A statistician analyzing real data (e.g., data coming from official registries or general public health surveys) will sooner or later face situations where it is necessary to go slightly beyond the traditional regression models taught in basic courses in statistics. The difficulty can be, e.g., unobserved heterogeneity between different parts of a population or overdispersion. In this talk (which consists of a few examples based on real data) we will briefly discuss how we can use the standard softwares SAS, R and BUGS (Bayesian inference Using Gibbs Sampling) to deal with some of these issues.