Combatting Caste based Discrimination in the Education Sector

The field experiment will be conducted on an intended sample of ~500 primary school teachers from several states in India.

2021-12-02

2021-12-31

We have two sets of primary outcomes:

Teachers’ behavior in two tasks: (i) bias in grading exams attributed to SC and GC students, where the objective quality of the exams is experimentally engineered; and (ii) donations to NGO’s promoting welfare of caste groups.

Measures of teachers’ attitudes towards SC individuals. We collect six intertwined and complementary measures: (i) willingness to forego monetary payoff to avoid supporting initiatives to employ more SCs in the capacity of cooks in schools; (ii) perceptions of gaps in cognitive outcomes between primary school children from scheduled caste (SC) and general category (GC); (iii) differences in perceived perseverance of SC and GC primary school children; (iv) associations between castes and work occupations; v) attitudes towards the use of caste slur; and vi) caste-based implicit association test scores

Teachers’ behavior:
(i) Bias in grading exams – teachers are provided exams to grade where quality is either held constant or made to vary by the experimenter and the caste identity is exogenously assigned. For instance, teachers will be provided exams which contain the following combinations: (i) identical quality with different caste names; (ii) high quality exam assigned to high caste name; and (iii) high quality exam assigned to low caste name. These combinations will also help us evaluate whether biases become stronger when perceived traditional hierarchies are overturned as represented in case.
(ii) Donation to NGO’s promoting the welfare of caste groups – at the end of the experiment teachers are asked whether they wish to donate any amount out of their show-up fee to a charity promoting the educational success of Brahmins (high ranked caste group) or SCs.
Measures of caste-based attitudes on six dimensions:
(i) Willingness to forego monetary payoff to avoid supporting initiatives to employ more SCs in the capacity of cooks in schools – this involves an incentivized decision by teachers about whether they hypothetically support an initiative that would hire SC as cooks for preparing meals served to school children as part of the mid-day meal scheme. The incentives are structured such that higher levels of support for such an initiative result in higher monetary payoffs.
(ii) Perception of cognitive outcomes on vocabulary and cognitive tests for primary school children from SC and GC – teachers are told the scores that 5 year old children from the GC achieved in vocabulary and cognitive tests conducted by the Young Lives Project. Then, they are asked to provide their incentivized guess for the scores achieved in the same tests by SC children of the same age.
(iii) Perception of perseverance among primary school children from SC and GC – teachers are asked to provide their estimates on how likely three students (caste identifiable by name) are to finish a new school activity that requires them getting up early and donating half-an-hour of their free time at least three days in a week. This question is not incentivized.
(iv) Assignment of individuals to occupations traditionally associated with castes – we ask teachers to assign each of five individuals to one of five occupations which have a clear socioeconomic hierarchy, as well as association with caste based occupations. The individuals are described as having the same level of education and diligence, and their caste is clearly identifiable from their assigned names. This question is not incentivized.
(v) Attitudes towards the use of caste slur: teachers are presented with a text describing an event in which a SC individual is addressed with a derogatory caste-slur. Teachers are asked to select one among several given options to best describe the event. One of these options describes the event as discriminatory, because of the use of caste-slur. This question is not incentivized.
(vi) Caste-based implicit association test scores – these are standard implicit association tests, using caste as the identity marker to measure implicit caste-bias.

We also intend to look at whether negative attitudes are relevant when considering the domain of gender. We hence also collect data on two gender related attitudes:
(i) Willingness to allow food menstruating women to cook for their family – this involves an incentivized decision by teachers about whether they hypothetically agree to allow women in their family who are menstruating to be able to work in their kitchen. Giving assent gives the teacher a positive monetary payoff, thus refusal to allow women in their family suggests a psychological cost for breaking gender-based norm related to taboos of menstruation.
(ii) Perception of cognitive outcomes on vocabulary and cognitive tests for primary school girls and boys – teachers are asked to provide incentivized guess about the scores of 5-year-old girls in a vocabulary and cognitive test when provided with the scores that boys achieved in the same assessment, conducted by the Young Lives Project.

The treatments are designed to mitigate widely held views among GC teachers. We refer to the two treatments as the success story treatment and empathy treatment.

Accordingly, our experiment incorporates two specific treatments and intends to measure their impact on the set of primary outcomes outlined above.

Success story treatment: subjects are presented a short text describing how three SC individuals managed to reach entrepreneurial success notwithstanding their humble origins and the discrimination they faced. The text is complemented by head shot pictures of the three entrepreneurs, as literature suggests that the messaging is enhanced by use of pictures and videos. In the control condition, subjects are presented with a similar short text (and headshots) describing how three western entrepreneurs managed to reach success, notwithstanding their humble or average origins.

Empathy treatment: The treatment involves showing teachers two short, animated videos. In the first animated video, a GC Indian doctor working at a foreign hospital faces discrimination from a patient due to their origins. In the second video, a SC doctor in India faces similar discrimination by a GC patient due to their caste. The objective is to highlight how discrimination is often arbitrary and could also be faced by someone similar to them, and thus generate empathy for lower caste individuals who often face discrimination.In the control condition individuals are shown a short animated video about rising costs in the hospital system.

Subjects are randomized into one of the two treatments – success or empathy treatment. The randomization is done using a computer.

The unit of randomization is the individual subject who participated in the experiment.

No

Around ~500 Teachers

Around ~500 Teachers

30% Success Video Treatment; 30% Empathy Treatment; 40% Control