FAMILY AND HOUSEHOLD STRUCTURE

The family system of the United States is often characterized as consisting of nuclear-family households—that is, households consisting of no more than the parent(s) and dependent children, if any (Lee 1999). This is certainly true of the great majority of family households. In fact, there has never been a point in American history in which extended-family households predominated statistically (Ruggles 1994a; Seward 1978). In 1997 only about 4.1 percent of all families in the United States were ”related subfamilies”—a married couple or single parent with children living with a related householder (U.S. Bureau of the Census 1998, Table 69). However, an analysis of census data from 1970 through 1990 by Glick and colleagues (1997) showed that the percentage of all households containing nonnuclear kin increased from 9.9 percent in 1980 to 12.2 percent in 1990, reversing a nearly century-long pattern of decline. In 1910 about 20 percent of the households of white families and 24 percent of those of black families contained nonnuclear kin (Ruggles 1994b). Apparently we have seen a long-term decline in the prevalence of extended-family households, very slightly counterbalanced by an increase in the 1980s; what happened in the 1990s is not yet known.

Not all of the of the households that do not contain extended families consist of the stereotypical nuclear family of two parents and their dependent children, however. There is great diversity among American families and households, and this diversity is increasing. Table 1 presents data on the composition of American households and families from 1960 to 1998. Even over this relatively brief period, substantial changes are apparent. The average size of both households and families decreased dramatically from 1960 to 1990, although they have both been stable in the 1990s. Many fewer households contain families and married couples in the late 1990s than in 1960, while the proportion of nonfamily households has more than doubled and the proportion of single-person households has nearly doubled. Female householders have increased substantially as a proportion of both all households and all families.

There are many factors responsible for these changes. To understand them, changes in marriage rates and age at marriage, divorce and remarriage rates, rates of nonmarital cohabitation, the departure of children from their parents’ homes, and the predilection of unmarried persons to live alone will be briefly examined. Each of these factors has affected family and household structure.

Marriage rates have declined considerably since 1960 (see Table 2). This is not readily apparent from the ”crude” marriage rate (the number of marriages per 1,000 population) because this rate does not take the marital status or age distributions of the population into account. The crude marriage rate was artificially low in 1960 because, as a result of the postwar baby boom, a large proportion of the population consisted of children too young to marry. The rates per 1,000 unmarried women (for both ages 15 and over and ages 15 to 44) show the frequency of occurrence of marriage for persons exposed to the risk of marriage, and here there is clear evidence of decline. Some of this, however, is attributable to increases in the median age at first marriage, which declined throughout the twentieth century until about 1960, but has been increasing rapidly since 1970. As age at marriage increases, more and more people temporarily remain unmarried each year, thus driving the marriage rate down. The best evidence (Oppenheimer et al. 1997) indicates that a major cause of delayed marriage is the deteriorating economic circumstances of young men since the 1970s. Perhaps the improving economy of the later 1990s will eventually produce some change in this trend.

Table 1:Changes in U.S. Households and Families Since 1960

The rising divorce rate has also contributed greatly to the declining proportion of married-couple households and the increases in female householders and single-person households. The crude divorce rate rose from 2.2 per 1,000 in 1960 to 5.2 in 1980 (reaching peaks of 5.3 in both 1979 and 1981) but has declined modestly since then to 4.3 in 1996. The rate of divorce per 1,000 married women 15 and older followed a similar pattern, reaching a high of 22.6 in 1980 and declining to 19.5 in 1996. Some of this decline is illusory, because the large baby boom cohorts are aging out of the most divorce-prone years (Martin and Bumpass 1989). However, although the divorce rate remains high, it has not been increasing since 1980.

Sweeney (1997) notes that, for recent cohorts, about half of all marriages have involved at least one previously married partner. However, rates of remarriage after divorce have been declining steadily. Annual remarriage rates were 204.5 per 1,000 divorced men and 123.3 per 1,000 divorced women in 1970; by 1990 they had decreased to 105.9 for men and 76.2 for women (U.S. Bureau of the Census 1998).

