Abstract
A common procedure in experiments is to use binary lotteries to induce in all subjects pre-specified risk preferences. The validity of this procedure has been established only for a subject performing a single task, yet the procedure is normally applied in multi-task settings. This article formally analyses the multi-task case and establishes necessary and sufficient conditions relating to experimental design. New guidance is provided for the design of experiments involving interdependent tasks.
Keywords:
Notes
1 Berg et al. (Citation1986) suggested that RPIP ‘… can be utilized in any experimental setting’, but this has not been formally demonstrated to date. This article shows that the claim needs to be qualified.
2 This is a stronger requirement than ‘statistical independence’. Statistical independence would merely allow EquationEquation 8(8) to be multiplicatively decomposed and written as .