Evaluating psychological distress data

Kessler k₆ psychological distress scores are analyzed using a count model and item response theory (IRT) models are applied to the items which produce the k₆ score and generate an alternative distress score, θ^* . Other ways of utilizing the constituent items are also examined. The data used in the analysis comes from the 2014 National Survey of Drug Use and Health. Three important results emerge. First, θ^* and k₆ are not highly correlated and their distributions are quite different. The k₆ score gives a much more favourable picture of mental health than θ^* . Second, k₆ does a much better job in explaining participation in treatment programs than θ^* suggesting a very limited role for IRT methods in the analysis of psychological distress data. As a diagnostic tool k₆ is an effective and simple way of summarizing the item data. Third, for researchers interested in which individual characteristics determine psychological distress better results are obtained by analyzing the six constituent items which are used to generate the k₆ score using ordered probability models rather than k₆ itself.