In very technical and older literature, there was much talk of “logit and probit” models. Today, there isn’t much talk of “probit” models, and the logit model is usually referred to in terms of “logistic regression.” The two methods differ in their probability distributions. The big reason that probit models fell out of favor was that probit coefficients are essentially uninterpretable. They don’t make “common sense” in any way. Most researchers prefer logistic regression because it results in coefficients that can be transformed into the familiar odds ratio by “exponentiating” the coefficient.
Last Modified: 02/14/2019