While living alone has risen to historic highs in the United States, the prevalence of living with
roommates, boarders and other nonrelatives has fallen. This historic transformation in living
arrangements reflects a growing prevalence across the 20th century for adults, notably the young and
unmarried, to live apart from family members and outside of boarding houses. Thus, the rise of living
alone is directly linked with the decline of roommates and boarders in the United States. To explore this
relationship, we use Integrated Public Use Microdata Samples (IPUMS) of decennial census data for
1850–2000 (Ruggles et al. 2010), and 2010 Census data. We use multinomial logistic regression to
examine the changing risks of living alone, with relatives, or with nonrelatives only since 1850.