Bayesian Updating: Meaning, Formula and Worked Examples
Learn what Bayesian updating means, how the prior, likelihood and posterior fit the formula, and follow worked examples that turn new evidence into a decision.
Contents
- The formula in plain terms
- Prior, likelihood and posterior
- Worked example with two suppliers
- Questions about Bayesian updating
- What is Bayesian updating?
- What is the difference between prior and posterior probability?
- Is likelihood the same as posterior probability?
- Can Bayesian updating support purchasing decisions?
Bayesian updating is a way to revise a probability when new evidence arrives, by combining what you believed before with how well the evidence fits each option. This page explains the meaning of Bayesian updating, gives the formula in plain terms, and walks through worked examples, including how prior, likelihood and posterior differ and how the update supports purchasing decisions.
The formula in plain terms
The Bayesian updating formula multiplies the prior probability of an option by the likelihood of the evidence under that option. Dividing by the total likelihood across all options turns those products into a posterior probability that adds up to one. In symbols, the posterior for an option equals its prior times its likelihood, divided by the sum of prior times likelihood over every option. Bayes theorem is the same relationship, and it lets you update beliefs with new evidence without starting from scratch.
Prior, likelihood and posterior
The prior probability is your starting belief about an option before the new evidence is seen. The likelihood is the chance of observing that evidence if the option were true, which is not the same as the chance the option is true. The posterior probability is the revised belief after the evidence is folded in, and it becomes the prior when the next piece of evidence arrives. Keeping these three separate prevents a frequent error, which is reading a likelihood as if it were already a posterior.
Worked example with two suppliers
Suppose a buyer splits belief evenly between two suppliers being reliable, then receives a test result that favors one of them. The likelihood of that result under each supplier is multiplied by the prior, and the two products are compared to produce the posterior for each supplier. This bayes update example shows that strong evidence shifts belief sharply, while weak evidence leaves the prior nearly intact. A second round of evidence repeats the process, and the earlier posterior becomes the new starting point for probability updating for decisions.
Questions about Bayesian updating
What is Bayesian updating?
Bayesian updating is the process of revising a probability for an option after new evidence appears. It starts from a prior probability, weights it by the likelihood of the evidence, and produces a posterior probability that reflects both. The method treats belief as something you adjust as information arrives rather than fix once.
What is the difference between prior and posterior probability?
The prior probability describes belief before the evidence is considered, and the posterior probability describes belief after it. The likelihood sits between them, giving the chance of the evidence under each option. A prior and posterior probability example makes the shift clear, because the same evidence can move a weak prior a long way or barely touch a strong one.
Is likelihood the same as posterior probability?
No, likelihood and posterior probability are different quantities in Bayesian updating. The likelihood is the probability of the observed evidence given an option, while the posterior is the probability of the option given the evidence. Confusing the two reverses the direction of the reasoning and distorts the decision.
Can Bayesian updating support purchasing decisions?
Yes, Bayesian updating supports purchasing decisions by revising the odds on each listing as new details arrive. A buyer sets a prior for each option, then updates it with evidence such as a test result, a specification or a review. Bayesian decision analysis in Excel is possible because the update is only multiplication and division across the options.