Missing the Target

Abstract

Koehler and Macchi (2009) criticize the experiments presented in Newell, Mitchell, and Hayes (2008) as being ''virtually irrelevant'' to exemplar cuing theory. This reply addresses that interpretation and argues that the experiments dealt with issues at the heart of the theory and provided evidence highly relevant to understand how people think about low-probability events. The role of the 'target' in probabilistic statements is examined, highlighting the need for further theoretical and empirical clarification of the concept. The remaining specific criticisms raised in the commentary are discussed as well. key words exemplar cuing; probability; frequency format; imaginability Newell, Mitchell, and Hayes (2008) (NMH) presented three experiments aimed at comparing two theories of the way people think about low-probability events: the statistical format account and exemplar cuing theory (EC). The statistical format account predicts that information presented as frequencies (e.g., 50 out of 5000) will be easier to imagine than equivalent information presented as probabilities (e.g., 1%) and this will have a subsequent effect on judgments (i.e., people will be more willing to participate in potentially positive events-(e.g., lotteries)-and less willing to participate in potentially negative events-(e.g., taking vaccinations). EC theory predicts that format per se will not impact judgments but that certain combinations of factors lead to exemplar generation and that these exemplars affect the weight people attach to the possibility that a low-probability event will occur. NMH's results supported the general notion that imaginability plays a role in thinking about low-probability events, but questioned one of the mechanisms of EC theory. NMH'S EXPERIMENTS Koehler and Macchi (2009) (KM) present a summary of NMH's experiments which fails to acknowledge that the overall conclusion is consistent with the main ideas behind exemplar cuing: namely that the imaginability of outcomes plays a key role in thinking about low-probability events. They also did not acknowledge the observed support for one mechanism 1 of EC theory: that integer numerators (e.g., 1 out of 1000) are more imaginable than fractional numerators 1 KM question the use of the words ''mechanism'' and ''multiplicative'' and the use of quotation marks. The quotation marks were not used to indicate direct quotes, merely to suggest a degree of uncertainty about the appropriateness of the term multiplicative mechanism. The term provided a shorthand way of describing the claim that reference class size and incidence rates are multiplied. KM use the word ''mechanism'' in their own conclusion so appear to be comfortable with the term even if it did not appear in their original article. Journal of Behavioral Decision Making about the imaginability of low-probability outcomes. Whether the exemplar-cuing scenarios really failed to cue exemplars remains open to debate and further empirical investigation.