Probit analysis is alternative of logit method. The main difference is that assume normal distribution of random
variables (independent variables in model). The difference lies in fact that logistic function has harder “fat tails”.
There are no significant differences in practise, only in the case that sample contains numerous observations with
extreme values. Parameters estimates obtained by logit and probit models cannot be compared directly because the
logarithmic distribution has variance equal π2
/3, therefore the estimates obtained by logit model have to be
multiplied by 31/2/π in order to be comparable with estimates obtained in the probit model