Field
Trial End Date
|
Before
May 31, 2023
|
After
August 31, 2023
|
Field
Last Published
|
Before
March 13, 2023 08:48 AM
|
After
May 31, 2023 11:27 AM
|
Field
Intervention End Date
|
Before
May 31, 2023
|
After
August 31, 2023
|
Field
Experimental Design (Public)
|
Before
There are two main treatment variations.
First, at the start of the main survey, participants are shown one of three different videos:
1. Control video - gives no information about transgender rights, and includes placebo information about voting rights in India
2. Messaging video - the participant is shown persuasive messaging, arguing that transgenders in India should have fundamental rights
3. Law video - the participant is shown a video that tells them that transgender people actually do have legal rights in India because of a Supreme Court ruling in 2014
In addition, participants are cross-randomised into one of two conditions:
1. Discussion - respondents take part in a discussion among a group of 3 neighbours, in which they discuss their preferred worker and make a collective hiring choice
2. No discussion - respondents do not take part in a discussion.
See pre-analysis plan for further details on the experimental design.
|
After
There are two main treatment variations.
First, at the start of the main survey, participants are shown one of three different videos:
1. Control video - gives no information about transgender rights, and includes placebo information about voting rights in India
2. Messaging video - the participant is shown persuasive messaging, arguing that transgenders in India should have fundamental rights
3. Law video - the participant is shown a video that tells them that transgender people actually do have legal rights in India because of a Supreme Court ruling in 2014
In addition, participants are cross-randomised into one of two conditions:
1. Discussion - respondents take part in a discussion among a group of 3 neighbours, in which they discuss their preferred worker and make a collective hiring choice
2. No discussion - respondents do not take part in a discussion.
See pre-analysis plan for further details on the experimental design.
In phase 2 of the experiment, two additional treatment arms will be added:
1. Pairwise discussion - 2 respondents in a group take part in a discussion, and the 3rd respondent listens
2. Public choices - respondents choose while knowing that their responses will be made public to the others in their group
|
Field
Planned Number of Clusters
|
Before
Approximately 750 groups of 3 respondents
|
After
Approximately 1133 groups of 3 respondents
|
Field
Planned Number of Observations
|
Before
Approximately 2250 respondents
|
After
Approximately 3400 respondents
|
Field
Sample size (or number of clusters) by treatment arms
|
Before
N = 300 for each of the following arms:
1. Control video + No Discussion
2. Messaging video + No Discussion
3. Law video + No Discussion
4. Control video + Discussion
5. Messaging video + Discussion
6. Law video + Discussion
For the 7th treatment arm (Mixed videos + Discussion), N = 450
|
After
N = 300 for each of the following arms:
In total:
1. No discussion: N = 1300
2. Discussion: N = 900
3. Public choices: N = 600
4. Pairwise discussion: N = 600
Each will be cross-randomised with the information treatment, so that:
1. Control video: N = 1133
2. Messaging video: N = 1133
3. Law video: N = 1133
|
Field
Power calculation: Minimum Detectable Effect Size for Main Outcomes
|
Before
Main comparison between 6 treatment groups:
Each participant states their preferences over 6 pairs of delivery options. 2 of these pairs includes 1 transgender delivery worker. Based on simulations from pilot data from 120 individuals, this yields an MDE of 15 percentage points (0.3SD) on pairwise comparisons between the 6 treatment arms (assuming 𝛼 = 5%, power = 80%).
|
After
Main comparison between 6 treatment groups:
Each participant states their preferences over 6 pairs of delivery options. 2 of these pairs includes 1 transgender delivery worker. Based on simulations from pilot data from 120 individuals, this yields an MDE of 15 percentage points (0.3SD) on pairwise comparisons between the 6 treatment arms (assuming 𝛼 = 5%, power = 80%).
Comparison of No discussion with Discussion. Similar simulations imply an MDE of 7.5 percentage points (0.15SD).
|