Non-parametric Item Response Theory, General Health Questionnaire, Psychological Distress
Background: Invariant item ordering (IIO) is defined as the extent to which items have thesame ordering (in terms of item difficulty/severity – i.e. demonstrating whether
items are difficult [rare] or less difficult [common]) for each respondent who completes a scale. IIO is therefore crucial for establishing a scale hierarchy that is replicable across samples, but no research has demonstrated IIO in scales of psychological distress. We aimed to determine if a hierarchy of distress with IIO exists in a large general population sample who completed ascale measuring distress.
Methods: Data from 4107 participants who completed the 12-item General Health Questionnaire (GHQ-12) from the Northern Ireland Health and Social Wellbeing Survey 2005-6 were analysed. Mokken scaling was used to determine the dimensionality and hierarchy of the GHQ-12, and items were investigated for IIO.
symptoms reflected the following hierarchy: anhedonia, concentration, participation, coping, decision-making and worthlessness.
Results: All items of the GHQ-12 formed a single, strong unidimensional scale (H=0.58). IIO was found for six of the 12 items (H-trans=0.55), and these
Limitations: The cross-sectional analysis needs replication.
Conclusions: The GHQ-12 showed a hierarchy of distress, but IIO is only demonstrated for six of the items, and the scale could therefore be shortened. Adopting brief, hierarchical scales with IIO may be beneficial in both a clinical and research context.
Doyle F, Watson R, Morgan K, McBridge O. A hierarchy of distress and invariant item ordering in the General Health Questionnaire-12. Journal of Affective Disorders, in press