Multiple switching behaviour in multiple price lists

Abstract

A common mechanism to elicit risk preferences requires a respondent to make a series of dichotomous choices. A recurring problem with this mechanism is a frequently observed tendency to switch from the less to the more risky choice multiple times, multiple switching behaviour. We introduce an instructional variation which our evidence suggests practically eliminates such behaviour. We read a script emphasizing only one decision will determine earnings before providing written instructions. Emphasizing the incentive compatibility of the payment rule reduces observed multiple switching behaviour from 13.3% to 2.3% in one format and from 25.8% to 6.7% in another.

Acknowledgements

This research was undertaken by the University of Calgary Behavioural and Experimental Economics Laboratory (CBEEL). I thank Mike McKee and Glenn Harrison for their encouragement and helpful suggestions.

Notes

¹Goeree et al. (2003) only reported that 6.0% of their subjects exhibit an identical pattern of inconsistent responses.

²Another difference between the two sets of sessions is the decision task separating the two formats. In the control sessions, subjects were asked their willingness-to-accept for the lottery in the PV format. However, this task was replaced in the treatment sessions by a task that required subjects to choose between the lottery in the PV format and the lottery in the RV format. We control for any possible influence this may have had on subject responses in the analysis.

³The selection of the each subject's decision that determined their payoff was presented as a compound lottery; the computer first selected the stage of the experiment (each had 1/3 chance of being selected) and then the decision of the selected stage was chosen (each had a 1/10 chance of being selected). Thus, we assume that preferences conform to the Independence Axiom (Samuelson, 1952). The evidence in the literature suggests that ‘random lottery selection’ is incentive-compatible for simple choice sets (Starmer and Sugden, 1991; Wilcox, 1993; Ballinger and Wilcox, 1997).

⁴The following script was read aloud to subjects. ‘Before we begin with the instructions, I would like to bring one thing to your attention. As you will read in the instructions, you are going to make several decisions in this experiment. However, only ONE of these will actually determine your earnings for this experiment! So, it is important that you take each decision seriously since a single mistake can be quite costly!’