Individuals' judgement of probabilities—the likelihood of different events to occur, or the attribution of likely causes to what has occurred—is plagued by numerous biases and errors. This note focuses on Bayesian updating and the base rate neglect, and also provides a brief overview of other biases, including overconfidence, reversion to the mean, the law of small numbers, the conjunction fallacy, confirming evidence bias, and hindsight bias. Numerous examples illustrate the implications of the biases for managerial decisions.ExcerptUVA-QA-0860Rev. Apr. 14, 2020Updating ProbabilitiesA man has two children. When asked if at least one of them is a girl, he answers,"Yes." What are the chances that the other child is also a girl?This is not a trick question⎯you can assume there is a 50-50 chance of a baby being born a boy or a girl, independent for each child. The answer to this riddle seems obvious: 50%. As it turns out, 50% is not the correct answer. Moreover, the mistake is not anecdotal; instead it reflects a flaw many individuals exhibit when incorporating new information. The flaw is systematic and leads to predictably bad decisions, and sometimes terrible ones. Incorporating new information is an essential task that business managers perform routinely, and avoiding systematic flaws in this part of the decision-making process can only lead to better outcomes. Examples include detecting the causes that produce quality defects sooner, avoiding projects with a knowable high probability of failure, and deciding when to request additional information.In order to demonstrate the logical process of updating probabilities and expectations, some groundwork is necessary. Specifically, we need to become familiar with the concepts of outcomes, probabilities, and events. Then, we must introduce the central concept of conditional probability. Finally, we will derive the law of total probability and use Bayes's rule to update probabilities
