Experimental methods: Learning your own risk preferences under different risk elicitation methods

This paper examines whether changes in risk preferences reflect genuine behavioural shifts or result from the elicitation method used. In a within-subject experiment, participants made initial risky choices, completed 24 unpaid learning rounds with real dice and recorded outcomes, and then made final choices. We compare four risk elicitation methods: two Ordered Lottery Selection (OLS) and two Multiple Price List (MPL) tasks. Our analysis focuses on how preferences evolve with experience, the role of reported errors during the learning phase, and the effect of method complexity (i.e., the number of potential payoff outcomes in a given lottery) on revealed risk attitudes. Results show that MPL tasks, particularly when more complex, produce stronger changes in behaviour. Participants who made errors during the MPL learning phase were more likely to display increased risk aversion. This suggests that task comprehension and cognitive load influence the stability of preferences. The findings contribute to the broader debate on preference stability and highlight the methodological relevance of comparing elicitation tools within the same experimental framework.