Comments on the Analysis of Twin Samples—I. The Parent Population

Abstract

Summary, Part I: In estimating twin frequencies from birth and mortality data, it is necessary to make some statistical corrections, the extent of which is demonstrated for the general population of the United States. In this population, twins constitute 2.19 per cent of all babies born since 1928. This rate is reduced to about 1.9 per cent by excess twin mortality within the first year of life, while the 2:1 ratio of same-sex to opposite-sex pairs observed at birth remains virtually unchanged. Later mortality reduces the proportion of males, but not to such an extent — even at age 75 — as to necessitate modification of the 1:1 sex ratio in estimating the proportion of monozygotic pairs by Weinberg's differential method.After the first year of life, there is no evidence for a significant difference between the mortality rates of twins and nontwins, so that the proportion of twin individuals in the population may be assumed to remain nearly the same at all ages. However, as pairs are broken by mortality and migration, the number of intact pairs is reduced at successive ages. Among people who survive to an advanced age, the relative frequency of pairs represented by at least one twin may be nearly twice as great as at birth. For comparative purposes, therefore, the most useful way to express twin data is in terms of individuals rather than in terms of twin pairs.