When Is a Fami ly a Family? Evidence from Survey Data and Impl icat ions for Family Pol icy

This article examines the correspondence between common assumptions about the American family and actual patterns. The assessment is based on national data on individuals, households, and families. Findings indicate that the coresident nuclear model should be considered a model rather than the model of family. Past as well as current marital ties need to be considered in defining "family," and divorce rather than death should be the expected cause of losing the main breadwinner in the family, except among elderly women. Parent-child ties to either young or adult children often span separate households. Coresidents can include individuals other than nuclear family members, and change rather than stability is the modal pattern in living arrangements. Rather than shaping concepts of the family from a single mold, policy makers and researchers are better advised to recognize the diversity and fluidity in family and household structures.