A Simple Model for the Statistical Analysis of Large Arrays of Mortality Data: Rectangular vs. Diagonal Structure

Wilmoth, J.R. & Caselli, G. (1987). A Simple Model for the Statistical Analysis of Large Arrays of Mortality Data: Rectangular vs. Diagonal Structure. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-87-058

Preview

Text WP-87-058.pdf Download (673kB) | Preview

Abstract

A simple descriptive model is proposed for the analysis of large, non-additive mortality arrays. Similar in form to additive-plus-multiplicative models discussed by other authors, the model goes one step further by introducing a diagonal term. An exemplary application of the model to French male post-War mortality data demonstrates three important characteristics of the data being analyzed: 1) the structure of the data matrix is largely additive; 2) some rectangular non-additivity exists, implying that mortality has declined with varying speed at different ages or, equivalently, that the shape of the age-curve of mortality has changed over time; and 3) residual non-additive diagonal structure exists, indicating that some "peculiar" cohorts have had mortality experiences which deviate by as much as 2 or 3% from levels which would be expected considering only the age and period of death.

Actions (login required)

View Item