Conference Paper BHPS-2007 Conference: the 2007 British Household Panel Survey Research Conference, 5 July -7 July 2007, Colchester, UK
Positive mental health versus emotional distress in the BHPS -abstract-
When applied in health care settings and clinical populations the General Health Questionnaire has a proven pedigree in case-finding for anxiety and depressive disorders. It is highly successful at identifying individuals who may currently be diagnosable with a psychiatric disorder when interviewed by a clinical professional. However instruments from the GHQ family are also widely used in general health survey and epidemiological research for ranking individuals by their level of psychological functioning or distress. Here the total score is often adopted as a useful summary measure of psychological health. In general use the GHQ is not intended to measure positive attributes, only the absence of distress. However, imaginative approaches to coding responses to some GHQ items have been employed to capture some suggested notion of positive mental health: the validity of these attempts is largely unknown, and they remain rather contentious. The measurement characteristics that resulted in the widespread success of the GHQ in its intended role, in screening applications, were integral to its design and development, and quite rightly have determined its mode of use: it was not designed to measure anything positive, and may do so only accidentally, with poor reliability since the instrument was optimized to discriminate between those who are ill and those who are not ill (with items selected on the basis of two-group (well vs ill) calibration studies. Under traditional binary scoring, caseness thresholds have been set to identify those individuals with likely disorder. Likert scores can also be summed to profile health variations on a single (GHQ-12, 30, 60) or multiple (GHQ-28) dimensions. But what scoring approaches are valid for different notions of positive mental health, and can psychometric evidence be marshaled to support this case? This talk will introduce these issues, illustrate them, and solve many of the remaining challenges, using contemporary theory from applied psychometrics.