Probit

Probit refers to a type of regression model used when the dependent variable is binary (having two outcomes, like success/failure). It models the probability of an outcome using the cumulative normal distribution function. Probit regression is similar to logistic regression, which uses a logistic (S-shaped) curve instead. Researchers might prefer probit when they believe the underlying relationship between predictors and the probability of an event follows a normal distribution, or when they want to link their analysis to areas like latent trait modeling where normality is a natural assumption. In practice, probit and logistic models often produce very similar predictions, but the choice can matter in fields like toxicology, economics, and psychology where theoretical considerations favor the normal curve.

For example, researchers studying whether patients adhere to a new medication regimen (yes or no) might use probit regression to predict the probability of adherence based on factors like age, health status, and prior adherence behaviors.