Decreasing marriage and remarriage rates and increasing divorce rates have combined to produce increases in single-person and single-parent households. This trend is mitigated, however, by the increasing prevalence of nonmarital heterosexual cohabitation. Most such unions (those without children) appear as nonfamily households in Table 1. Evidence from the National Survey of Families and Households (Bumpass 1994; Waite 1995) shows that, in the early 1990s, nearly one-quarter of all unmarried adults aged 25 to 29 were cohabiting. This percentage declines with age, but still exceeded 20 percent for those in their late thirties. The National Survey of Family Growth found that, in 1995, more than 41 percent of all women aged 15 to 44 had cohabited or were currently cohabiting (National Center for Health Statistics 1997). Of course many of the women who had not cohabited at the time of the survey will do so in the future. The best estimates suggest that more than half of all couples who marry now cohabit prior to marriage; further, about 60 percent of all cohabiting unions eventuate in marriage (Bumpass 1994; Bumpass et al. 1991).

To a considerable extent the increase in cohabitation has offset the decline in marriage. This is particularly the case among blacks, for whom the decrease in marriage rates over the past several decades has been much more precipitous than it has been for whites (Raley 1996; Waite 1995). Although cohabiting unions are less stable than marriages, ignoring cohabitation results in substantial underestimates of the prevalence of heterosexual unions in the United States.

In spite of the increase in cohabitation, changes in marriage and divorce behavior have had substantial effects on household and family structure in the United States over the past four decades. Fewer people are marrying, those who marry are doing so at later ages, more married people are divorcing, and fewer divorced people are remarrying. This means that Americans are living in smaller households than they did in 1960, but there are more of them. The rate of growth in the number of households has substantially exceeded the rate of growth in the number of families. Referring back to Table 1, from 1960 to 1998 the number of households increased by more than 94 percent, while the increase in the number of families was only about 57 percent. Over the same time period, the total population of the United States increased by just under 50 percent (U.S. Bureau of the Census 1998, Table 2). Our population, therefore, is distributed in a larger number of smaller households than was the case in 1960.

Table 2:Changes in U.S. Marriage and Divorce Since 1960

note: 1. Monthly Vital Statistics Report, Vol. 43, No. 12 (S), July 14, 1995 for 1960-90; and U.S. Bureau of the Census (1988): Table 156 for 1996.

2. U.S. Bureau of the Census, Current Population Reports, P20-514 (March 1998) and earlier reports for 1960-90; and U.S. Bureau of the Census web site for 1996.

3. Monthly Vital Statistics Report, Vol. 43, No. 9 (S), March 22, 1995 for 1960-90; and U.S. Bureau ofthe Census (1998): Table 156 for 1996.

One cause of the decline in household size shown in Table 1 is decreased fertility. The fertility rate (number of births per 1,000 women aged 15 to 44) was 118.0 in 1960; by 1997 it had decreased to 65.0, although most of the decrease occurred prior to 1980 (National Center for Health Statistics 1999). The trend toward smaller households and families is reflective to some extent of decreases in the number of children per family.

A larger cause of the decrease in household size, however, is the proliferation of single-person households, as shown in Table 1. Single-person households consist of three types of persons: the never-married, who are primarily young adults; the divorced and separated without coresident children, who are primarily young and middle-aged; and the widowed who live alone, who are primarily elderly. Each of these types has increased, but for somewhat different reasons. Each must therefore be examined separately.

As shown in Table 2, average ages at marriage have risen markedly since 1960, and the percentage of young adults who have never married has increased proportionately (Waite 1995). This has been accompanied by a long-term decline (since prior to World War II) in the average age of leaving the parental home (Goldscheider 1997). Prior to 1970 most of this decline was driven by decreasing ages at marriage, but since then it has reflected an increasing gap between leaving the family of orientation and beginning the family of procreation. More young adults are living independently of both parents and spouses. Some of them are cohabiting, of course, but increasing numbers are residing in either single-person or other nonfamily households (Goldscheider 1997; White 1994).

Since about 1970 there has been some increase in the proportion of young adults who live with their parents. This marks the reversal of a long-term decline in age at leaving home (White 1994). This is, in part, a by-product of increasing age at marriage. However, decreases in exits from parental homes to marriage have been largely offset by increases in exits to independent living, so this recent increase in young adults living with parents is actually very small (Goldscheider 1997). On the other hand, there is increasing evidence that the process of launching children has become much more complex than in previous years. Goldscheider (1997) also shows that the proportion of young adults who return to their parents’ homes after an initial exit has more than doubled from the 1930s to the 1990s; increases have been particularly striking since the early 1960s. This is a response, in part, to the rising divorce rate, but also an indication that it has gotten increasingly difficult for young adults, particularly young men, to make a living (Oppenheimer et al. 1997). Nonetheless, the proportion of young adults living independently of both parents and spouses continues to increase, contributing to the prevalence of nonfamily households.

