Field study of choice-matching for eliciting stated preferences

How to cite

Zawojska, Ewa; Krawczyk, Michał. 2021. "Field study of choice-matching for eliciting stated preferences." AEA RCT Registry. AEARCTR-0006705. https://doi.org/10.1257/rct.6705-1.0

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

Ewa Zawojska* , Michał Krawczyk

* Primary investigator

Registry ID
AEARCTR-0006705
DOI
https://doi.org/10.1257/rct.6705-1.0
First registered
Last updated
Status
In development
Trial dates
2021-01-10 – 2021-03-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/6705

Abstract

The study aims at testing empirically the theoretical approach of choice-matching proposed by Cvitanić et al. (2019, AER: Insights, 1(2), 179-192) for incentivizing truthful responding. We apply the proposed framework to the context of stated preferences. While the choice-matching approach is originally designed for incentivizing responses to a multiple choice question, we illustrate its possible application to an open-ended question. To that end, we conduct an online experiment, mirroring a standard stated preference survey as used for valuation of public goods. We implement two versions of the survey questionnaire: one employing the incentive-compatible choice-matching approach and another representing a common hypothetical setting. The two survey versions differ only with respect to necessary characteristics for implementing the choice-matching. In particular, both versions employ the same open-ended question for eliciting individual's willingness-to-pay values for a public good. The good is a policy project of installing solar panels in public spaces of the city of Warsaw in Poland (e.g., railway stations) so that these places could increase their green energy use. We inquire whether willingness-to-pay values are different when stated under the choice-matching approach than when expressed in standard conditions without any incentivization to truthful responding.

Experimental design

The economic experiment will be conducted online on a sample of about 200 respondents randomly invited to participate in the study from the panel of registered participants to economics experiments of the Laboratory of Experimental Economics at the Faculty of Economic Sciences of the University of Warsaw. The experiment will have a form of a short online survey (lasting about 20 minutes), and two versions of the questionnaire will be developed: one employing the choice-matching approach and the other one representing a common hypothetical setting without any incentives for truthful responding. Other than the differences required by the implementation of the choice-matching approach, the questionnaires will be identical.
Randomly selected participants from the panel will receive email invitations with a link to the survey. Specific time (about a week) will be allotted for the participants to complete the survey. Upon starting the survey, each participant will be randomly assigned to one of the two versions of the questionnaire. Participants will not be made aware of the two versions.

Design details

Randomization method
Randomization is done by a computer for both randomized assignment of individuals to treatments (survey versions) and randomized selection of individuals for whom the decisions are binding
Randomization unit
Randomization at the individual level
Clusters
200 individuals
Observations
200 individuals
Arms
100 individuals in the standard survey version and 100 individuals in the choice-matching approach
Primary outcomes
The key outcome variable is willingness-to-pay statements in the open-ended preference elicitation question. The study will reveal whether preferences stated in an open-ended valuation question are statistically different or not between the theoretically incentive-compatible choice-matching approach and a standard, not incentive-compatible, hypothetical setting. While there could be many interpretations behind any of the two outcomes, we will try to control for the interpretations, using additional questions included at the end of the survey.

Cited by

No citation data available for this registration yet.

Keywords