Unraveling the Choice Format Effect: A Context-Dependent Random Utility Model

A few recent studies report that preferences elicited from a two-alternative choice format are different from those elicited from a three-alternative choice format, when both choice set designs contain a status quo option. We hypothesize that the choice format effect might disappear when we estimate a random utility model (RUM) that allows for context-dependent preferences. Based on the estimated context-dependent RUM, we find that the choice format effect is likely the result of two offsetting effects: a task complexity effect that increases the probability of choosing the status quo option, and a preference matching effect, which decreases the probability.