a type of regression analysis in which a dichotomous outcome variable is related to any of a number of different predictor or independent variables. The technique assumes that a latent process generated the observed binary data and often produces very similar results to logistic regression. Also called probit regression.