Whom do we listen to? Preferences in social learning

Last registered on October 04, 2023

Pre-Trial

Trial Information

General Information

Title
Whom do we listen to? Preferences in social learning
RCT ID
AEARCTR-0011924
Initial registration date
September 15, 2023

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
October 04, 2023, 12:51 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Region

Primary Investigator

Affiliation
Faculty of Economics and Business, University of Lausanne

Other Primary Investigator(s)

Additional Trial Information

Status
Completed
Start date
2023-09-15
End date
2023-09-25
Secondary IDs
Prior work
This trial is based on or builds upon one or more prior RCTs.
Abstract
People are known to egocentrically discount instrumentally valuable information acquired through social interactions. In this paper, I explore whether this is driven by preferences over the identity of information sources using online experiments where the identity of information sources and the value of information are exogenously assigned. The study focuses on caste identity in India, and asks whether people have preferences for information that are based on the caste identity of the information source.
External Link(s)

Registration Citation

Citation
Moorthy, Akshay. 2023. "Whom do we listen to? Preferences in social learning." AEA RCT Registry. October 04. https://doi.org/10.1257/rct.11924-1.0
Experimental Details

Interventions

Intervention(s)
This is an experimental study on the role of social identity on the demand for information. The main idea is that information acquisition is driven by concerns for the accuracy for information, which is difficult to observe in many settings. In the context of information acquisition through social interactions, people may use an information provider's social identity to draw inferences about the quality or ability of the information. In this experiment, we aim to determine whether people possess preferences for information acquisition in a setting where social identity a salient and important part of everyday life - caste identity in India. The experiment is designed to control beliefs about the quality of information sources, which will allow for the identification of preferences for the identity of information sources.

Intervention Start Date
2023-09-20
Intervention End Date
2023-09-25

Primary Outcomes

Primary Outcomes (end points)
A binary variable, SWITCH, indicating whether an individual sticks to their own decision or switches to the shown number. This is a discrete choice that is directly elicited in the experiment.

A continuous variable FIRSTGUESS, which is the probability (0-100) estimate that an individual independently makes on a given task.

Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Individual's subjective certainties in their own decisions, individual beliefs about the ability of randomly chosen individuals from different identity groups, individual religiosity, individual exposure to people from social groups that are different from their own. These outcomes will be used in secondary analyses, and are explained in more detail in the analysis plan.
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Experimental task:
Participants need to estimate a probability (0-100) given the necessary information in a balls-and-urns paradigm. First, they make an independent estimate. Next, they are shown a number and asked if they want to stick to their decision or switch to the shown number. The shown number is correct with a given probability (the QUALITY of the information), else it is incorrect.

Participants complete 6 of these tasks, and all estimates are incentivised. Incentives are clearly communicated to participants, and all participants undergo a short training and comprehension module before starting the main experimental tasks.

Treatments:
The treatments manipulate the manner in which this number is shown in a 2x2 design. In treatment group Quality, information about the quality of the information (the probability with which the information is incentive maximising) is shown. In treatment group No Quality, this information is not shown. The quality is either 50%, or 90%, and is randomly chosen for each task. The values are selected from the responses to the same tasks in a previously conducted study.

The other arm of the 2x2 manipulates the caste identity. In treatment group G, the identity of the information source is a General caste category individual. In treatment group O, the identity of the information source is an individual who belongs to a caste group other than the general caste category.

Randomisation:
Randomisation is conditional only on the target sample size for each cell of the 2x2, details and sample sizes for each cell are provided below.

Incentives:
Incentives are calculated as follows: Participants make 6 independent estimates, and 6 SWITCH decisions. One of these 12 outcomes is randomly chosen for an incentive. If the estimate represented in the randomly chosen decision is within 2 percentage points of the Bayesian posterior, participants receive a bonus.
Experimental Design Details
Randomization Method
Pseudo-randomly assigned to treatment upon entering the online experiment environment using a random number generator method in Python.
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
We plan to recruit 1200 individuals for the study through an online survey provider.
Sample size: planned number of observations
We plan to recruit 1200 individuals for the study through an online survey provider.
Sample size (or number of clusters) by treatment arms
There are four main treatments, and participants are randomly assigned to these treatments by the experimental software when they enter the survey experiment.Randomisation is weighted by the sampling requirement for each treatment group, and is implemented algorithmically.

Treatments and target participants:
No Quality treatment + General caste: 150.
No Quality treatment + Non-general caste: 150.
Quality treatment + General caste: 450.
Quality treatment + Non-general caste: 450
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB

Institutional Review Boards (IRBs)

IRB Name
NHH IRB
IRB Approval Date
2023-08-08
IRB Approval Number
NHH-IRB 44/22
Analysis Plan

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials