Experimental Design
In the experiment, each subject faces a menu that consists of 2-12 alternatives. Each alternative is a risky asset that specifies a monetary payment with a given probability. There are 20 budgets per subject presented in random order. Each subject implements a procedure to compare the alternatives and then chooses a unique alternative from the menu based on her preferences. Procedure is implemented by opening (showing) the alternative, selecting (temporary choosing) the alternative, or discarding (closing) the alternative.
We use a between-subject experimental design where each subject participates in one of the following four treatments:
Treatment 1: Delegated Procedure
In this case, the computer will generate a sequence of binary comparisons from the menu for the DM to perform. All discarded alternatives are discarded permanently and cannot be re-opened.
Treatment 2: Assisted Binary Procedure
In this case, the DM can implement any procedure she wishes with the caveat that she is restricted to binary comparisons. That is, the DM can only consider two alternatives from the menu at the same time. All discarded alternatives are discarded permanently and cannot be re-opened.
Treatment 3a: Assisted Procedure I
In this case, the DM can implement any procedure she wishes with no restrictions. That is, the DM can simultaneously consider any number of alternatives from the menu. All discarded alternatives are discarded permanently and cannot be re-opened.
Treatment 3b: Assisted Procedure II
In this case, the DM can implement any procedure she wishes with no restrictions. That is, the DM can simultaneously consider any number of alternatives from the menu. All discarded alternatives are not discarded permanently and can be re-opened.
Treatment 4: Free Procedure
In this case, the DM faces the full menu of 10 alternatives and she chooses one alternative. Since there is no pre- defined procedure, the DM can employ any heuristic to assist her choice. All discarded alternatives are not discarded permanently and can be re-opened.
Our experimental treatments allow us to answer our research questions by gradually increasing the complexity of the decision problem. The outcome variable of interest is the consistency of the DM's choices: that is, the extent to which the DM's observed choices can be generated by the maximisation of her preferences. Any observed inconsistencies in Treatment 1 are due to the DM's failure to accurately express her preferences. Any differences between Treatments 1 and 4 can be attributed to the procedural costs or burdens imposed on the DM. The comparison between Treatments 2 and 3 will shed light on the types of procedures that the DM uses to implement the maximisation of her preferences. In particular, any differences between Treatments 2 and 3 will illustrate the extent to which the DM's optimal procedure requires binary comparisons versus comparisons from larger sets.