(April 2006) Trends in family households in the United States have received a great deal of media attention because of the long-term and dramatic decline in the share of married couple families (and the corresponding increase in the share of single-parent families) since 1960.

The Census Bureau’s Current Population Survey (CPS) has chronicled the decline in family households between 1960 and 2003: Family households (a household with two or more people related by birth, marriage, or adoption) decreased from 85 percent to 68 percent of all U.S. households, while the proportion of families with children that were headed by a single parent increased from 9 percent to 28 percent.1 These changes resulted from a combination of factors, including population aging, rising average age at first marriage, high divorce rates, improvements in the health and financial status of the older population, and changing residential preferences.

The CPS is the primary source of information about national trends in marriage and family structure. But the American Community Survey (ACS)—a relatively new Census Bureau monthly survey that provides communities reliable and timely demographic, housing, social, and economic data—can provide information about the wide variation in family structure in different parts of the United States.2

For instance, in some large U.S. cities (defined here as cities with 250,000 or more population), the share of female-headed families approaches 50 percent (for example, in Cleveland, Detroit, and Newark, N.J.), while the share is much lower in other cities, including Mesa, Ariz.; San Jose, Calif.; and Wichita, Kan. (see Figure 1). There is a strong racial and ethnic overlay to these geographic differences, with high rates of female headship among African Americans and considerably lower rates among other racial and ethnic groups.3 The detail available from ACS data for these specific groups and metro areas will enhance the analysis of these differences.

Figure 1
Female-Headed Families as a Percent of All Families in Selected Cities, 2003

Note: The vertical bars illustrate the margin of sampling error around each estimate, based on a 90 percent confidence interval.
Source: Mark Mather, Kerri L. Rivers, and Linda A. Jacobsen, “The American Community Survey,” Population Bulletin 60, no. 3 (2005).

The increase in female-headed families is of concern because people living in female-headed families typically have access to fewer economic or human resources than people in married-couple families.4 There are fewer potential earners in female-headed families, which partially explains their lower household income. Delinquent child support payments from absent fathers also erode economic resources available to many female-headed families. In 2000, only about 35 percent of female-headed families with children reported receiving child support or alimony payments.5

Figure 2
Poverty Rate of Female-Headed Households With Children by Race/Ethnicity, 2003

Source: The Population Reference Bureau analysis of the 2003 American Community Survey.

ACS data also indicate that children living in female-headed families are particularly vulnerable to poverty. The high rate of female-headship among African American families is one of the major contributing factors to the high poverty rates for African American children. In 2003, about 37 percent of families maintained by women with children were poor, nearly six times the rate for married couples with children.

Poverty rates were highest for female-headed families headed by American Indian and Alaska Native women, Latino women, and African American women, and lowest for Asian American and white non-Hispanic women (see Figure 2). In the case of American Indians, the high poverty rates reflect the geographic isolation of many American Indian tribes in rural communities and reservations that are cut off from employment opportunities.

The ACS can also provide data for research on gay and lesbian households. As with the decennial census, the ACS does not ask about sexual orientation, but does include information about households headed by same sex, unmarried partners. These results can be used to derive indirect estimates of the number and characteristics of gay and lesbian couples in the United States. The ACS offers a distinct advantage over other national surveys such as the CPS because its large sample size will allow data users to produce reliable estimates for relatively small population groups.

Geographical Variations in Divorce Patterns

As with the CPS and the decennial census, the ACS collects information on five categories of marital status: married, separated, widowed, divorced, and never married. The ACS adds rich detail to the information from these other sources.

One of the biggest demographic stories of the past several decades has been the increase in the proportion of adults who are not currently married—either because they have never been married or because they are separated or divorced. Although the proportion of the population that is divorced has leveled off at 10 percent since the late 1990s, the ACS indicates that there is wide variation in this proportion among different parts of the United States. In 2003, many of the cities with the highest proportion of people who are divorced were located in Florida; while Boston, along with California cities Santa Ana and San Jose, had among the lowest proportion of adults who are divorced (see Figure 3). Variations in age structure, education levels, and religion probably contribute to these geographic differences.6

Figure 3
Percent Divorced Among People Ages 15 and Older in Selected Cities, 2003


Note: The vertical bars illustrate the margin of sampling error around each estimate, based on a 90 percent confidence interval.
Source: Mark Mather, Kerri L. Rivers, and Linda A. Jacobsen, “The American Community Survey,” Population Bulletin 60, no. 3 (2005).

However, the ACS does not collect information about marital history—for example, age at first marriage or previous marriages and divorces—that would allow researchers to study causal relationships between marriage and other life course events.7


The ACS includes a single question about fertility: “Has this person given birth to any children in the last 12 months?” A similar question appeared on the 1990 Census long form, but was dropped from the 2000 Census long form because the data were not required by federal law. This question is included in the ACS to improve population estimates and controls. The question is also used by the Department of Health and Human Services, state and local governments, and other agencies and organizations to assess current fertility patterns and to determine the need for programs and services targeting families with children.

Although the National Center For Health Statistics collects much more detailed information than the ACS about U.S. fertility, the ACS data are useful because they provide information about the characteristics of mothers with young children, including age, marital status, nativity status, education level, poverty status, and labor force status.8 The 2003 ACS estimated that 4 million women ages 15 to 50 had one or more births during the 12 months prior to the survey.

Mark Mather is deputy director of Domestic Programs at the Population Reference Bureau. Kerri L. Rivers is a research associate and program administrator for Domestic Programs at the Population Reference Bureau. Linda A. Jacobsen is director of Domestic Programs at the Population Reference Bureau.


  1. U.S. Census Bureau, “America’s Families and Living Arrangements: 2003,” Current Population Reports P20-553 (September 2004): table HH-1, accessed online at www.census.gov, on July 19, 2005; and U.S. Census Bureau, “America’s Families and Living Arrangements: 2003”: table FM-1, accessed online at www.census.gov, on July 19, 2005.
  2. ACS data on family structure are derived primarily from the ACS question on the relationship of each person to the householder. Based on respondents’ answers to this question, the Census Bureau classifies all households into two types: family households and nonfamily households. A family household consists of the householder (typically the person in whose name the home is owned or rented) and one or more individuals related to him or her by birth, marriage (a stepchild, for example), or adoption. A nonfamily household consists of a householder living alone or with nonrelatives only (for example, a foster child, housemate, or unmarried partner).
  3. The Annie E. Casey Foundation, 2005 KIDS COUNT Data Book (Baltimore: Annie E. Casey Foundation, 2005).
  4. The Annie E. Casey Foundation, 2005 KIDS COUNT Data Book: 38.
  5. The Annie E. Casey Foundation, 2003 KIDS COUNT Data Book (Baltimore: Annie E. Casey Foundation, 2003).
  6. William V. D’Antonio, “Walking the Walk on Family Values,” The Boston Globe, Oct. 31, 2004.
  7. Federal Interagency Forum on Child and Family Statistics, “Summary of Working Group Recommendations” (proceedings from Counting Couples: Improving Marriage, Divorce, Remarriage, and Cohabitation Data in the Federal Statistical System, Bethesda, MD, Dec. 13-14, 2001), accessed online at www.childstats.gov/pubs.asp, on July 19, 2005.
  8. Public Use data files would be required to analyze these results. See Jane Lawler Dye and Tavia Simmons, “Measuring Fertility Using the American Community Survey” (paper presented at the annual meeting of the Population Association of America, Washington, DC, March 29-31, 2001), accessed online at www.census.gov/acs, on July 19, 2005.