Gender Differences in Measures of Health in a UK Population Based Survey of Older People: A Latent Variable Modeling Approach

George B Ploubidis, London School of Hygiene and Tropical Medicine

We employed a latent variable model and decomposed six health indicators, three objective (measured by a nurse) and three self-reported, into three components: 1)valid health; 2)method and 3) random error variance using data from the English Longitudinal Study of Ageing. Among the indicators used, functional limitations appeared the best indicator of health i.e. contaminated less by method and error variance than all other indicators. Analysis using the three latent factors confirmed the gender gap in health, with women having poorer health than men when external bias in the form of method variance and random error was controlled. However, men appear prone to ‘overeporting’ of health problems when self report measures are used. Age was also associated with worse health and self report method variance. Older participants’ responses to self report health indicators appear to be influenced by external factors to a larger extent than those of their younger counterparts.