Does the messenger matter? Race and gender effects in stated preference experiments

Stated‐preference surveys are delivered by someone, yet the messenger's identity is rarely reported. This paper develops a theoretical model that extends the consumer choice model and shows that willingness to pay (WTP) falls as the messenger becomes more disfavored. We test this in an online randomized experiment varying the messenger's perceived race and gender. Against a no‐messenger control, a White‐male messenger raises WTP for a hemp T‐shirt and a Black‐female messenger lowers it for child‐labor‐free chocolate; the favored‐versus‐disfavored contrast is one‐sixth of the control median in both goods. Messenger identity is an unmeasured source of variation in welfare estimates.