"The Use of the Census to Estimate Childhood Mortality: Comparisons fromthe 1900 and 1910 United States Census Public Use Samples"
Michael R. Haines, Samuel H. Preston
NBER Historical Working Paper No. 85
This paper estimates child mortality by race and nativity for the U.S. as a whole and the Death Registration Area based on the public use micro- samples of the 1900 and 1910 censuses. We compare indirect estimates to mortality rates and parameters based on published census and vital statistics data. The censuses of 1900 and 1910 both asked adult women about children ever born and children surviving which, when tabulated by age or marriage duration can be used to estimate probabilities of their children dying at various ages up to 25. Data on children ever born for 1910 were partially tabulated and published in conjunction with the 1940 federal census but the information on children surviving was never tabulated and published; nor was information from 1900. The public use micro samples of the 1900 census permit the application of these well-established indirect methods. This paper applies the basic indirect age and marriage duration methods, and a method using the backward projection of age distribution of surviving own-children of younger adult women. The results match well to life tables calculated from aggregaed census and vital statistics for the total white, native white and foreign-born white populations. The results are less definite for African-Americans but it seems that mortality was substantialy better than indicated by the widely cited Glover life tables for 1900/02, 1901/10, and 1909/11 for the original the original Death Registration Area of 1900. Overall, however, it appears that calculated life tables from published vital statistics and census popula- tions for the Death Registration Areas of 1900 and 1910 describe the remainder of the population relatively well.
Published: Historical Methods, Vol. 30, no. 2 (Spring 1997): 77-96.