I use the [U.S.] Integrated Public Use Microdata Series to assess the potential effects of local labor-market conditions on long-term trends and race differences in marital instability. The rise of female labor-force participation and the increase in nonfarm employment are closely associated with the growth of divorce and separation. Moreover, higher female labor-force participation among black women and lower economic opportunities for black men may account for race differences in marital instability before 1940, and for most of such differences in subsequent years. However, unmeasured intervening cultural factors are probably responsible for at least part of these effects.

The end of welfare as we know it in the United States was predicated on the belief that the welfare system was responsible for dramatic upsurge of single-parent families. This article addresses the issue historically, examining the potential impact of interstate differences in Aid to Families with Dependent Children (AFDC) benefit levels on family structure from 1940 to 1990. The author's results reveal that the impact of AFDC on family structure was considerably smaller in the period from 1940 to 1970 than in 1980 or 1990. It is concluded that increasing welfare benefits cannot account for a significant portion of the increase in illegitimacy, divorce, or separation in the postwar period. Nevertheless, rising benefit levels are significantly associated with changes in the living arrangements of unmarried mothers.

Examines the Integrated Public Use Microdata Series census material project at the University of Minnesota. The project aims to provide a unified interface for researchers and to make the interface and data available on the World Wide Web.

Discusses the methods by which the effects of sampling errors that occur in Public Use Microdata Samples (PUMS) are reduced. PUMS are random samples taken from the decennial censuses since 1850 that can be used to determine American social trends over time. Because the samples are entire household units, they are referred to as cluster samples, which sometimes produce a relatively high incidence of error. This problem is solved by providing design factors for variables such as age, sex, and marital status. Design factors are based on the standard error for each cluster, which in turn is based on standard deviation. These design factors allow for more accurate use of the PUMS, as does the technique of stratification. This involves separating the cluster populations into strata by certain key characteristics, such as household size, home ownership, and race. Stratification has the beneficial effect of further reducing the standard error in cluster samples. An appendix discusses the process by which the PUMS cluster samples for each census were created.

The Public Use Microdata Samples (PUMS) consist of random information samples of households that have been taken from the US decennial censuses since 1850. Because the samples also include information about individuals in a household, the PUMS can provide information about household size, composition, and fertility. The PUMS from the various censuses are incompatible due to differences in census-taking techniques, however, and for this reason the Social History Research Laboratory at the University of Minnesota has developed the Integrated Public Use Microdata Series (IPUMS) to help researchers gauge changes over time. A series of IPUMS variables can be used to identify familial relationships as well as information about individuals in the samples. The original samples often do not have this information. The IPUMS variables can be used to identify parents, stepparents, spouses, and siblings, while others can provide information on household composition and size. The appendix discusses the procedures that were used to construct these IPUMS variables.

Discusses the process by which the Social History Research Laboratory at the University of Minnesota designed the Integrated Public Use Microdata Series (IPUMS). The Public Use Microdata Samples (PUMS) consist of random selections of information that have been taken from the US decennial censuses since 1850. These samples are useful sources of information for the decade from which they were taken, but because each census records data differently, much of the information is incompatible with other censuses. The Social History Research Laboratory has converted these PUMS into an integrated form called IPUMS. By designing common coding systems for such items as household composition, familial relationships, occupations, and geographical location, the IPUMS allows a researcher to chart social changes over time. Versions of the IPUMS data series were released in 1993 and 1994, and the final version will be released in November 1995 by the National Archives. A three-volume set will also be released that provides detailed documentation for the data series and user information.

This essay reexamines the revisionist argument about the history of the family in light of new evidence about long-run changes in American family structure. In particular, I use the new Integrated Public Use Microdata Series, a national database incorporating consistent individual-level data from the U.S. Census over the period 1850 to 1990. I also report findings from the only eighteenth-century American census of sufficient size and quality to permit a consistent analysis of family composition, the 1776 census of Maryland. The evidence suggests that the revisionist interpretation needs revising. In fact, a form of extended family structure was dominant in nineteenth-century America and quite probably in the eighteenth century as well. The American preference for extended family structure disappeared in the twentieth century, and I...offer a brief analysis of some explanations for this change.

I use a new data source, the Integrated Public Use Microdata Series, to trace race differences in family structure between 1880 and 1980. Analysis confirms recent findings that the high incidence among African-Americans of single parenthood and children residing without their parents is not a recent phenomenon. From 1880 through 1960, black children were two to three times more likely to reside without one or both parents than were white children. In recent years, however, the race differential in parental absence has grown. Also, blacks have had a consistently higher percentage of extended households than have whites, but until 1940 this was the result of single parenthood and parentlessness among children: Extended households were more common among whites once the effects of absent parents were controlled

The author discusses problems involved in modeling the effects of demographic factors on historical kinship patterns, with a focus on microsimulation models. "Since microsimulations of kinship ignore the correlations in demographic behavior within kin groups, they ordinarily understate the variance of kinship distributions; for many kin types, they also underestimate the expected number of kin." The author concludes that "those who design demographic models of kinship should be sensitive to the potential for systematic error."