What is a family?
At various times in my career I have returned to a perennial question: What is a family? As a researcher, I claim to study families. But if you were to ask me for a definition of family, I would probably get tongue-tied and mumble something like, "Well, there are different ways of thinking about it…." A perusal of family textbooks reveals that many authors also have struggled with this question. Presumably, the teachers who use these textbooks also have struggled. This is a curious problem, because as researchers, we try to measure the characteristics of families. Yet, the unit of analysis is not always clear.
A common approach is to rely on an objective definition of family. In doing so, researchers impose their definitions on the people being studied. For example, the U.S. Census Bureau defines a family in the following manner: "A family is a group of two people or more (one of whom is the householder) related by birth, marriage, or adoption and residing together." This definition is useful for obtaining a current snapshot of various family configurations (such as the percentage of children living with single parents in a given year), and to track changes over time in family living arrangements. The main limitation of this definition is that most people do not define families in this fashion.
A key feature of an objective definition is that it specifies (and restricts) who is (and who is not) in a person's family. An alternative, subjective approach is to let people make their own decisions about who is in their families. Drawing on people's implicit definitions leads to some interesting implications. Consider the common case of a divorced mother living with her child. The mother is unlikely to consider her ex-husband to be a family member—at least since the divorce occurred. But if the child continues to spend a significant amount of time with her father, she probably will continue to view him as a family member. In cases like these, the mother and child do not share the same family, although their families overlap. A researcher who adopts the U.S. Census Bureau definition would say that the daughter is in a single-parent family, whereas the daughter would claim to be in a two-parent family.
This example illustrates three curious features of subjective definitions of families. First, residence and family are different. In fact, when asked to list family members, most people name one or more individuals who do not reside with them. Moreover, some people exclude individuals who live with them from their family circles. For example, research shows that many adolescents do not consider their residential stepfathers to be family members. Relying on household-centered definitions has led researchers to over-identify the home as the site of a person's most meaningful family relationships and to give insufficient attention to relationships and interactions with family members outside the home.
Second, family membership is asymmetrical. That is, person A might claim person B as a member of his family, but person B might not claim person A as a member of her family. Examples would include cases in which a stepfather views a stepdaughter as a family member, but the stepdaughter does not reciprocate this claim. When families are defined objectively, in contrast, membership is always reciprocal, and family membership can be determined by obtaining data from a single member.
Finally, subjective family members are not always related by blood, marriage, or adoption. Many cohabiting partners, for example, consider themselves to be family. Similarly, many people consider their best friends to be family members. Family scholars have noted the importance of fictive kin in many people's lives—a phenomenon that is especially important for some racial and ethnic groups, such as African Americans. From a subjective perspective, of course, these individuals are not "fictive." Other people may choose to reject close biological relatives (such as a parent, a child, or a sibling) from their family definition. Children who rarely (or never) see their biological fathers, for example, often say that they have no father. When it comes to perceptions of family, biology is not destiny.
Accepting people's subjective definitions of families moves us closer to the lived reality of people's lives. Families are more difficult to study, however, if we rely on people's subjective definitions. Nevertheless, adopting a subjective approach does not create insurmountable problems for researchers. Current methods of social network analysis, in particular, can be adapted to study people's subjective families. Using this approach, researchers can identify family networks by asking focal respondents to list all the members of their families. A frequent outcome of this exercise would be that members of the same household produce different family networks. But that would not stop researchers from inquiring about the perceived characteristics of these networks. For example, respondents can report on the extent to which family members argue or disagree with one another, despite the fact that each person in a household might be referring to a different (but overlapping) group of people. Although this approach might seem unwieldy, it would more accurately reflect people's understandings and experiences.
Other characteristics that may be applicable to family networks include the extent to which members communicate with one another, are emotionally close, exchange assistance, and are available as potential help givers in times of need. Obtaining this information would allow researchers to characterize family networks as fragmented versus cohesive. An example of a testable hypothesis would be that children with divorced parents report more fragmented family networks than do children with continuously married parents. This type of data also would allow researchers to assess the extent to which family networks are sources of social capital (resources exchanged through relationships). In addition to assessing links within networks, researchers could assess the extent to which network members have relations with or exchange resources with members of different networks.
A potentially interesting extension would be to assess the links between overlapping networks within the same household. For example, how much overlap exists between the networks of husbands and wives or parents and children? Researchers could relate these data to various individual-level outcomes, such as spouses' marital satisfaction or children's educational achievement. The assumption underlying the analysis would be that households with a high level of family network overlap confer certain benefits on their members. Or one might argue for the alternative hypothesis when members of overlapping networks transmit tension and negative emotions.
To study families as families and not as aggregates of individuals, researchers must define the unit of analysis. Most current research is based on objective definitions of families—definitions that are not shared by the majority of individuals that we study. Viewing families as overlapping networks that extend across multiple households, with each network having at its nucleus a reference person, might yield new insights, especially in an era when families are becoming more complex and difficult to classify. Because only a small number of family scholars have adopted this approach, the potential of taking people's families-of-choice seriously is currently unrealized.