A Field Study of Donations at Supermarket Cash Registers

2016-07-01

2016-07-15

From data collected by the supermarket:
1. Daily total donations by store (before, during and after the intervention).
2. Daily data at the store level (number of costumers and purchases)
From data collected from surveys (as reported by the customers):
1. Asking frequency by cashiers
2. Donation behavior (extensive margin)
3. Reasons for donating
4. Reasons for not donating
5. Beliefs about supermarket keeping part of donation
6. Perceptions about supermarket
7. Perceptions about charitable organization
8. Measures on a 5-pt scale of guilt (self-reported), pressure (self-reported), altruism (self-reported), principle of care (self-reported) and image concern (actual behavior measured by wearing sticker, see below)

Half of the sample of stores we study are randomly selected to receive the campaign, and the other half do not receive it at the time of the study. We measure the impact of the campaign by surveying costumers as they exit the store, for both treatment and control stores. We previously collect baseline data from similar surveys for a subset of all stores in our sample.

In July 2016 the supermarket and the charity will be launching a transparency campaign. The campaign consists of placing signs at the stores and distributing flyers to costumers clarifying that (1) 100% of the donations made at the supermarket cash registers go to the charity, (2) the supermarket cannot legally get a tax break from the donations it collects at its registers, and (3) donations can be made with any payment method. In collaboration with the charity and the supermarket, we selected a subgroup of 40 stores within the Metropolitan area in Santiago de Chile to evaluate the efficacy of the campaign at increasing donations and to explore the underlying mechanisms for any effect. We selected the 40 stores randomly from the total of approximately 110 stores in the region. The randomization was stratified at the type of store (3 types as defined by the supermarket chain company) and, within each type, at whether they were below or above the median donation rate.

In June 2016 we conducted the first wave of the survey to obtain baseline data. We could not visit 4 of the 40 stores at this wave because the 4 stores were either closed or prohibitively distant for the surveyors to get to them. To conduct the survey we approached costumers inside the store (as they finished their payment) or as they exited the store (if we lacked permission from the store manager to go inside). To ensure anonymity and confidentiality, the surveys were answered on iPads that were handed to costumers who agreed to participate. No personal information was collected or requested. To avoid demand effects, the surveyors wore T-shirts from a local university, and an association with the charity or the supermarket was never mentioned. At the end of the survey, respondents were offered a sticker with the text "I helped" and the University logo. This first wave lasted 7 days.

We plan to conduct the second wave in July 2016, when the campaign is expected to start. Before conducting the first wave, we selected at random (stratified as described before) half of the 40 first-wave stores to receive the campaign, and the other half to not receive it at the time of the study. At the request of the charity (request coming after the first wave finished), we have increased the number of stores that will participate in the second wave to 60. The additional stores were randomly selected from the total population; half of them were randomly assigned to receive the campaign, and the other half to not receive it at the time of the study, using the same stratification as before. The surveys for this second wave will be conducted as before, and the wave is expected to last 5 to 15 days depending on logistics.

Finally, pending availability of resources, we plan to conduct a follow-up third wave on the 60 second-wave stores. The surveys for this second wave will be conducted as before, and the wave is expected to last 5 to 15 days depending on logistics.

Randomization done in office by a computer.

The unit of randomization is the store.

Yes

40 stores for first wave (baseline), 60 stores for second wave.

1200 costumers for first wave, 2000 costumers for second wave.

30 stores control, 30 stores campaign.