How do urn models help explain opinion polls and sample surveys?

Urn models serve as a foundational framework in probability theory for conceptualizing how sample surveys and opinion polls function. In these models, an urn containing a mixture of two types of items—such as red and black balls—represents a population divided by a specific characteristic, such as voters supporting candidate A versus candidate B.1

By drawing a sample from the urn, researchers can make inferences about the total composition of the urn, which is analogous to using a sample of voters to predict an election outcome. When the composition of the population is unknown, statisticians often use Bayes’s theorem to combine prior information—such as historical voting patterns—with the results of the current sample to produce an a posteriori distribution. This allows for a probabilistic estimate of the population's preferences even when the underlying "urn" cannot be observed directly.1

While simple models assume replacement, real-world sampling often involves selecting individuals without replacement, which is mathematically described by the hypergeometric distribution. These models provide a rigorous way to quantify uncertainty and guide decision making when faced with incomplete information.21