Strategic bias in discrete choice experiments

Abstract An induced value laboratory experiment is conducted to explore the vulnerability of discrete choice experiments to strategic misrepresentation of preferences. We consider strategic behaviour to arise when an agent: (i) believes the choice experiment will be used to determine a provision decision over a discrete set of alternatives; and (ii) has expectations about the relative likelihood of those alternatives being selected and delivered. In the experiment, agents receive induced values for the discrete set of provisioning alternatives. In treatments where agents receive information that their first best outcome is unlikely to win, we investigate the extent to which their choices change, in a manner consistent with them seeking to deliver their second best outcome in the provisioning decision. We find that 27% of respondents misrepresent their preferences and reveal evidence of strategic bias. We find that this behaviour is sufficient to change inferences about preferred provision at the aggregate level.