This paper assesses the impact of changing demographic behavior—fertility levels, mortality,
generation length, and nuptiality—on the frequency of multigenerational families among whites in the
United States between 1880 and 1980.1
My focus is methodological. Demographers and family
historians have proposed and applied a variety of different techniques for assessing the relationship
between demographic conditions and household composition. I argue that most of these previous
efforts—including my own—have yielded inconclusive results. The methods of historical family
demography have been too ambitious and too complex, have involved too many assumptions, and
have been based on insufficient data. I propose some simpler approaches that allow us not only to
control for the overall effects of demographic change, but also to estimate the demographic
components of change in kinship and family structure. I have applied these simplified methods to data
from the new Integrated Public-Use Microdata Series (IPUMS), a national census microdatabase
spanning the period 1850 through 1990 currently in preparation at the University of Minnesota
(Ruggles et al. 1993). The results clearly demonstrate that the demographic transition had profound
implications for the frequency of multigenerational families, and that multigenerational family structure
was preferred among whites in the United States until the mid-twentieth century.