Abstract
Accurate vital statistics are required to understand the evolution of racial disparities in infant health and the causes of rapid secular decline in infant mortality during the early twentieth century. Unfortunately, U.S. infant mortality rates prior to 1950 suffer from an upward bias stemming from a severe underregistration of births. At one extreme, African American births in southern states went unregistered at the rate of 15 % to 25 %. In this study, we construct improved estimates of births and infant mortality in the United States for 1915–1940 using recently released complete count decennial census microdata combined with the counts of infant deaths from published sources. We check the veracity of our estimates with a major birth registration study completed in conjunction with the 1940 decennial census and find that the largest adjustments occur in states with less-complete birth registration systems. An additional advantage of our census-based estimation method is the extension backward of the birth and infant mortality series for years prior to published estimates of registered births, enabling previously impossible comparisons and estimations. Finally, we show that underregistration can bias effect estimates even in a panel setting with specifications that include location fixed effects and place-specific linear time trends.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Examples of recent studies that used underregistration-biased births estimates include Bhalotra and Venkataramani (2011), Clay et al. (2013), Collins and Thomasson (2004), Cutler and Miller (2005), Eriksson and Niemesh (2016), Hansen (2014), Jayachandran et al. (2010), Moehling and Thomasson (2014), and Thomasson and Treber (2008).
States entered the DRA as early as 1880, but the BRA did not begin until 1915. Table A3 in the online appendix lists the entry dates for each state into the BRA and the DRA.
Researchers in the early twentieth century understood the biases present in infant mortality rates. For example, in 1934, a former president of the Population Association America wrote, “If birth registration is equally deficient in various states, only absolute values for birth rates and infant mortality are affected. However, if there are large differences in completeness between states, the comparative standing of states in these respects will vary correspondingly when they are ranked on an adjusted instead of an unadjusted basis” (Whelpton 1934:125).
City-level infant mortality estimates are also biased from birth underregistration. However, we are loath to use our procedure to adjust rates at the city level. The census records only the state of birth, not the city. Any allocation of census enumerations to a city of birth would be plagued by bias from cross-city migration. Because state of birth is known, this migration bias does not affect our state-based revised rates.
Vital registration systems were and remain the responsibility of several states. The federal government’s role is limited to the promotion of state registration systems and to working with the states to produce national-level statistics. An act of Congress in 1902 put the national system on a firm footing by making the Census Bureau a permanent agency and providing the authority to collect information on births and deaths.
The recall period for the infant postal cards was limited to four months to reduce bias from memory lapses.
Table A1 in the online appendix reports results from the test by region and whether delivery occurred in a hospital.
A subsequent test conducted in 1950 showed major improvements over the decade, with 97.8 % registered for the nation as a whole, although some states lagged behind (Shapiro and Schachter 1952). The likely explanation for this rapid improvement is that registration completeness is highly correlated with the percentage of births delivered in a hospital. Completeness eventually reached close to 100 % by the mid- to late 1960s, when hospital deliveries approached 100 % of all births after the integration of hospitals in the South (U.S. Census Bureau 1973).
Figure A1 in the online appendix plots the proportion of all births registered from the 1950 test against that from the 1940 test. The figure clearly shows that all states increased the quality of their published birth data. However, some improved more than others. One plausible reason for this variability is that low rates of out-of-hospital births persisted for nonwhites in the South and West.
The Mountain census region includes the following states: Arizona, New Mexico, Nevada, Utah, Colorado, Wyoming, Idaho, and Montana.
Table A4 in the online appendix reports results from this exercise.
The white rates are for native-born whites only, and the nonwhite rates are for blacks. Foreign-born whites and nonblack nonwhites are excluded from the current estimates.
The 1930 and 1940 censuses were conducted on April 1, so the census counts do not align with the vital statistics data reported by calendar year. Our estimates of births for a given year are underestimated when actual births are declining and are overestimated when actual births are increasing. In practice, reallocating “births” so that the census counts follow the calendar year does not meaningfully change IMR estimates. Moreover, year and state fixed effects in a panel setting account for any of the differences. An explanation of the allocation procedure and full set of results are available upon request from the authors. The 1920 census questions referred to January 1, and thus the 1920 census counts do not suffer from this problem.
Rates with noninfant deaths allocated to state of birth are the preferred revised rates and correspond to Adjustment 4 in the online appendix. The procedure allocates the number of age-specific reported noninfant deaths in each state of occurrence to states of birth using the age-birth-state breakdown in the complete count censuses. For example, if 10 % of black 8-year-olds living in Illinois in the 1940 census were born in Mississippi, then 10 % of black noninfant deaths in Illinois are apportioned to black births in Mississippi for the 1932 birth year. Rates with noninfant deaths recorded in the state of occurrence corresponds to Adjustment 2 in the online appendix. Figure A2 in the online appendix plots the relationship for noninfant deaths.
