Hazard Perception and Anchoring: A Comparison of the Three Models Explaining the Anchoring Effect

Abstract

When individuals estimate something numerically, their estimation tends to be close to a value perceived beforehand, called an anchor. This tendency is called “the anchoring effect.” We introduce three hypotheses – the numeric priming hypothesis, the semantic priming hypothesis, and the magnitude priming hypothesis – that explain the anchoring effect. We apply them to participants’ estimation of the number of sufferers in order to examine which model explains the anchoring effect best. Experimental results support the numeric priming hypothesis, indicating that the anchoring effect occurs even when no semantic relatedness exists between the number presented as the prime and the successive numerical estimation. Implications for disaster risk communication are discussed based on the results we obtained.