Cover: Distinguishing the Ratio Bias from Unsystematic Error

Distinguishing the Ratio Bias from Unsystematic Error

Situation and Individual-difference Effects

Published in: Journal of Behavioral Decision Making [Epub January 2018]. doi: 10.1002/bdm.2068

Posted on on February 09, 2018

by Eric R. Stone, Andrew M. Parker, Lauren D. Townsend

People often perceive the occurrence of events to be less likely when the likelihood of the event is expressed in ratios consisting of smaller numbers versus larger numbers, an effect known as the ratio bias. This work presents a theoretical framework for the conditions that need to be met for the ratio bias to occur. In doing so, we contrast effects on the ratio bias to those on unsystematic error, which have often been confounded in previous research. We find that the ratio bias is weaker (1) when both sets of numbers are relatively large than when both sets of numbers are relatively small; (2) for scenarios involving lottery tickets than for scenarios involving drawing balls from a bin; and (3) when a physical display depicting the numbers is provided to participants. Each of these factors reduced the ratio bias without reducing unsystematic error. Additionally, we show that unsystematic error is lower among people who (1) reason on the basis of proportions rather than on the basis of the numerator and denominator individually; (2) score higher on the rational scale of the Rational-Experiential Inventory; and (3) are of higher numeracy. We use these results to distinguish causes of error generally from those on the ratio bias specifically and discuss the implications for our understanding of when the ratio bias occurs.

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.