Among the questions sociology professor Frances Goldscheider asks her undergraduates is: would you let your newly widowed uncle move in with you and your spouse? African American students, she has found, are more likely to answer yes, figuring, “How much room can an old man take?” White students more often object to what the uncle’s presence would do to their privacy.

The exercise demonstrates a truism in the field of sociology: that African American adults are more likely than whites to live with extended family. Single black adults more often live with their parents, for example, and elderly blacks more often live with their adult children. Sociologists, Goldscheider says, have long known that the difference cannot be explained by the fact that African Americans, on average, have lower incomes than whites.

So what causes the disparity? Goldscheider and a colleague examined U.S. census data from 1900 to 1990 with the hypothesis that the trend would be somehow tied to African American migration from southern farms to northern cities—that people from farm backgrounds are more likely to live with family. Because African Americans made the transition from farm to city a generation or so later than whites, the hypothesis seemed to make sense.

“That’s not what we found at all,” Goldscheider says; in fact, she concluded, migration patterns seem unrelated to family living arrangements. Until fairly recently, she discovered, the trend was the exact opposite of the norm today: in 1940, unmarried black adults were significantly less likely than their white counterparts to live with extended family. Barely 50 percent of unmarried African Americans lived with family, compared to nearly 70 percent of unmarried whites. “Nobody knew about this,” she says. “It’s sort of like finding a new planet.” Goldscheider’s findings were published in the August issue of Demography.

Extended-family living declined steadily in the United States between 1940 and 1990. But it declined primarily among whites, the census data revealed. Sometime in the 1960s a crossover took place, with African Americans becoming more likely than whites to live in extended-family households. By 1990, 39 percent of unmarried African Americans, but less than 30 percent of unmarried whites, lived with extended family.

Goldscheider says more research may reveal the reason for the crossover, but census data do shed light on what is not to blame. The difference is not linked to higher rates of single parenthood among African Americans, she says. And the sharp decline among whites goes far beyond what would be expected as a result of increases in income or education. Attending college reduced the odds of living with extended family by 44 percent for whites, for example, but only 14 percent for blacks.