The logistic model (also called logit model) is a natural candidate when one is interested in a binary outcome. For instance, a researcher might be interested in knowing what makes a politician successful or not. For the purpose of this blog post, “success” means the probability of winning an election. In that case, it would be sub-optimal to use a linear regression model to see what factors are associated with successful politicians, as the outcome variable is binary (a politician either wins or loses an election).