Referring into opportunities III: Field experiment on reactive help

Last registered on August 06, 2024
View Trial History

Pre-Trial

Trial Information

General Information

Title
Referring into opportunities III: Field experiment on reactive help
RCT ID
AEARCTR-0014136
Initial registration date
August 05, 2024

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
August 06, 2024, 4:03 PM EDT

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

Locations

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

Request Information

Primary Investigator

Name
Manuel Munoz
Affiliation
Luxembourg Institute of Socio-Economic Research
Contact Primary Investigator

Other Primary Investigator(s)

Additional Trial Information

Status
On going
Start date
2024-08-05
End date
2024-09-30
Keywords
Behavior, Education
Additional Keywords
JEL code(s)
Secondary IDs
Prior work
This trial is based on or builds upon one or more prior RCTs.
Abstract
In this study, I investigate the role of social networks and biases in access to opportunities at a university setting, where such opportunities require third-party referrals, such as nominations and recommendations. I focus on university students as candidates and faculty members as referrers. Through a field experiment, I explore how these networks influence the request and receipt of referrals, potentially perpetuating biases that disadvantage certain groups in reactive help settings.

In the experiments, I inform students that they are offered an opportunity based on both their demographics and academic merit, requiring them to seek support from a faculty member to access the opportunity. I experimentally vary the information students disclose when requesting help—either focusing solely on academic merit (control condition) or including demographics. This variation allows me to evaluate how the type of information affects students' likelihood of seeking help, their choice of faculty members, and the likelihood of receiving assistance.

I conduct the study in two waves, one with only high merit student. The other with only low merit students. In each wave I use the same treatment variations.

This study complements a previous study on proactive help conducted in 2022, where faculty members where invited to nominate students to the same combined opportunity, access to an international training program plus participating in the lottery for a valuable prize. Data from the previous experiment allows me to identify proactive helpers, those who chose to make a referral in the first study. This consistency enables me to examine the relationship between proactive and reactive help. Specifically, I investigate whether faculty members who initiate help in the proactive experiment are also those whom students seek out in the reactive experiments.

The findings from these experiments will provide insights into the presence and extent of biases in referral processes and the effectiveness of awareness interventions in mitigating these biases. I aim to inform strategies for enhancing access to opportunities for underrepresented groups by increasing awareness of potential biases.
External Link(s)

Registration Citation

Citation
Munoz, Manuel. 2024. "Referring into opportunities III: Field experiment on reactive help." AEA RCT Registry. August 06. https://doi.org/10.1257/rct.14136-1.0
Sponsors & Partners

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

Request Information
Experimental Details

Interventions

Intervention(s)
Students are invited to register to participate in a beneficial opportunity: an international training program and a lottery for a valuable prize. To register, students are required to get an endorsement from a faculty member. The intervention varies the information students receive about the selection criteria for the opportunity. All students can be informed that they are selected because of their academic merit (high or low). In addition some are also informed that they are selected because of their demographics (e.g., gender, social class, etc). Students seeking an endorsement are provided a pre-defined message they use to request the support of a faculty member. The message may only disclose that the student is selected because of her academic merit, or it may also disclose that selection is based on demographics.
Intervention Start Date
2024-08-05
Intervention End Date
2024-09-30

Primary Outcomes

Primary Outcomes (end points)
Help-seeking, selection, help receipt, Quality of the candidate
Primary Outcomes (explanation)
Help-seeking is measured as the binary choice of requesting a faculty member for an endorsement for the opportunity or not.
Selection is measured as the binary choices of asking a faculty member among all potential endorses a student can seek for help. The choice set is the entire set of connections between a student and all the faculty members she has taken a class with. The selection is the specific faculty member that is chosen, in contrast to the set of faculty members that are not.
I also measure in selection the likelihood of choosing a faculty member that was a proactive helper in the initial experiment of 2022.
Help receipt is measured as the likelihood that a requested faculty member makes the endorsement and helps the help-seeking student.
Quality of the candidate is measured by the likelihood that the student receiving help registers to the program, starts it and completes all sessions of the training program.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
I will conduct a randomized control trial (RCT) with 4831 students and six groups, randomized at the level of the individual participant.

2561 students have high GPA (high academic merit) and are assigned into treatments as follows: 864 students in the first treatment arm will receive an invitation to take-up the beneficial opportunity and seek the endorsement of a faculty member. These students are informed they are chosen both because of their academic merit and their demographics, and the endorsement message makes this explicit. 864 students in the second treatment arm will receive the same information as in the first treatment, but the endorsement message used to request the help from a faculty member only states that selection is based on academic merit (and does not disclose any information on identities). 833 students in the third treatment receive the same invitation, but neither the students nor the faculty members receive any information that selection is identity-based and are only informed that it is based on the student's high academic merit.

2270 students have low GPA (low academic merit) and are assigned into the same treatments as follows: 776 in the first treatment, 757 in the second treatment, and 737 in the third treatment.

The experimental design allows me to address the following research questions:
(1) What is the impact of the information that selection is identity-based on the likelihood of seeking help?
(2) What is the impact of the information that selection is identity-based on the selection of faculty members to request the endorsement from?
(3) What is the impact of the information that selection is identity-based on the likelihood of receiving an endorsement?
(4) What is the impact of the information that selection is identity-based on the quality of the candidate?

Experimental Design Details
Not available
Randomization Method
Randomization done in office by a computer.
Randomization Unit
Individual level. I use block randomization to balance individual characteristics of the students such as gender, social class and academic merit (GPA).
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
4831 students
Sample size: planned number of observations
4831 students
Sample size (or number of clusters) by treatment arms
For the high merit students, sample size is as follows:
864 in treatment with two-sided information (both student and faculty endorser are informed that selection is identity-based)
864 in treatment with one-sided information (only the student is informed that selection is identity-based)
833 in treatment with no information

For the low merit students, sample size is as follows:
776 in treatment with two-sided information (both student and faculty endorser are informed that selection is identity-based)
757 in treatment with one-sided information (only the student is informed that selection is identity-based)
737 in treatment with no information
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials

Documents

Document Name
Prior Pre-registration
Document Type
other
Document Description
The first data collection took place in the fall of 2022. This is the wave with the high GPA students. At the time of the study the partner university was not open to sharing the data for publication purposes. As such, I did not pre-registered this wave of the experiment.

The second wave took place in the spring of 2023. This was the study with low GPA students. At this moment the partner university was opened to share the entire data for publication purposes. So, I did pre-register the second wave. It was pre-registered in AsPredicted with reference number #124501, using the title "The effect of signaling on take-up". However, at this time, I only had access to the data of the natural field experiments, who requests help from whom and what choices they make in terms of participation in the program.

The partnership with UNAB will allow me to access co-enrollment measures. This includes which courses students take, when, and who teaches those courses. This will allow me to build a rich dataset on network measures, defined by the connections between a student and each faculty member through the classes they shared together. Such network data complements the experimental data and allows me to go beyond simpler analyses such as OLS, which focus only on the help-seeking choices made, and to use more elaborate measures to estimate selection, such as Random Utility Choice models, which take into account not only the faculty endorser chosen but also the entire set of faculty members that could have been requested but were not, for each student invited to the program.
File
Prior Pre-registration

MD5: c02c7529d3327f888000bfe89106dd1a

SHA1: f1e113c442c36b12861b83744f240454e94f00f4

Uploaded At: August 05, 2024

IRB

Institutional Review Boards (IRBs)

IRB Name
Universidad Autonoma de Bucaramanga
IRB Approval Date
2022-07-01
IRB Approval Number
N/A
Analysis Plan