Response Mode Bias Revisited - The "Tailwhip" Effect

Summary Expected utility theory is an important instrument for decision making under risk. To apply it in a prescriptive context, the decision maker‘s utility function needs to be elicited. Several methods for utility elicitation have been proposed in the literature which, from a theoretical point of view, should lead to identical results. However, beginning in the late 1970s, researchers have discovered that theoretically equivalent methods of utility elicitation lead to different results [4, 6, 7]. One important bias phenomenon which was identified in this context is the \Response mode bias", which describes an inconsistency in elicitation results of the Certainty Equivalence (CE) and the Probability Equivalence (PE) methods. Both methods are based on the comparison between a certain payoff xs and a lottery, which yields a better outcome xh with probability p and a worse outcome xl with probability 1??p. In the CE method, the decision maker is informed about xh, xl and p and has to provide a value xs which makes him or her indifferent between the lottery and the certain payment. In the PE method, the decision maker provides a value for p in response to the three other parameters. Using a set of 10 questions, Hershey et al. [6] found that in problems involving only losses, the PE method led to a significantly higher number of subjects exhibiting risk seeking behavior than the CE method. This phenomenon, which was later on labeled the \Response mode bias" was particularly noticeable in lotteries involving a comparatively high probability of loss.