Phantoms Never Die: Living with Unreliable Mortality Data
Andrew Cairns, David Blake, Kevin Dowd & Amy Kessler
The analysis of national mortality trends is critically dependent on the quality of the population, exposures and deaths data that underpin death rates. This paper, using England & Wales population data by way of example, develops a framework that allows us to assess data reliability and identify anomalies. First, we propose a set of graphical diagnostics that help to pinpoint anomalies. Second, we develop a
simple model that allows us to quantify objectively the size of any anomalies. An important conclusion is that bigger anomalies can often be linked to uneven patterns of births in cohorts born in the distant past, leading to errors of more than 9% in the estimated size of some England & Wales birth cohorts. We propose methods that can use the births data from these cohorts to improve estimates of the
underlying population exposures.
Keywords: Mortality data, deaths, population, exposures, cohort-births-deaths, exposures methodology, convexity adjustment ratio, graphical diagnostics.