Logit

The logit is the natural logarithm of the odds that an event occurs or a state exists. For example, if P(Y) is the probability that someone is a computer user or that someone commits a crime, then logit(Y) = ln[P (Y)/(1−P(Y))]. The logit is negative whenever P(Y) is less than .5, and is negative and infinitely large in magnitude when P(Y) = 0; it is positive whenever P(Y) is greater than .5, and positive and infinitely large in magnitude when P(Y) = 1. If P(Y) =.5, because the natural logarithm of 1 is zero, logit(Y) = ln[(.5)/(.5)] = ln(1) = 0. The logit of a dichotomous or dichotomized dependent variable is used as the dependent variable in logistic regression and logit models.