Do Choice Experiments Generate Reliable Willingness to Pay Estimates Theory and Experimental Evidence

Abstract

In this paper we set up a three-stage experimental and theoretical framework to investigatestrategic behaviour and design induced status quo bias in choice experiments. The researchdemonstrates that: (1) Repeated multiple choice experiments are not demand revealing. Thus,they do not generate reliable estimates of willingness to pay. (2) Most non-demand revealingchoices are for the second-best option in a choice set. Consistent with the predictions of votingtheory, these choices of the second-best option occur when there is an undesirable third option ina choice set. (3) As a result of the mathematics of combinatorial choice set design, the status quooption frequently occupies the second-best position in a choice set. (4) Experimental subjectchoices of the status quo in the second-best position are consistent with theoretical predictionsderived from the mathematics of combinatorial choice set design. Although choice experimentsas currently used in the field cannot be assumed to generate unbiased estimates of willingness topay, this study demonstrates that the bias is of a predictable nature and direction.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 2015
Accession Number
AD1013448

Entities

People

  • Katherine S. Carson
  • Riccardo Scarpa
  • Susan M. Chilton
  • W. G. Hutchinson

Organizations

  • United States Air Force Academy

Tags

Communities of Interest

  • Biomedical

DTIC Thesaurus Topics

  • Air Force
  • Biological Sciences
  • Commerce
  • Computational Science
  • Data Science
  • Demography
  • Economics
  • Elections
  • Experimental Design
  • Factorial Design
  • Information Science
  • Marketing
  • Mathematical Models
  • Mathematics
  • Simulations
  • Social Sciences
  • Surveys

Readers

  • Applied Combinatorial Optimization and Logic Circuit Design.
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.