Modelling within-household associations in household panel studies
November 15, 2017 - Presented by Professor Fiona Steele from the London School of Economics, hosted by THE UNIVERSITY OF QUEENSLAND
Date / Time
11:15 am 15/11/2017 -
12:15 pm 15/11/2017
Seminar 201, Level 2 - Cycad Building #1018, Long Pocket
The University of Queensland, Saint Lucia QLD, Australia
rsvp to: firstname.lastname@example.org by Monday, 13 November
Household panel data provide valuable information about the extent of similarity in coresidents’ attitudes and behaviours. However, existing analysis approaches do not allow for the complex association structures that arise due to changes in household composition over time. In particular, we show that non-hierarchical random effects models impose unrealistic constraints on between-individual covariances. We propose a more flexible marginal modelling approach where the changing correlation structure between individuals is modelled directly. A key component of our correlation model specification is a ‘superhousehold’, a form of social network in which pairs of observations from different individuals are connected (directly or indirectly) by coresidence. These superhouseholds partition observations into clusters with a nonstandard and highly variable correlation structure. We therefore conduct a simulation study to evaluate the performance of the generalised estimating equations estimator for irregular within-cluster correlation matrices. Our approach is applied in an analysis of individuals’ attitudes towards gender roles using British Household Panel Survey data. We find strong evidence of between-individual correlation before, during and after coresidence, with large differences among spouses, parent-child, other family, and unrelated pairs. Our results suggest that these dependencies are due to a combination of non-random sorting and causal effects of coresidence.
Professor Fiona Steele is Professor of Statistics at the London School of Economics. Her research interests are in developments of statistical methods that are motivated by social science problems. Her areas of expertise include longitudinal data analysis, multilevel modelling, survival analysis, and simultaneous equations modelling. She has worked on a range of applications in demography, education, family psychology and health. Fiona was elected a Fellow of the British Academy in 2009.