In the machine-readable files, we provide an additional adjusted rate that does not scale up census counts by the extent of underenumeration, as in our preferred estimate. In some sense, underenumeration that enters the denominator provides a balance against error in the numerator from unregistered infant deaths. However, the differences in the two series may not be important in some contexts. Scaling by the extent of underenumeration causes a level shift down in IMR but does not affect the overall trend (see Fig. A4 in the online appendix). Moreover, the scaling does not affect results in a panel setting with year and state fixed effects.
This scaled rate corresponds to Adjustment 5 in the online appendix.
For births during the six months prior to the April census date in 1940, the estimates of registration completeness contained in Grove (1943) provide an accurate measure of the bias in infant mortality calculations. As such, we are confident in their use to make adjustments at the state level for 1939 and 1940. See the discussion of Adjustment 3 in the online appendix for more information.
The proportion of births registered clearly varies over time within a state. A simple way to argue the point is to notice the large differences in registration rates by whether the birth occurred in a hospital as well as the rapid increase in the proportion of hospital births over time. The 1940 test showed that 98.5 % of all hospital births were registered, and 86.1 % of births were outside hospitals. Moriyama (1946) estimated that only 36.9 % of births were hospital deliveries in 1935 but that this figure increased to 55.8 % in 1940 and to 75.6 % in 1944.
Table A2 in the online appendix lists the years and states for which new estimates are available.
We do not observe a similar regional convergence for whites as the urban penalty for infant whites had disappeared by 1920.
According to the 1940 decennial census, 89 % of blacks lived in urban areas in the North census region, whereas 34 % were urban dwellers in the South census region. Data underlying these calculations come from the full count 1940 census microdata from IPUMS.
The authors were well aware of the underregistration of births and discussed how potential bias might enter their estimates. However, at the time, no direct way of accounting for the bias was available.
References
Anderson, D. M., Brown, R., Charles, K. K., & Rees, D. I. (2016). The effect of occupational licensing on consumer welfare: Early midwifery laws and maternal mortality (NBER Working Paper No. 22456). Cambridge, MA: National Bureau of Economic Research.
Bhalotra, S., & Venkataramani, A. (2011). The captain of the men of death and his shadow: Long-run impacts of early life pneumonia exposure (IZA Discussion Paper No. 6041). Bonn, Germany: Institute for the Study of Labor.
Cassedy, J. H. (1965). The registration area and American vital statistics: Development of a health research resource, 1885–1915. Bulletin of the History of Medicine, 39, 221–231.
Clay, K., Troesken, W., & Haines, M. R. (2013). Lead and mortality. Review of Economics and Statistics, 96, 458–470.
Collins, W. J., & Thomasson, M. A. (2004). The declining contribution of socioeconomic disparities to the racial gap in infant mortality rates, 1920–1970. Southern Economic Journal, 70, 746–776.
Cutler, D., & Miller, G. (2005). The role of public health improvements in health advances: The twentieth-century United States. Demography, 42, 1–22.
Davis, W. H. (1925). Necessity for completing the registration area by 1930. American Journal of Public Health, 15, 399–404.
Deacon, W. J. V. (1937). Tests and promotion of registration of births and deaths. American Journal of Public Health, 27, 492–498.
Eriksson, K., & Niemesh, G. T. (2016). Death in the promised land: The Great Migration and black infant mortality. Retrieved from https://ssrn.com/abstract=3071053
Eriksson, K., Niemesh, G. T., & Thomasson, M. (2018). Revised infant mortality rates and births for the United States 1915–1940 (ICPSR 37076, version 1) [Computer file]. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor]. https://doi.org/10.3886/ICPSR37076.v1
Greville, T. N. E. (1947). United States life tables and actuarial tables, 1939–1941. Washington, DC: National Office of Vital Statistics.
Grove, R. (1943). Studies in the completeness of birth registration, part I. Completeness of birth registration in the United States, December 1, 1939 to March 31, 1940 (Vital Statistics—Special Reports Series, Vol. 17, No. 18). Washington, DC: National Office of Vital Statistics.
Hacker, J. D. (2013). New estimates of census coverage in the United States, 1850–1930. Social Science History, 37, 71–101.
Hansen, C. W. (2014). Cause of death and development in the US. Journal of Development Economics, 109, 143–153.
