Previous studies of prevalence and psychometric properties of substance use primarily focus on individual substances, treating each as distinct (e.g. Kahler et al., 2004; Kashdan et al., 2005; Krueger et al.; 2004, Saha et al., 2006; Wallace et al., 2003). However, use of different substances is highly inter-correlated, as are the genetic and environmental contributions to the use of specific substances (e.g. Hall & Lynskey, 2005; Kendler et al., 2003). It is therefore likely that use of multiple substances forms a coherent dimension of overall poly-substance involvement. The application of multigroup IRT methodology allowed us to investigate individual substances in terms of how they relate to the dimension of poly-substance involvement, as well as estimating how age and sex effects operate at the level of the underlying dimension.
