Logit and probit models are used to model categorical dependent variables (ranging from 0 to 1), with the aim of evaluating these probabilities based on independent variables. These models help researchers in making binary predictions, such as buying a product or paying off a loan, and analyzing probabilities to improve decision-making.