The increase in divorce and decrease in remarriage have contributed to the rise in single-person households, as formerly married persons establish their own residences and, increasingly, maintain them for longer periods of time. They have also contributed to the rise in family households that do not contain married couples. As shown in Table 1, families headed by females (without husband present) increased from 10 percent of all families in 1960 to nearly 18 percent in 1998. Families headed by males (without wife present) also increased, from 2.8 percent of all families in 1960 to 5.5 percent in 1998. Among families with children under 18 in 1998, 20 percent were headed by women without spouses and 5 percent by men without spouses (U.S. Bureau of the Census 1998, Table 70).

As a consequence of these changes plus the rise in nonmarital childbearing, the proportion of children under 18 living with both parents decreased from 88 percent in 1960 to 68 percent in 1997 (U.S. Bureau of the Census 1998, Table 84). In addition, there is a large race difference in the living arrangements of children. Only 35 percent of black children lived with both parents in 1997, compared to 75 percent of white children. More than half (52 percent) of all black children lived with their mothers only, as did 18 percent of white children. Further, 8 percent of black children and 3 percent of white children lived with neither parent. Some of these children are living with, and being cared for by, their grandparents (Pebley and Rudkin 1999). This raises the issue of the living arrangements of older persons.

A somewhat longer perspective is necessary to observe changes in the living arrangements of older persons. Ruggles (1994a) has shown that, in 1880, nearly 65 percent of all elderly whites and more than 57 percent of all elderly nonwhites lived with a child. Since about 18 percent of all older persons had no living children, Ruggles estimates that about 78 percent of whites and 70 percent of nonwhites who had children lived with a child. By 1980 the percentages living with children had decreased to 16 for whites and 29 for nonwhites. There is little evidence of major changes in the proportion living with children since 1980. Further, Ruggles (1996) found that only 6 percent of all elderly women and 3 percent of elderly men lived alone in 1880. By 1997 the percentages living alone had increased to 41 for women and 17 for men (U.S. Bureau of the Census 1998, Table 50). The growth of single-person households among older people has been particularly rapid since about 1940.

Two sets of factors appear to be primarily responsible for the ”migration” of older people from typically sharing households with their children in the late nineteenth century to living alone (or with their spouses only) in the late twentieth century. First, the family life cycle was quite different in 1900 than today. People married a bit later (and markedly later than in the 1960s and early 1970s), had more children, and had children later in life. Consequently, a significant proportion of people in their sixties had unmarried children who simply had not yet left the parental home. Ruggles (1994a) shows that, in 1880, about 32 percent of all unmarried elders and 57 percent of the married resided with a never-married child. Of course many of these children may have remained home precisely in order to care for their aging parents. Unmarried elders were more likely to live with married children.

Second, economic factors played a major role. Social Security did not exist until 1940. In 1900, 85 percent of all men between the ages of 65 and 69 were in the labor force, as were 49 percent of all men 85 and over (Smith 1979). However, this option was much less available to women; the comparable proportions in the labor force were 12 and 6 percent. Many older persons, particularly women, had no means of support other than their children. Rates of coresidence of aging parents with their adult children have decreased as the prosperity of the elderly has increased; more can now afford to live independently.

However, Ruggles (1994a) found that wealthier older people were more likely to share a household with children than were poorer elders in the nineteenth century, and the majority of multigenerational families lived in households headed by the elderly parent(s). These facts suggest that adult children benefited economically from coresidence and that the possibility of inheriting a farm or business from aging parents may have motivated many adults to coreside with parents. Today coresidence is more common among poorer than wealthier people (Ruggles 1994a, 1996).

As of March 1998, 41 percent of all women aged 65 and older lived alone, as did 17 percent of all older men. These percentages increase to 53 percent and 22 percent for women and men, respectively, for those age 75 and over (U.S. Bureau of the Census Web site). The reason for this large gender difference, of course, is the difference in marital status between men and women. Among men 75 and over, nearly two-thirds are married and less than one-quarter are widowed; among women these figures are almost exactly reversed. According to 1980 census data, the proportion of all elderly persons living alone increases from 22 percent among those 65 to 69 to more than 41 percent in the 85-89 age category, then drops to 33 percent for those 90 and over (Coward et al. 1989), after which the modal category becomes living with children. Older persons who have lost their spouses through death are clearly exhibiting a tendency to live alone as long as possible, which for many of them extends into the latest years of life.

