Phone Tree Sampling to Measure Food Security

Last registered on October 22, 2024

Pre-Trial

Trial Information

General Information

Title
Phone Tree Sampling to Measure Food Security
RCT ID
AEARCTR-0013116
Initial registration date
March 07, 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
March 15, 2024, 4:38 PM EDT

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

Last updated
October 22, 2024, 4:13 PM EDT

Last updated is the most recent time when changes to the trial's registration were published.

Locations

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

Request Information

Primary Investigator

Affiliation
Cornell University

Other Primary Investigator(s)

PI Affiliation
Cornell University

Additional Trial Information

Status
On going
Start date
2020-08-01
End date
2024-12-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
We are studying a novel way to measure food security in hard-to-reach areas. We will compare food security measures collected from four different groups: (i) in-person respondents sampled in the traditional fashion (village listing + random selection), (ii) in-person respondents sampled in the traditional fashion that have a phone and agree to be surveyed by phone, (iii) phone survey respondents sampled in the traditional fashion but interviewed by phone, and (iv) phone survey respondents sampled using a novel "phone tree" method of snowball sampling, which is fully remote and does not require travel to the study area. The experimental variation arises between groups (ii) and (iii), where households will be randomly assigned to in-person or phone surveys. The phone tree group simulates what it would be like to collect food security data in a new location, without traveling there or sending a team. Our goal is to understand the magnitude of the biases that emerge from using the phone tree method to measure food security in hard-to-reach areas, and to explore whether certain types of questions can be useful in reducing those biases.
External Link(s)

Registration Citation

Citation
Dillon, Brian and Joanna Upton. 2024. "Phone Tree Sampling to Measure Food Security." AEA RCT Registry. October 22. https://doi.org/10.1257/rct.13116-2.0
Experimental Details

Interventions

Intervention(s)
This is a methods experiment. We are randomly assigning households in Malawi to be interviewed in-person or by phone. Findings from these two groups will be compared to those from two other groups: those without phones, and those with a phone or access to a phone that are reached via phone tree sampling. This experiment runs alongside an existing food security monitoring survey that is run in many districts of Malawi. Our focus is on the Chikwawa and Nsanje districts.

Update October 2024: for the upcoming fourth survey round we will randomly assign respondents in group 3 -- those that were sampled with a traditional face-to-face listing but who have been interviewed over the phone -- to either in-person or phone surveys.
Intervention Start Date
2024-03-11
Intervention End Date
2024-06-30

Primary Outcomes

Primary Outcomes (end points)
Primary outcome 1: monthly household food security score, constructed using the Household Hunger Scale (HHS)
Primary outcome 2: household perceptions about the village food security score, constructed using the HHS questions modified to relate to village- rather than household-level experiences
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We are randomly assigning households to in-person or phone surveys. We are comparing food security measures from those two groups to non-experimental groups of households that do not have phones, and households that have phones and are found via snowball sampling rather than a traditional listing.
Experimental Design Details
Not available
Randomization Method
Randomization conducted on a computer.
Randomization Unit
Households that have phones and are willing to be interviewed by phone are randomly assigned to an in-person survey or a phone survey.

Update October 2024: the households in the phone survey group (not the phone tree group) will be randomly assigned to either continue with phone surveying or to be interviewed in person during survey round 4.
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
For the experimental group we are working in 32 clusters (16 per district in two districts).
Sample size: planned number of observations
Total is between 1200-1600 for the two experimental groups. For the phone survey group: 25 per cluster (800 total -- 400 per district in 2 districts) For the in-person group we are interviewing 25 households per cluster in 16 clusters (800 total -- 400 per district in 2 districts), but not all of these households have phones and so only a subset will be randomly assigned to the in-person rather than face-to-face survey.
Sample size (or number of clusters) by treatment arms
800 observations in the phone survey group.

800 observations in the face-to-face survey group, but only a subset of those will have phones and be eligible for the phone survey, and we will not know the exact number until we begin data collection.

There are two other groups involved:
-- The phone tree / snowball group consists of 800 households (400 per district) that we reach through referral sampling with no listing
-- Members of the in-person survey group who do not have phones or consent to phone surveying constitute their own comparison group

Update October 2024: across rounds 1-3 we have interviewed 896 households in the phone survey group, slightly above our initial target. For round 4 we are 50/50 randomly assigning those 896 households to another round of phone surveying (448 households) or an in-person survey (448 households). This randomization is stratified by enumeration area, roof type (metal vs. not metal), and mean value of food security measure (HHS) over rounds 1-3.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB

Institutional Review Boards (IRBs)

IRB Name
Cornell University IRB
IRB Approval Date
2021-12-01
IRB Approval Number
1605006348