Risk Preferences Revealed Through Non-Forced Choices

How to cite

Gerasimou, Georgios. 2019. "Risk Preferences Revealed Through Non-Forced Choices." AEA RCT Registry. AEARCTR-0004038. https://doi.org/10.1257/rct.4038-2.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Georgios Gerasimou*

* Primary investigator

Registry ID
AEARCTR-0004038
DOI
https://doi.org/10.1257/rct.4038-2.1
First registered
Last updated
Status
Withdrawn
Trial dates
2019-03-24 – 2020-07-31
Countries
United Kingdom of Great Britain and Northern Ireland (Scotland)
JEL codes
C91, C8, D9
Registry record
http://www.socialscienceregistry.org/trials/4038

Abstract

This project aims to understand if peoples' choices under risk are more likely to reveal preferences that conform to some foundational theoretical properties of risk-preference consistency when -contrary to what is the norm in decision experiments- people are not forced to choose. The incentivized laboratory experiment that will be conducted will feature two rounds of decisions over the same menus of lotteries. Subjects in the control group will be forced to choose in both rounds and at every menu they are presented with. Subjects in the treatment group will be allowed to delay their choice in the first round. Comparing the consistency of subjects' behaviour across groups has the potential to offer a novel robustness check for existing forced-choice methods of risk-preference elicitation.

Experimental design

Subjects will be asked to make decisions from a series of binary menus of money lotteries in an incentivized laboratory experiment. The same menus of lotteries will be presented in two rounds and subjects will be asked to make a decision at each menu in each round. The order of presentation will vary randomly between subjects and between rounds. Subjects will be reminded of the decision they made at each menu in the first round before they are asked to make a decision from it again in the second round.

Subjects in the control group ("forced-choice") will be asked to choose one of the available lotteries from each menu in both rounds. Subjects in the treatment group ("non-forced-choice") will be asked in the first round to either choose one of the available lotteries or to choose "I'm not choosing now". In the second round, these subjects will also be asked to choose a lottery from each menu.

Subjects will be allocated an amount "a" at the beginning of the experiment. Subjects in the control group ("forced-choice") will be asked to choose one of the available lotteries from each menu in both rounds. Subjects in the treatment group ("non-forced-choice") will be asked in the first round to either choose one of the available lotteries or to choose "I'm not choosing now". In the second round, these subjects will also be asked to choose a lottery from each menu.

At the end of the experiment, one menu will be randomly selected for each subject. Subjects will win the lottery they chose from that menu in the second round, which will then be played out for them. In addition to their lottery winnings, they will receive: (a) their initially allocated amount "a" if they chose the same lottery at that menu in both rounds; (b) a fraction "d" of their initially allocated amount if they chose different lotteries at that menu in the first and second rounds; (c) [relevant in the "non-forced-choice" group] fractions "b" or "c" of their initially allocated amount "a" if they chose "I'm not choosing now" at that menu in the first round, with a > b > c > d.

Design details

Randomization method
The payoff-relevant randomly selected menu will be chosen uniform-randomly by a computer. The lottery to be won by each participant will be played out publicly by randomly drawing an integer number using a certified random-number generator (www.random.org). It will be explained to subjects beforehand how each number that could be drawn is mapped to one of the possible monetary outcomes for each lottery.
Randomization unit
Experimental sessions.
Clusters
11 experimental sessions.
Observations
260 (on average, 23.6 per session).
Arms
100 subjects in the control group and 160 subjects in the treatment group (80 per treatment).
Primary outcomes
1) Are there significant differences between forced- and non-forced-choice subjects' behaviour in their degree of conformity with: (a) First-order stochastic dominance. (b) Second-order stochastic dominance. (c) The Independence axiom of preferences under risk. (d) The Betweenness axiom of preferences under risk. 2. Are there significant differences in the magnitute of the "Common Ratio effect" and a variant of the Kahneman and Tversky (1979) version of the "Allais paradox" between the forced- and non-forced-choice groups? 3. Are there significant differences in the above outcomes between subjects in the forced-choice group and subjects in the non-forced-choice group who never delayed any choices? 4. Are there significant differences in the above outcomes between subjects in the forced-choice group and subjects in the non-forced-choice group who delayed choice at least once? 5. Are there significant differences in the active-choice intertemporal consistency between subjects in the forced- and non-forced-choice groups? 6. Are reversals in the subjects' revealed attitude to risk more likely in the forced-choice group (including self-forced)? 7. In the non-forced-choice group, does the decision to delay choice at a menu depend on the expected value and the maximum prizes of the lotteries in that menu? 8. In the non-forced-choice group, does the decision to delay choice at a menu depend on the cost of delaying choice? 9. In the non-forced-choice group, is it more likely for subjects to delay choice at menus in which the lotteries are not ranked by second-order stochastic dominance or, more generally, when the complexity of the decision problem increases? 10. Does the incidence of stochastic-dominance violations in menus that feature equivalent lotteries depend on the way in which lotteries are described?

Cited by

No citation data available for this registration yet.

Keywords