Dealing with unmeasured confounders in studies of health inequalities:
George Ploubidis, London School of Hygiene and Tropical Medicine
Date: Thu, 7 April 2011 , Time: 16:00-17:00
Recent years have seen developments in the conceptualization of the possible pathways linking Socio-Economic Position (SEP) and health over the life course, as well as across generations. The assessment of the validity of these hypothesised pathways requires suitable tools. However, standard analytical methods based on regression modelling, may not be adequate for this task since the hypothesised pathways may be extremely complex, involving several mediators, each of which may be associated with the outcome of interest via other, confounding or mediating, factors. Even in situations where appropriate models such as structural equation models are employed, the robustness of parameter estimates depends on the correct conceptual specification of the model, which in practice means that all potential confounders must be included in the analysis. An attractive feature of structural equation models is that they allow the specification of several scenarios for the unmeasured confounders –defined here as latent variables- and the relative plausibility of each scenario can be quantified. Results from sensitivity analyses using latent variables to capture the unobserved heterogeneity caused by unmeasured confounders will be presented. Preliminary findings suggest that latent variables were able to successfully mimic the effect of unmeasured confounders in situations encountered frequently in the study of health inequalities.
About the presenter
After completing a B.Sc (Hons) at Lancaster University and a M.Sc at Bristol University he was awarded a European Union Erasmus scholarship for a PhD in psychometrics in the University of Athens (Greece) with a year of formal training in statistics at Tilburg University (Netherlands). He completed his PhD in 2004 (with distinction) and soon after he was appointed as a postdoctoral research associate at the department of psychiatry of the University of Cambridge. In 2007 he accepted the position of a lecturer at the Centre for Population Studies at the London School of Hygiene and Tropical Medicine (LSHTM). His research interests are life course analysis on mental health, social and structural determinants of population health, and developing and evaluation of measurement tools for population based health surveys.