Chapter 16, Causes Trump Statistics (Kahneman, 2011), explores Bayesian inference on judgement. Again, interesting viewpoint of statistics from the judgement perspective instead of strictly from an applied experimental perspective. This chapter looks at a priori information’s influence on judgement. The influence of information that can psychologically influence what information is considered or given more weight. A priori information may automatically trigger the System 1 stereotype. 

Kahneman provides examples comparing “statistical base rates” vs. “causal base rates”. He defined statistical base rates as “facts about a population to which a case belongs, but… [which] are not relevant to the individual case” and may be “underweighted, and “sometimes neglected altogether” (P. 168). For example, the “statistical base rate” information provided was that there were many more of one color of cab (85% green) than the other color cab (15% blue) in a city. An eyewitness testified that a blue cab was involved in a particular accident. The reliability of the eyewitness was determined to be 80% reliable or correct. With this information, the decision-makers looked toward the reliability of the eye witness (80%) and determined that “the probability that the cab involved in the accident was Blue rather than Green” was 80%. The statistical base rate about the percentage of green vs blue cabs in the city was ignored.

Kahneman defined the “causal base rates” as information that “change your view of how the individual case came to be” which “are treated as information about the individual case” and “are easily combined with other case-specific information” (p.168). The  “causal base rate” experiment stated that there were the same number of each color of cabs (50% green; 50% blue) but one color was involved in the majority of accidents (85% green; 15% blue). When given that information along with the reliability of the eyewitness (80% reliable), the decision-maker gave the percent involved in accident (85% Green) more weight because they instantly created a stereotype of the group involved in the majority of accidents as more accident prone without considering any other reasons which may have influenced the accidents (e.g., where they serviced, when they serviced, and who was deemed at fault in those accidents).

The causal base rate information seemed to trigger System 1 and the automatic creation of stereotypes to make decisions easier (easier to place blame or cause). With only statistical base rates people tended to ignore the base rate information in determining the probability and go with the eyewitness reliability (which is easier). Basically, humans are emotionally swayed, quick to create stereotypes, and tend toward the easiest choice. No matter how well versed one may be in statistics and probability testing it is hard to fight this urge. 

Despite the concept of stereotypes being frowned upon, Kahneman points out that the use of stereotypes which “represent categories as norms and prototypical exemplars” can aid in judgement (p.168). Although, he goes on to state that, “some stereotypes are perniciously wrong, and hostile” and “can have dreadful consequences” (p. 168-169). The conversation makes me think of the area of expertise in which there is an instant feeling that directs judgement and tends to be overwhelmingly correct and difficult to explain to anyone else. There might be a distinction to be made about whose System 1 judgement should be relied upon. One might be stereotyping based on ill conceived notions and one might be categorizing based on experience and both will be flooded with strong feelings.

Kahneman goes on to discuss another study that illustrates how people tend to have an “unwillingness to deduce the particular from the general was matched only by their willingness to infer the general from the particular” (p.174). A common reminder in college classes is that a case you have experienced personally doesn’t make it more likely to have occurred in the general population. Nevertheless, people feel stronger about something they have personally experienced or have heard about one person experiencing, than a statistic that states the probability of the occurrence in a large population. He goes on to state that “even compelling causal statistics will not change long-held beliefs [e.g., stereotypes] or beliefs rooted in personal experience” (p.174). Nevertheless, be-it our narcissistic tendencies or evolutionary mechanisms, we are more likely to remember and learn something that we find surprising in our own behavior or experience than hearing about others and this will effect our judgement. 

* A little background definitions: “Bayes’s theorem, in probability theory, [is] a means for revising predictions in light of relevant evidence, also known as conditional probability or inverse probability “(https://www.britannica.com/topic/Bayess-theorem). “Related to the theorem is Bayesian inference, or Bayesianism, based on the assignment of some a priori distribution of a parameter under investigation. In 1854 the English logician George Boole criticized the subjective character of such assignments, and Bayesianism declined in favor of “confidence intervals” and “hypothesis tests”—now basic research methods” (https://www.britannica.com/topic/Bayess-theorem).

 

Reference

Kahneman, D. (2011). Causes trump statistics. In Thinking Fast and Slow (pp. 156-165). New York, NY: Farrar, Straus and Giroux.