Purpose: The general health questionnaire-12 (GHQ-12) is a self-report instrument for measuring psychological morbidity. Previous work has suggested several multidimensional models for this instrument, although it has recently been proposed that these may be an artefact resulting from a response bias to negatively phrased items. The aim here was to explore the dimensionality of the GHQ-12.
Methods: Cluster analysis, exploratory factor analysis and confirmatory factor analysis were applied to waves of data from the English longitudinal study of ageing (ELSA Waves 1 and 3), in order to evaluate fit and factorial invariance over time of the GHQ-12.
Results: Two categories of respondents were identified: high and low scorers. Item variances were higher across all items for high scorers and higher for negatively phrased items (for both high and low scorers). The unidimensional model accounting for variance observed with negative phrasing (Hankins in Clin Pract Epidemiol Ment Health 4:10, 2008) was identified as having the best model fit across the two time points.
Conclusions: Item phrasing, item variance and levels of respondents' distress affect the factor structure observed for the GHQ-12 and may perhaps explain why different factor structures of the instrument have been found in different populations.