PREFERENCE ELICITATION RESPONSES:
DIFFERENT ANSWERS TO FUNDAMENTALLY IDENTICAL QUESTIONS OR
DIFFERENT ANSWERS TO FUNDAMENTALLY DIFFERENT QUESTIONS?*

Richard T. Carson

Department of Economics
University of California, San Diego
La Jolla, CA  92093

*Co-authors:  Theodore Groves and Mark Machina

Abstract

In empirical work involving preference questions a typical result is 
that if a question is asked in two different ways responses consistent 
with two different sets of preferences will be obtained.  This general 
result appears to be robust to moving away from the realm of surveys 
and into the realm of experimental economics and consumer marketing.  
This finding of the lack of procedural invariance has often been taken 
by cognitive psychologists and others as evidence against the 
rationality assumption which underlies much of economic analysis.  
We examine the issue of whether different ways of posing preference 
questions are informationally and strategically equivalent.  In 
general we find that they are not.  A simple model based on the 
standard neoclassical assumptions concerning consumer preferences and 
mechanism design theory yields a set of testable predictions as to the 
direction of the difference that should be seen when comparing 
implications about preferences derived from two preference revelation 
opportunities.  Reference to a large body of existing empirical 
evidence provides support for the relevance of the model.