Rethinking the Probative Value of Evidence: Base Rates, Intuitive Profiling and the "Postdiction" of Behavior

• NCJ Number: 194756
• Journal: Law and Human Behavior Volume: 26 Issue: 2 Dated: April 2002 Pages: 133-158
• Author(s): Deborah Davis; William C Follette
• Date Published: 2002
• Length: 26 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This article evaluates the probative value of evidence vis-à-vis the likelihood of having committed a crime, or the possibility of engaging in future crimes.

Abstract

In thousands of American courts, intuitive profiling evidence is presented to judges and juries. An example of intuitive profiling is if a person is in debt, they are likely to be an embezzler and vice-a-versa. This assumption has flaws and this article will explore the discrepancy between predictor variables (profiling characteristics) and the associated behavior (embezzlement) and the number of false positive outcomes. This article uses base rate data and its probative value (or lack of) as it relates to infidelity and the profile of a spouse murderer. The question begging an answer is: What is the likelihood a married man who is unfaithful to his wife is likely to kill her? The theoretical maximum probative value of information regarding infidelity is calculated using certain assumptions. These include probability of murder given infidelity and no infidelity, and the length of marriage and yield a base rate of the probability of murder if unfaithful and the probability of murder if faithful. This article's analysis revealed the maximum probative value of infidelity was less than one tenth of one percent more likely that an unfaithful husband would murder his wife than a faithful man would kill his wife at some point in their marriage. The false positive rate analysis showed that murder due to infidelity would be incorrect more than 99 percent of the time. In conclusion, the base rates of the predictors in the legal system far exceed the base rate of the crime. Thus, the probative value of evidence does not outweigh its potential prejudicial value. The use of probability and statistical techniques used to support (or not) a particular position needs to be further understood by judge and jury. Tables, references