Hedrich, A. W., Collinson, J., & Rhoads, F. D. (1939). Comparison of birth tests by several methods in Georgia and Maryland (Vital Statistics—Special Reports Series, Vol. 7, No. 60). Washington, DC: National Office of Vital Statistics.
Jayachandran, S., Lleras-Muney, A., & Smith, K. V. (2010). Modern medicine and the twentieth century decline in mortality: Evidence on the impact of sulfa drugs. American Economic Journal: Applied Economics, 2(2), 118–146.
Komisarow, S. (2017). Public health regulation and mortality: Evidence from early 20th century milk laws. Journal of Health Economics, 56, 126–144.
Land, K. C., Hough, G. C., Jr., & McMillan, M. M. (1984). New midyear age-sex-color-specific estimates of the U.S. population for the 1940s and 1950s: Including a revision of coverage estimates for the 1940 and 1950 censuses. Demography, 21, 623–645.
Lenhart, R. F. (1943). Completeness of birth registration in the United States in 1940. American Journal of Public Health, 33, 685–690.
Lewis, J. (2018). Infant health, women’s fertility, and rural electrification in the United States, 1930–1960. Journal of Economic History, 78, 118–154.
Lunde, A. S. (1980). The organization of the civil registration system of the United States (Technical Report No. 8). Bethesda, MD: International Institute for Vital Registration and Statistics.
Miller, G. (2008). Women’s suffrage, political responsiveness, and child survival in American history. Quarterly Journal of Economics, 123, 1287–1327.
Moehling, C., & Thomasson, M. (2014). Saving babies: The impact of public health education programs on infant mortality. Demography, 51, 367–386.
Moriyama, I. (1946). Estimated completeness of birth registration: United States, 1935 to 1944 (Vital Statistics–Special Reports Series, Vol. 23, No. 10). Washington, DC: National Office of Vital Statistics.
Moriyama, I. (1990). Measurement of birth and death registration completeness (Technical Report No. 43). Bethesda, MD: International Institute for Vital Registration and Statistics.
Preston, S. H., Elo, I. T., Hill, M. E., & Rosenwaike, I. (2003). The deemography of African Americans 1930–1990. Dordrecht, the Netherlands: Kluwer Academic Publishers.
Ruggles, S., Genadek, K., Goeken, R., Grover, J., & Sobek, M. (2017). Integrated public use microdata series: Version 7.0 [Machine-readable database]. Minneapolis: University of Minnesota. https://doi.org/10.18128/D010.V7.0.
Shapiro, S. (1950). Estimating birth registration completeness. Journal of the American Statistical Association, 45, 261–264.
Shapiro, S., & Schachter, J. (1952). Birth registration completeness, United States, 1950. Public Health Reports (1896–1970), 67, 513–524.
Thomasson, M., & Treber, J. (2008). From home to hospital: The evolution of childbirth in the United States, 1928–1940. Explorations in Economic History, 45, 76–99.
Troesken, W. (2008). Lead water pipes and infant mortality in turn-of-the-century Massachusetts. Journal of Human Resources, 43, 553–575.
U.S. Census Bureau. (1973). Test of birth registration completeness 1964 to 1968 (Evaluation and Research Program Series PHC(E)-2). Washington, DC: U.S. Government Printing Office.
Whelpton, P. K. (1934). The completeness of birth registration in the United States. Journal of the American Statistical Association, 29, 125–136.
Wilbur, C. (1916). The federal registration service of the United States: Its development, problems, and defects. Washington, DC: U.S. Government Printing Office.
Wilcox, W. F. (1933). Introduction to the vital statistics of the United States, 1900 to 1930. Washington, DC: U.S. Department of Commerce, U.S. Census Bureau.
Acknowledgments
A set of machine-readable files of revised births and infant mortality rates is available online through the Inter-university Consortium for Political and Social Research (ICPSR 37076, version 1). We are particularly grateful to William Even, Analisa Packham, the editors of this journal, and three anonymous referees for helpful suggestions. We also thank Jeremy Atack, William J. Collins, Dora Costa, Gordon Hanson, Adriana Lleras-Muney, Seth Sanders, Marianne Wanamaker, and Sven Wilson for comments when portions of this work were included in the paper “Death In the Promised Land: The Great Migration and Black Infant Mortality.” Brian Lee and Man-Ting Chang provided excellent research assistance.
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
ESM 1
(PDF 433 kb)
Rights and permissions
About this article
Cite this article
Eriksson, K., Niemesh, G.T. & Thomasson, M. Revising Infant Mortality Rates for the Early Twentieth Century United States. Demography 55, 2001–2024 (2018). https://doi.org/10.1007/s13524-018-0723-2
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13524-018-0723-2