The frequency presentation effect : determining its nature and consistency
Tracy L Lindberg · Summit (Simon Fraser University) · 1995
This research replicates and extends the recent work of Cosmides and Tooby (in press) on probabilistic reasoning in judgments under uncertainty.Using a problem well-known for eliciting base-rate neglect, Cosmides and Tooby concluded that subjects were good intuitive statisticians when ?he information was presented in frequency form.Four experiments were designed as a threepart investigation into the frequency presentation effect reported by Cosmides and Tooby.The goals of the investigation were as follows: (I) to test the replicability of the frequency effect; (2) to isolate the effect of the frequency presentation from the effects of confounding variables; and (3) to re-examine how well subjects' reasoning reflects aspects of a calculus of probability, such as Bayes' Theorem.Results from Experiments 1 and 2 replicated the frequency effect, however the effect was inconsistent with clarified versions of the problem.The effect disappeared in Experiments 3 and 4, once aspects of mathematical difficulty were controlled.In addition, examination of subjects' descriptions of their solutions revealed poor understanding of Bayes' Theorem.These results are discussed in relation to recent research on cognitive biases.hours of statistical assistance donated by Michael Maraun.Special thanks are directed to Debra Quinn and Bryan Kraushaar, for their respective efforts in the daunting task of coding the questionnaires.I appreciate your dedication.Thanks to Timmy and Kalilah, for their calming influence.