Choice models are commonly used in transport studies. Logit models are good for some purposes but struggle with significant correlations between options. In these cases, probit models based on normal cost distributions offer a better alternative. However, problems with more than two choices require approximate solutions, leading to long computation times or inaccuracies.