriptive norms) provided a unique opportunity to test whether communication could activate latent anti-discrimination sentiment.

Details of the intervention

The researcher conducted a randomized evaluation to measure the impact of group discussions between non-transgender individuals and video screenings about legal rights on hiring discrimination against transgender individuals. The researcher randomly assigned 3,397 non-transgender individuals into variations of group discussions described below. The experiment was framed as market research for a grocery delivery service to measure real hiring decisions while minimizing the risk that participants would guess the study's true purpose. The primary outcome was measured through a series of 10 private, binary hiring choices. In each choice, participants were shown photos of two potential delivery workers—one always a cisgender man and the other a cisgender man, a cisgender woman, or a transgender woman—and had to choose one. Their choice involved real trade-offs between the worker they preferred and the groceries they would receive. Participants were assigned to the following groups:

  1. 3-person discussion (890 participants): All three participants in the group engaged in a 10-minute discussion about their preferred hiring choices and then made collective decisions about which workers to select.
  2. 2-person discussion (549 participants): Two participants were randomly selected to discuss and make choices for the group while the third participant listened to the discussion.
  3. Public but no discussion (599 participants): All three participants made their individual choices with the knowledge that the choices would be later revealed to the group. This group did not engage in a discussion prior to decision making.
  4. Comparison group (1,365 participants): Participants made all ten choices individually and privately.

To compare the impact of the different group discussions with information sharing through video screenings, the researchers also randomly assigned the same 3,397 individuals into three additional groups:   

  1. Legal rights video (1,135 participants): Participants watched a 2-minute video showing information on the Supreme Court of India’s ruling giving transgender people the same fundamental rights as others.
  2. Rights messaging video (1,135 participants): Participants watched a video in which they were told that transgender people should have the same fundamental rights as others without mentioning that they have these rights.
  3. Comparison group (1,135 participants): Participants watched a video that discussed voting rights but did not mention transgender rights. 

To measure the impact of the group discussions and the videos, researchers examined private participants’ hiring choices after the intervention to see if they selected transgender workers more often. On average 35 days after the initial survey, participants were given another survey to measure the persistence of their choices.

Results and policy lessons

Both participating in group discussions and being informed about the legal rights of transgender individuals through videos reduced hiring discrimination against transgender women among residents of Chennai. One month after the intervention only individuals in the group discussions continued to discriminate less. 

Hiring choices: Compared to individuals in the comparison group that did not take part in group discussions, participants in 3-person discussions (group one) were 17 percentage points more likely to select a transgender individual for their delivery (a 42 percent increase). Participants in the 2-person discussion arm (group two) were 13 percentage points more likely while participants in the public but no-discussion arm (group three), were not more likely than the comparison group to hire transgender individuals. Participants who were offered the legal rights video and the rights messaging video were 10.3 percentage points (a 23 percent increase) and 5.8 percentage points (a 13 percent increase) more to hire a transgender individual, respectively.  

Benchmarking discrimination: Group discussion reduced participants’ willingness to sacrifice grocery items to avoid hiring a transgender worker. Participants in the comparison group were willing to forgo items worth nearly twice the median daily per capita food expenditure whereas those in the 3-person discussion arm (group one) were not willing to sacrifice any items, on average.  

Persistence of reduced discrimination: Participants in the 3-person discussion arm (group one) continued to discriminate less for 2-9 weeks after the discussions took place. These participants were four percentage points (an eleven percent increase) more likely to hire transgender individuals in a hypothetical follow-up decision compared to participants in the comparison group. In other words, 25 percent of the short-term impact persisted after approximately one month. On the other hand, participants who viewed transgender rights videos did not discriminate less one month after the intervention.

Participants’ behavior likely shifted because the group discussions changed the perceived social norms around hiring. Pro-transgender participants spoke up more often, making supportive statements 2.5 times more common than anti-transgender ones. Their arguments were morally persuasive, emphasizing values like opportunity, rights, and equality. This imbalance appears to have led participants to revise their views of how much others discriminate—raising their estimates of neighbors’ willingness to hire transgender workers by 24 percentage points.

 

Policy Lessons:

  1. Promote peer-to-peer dialogue: In the right circumstances, communication within existing social networks can be an inexpensive way to reduce discrimination. However, this might not always work.. In cases where people claim to already be against discrimination, governments and NGOs could consider designing and testing programs that facilitate structured but undirected conversations in community centers, schools, and workplaces.
  2. Target contexts where prescriptive norms are favorable: This approach is most likely to succeed in settings where a gap already exists between widespread discriminatory behavior and a widely held belief that discrimination is wrong. Attitude surveys can help identify these promising contexts.
  3. Empower internal advocates: The study shows a vocal minority can drive change. Rather than relying on outside experts, policies can be designed to identify and empower members within a community or organization who already promote fairness training them to become effective advocates for shifting group norms.

"Silence to Solidarity: Using Group Dynamics to Reduce Anti-Transgender Discrimination in India." Working Paper 2024.

1.

Glover, Dylan, Amanda Pallais, and William Pariente. 2017. “Discrimination as a Self-Fulfilling Prophecy: Evidence from French Grocery Stores.” Quarterly Journal of Economics 132 (3): 1219–60.; Charles, Kerwin K., and Jonathan Guryan. 2008. “Prejudice and Wages: An Empirical Assessment of Becker's The Economics of Discrimination.” Journal of Political Economy 116 (5): 773–809.

2.

Fernández, R., Parsa, S., & Viarengo, M. (2019). Coming out in America: AIDS, Politics, and Cultural Change. 

3.

Rao, G. (2019). Familiarity does not breed contempt: Generosity, discrimination, and diversity in Delhi schools. American Economic Review. doi: 10.1257/aer.20180044

4.

Scacco, Alexandra, and Warren Shana S. 2018. “Can Social Contact Reduce Prejudice and Discrimination? Evidence from a Field Experiment in Nigeria.” American Political Science Review , vol. 112, no. 3 (August): 654-677.

5.

 Jackson, M. O. (2011). An Overview of Social Networks and Economic Applications. In Handbook of Social Economics (Vol. 1, pp. 511–585). Elsevier. doi: 10.1016/B978-0-444-53187-2.00012-7