Academic journal article Memory & Cognition

Use of Base Rates and Case Cue Information in Making Likelihood Estimates

Academic journal article Memory & Cognition

Use of Base Rates and Case Cue Information in Making Likelihood Estimates

Article excerpt

In five experiments, we investigated college students' use of base rate and case cue information in estimating likelihood. The participants reported that case cues were more important than base rates, except when the case cues were totally uninformative, and made more use of base rate information when the base rates were varied within subjects, rather than between subjects. Estimates were more Bayesian when base rate and case cue information was congruent, rather than contradictory. The nature of the "witness" in case cue information (animate or inanimate) did not affect the use of base rate and case cue information. Multiple trials with feedback led to more accurate estimates; however, this effect was not lasting. The results suggest that when base rate information is made salient by experience (multiple trials and within-subjects variation) or by other manipulations, base rate neglect is minimized.

Base rate neglect refers to the robust finding that people often underweight the importance of base rates in a decision task involving two or more sources of information (see, e.g., Goodie & Fantino, 1996; Koehler, 1996; Tversky & Kahneman, 1982). In base rate experiments, participants are typically provided with base rate information, which concerns how often each of two outcomes occurs in the general population, and case-specific information, such as the results of a diagnostic test or witness testimony. The participants' task, typically, is to select the more likely of two outcomes or to provide a verbal estimate of the probability of one or both outcomes. The taxicab problem, described by Tversky and Kahneman, is one of the most recognizable examples:

A cab was involved in a hit and run accident at night. Two cab companies, the Green and the Blue, operate in the city. You are given the following data:

(a) 85% of the cabs in the city are Green and 15% are Blue.

(b) A witness identified the cab as Blue. The court tested the reliability of the witness under the same circumstances that existed on the night of the accident and concluded that the witness correctly identified each one of the two colors 80% of the time and failed 20% of the time.

What is the probability that the cab involved in the accident was Blue rather than Green?

If both pieces of information were considered and were combined according to Bayes's theorem, the probability estimate would be close to .41; however, in most studies, adult participants overweight the case-specific information (b) and neglect the base rate information (a). This has been found when the probabilistic information has been conveyed verbally (as in Tversky & Kahneman, 1982) and also when it has been directly experienced (as, e.g., in the behavioral studies of Goodie & Fantino, 1996).

Underutilization of base rate information is of interest in that its occurrence bears on the general question of whether normative models of inference are also descriptive of human reasoning (see, e.g., M. S. Cohen, 1993; Doherty, 2003; Stanovich & West, 2000) and the more specific question of the extent to which people understand principles of probability and apply them in making judgments and decisions (e.g., Gigerenzer, 1998; Hertwig & Gigerenzer, 1999; Johnson-Laird, Legrenzi, Girotto, Legrenzi, & Gaverai, 1999). More particularly, there is discussion and some disagreement as to when it is normative to use base rates as a factor in making likelihood judgments and decisions (e.g., Bar-Hillel, 1990; Birnbaum, 1983; L. J. Cohen, 1979).

Does it matter whether or not people make use of base rates when they judge likelihood? The use of base rates is of practical importance in, for example, the area of medical diagnosis. In a study in which physicians were asked to estimate the probability of a woman's having breast cancer, given a positive mammogram, Eddy (1982) reported that most estimated the probability at around 75%-close to the sensitivity of the test, which had been reported to the participants as 79%, and far from the correct answer of about 8%. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.