Messaging and Tangible Donations in a Supermarket Setting

How to cite

Almond, Rhosyn. 2019. "Messaging and Tangible Donations in a Supermarket Setting." AEA RCT Registry. AEARCTR-0004894. https://doi.org/10.1257/rct.4894-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Rhosyn Almond*

* Primary investigator

Registry ID
AEARCTR-0004894
DOI
https://doi.org/10.1257/rct.4894-1.0
First registered
Last updated
Status
On going
Trial dates
2019-10-04 – 2019-12-19
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/4894

Abstract

This research aims to use a field experiment to investigate whether reminders to donate at the time of purchase can affect the number and type of in-kind charitable donations made to a local food provision organisation, and to see how these effects change over time. The signs use messaging which makes the local aspect of the recipients salient.
A commonly cited reason for not donating to the FoodBank is that visitors to a supermarket forget to make a donation until they have already reached the end of their shopping and paid for their items. In conjunction with key items not being donated, this leaves great room for improvement in the type and level of donations made to the FoodBank. FoodBanks receive around half of their total donations from donation points in supermarkets, so getting this element of their strategy right would have large effects on their overall stock levels.
The use of shelf level signs to remind individuals to purchase an item for donation during their shop, and highlight which items are most needed, would appear to solve these two barriers to donation. These have been trialled in some stores in the UK, but as of yet the effect of these signs on donation rates has not been investigated.

Intervention

There are two treatments in this intervention. They take place in 18 local grocery stores which collect donations for the local FoodBank. 6 are assigned to each arm, and 6 to the baseline control group. The treatments involve highlighting which items are most needed by the Norwich FoodBank with the use of shelf level labelling and posters. The resulting donations made in store are measured weekly for the 10 week duration of the intervention. The percentage of donations of the highlighted item types will be compared between stores to find the effect of the treatment.

Experimental design

This experiment is a randomised control experiment. It is unblinded, however the subjects will not know that they are in an experiment as they are likely to only shop at one of the branches. There are 2 Treatments and one control group. It is a between groups design - each store is exposed to only one treatment for the duration of the experiment.

Design details

Randomization method
Randomisation is done through the use of a Python code with a set seed and is replicable. Block randomisation was used on three variables. The three variables were: local income (a proxy measure for this is used: the percentage of free school meals recipients in the nearest primary or infant school); store characteristics (the presence of a Post Office or a Subway inside the store - this could affect footfall); accessibility of donation point (whether the donation basket was located after the checkout, or requires a customer to double back on their journey out of the store). There are guiding hypotheses for these three variables leading to the use of them in the blocking randomisation.
Randomization unit
Individual Store
Clusters
Individual Stores
Observations
900 "most needed item" weekly totals. (18 stores, 5 items per store, for 10 weeks)
Arms
6 grocery stores per arm, 3 arms. 1 arm control, 2 arms of treatments.
Primary outcomes
The percentage of 'treated' or "most-needed" items that are donated each week. The overall donation level (in weight, £, and number of items).

Cited by

No citation data available for this registration yet.

Keywords