Dwuwymiarowe zmienne licznikowe – bayesowskie modelowanie selekcji próby

Abstract

The article presents a joint statistical model of two count variables, one of which can be degenerated at zero. We consider a modelling framework based on switching between a bivariate Poisson regression model and a univariate one, where the switching depends on the observed outcome of the third, dichotomous variable. Bayesian analysis is advocated; in two special cases of our Bayesian model, important consequences for inference are stated. In the empirical section we consider joint modelling of the number of cash and bank card transactions in Poland with the use of data for both cardholders and non-holders.