Abstract
Data from this study and five earlier studies are analyzed to determine the magnitude and significance of the influence of sex combinations of prior children upon the ultimate (or completed) number of children in the family. In addition to confirming the observation of earlier investigators that the influence of combinations of sexes in the first two children differs by geographic area, it is shown that a difference has occurred in a single region over time, and the data for the most recent cohort suggest an effect of social class as well. Three measures of association (the corrected odds ratio, the natural logarithm of this ratio, and the fourfold point correlation) point to a small magnitude of the influence studied even when it is present at a statistically significant level. Notes on sample size relative to levels of α and β errors are included, as is discussion of a valid method of combining contingency tables.