Study 2b: Employee Perspective.

We conducted a power analysis based on a pilot test, which showed a small to medium effect and ∼20% exclusions. In line with our preregistration (https://aspredicted.org/f7vc2.pdf), we therefore attempted to recruit 470 individuals after exclusions. We recruited 648 participants who indicated in their Prolific intake survey that they have customer-service experience. Consistent with our preregistration, we excluded participants who met one or more exclusion criteria (did not expect the employee to be positioned behind the register, n = 109; answered an attention check incorrectly, n = 30; suspected bots, n = 52; and those who reported no customer-service work experience in our survey, n = 101). Results are consistent with those reported here when excluded observations are included in the model. The final sample consisted of 414 participants (age: M = 31.1 y, SD = 10.3 y; 51.9% men, 47.3% women, and 0.07% other; 62.1% Caucasian, 8.9% African American, 16.9% Asian American, 7.5% Latinx, and 4.6% other).

We instructed participants to imagine that they worked as a cashier at a convenience store and that one night a robber comes into the store (see OSF for full description). Then, participants were randomly assigned to one of two employee-position conditions as in study 2a using the same images.

We measured to what extent (from 1 = not at all to 7 = perfectly) participants expected that the robbery would go smoothly (i.e., “without anyone getting hurt,” reverse scored as anticipated injury) and also included the same measure for anticipated violence as study 2a (α = 0.73). We included an attention check (“in the image you saw, where was the employee standing?”) and a manipulation check (“to what extent were your expectations about where you would be standing aligned with what happened in this scenario?”) (from 1 = completely aligned to 7 = NOT aligned) and measured demographics at the end.