Modelling the effect of social mobility on health

Heather Turner and David Firth, University of Warwick

[Project number 30129]

This project aims to explore the use of diagonal reference models in describing the effect of social mobility on health outcomes. Data from the 1991 and 2001 censuses as analysed by Bartley and Plewis (JRSS-A, 2007) will be used as a case study.

This study forms part of a larger research project, "Generalized Nonlinear Models: Theory, Computation and Extensions", supported by EPSRC grant EP/G056323/1. As part of this work, the applicants are preparing a manuscript entitled "Generalized Nonlinear Models in Practice". This manuscript will be part methodological discussion and part demonstration through practical case studies. The project proposed here is planned to form one of these case studies.

Diagonal reference models were specifically proposed for modelling the effects of social mobility (Sobel, 1981). The effect on the outcome of interest of moving from an origin class to a destination class is modelled as a weighted sum of the effects for stable individuals in those classes. Thus the effects of origin and destination are taken into account, whilst stable and mobile individuals are distinguished from each other, all in a parsimonious model. Such models have been applied in several sociological studies, but are relatively unknown in public health research. Therefore it is of particular interest to explore the use of diagonal reference models in this setting.

This project will consider data from the 1991 and 2001 census as analysed by Bartley and Plewis (2007). In their research, Bartley and Plewis used limiting long-term illness as an indicator of health and fitted logistic regression models to study the effects on this outcome of age, origin or destination class and social mobility status, for men and women separately. This study will replicate their analysis and compare it to an analysis using diagonal reference models.