topic-unique styles??that enable a complete specification of the distribution of the result by using random consequences to capture in-issue correlations. The selection as to which approach to decide on partly depends on the purpose with the investigate and the specified interpretation with the believed consequences (inhabitants-regular versus subj