Content area
Full Text
(ProQuest: ... denotes non-USASCII text omitted.)
Introduction
The General Health Questionnaire (GHQ) was developed by Goldberg & Hillier (1979) as a screening device for minor, non-psychotic, psychiatric disorder, or psychological morbidity. Originally devised as a 60-item questionnaire measuring physical and psychological symptoms, the GHQ is also available in 30-, 28- and 12-item versions that focus on the psychological symptoms. For example, the 28-item questionnaire covers four main elements of distress: depression, anxiety, social impairment and insomnia (Goldberg & Hillier, 1979). The GHQ is widely used clinically and in research and has been translated into several languages (http://shop.nfer-nelson.co.uk/icat/generalhealthquestionnair; accessed 9 January 2008). Each version of the GHQ is related through having common items and, although different scoring systems can be applied, including a modified Likert-type scoring system, a total score is generated. The total score indicates the level of psychological morbidity, with higher scores indicating greater levels of morbidity, or poorer general health, with suggested thresholds, indicating psychological distress, for each version of the scale.
As far as it is possible to discern from an online search of the Web of Science using 'GHQ' and 'Mokken' as search terms and no date restrictions, the GHQ, which is strong psychometrically, has not been subjected to analysis for a hierarchy among its items, although it has been used to validate other scales that have been developed using this procedure (Barkow et al. 2001; Mergl et al. 2007). One hierarchical scale for general health has been developed but does not seem to be widely used (Moorer & Suurmeijer, 1994; Moorer et al. 2001). Hierarchical scales are used frequently in social, psychological, medical and nursing research (Kempen & Suurmeijer, 1991; Watson 1996; Kingshott et al. 1998; Ringdall et al. 2003), and establishing whether a scale has hierarchical properties adds a new dimension to its use other than simply using the total score obtained by summing, for example, Likert-type responses. If a scale is demonstrated to have hierarchical properties, it indicates that the items are ordered relative to one another and, by implication, ordered along the latent trait that is being measured. Therefore, although a total score from a set of hierarchically ordered items indicates the extent to which the latent trait is present (or absent), just as with any...