Issues in the Analysis of Small Area Mortality

Abstract

This paper considers mortality variation between small areas in London during the 1980s in both a cross-sectional and a temporal perspective. The analysis takes account of the special nature of small area mortality data and incorporates sampling variation in addition to extraneous variation, as well as investigating the extent of autocorrelation, both spatial and temporal. The strongest association between socio-economic structure and mortality is apparent in the cross-sectional analysis and remains when spatial autocorrelation is allowed for. However, socio-economic concomitants of mortality are also significant in the temporal analysis based on conditional (two subperiod) and panel (time series) analyses.