Household panel studies are a rich resource for studying residential mobility. However, there is a lack of consistency in previous research on the choice of the unit of analysis. Some authors have followed the moves of households, while others have followed individuals, but neither approach is wholly satisfactory. A household-based analysis treats households as single decision-making units, and therefore does not allow for changes in household composition like union dissolution, union formation and children leaving the parental home. An individual-centred approach allows for individuals to act independently of others in their household, but does not allow for moves that are the result of decisions made jointly by the household members.
We propose a general framework for analysing residential mobility that (a) distinguishes between the moves made by couples from those made by individuals who are not in a co-residential union (singles), and (b) tracks individuals as they change between single and couple status over time. In a joint model for singleton and couple mobility, we allow the probability of a single person moving between waves t and t + 1 to depend on his/her individual-specific random effect, and the corresponding transition probability for couples to depend on a weighted combination of both partners’ random effects. The model can be framed as a multilevel multiple-membership model.
The proposed method is applied in an analysis of residential mobility using data from the British Household Panel Study, and the results compared with those from other approaches adopted in previous research.
Fiona Steele (University of Bristol)
Date & time:
28 Feb 2011 16:00 pm
External seminars home