Accounting for Variation in the Explanatory Power of the Psychometric Paradigm

The Effects of Aggregation and Focus

Published In: Journal of Risk Research, v. 10, no. 4, June 2007, p. 527-554

Posted on RAND.org on December 31, 2006

by Nicolas C Bronfman, Luis Ardon Cifuentes, MIchael deKay, Henry H. Willis

Read More

Access further information on this document at www.informaworld.com

This article was published outside of RAND. The full text of the article can be found at the link above.

Most psychometric studies of risk perception have used data that have been averaged over participants prior to analysis. Such aggregation obscures variation among participants and inflates the magnitude of relationships between psychometric dimensions and dependent variables such as overall riskiness. However, most studies that have not averaged data over participants have also shifted the focus of analysis from differences among hazards to differences among participants. Hence, it is unclear whether observed reductions in the explanatory power of psychometric dimensions result from the change in the level of analysis or from the change in the focus of analysis. Following Willis et al.'s (2005) analysis of ecological risk perceptions, we unconfound these two variables in a study of risk perceptions in Santiago, Chile, although we use more traditional hazards, attributes, and statistical procedures. Results confirm that psychometric dimensions explain less variation in judgments of riskiness and acceptability at the disaggregate level than at the aggregate level. However, they also explain less variation when the focus of analysis is differences among participants rather than differences among hazards. These two effects appear to be similar in magnitude. A simple hybrid analysis economically represents variation among participants' judgments of hazards' riskiness by relating those judgments to a common set of psychometric dimensions from a traditional aggregate-level analysis.

This report is part of the RAND Corporation external publication series. Many RAND studies are published in peer-reviewed scholarly journals, as chapters in commercial books, or as documents published by other organizations.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.