Biplots of 1959–1961 United States Metropolitan Mortality

Abstract

National and international public-health bodies compile detailed age—race—sex-specific mortality data for a variety of geographic areas. Epidemiologists make extensive use of this information—but often in a piecemeal, summary fashion—to investigate questions of disease etiology. Exploratory data-analytical methods can be applied to these highly disaggregated mortality tables in their entireties to reveal broad patterns of the interrelationships between cause, place, and time of death, and age, race, and sex. One effective procedure for uncovering this multitude of potentially significant interaction effects is the biplotting of the deviations from various row-and-column independence models of the mortality tables. Biplots of two 201 × 56 tables are presented. These tables give the number of 1959–1961 deaths from 55 causes plus the number of survivors in 201 standard metropolitan statistical areas (SMSAs) of male whites aged 65–74 and of female whites aged 65–74. It is clear from the results that SMSAs that attract relatively large numbers of retirees have distinctive mortality characteristics of an advantageous nature. Metropolitan variations in mortality from intestinal neoplasms are highlighted in the biplots. Strong indications that cause-of-death classification procedures are not geographically uniform appear.