ABSTRACT
In this chapter I shall attempt to outline the ideas behind, and uses of, a range of statistical techniques which can be lumped together under the general term of structural equation modelling (SEM). This is a set of techniques which have been developed more recently than the other methods discussed so far in this book (originating in the 1960s), and have only just started to become commonplace in psychological research. Indeed, at the time of writing, they are undeniably trendy – it has become fashionable for journal reviewers, for example, to advise authors to re-analyse multiple regression data in the style of a path analysis or even a structural equation model. This enthusiasm for sophisticated and impressive-looking analysis may be misguided at times, however, and there are many statisticians and psychologists who are somewhat cautious about the routine application of SEM to psychological data.
