“Lexis cohorts”: extracting information on half-year cohorts from 1-year format Lexis data
Alan A. Cohen, Centre for Global Health Research
John Tillinghast, Johns Hopkins Bloomberg School of Public Health
Vladimir Canudas-Romo, Johns Hopkins University
Often, demographic studies across countries may require cohorts at a finer scale than the 1-year cohorts available in sources like the Human Mortality Database. Examples include effects of season of birth on life expectancy or the discrete cohort effects of events such as pandemics and wars. We have developed a method to test differences between half-year cohorts when data are available at a 1-year time scale in Lexis format. For every birth year and death age, there are two possible death years. Those dying in the earlier year have a ~75% chance of being born in the first half of the birth year; the converse is true for those dying in the later year. We consider those dying in the earlier and later year separate “Lexis” cohorts for statistical purposes and test differences between them. Effect sizes cannot generally be estimated, but qualitative differences can be detected.
Presented in Poster Session 5: Contexts