Dynamic households? Assessing the impact of changing households on smoking behaviours using multi-level multiple membership modelling.

It is well-known that smoking is strongly associated with the number of smokers within an individual’s social network, in particular the people whom he/she lives with. There is some qualitative evidence to suggest that a change of job or place of residence may result in changes in an individual’s social network, leading to changes in routines and thus increasing the changes of quitting. In this work-in-progress, I explore the impact of changing households on smoking using multi-level multiple membership modelling. In doing so, I wish to examine and stimulate discussion on the issues associated with developing a unique ‘household’ indicator over time and incorporating changing compositional household characteristics.