Intervention (Hidden)
Our research will involve 3,000 local South African consumers who will complete a 20-minute survey featuring a choice-based conjoint experiment. The survey will proceed as follows:
First, participants will complete a screener to confirm eligibility. They will be asked two questions: (1) Are you South African? and (2) Have you shopped at a spaza shop (a small local retail store) in the past three months? Only those who answer “yes” to both questions will be eligible to participate.
Second, eligible participants will be introduced to a hypothetical shopping scenario. They will be asked to imagine purchasing a basic grocery item—such as bread—from a retail store. They will be shown different versions of bread packets, each varying in specific features. The instructions will read:
“You will see bread packets with different features such as price and store characteristics. The packet size is standard across options. On each screen, please select the option you prefer.”
In each choice task, participants will choose between four product options. Each option will vary along four randomized attributes:
Price – multiple levels: e.g., average market price ±5%
Ethnicity of shop owner – South African or Foreign (operationalized via a short shopkeeper bio)
Customer service – High (longer hours, WhatsApp-based customer service) vs. Low (shorter hours, no WhatsApp channel)
Brand – Varying product brands
This constituted the first version of the conjoint experiment. In a second version, we introduced a fifth attribute to better explore the distinction between statistical discrimination and animosity-based discrimination:
Store Safety – Whether the store has been (a) targeted by crime, (b) involved in criminal activity, or (c) neither.
In both versions, participants will also complete demographic and attitudinal questions, including age, gender, employment status, occupation, and education. Additionally, we will ask about their views on immigration policy, perceptions of product and store hygiene, experiences with crime, and whether they have received government benefits (e.g., unemployment support). These data will help us examine heterogeneity in attitudes toward foreign-owned retailers.
Finally, participants will receive 40 Rand as compensation, comprising the cost of one product randomly selected from their choices and the remaining balance as change. The selected product will be provided at the end of the survey to maintain incentive compatibility.