Displaying expected values in risk elicitation tasks: the effect of information in experimental risk research

Abstract

We investigate the influence of providing expected lottery results to experiment participants in two common risk attitude elicitation tasks. In a between-subject design, either the Holt and Laury task or the Eckel and Grossman task is carried out by a sample of 208 students. We find no significant effect of shown expected values on the risk attitude measured by the tasks. This result even holds true if we divide the experiment participants into specific sub-groups, i.e. female and male, or lower numeracy and higher numeracy participants. Furthermore, comprehension and processing time are not significantly influenced by presented expected values. Therefore, we conclude that providing information on expected values does not influence decision-making in tasks involving risk. This result indicates the robustness of elicited risk attitudes to variation in common experimental methods, and demonstrates that more information could be provided without creating bias in the results.

Keywords:

• Risky choice
• risk attitudes
• elicitation methods
• expected values

JEL classification:

• C 91
• D 81

Notes

1. Gustafson (1998) also pointed out that qualitative approaches on the perception on different risks indicate a different hierarchy for females and males and that the differences often left unexplained.

2. For demographics, we refer to Table A.2 in Appendix.

3. We assumed that the lottery payments are integrated into consumption within one day.

4. For an overview of modeling approaches for the stochastic components of behavior in experiments, we refer to Loomes (2005).

5. The estimation results are qualitatively equal if a different error specification, according to Luce (1959), is used.