Ordinal regression was used to test for associations between the dependent variable, number of lifetime other substance dependence diagnoses (zero through to five dependence diagnoses: alcohol, cannabis, sedatives, stimulants and cocaine) and the independent variables. Ordinal regression assumes that the slope coefficients in the model are the same across response categories (and lines of the same slope are parallel: the proportional odds model). Since the ordered logit model estimates one equation over all levels of the response variable, the test of parallel lines assesses whether the one-equation model is valid. If the null hypothesis is not rejected, it can be concluded that the proportional odds assumption holds. For both males and females, the test of parallel lines was non-significant (males χ2 60.70, df 48, p=0.10; females χ2 22.26, df 52, p=1.00), indicating that the slope coefficients in the model were the same across number of dependence diagnoses.
