Getting rid of dichotomous sex estimations: why logistic regression should be preferred over discriminant function analysis