Confidence Preferences

2020-10-05

2021-12-31

Matching probabilities for ambiguous prospects

Subjects indicate their choices between an ambiguous prospect (with and without samples) and a list of risky prospects. Subjects’ switching points indicate the matching probability for an ambiguous prospect (Dimmock et al. 2015).

Certainty equivalents for risky prospects

We use physical Ellsberg-type urns as stimulus. Choices are incentivized monetarily. All urns contain 100 balls with each ball either red or black. Subjects bet on both colors. The color compositions of the risky urns are known. The color compositions of the ambiguous urns are unknown. There are three types of choice sets.
In Round A, we elicit risk attitudes through certainty equivalents for 3 risky urns containing 10 red (90 black), 30 red (70 black) and 50 red (50 black).
In Round B, we elicit ambiguity attitudes through matching probabilities for a fully ambiguous urn. Subjects choose without samples.
In Round C, we elicit ambiguity attitudes through matching probability for varyingly ambiguous urns. We randomly draw some sample balls (with replacement) from the urn and subjects can see the sample balls. We observe how the matching probabilities change with different samples.

The offline version of the experiment uses real urns. The experimenters fill urns and draw samples before subjects enter the room. We randomize the content of each ambiguous urns by throwing a 12-sided die. The number shown by the die determines color composition in the urn. For example, if the die shows 1, the urn will contain 10 red balls and 90 black balls. If the die shows 2, the urn will contain 20 red balls and 80 black balls. The number of the die increase 1 will add 10 to the number of the red balls. If the die shows 11, the urn will contain 100 black balls and 0 red balls. If the die shows 12, we throw the die again until it gives a number smaller than 12. In total we have 11 possible color composition of the urn.
We randomize the sample size for each urn by throwing the same die. The number shown by the die determines the sample size. For example, if the die shows 1, the sample size will be 1 exactly. If the die shows 2, the sample size will be 2. The same rule follows until a sample size of 12. Therefore, the sample size can be any number between 1 to 12.
To determine the payoff, we randomly select one choice for payment. For this purpose, we set up 3 boxes of envelopes from which subjects draw one right after they enter the room (before they make any decisions). The subjects can inspect the boxes and envelopes if they like. The envelope from the first box will determine the ball color (red/black), the envelope from the second box the choice set (1-14), and the envelope from the third box the menu item (1-11). We open the envelopes only after the subjects made all the decisions. According to his/her choice for that particular color, choice set, and the menu item, either (s)he gets paid by a certain cash amount or (s)he draws a ball from the chosen urn, whose color will then determine the payoff.
The online version of the experiment uses computers for visualizing urns. The composition of the urns, as well as the sample balls drawn from the urns, are randomly drawn by the computer program. All other aspects of the stimuli mirror the offline version of the experiment.

Throwing a die. There is no manipulation. Online experiment is randomized by a computer.

Decisions

No

n/a

Offline about 150 subjects, Online about 150 subjects

n/a