Older persons now constitute nearly 13 percent of the total population of the United States, compared to about 4 percent in 1900. With so many of them maintaining their own residences, either with their spouses or alone following widowhood, their contribution to the proliferation of small and single-person households is substantial.

If so many older persons lived with their children in the late nineteenth century, why were there so few extended-family households? Ruggles (1994a) shows that just under 20 percent of the households of whites contained extended families in both 1880 and 1900; this compares to less than 7 percent in 1980, but it was still very much a minority statistical pattern. There were three primary reasons. First, because of more limited life expectancies and relatively high fertility rates, there were proportionally few older people in the population, so where they lived made less difference to the nation’s household structure. Second, as noted above, many older persons lived with an unmarried child; unless other relatives are present, this arrangement constitutes a nuclear-family household regardless of the age of the parent. Third, while these cohorts of older persons typically had many children (an average of 5.4 per woman in 1880), these children did not live together as adults, so older persons could live with only one; their remaining children lived in nuclear families. Ruggles (1994a) estimates that more than 70 percent of all elders who could have lived with a child actually did so in 1880; the comparable percentage in 1980 was 16. In comparison to the last century, older persons today are much less likely to live with children and much more likely to live alone, contributing to the proliferation of small and single-person households.

To this point, factors that have contributed to long-term decreases in household and family size, and consequent increases in the numbers of households and families, have been elucidated. There is evidence of changes in these directions in all age segments of the population. These trends do not mean, however, that more complex family households are not part of the contemporary American experience.

As noted at the beginning of this entry, the United States has never been characterized by a statistical predominance of extended-family households, although it appears that the preference was for intergenerational coresidence in the form of stem families (families containing an older parent or parents and one of their married children) until the early years of the twentieth century. But extended family households do occur today. At any single point in time, they constitute less than 10 percent of all households (Glick et al. 1997; Ruggles 1994a). However, a dynamic perspective presents a somewhat different picture.

Beck and Beck (1989) analyzed the household compositions of a large sample of middle-aged women who were followed from 1969 to 1984. The presence of nonnuclear kin in their households was noted for specific years and was also calculated for the entire fifteen-year period. In 1984, when these women were between the ages of 47 and 61, only 8 percent of white married women and 20 percent of white unmarried women lived in households containing their parents, grandchildren, or other nonnuclear kin. The proportions were higher for comparable black women: 27 percent of the married and 34 percent of the unmarried. However, over the fifteen years covered by the survey, about one-third of all white women and fully two-thirds of the black women lived in a household containing extended kin at some point.

These and other data (Ruggles 1994a, 1994b) show that today blacks are more likely than whites to live in extended-family households. This was not the case until about 1940. What has happened is that the decrease in intergenerational coresidence since the late nineteenth century has been much steeper for whites than for blacks. This is probably connected to much lower rates of marriage among blacks; living in multigenerational households is much more common for unmarried than for married persons. It may also reflect the shift in the distribution of extended families from the wealthier to the poorer segments of the economic structure. Rather than serving as a means of ensuring inheritance and keeping wealth in the family, extended family living today is more likely to be motivated by a need to share and conserve resources.

The family and household structure of the United States has changed dramatically over the past century, in spite of the fact that our family system has remained nuclear in at least the statistical sense. More and more Americans are living in single-person households before, between, and after marriages. More are living in single-parent households. Collectively Americans are spending smaller proportions of their lives in families of any description than they did in the past (Watkins et al. 1987). However, they are more likely than ever before to live in nonmarital heterosexual unions, and many of them live in households that contain nonnuclear kin at some point in their lives. In fact, there is evidence (Glick et al. 1997) that the proportion of extended-family households increased between 1980 and 1990.

The growth of small and single-person households is in many ways indicative of the fact that more Americans can now afford to remain unmarried, leave unhappy marriages, and maintain their own residences in later life. The proliferation of households represents the proliferation of choices. The consequences of these choices remain to be seen.