ABSTRACT
In this final chapter we take up a number of matters that quite commonly arise i n applicat ions. F i r s t we deal i n t u r n w i t h the analysis of variables that are derived from the in i t i a l p r imary mea surements i n a s tudy and then w i t h the related issues of hidden or latent variables and errors of measurement and the connection w i t h the large l i terature on linear s t ructura l models. W e then de velop further the use of graphical representations addressing the p rob lem of deducing the impl icat ions of a given model for a new set of variables formed by margina l iz ing and condi t ioning. T h e i m portant compl ica t ion of miss ing values is then discussed. F ina l ly , we re turn to the broad topic, touched on i n Section 2.12, of the role of notions of causal i ty in interpret ing da ta arising i n the types of s tudy w i t h wh ich we are deal ing in this book.
