Estimating Spatial Differentials in Life Expectancy in Greece at Local Authority Level Cleon Tsimbos 1 , Stamatis Kalogirou 2 and Georgia Verropoulou 1,3, * 1 Department of Statistics and Insurance Science, University of Piraeus, Piraeus, Greece 2 Department of Geography, Harokopio University, Athens, Greece 3 Institute of Education, University of London, London, UK ABSTRACT The aims of this paper are to estimate, for the first time for Greece, life expectancy at birth by gender at local authority level and to explore spatial patterns. The data used in the analysis come from the vital registration system of Greece and the 2001 population census. For areas with male/female population 5,000 or more, representing 97% of the total, abridged life tables are constructed by employing the Chiang methodology. For local areas of less than 5,000, estimates of expectation of life at birth are obtained by employing regression models. Standard errors of life expectancy are estimated using the Chiang approximation as well as the Scherbov–Ediev reference tables. The results are presented in thematic as well as cluster maps; the latter are based on local Moran’s I spatial autocorrelation statistics. Local populations are ranked by level of deprivation in three groups, low, medium, and high, and differences in mean life expectancies are assessed. The findings indicate that across localities, life expectancy ranges from 70.7 to 79.6 for men (8.9-year difference) and from 76.1 to 82.5 for women (6.4-year difference). More deprived areas exhibit lower life expectancy but greater sex difference. Comparatively high life expectancy is found in Crete, the Aegean and Ionian Islands, the Peloponnese, Central-Western Greece and in Athens and Thessaloniki metropolitan areas; conditions are unfavourable in North-Eastern Greece (particularly Thrace). Life expectancy standard errors, based on the two aforementioned procedures, are close, but Chiang approximation tends to underestimate to some extend standard errors particularly for populations 5,000–10,000. Copyright © 2013 John Wiley & Sons, Ltd. Accepted 02 May 2013 Keywords: local life expectancy; estimation for small area populations; Chiang method; Scherbov– Ediev reference tables; spatial patterns; Greece INTRODUCTION S mall area estimation of mortality and evaluation of the causes of differentials at local authority level are very important in formulating appropriate public health policies aiming at reducing inequalities in health and increasing the length of life of the population. Expectation of life at birth is an appropriate summary measure of mortality extensively used in the assessment of regional disparities. It is preferable to directly or indirectly standardised mortality indices as it is more intuitive, immedi- ate, easily interpretable, and understood by non- demographic or statistical lay people or competent health authorities (Silcocks et al., 2001; Tobias and Cheung, 2003). Estimation of life expectancy at birth (e 0 ) and subsequent ages initially requires the construction of a life table. The main problem in constructing life tables at local level is the small size of the populations exhibited in census counts. Small population size means a small number of deaths and, in several instances, zero or insufficient number of vital events in some age groups for *Correspondence to: Correspondence to: Georgia Verropoulou, Department of Statistics and Insurance Science, University of Piraeus, Piraeus, Greece. E-mail: gverrop@unipi.gr Copyright © 2013 John Wiley & Sons, Ltd. POPULATION, SPACE AND PLACE Popul. Space Place (2013) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/psp.1800