Reconstruction of Birth Histories for the Study of Fertility in the United States, 1830-1910

Abstract

This paper describes a method to reconstruct complete birth histories for women in the 1900 and 1910 U. S. census IPUMS samples. The method is an extension of an earlier method developed by Luther and Cho (1988). The basic method relies on the number of children ever born, number of children surviving, number of children coresident in the household and age-specific fertility rates for the population to probabilistically assign an "age" to deceased and unmatched children. Modifications include the addition of an iterative Poisson regression model to fine-tune age-specific fertility inputs. The potential of complete birth histories for the study of the U.S. fertility transition is illustrated with a few examples.

Figure 1.

Children ever born by mother’s birth cohort, white women in the United States

Figure 2.

Assumed age-specific fertility schedule for example mother age 48 in the 1900 census with notches for coresident children ages 25, 17, 15, 13, 11 and 5

Figure 3.

Assumed age-specific fertility schedule for example mother age 48 in the 1900 census with notches for coresident children and notch for first imputed child

Figure 4.

Age-specific birth rates for white women with reconstructed birth histories, 1900 and 1910 IPUMS samples.

Figure 5.

Age-specific birth rates, white women born 1840–49 currently married to spouses in selected occupations and regions

Figure 6.

Age-specific birth rates, native-born white women born 1840–49 living in the Midwest region and currently married to farmers, by baseline imputation model and iteration number

Figure 7.

Age-specific birth rates, native-born white women born 1840–49 living in New Engand and currently married to men with professional occupation, by baseline imputation model and iteration number

Figure 8.

Birth interval survival curve for currently-married women age 15–68 in the 1900 and 1910 IPUMS samples