Abstract
This study aimed to identify predictors of residential mobility in 55+ Canadians, to characterise neighbourhood changes following mobility, to assess whether such changes differ according to income, and to evaluate for cross-sectional estimations of place-health relationships the extent of bias associated with residential mobility. Using longitudinal data from the Canadian National Population Health Study (NPHS), residential mobility was operationalised by a change in postal code between two consecutive waves. Individuals' sociodemographic factors and neighbourhood characteristics were analysed in relation to mobility. Bias in cross-sectional estimations of place-health associations was assessed analysing neighbourhood-level deprivation and housing quality in relation to self-assessed health. Multiple age-related events were predictive of moving. Three out of 10 individuals moved at least once. Two thirds of movers experienced a change in neighbourhood type and such changes were not associated with income. No systematic biases in estimating place effects on health using cross-sectional data were observed. Given that individual-level socioeconomic status (SES) was neither a predictor of moving nor of its consequences in terms of neighbourhood type, controlling for SES could potentially lead to biased estimations of place-health associations. Results suggest that cross-sectional data can yield valid estimations of place-health associations among older adults.