Accidents, Disease, and the Greenhouse Effect: Effects of Response Categories on Estimates of Risk

Two experiments are reported that demonstrate that estimates of risk can be influenced by the range of response categories presented to subjects. In Experiment 1, 146 members of the public estimated the likely number of deaths from each of 20 causes. Subjects presented with five categories covering a low range of frequencies gave lower absolute estimates that those presented with a higher range scale. In Experiment 2, 106 students estimated the consequences of the greenhouse effect. Estimates of future rise in sea level and, independently, future flooding were higher when the categories of the relevant response scales covered a high range rather than low range of possible effects. However, estimates of sea-level rise were not influenced by the response scale used for flood estimates, or vice versa. In neither experiment was there any generalization of such effects to other nonmanipulated dimensions. It is concluded that (a) people's notions of risk are poorly anchored numerically but do not necessarily re...