Kahneman 1972 - Subjective Probability - A Judgment of Representativeness

Google Scholar Link

Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness. Cognitive psychology, 3(3), 430-454.

Summary

Representativeness is a heuristic wherein the characteristics of a sample are evaluated against the supposed characteristics of the population. For example, a family with 3 boys and 3 girls will be judged to be more representative than a family of 5 boys and 1 girl. The supposed distribution of genders is 50/50, so the first family, with a matching gender distribution, is seen as more representative. Samples that are judged more representative are also judged to be more probable. Samples that preserve majority-minority relations are perceived as more representative and more probable. Samples that contain all possible outcomes of the population are viewed as more representative and more probable.

A second aspect of representativeness is that a representative sample “reflects the salient features of the process by which it is generated.” Returning to the previous example of a family, if the birth order of the children is B G B G B G, that sample will appear less representative than a family whose birth order is B G B B G G. Consequently, the scenario represented by the second family will appear to be more probable. This is because the first example does not appear random. In reality, both are equally likely. Lay understanding of randomness consists of both irregularity and local representativeness. The first family’s birth order does not appear irregular, and therefore not random. A locally representative sample maintains all the characteristics of the population. For example, if a population consisted of 30 red blocks, 30 blue blocks, and 5 yellow blocks, the sample R B Y would appear to be more random (and representative and probable) than the sample R B B.

This heuristic is present even amongst statistically trained students – even amongst psychologists. The authors state that the prevalence of representativeness is due not only to its ability to preserve cognitive resources, but also because it “often work[s] with a rate of success that is more than fortuitous.” Availability is defined as the ease with which relevant instances are recalled. Highly available samples are also viewed as more probable, though the construct is different from representativeness. Availability is likely to be used when specific instances are focused on, while representativeness is more likely to be used when general properties are assessed.

Application

Even after reading this paper, I find it hard to divest myself of this heuristic. Overcoming it takes conscious effort and a real understanding of probability. As Feller is quoted in the paper as saying, “To the untrained eye, randomness appears as regularity or tendency to cluster.”