This paper reports the results of research and analysis undertaken by Census Bureau Staff. It has undergone a more limited review than official Census Bureau publications. This report is released to inform interested parties of research and to encourage discussion.
The American Community Survey (ACS) is a new approach to collecting reliable, timely information needed by local communities. It will replace the decennial census long-form questionnaire in future censuses and is a critical element in the Census Bureau’s 2010 Decennial Census Program. Like the long form it is designed to succeed, the ACS collects detailed demographic, socioeconomic, and housing information.
Fully implemented in 2005, the ACS is the largest household survey in the United States, with a sample size of about 3 million housing unit addresses throughout the country. Release of annual estimates from the ACS began in 2006 for all geographic areas with a population of 65,000 or more; 3-year average estimates will begin in 2008 for areas and subpopulations as small as 20,000; and 5-year average estimates will start in 2010 for census tracts, block groups, and small subpopulations. All estimates, including the 3-year and 5-year average estimates, will be updated every year.
During the testing program (2000 to 2004), the ACS collected information from approximately 800,000 addresses per year and produced estimates for the United States, states, and essentially all places, counties, and metropolitan areas with at least 250,000 people.
The data contained in this working paper are based on the ACS sample interviewed in 2004. The population represented (the population universe) is limited to the household population and excludes the population living in institutions, college dormitories, and other group quarters. For information on the ACS sample design and other ACS topics, visit http://factfinder.census.gov/home/en/datanotes/exp_acs2004.html
Figure 1. Median Age at First Marriage: 2004 [PDF]
Figure 2. Unmarried Men Per 100 Unmarried Women by Age Group: 2004 [PDF]
Figure 3. Unmarried Men Per 100 Unmarried Women, Age 15 to 44: 2004 [PDF]
 The ACS is designed to provide detailed profiles of the population at the state and local levels. For the 2010 Decennial Census Program, it will gather the type of detailed social, economic, and housing data that have been collected on a sample basis in the decennial censuses since 1940.
This working paper includes data that illustrate the geographical variations in the median age at first marriage for men and women and the ratio of unmarried men to unmarried women at the state level. Previous Census Bureau reports based on the Annual Social and Economic Supplements (ASEC) to the Current Population Survey presented marital status statistics only at the national level.
Table 1 shows the percent distribution of marital status categories by age and sex for the population 15 and older. Among the 224.9 million people 15 and older in the United States in 2004, 56 percent of men and 51 percent of women indicated they were currently married, 9 percent of men and 11 percent of women were divorced, and 2 percent of men and 3 percent of women were separated.
 A higher percentage of men (31 percent) than women (25 percent) had never married. A lower percentage of men (2 percent) than women (10 percent) were widowed, reflecting higher death rates for men, which result in higher percentages of widowed women.
 The higher likelihood that men will remarry also contributes to a lower percentage of widowed men than women.
These statistics present a profile of the population as people age and get married, experience marital disruption and dissolution, and progress to older ages, when mortality rates increase among both sexes.
Women marry earlier than men on average. By age 25 to 29, 58 percent of women had ever been married, 10 percentage points higher than for men of the same age. Higher percentages for women remain evident until ages 55 to 64, when 94 percent of both men and women reported that they had ever been married.
As people enter the middle of the life cycle, separation and divorce become more common. In each of the 10-year age groups from 35 to 74, 10 percent or more of both men and women were divorced or separated. At all ages from 20 to 24 onward, women had higher percentages divorced or separated than men, probably due to men having a higher likelihood of remarrying than women.
 At ages 75 and over, the majority of women were widowed, while the majority of men were married and living with their spouse.
Marital patterns by race and Hispanic origin are shown in Table 2.
 Differences in marital status by race and Hispanic origin may be affected by variations in the age structure for race groups and Hispanics. For example, populations characterized by a more youthful age structure may record a higher proportion of people who have never married, while populations with older age structures may have higher proportions who are widowed.
 men were currently married (40 percent and 44 percent, respectively) than Asian men and non-Hispanic White men (62 percent and 60 percent respectively). The same pattern holds for women, with 29 percent of Black women and 41 percent of AIAN women currently married, compared with 62 percent of Asian women and 55 percent of non-Hispanic White women.
The pattern for the percentage never married across race groups and Hispanics is converse that for the percentage currently married. Black and AIAN men and women had higher percentages never married than Asian and non-Hispanic White men and women.
Black and AIAN men and women had a higher percentage divorced or separated than other men and women. While about 14 percent to 15 percent of Black men and AIAN men were divorced or separated, 11 percent of non-Hispanic White men, and 5 percent of Asian men were divorced or separated. Nineteen percent of Black women and AIAN women were divorced or separated
, while 14 percent of non-Hispanic White women, and 7 percent of Asian women were divorced or separated.
The American Community Survey is the only data set collected by the Census Bureau on a recurring, annual basis that permits the calculation of reliable estimates of the median age at first marriage for men and women for all 50 states and the District of Columbia.
 Data on median age at first marriage from the annual Current Population Survey are published at the national level, and the National Center for Health Statistics no longer publishes the median age at first marriage.
In 2004, the median age at first marriage was 26.8 years for men and 25.4 years for women. For each state and the District of Columbia, Figure 1 shows whether the median age at first marriage for men was above, below, or not statistically different from the median age for the nation. Figure 1 also shows the median age at first marriage for women. The eleven states in which the median age at first marriage for men was higher than the U.S. median include a cluster in the Northeast, as well as Hawaii, California, Montana, and Florida. A grouping of states in the middle of the country extending from Texas to Minnesota had a median age at first marriage for men that was lower than the national median. The pattern for women is similar. A group of Northeastern states had a median age at first marriage higher than that of the United States, and a group of states in the middle of the country had a median below the national median.
The states with some of the highest median ages at first marriage included Massachusetts, Rhode Island, New York, and New Hampshire for men and Massachusetts, Hawaii, and New Jersey for women, indicating geographical areas where first marriage is being delayed or where a higher percentage of never married people have migrated (Table 3). The District of Columbia recorded median ages at first marriage for both men and women that are 2 years higher than the respective national medians. The states with some of the lowest median ages at first marriage included Utah, Idaho, Oklahoma, and Arkansas, each of which had a median age at first marriage of 25 years or less for both men and women.
While no single indicator can capture the marital status profile of an area, a frequently used index, the ratio of unmarried men to unmarried women, summarizes the numbers of men and women who could marry.
The ratio is affected by the age distribution in a given area. Figure 2 illustrates the way the ratio of unmarried men to unmarried women shifts over the life course. The ratio is higher at younger ages, with the highest ratio in 2004 at ages 25 to 29 (116 unmarried men per 100 unmarried women). The ratio is higher at younger ages because the age at first marriage is slightly higher for men than it is for women. At ages 45 to 54, the ratio dips below 100 to 83, meaning that unmarried women outnumber unmarried men. The ratio continues to decline into the older age groups. For those aged 85 and older, in 2004, there were 25 unmarried men per 100 unmarried women. Higher death rates for men, as well as a higher likelihood of remarriage for men, contribute to the higher proportion of unmarried women in the older age groups.
Figure 3 and Table 3 show the ratio of unmarried men to unmarried women for those aged 15 to 44, since most marriages are among people in this age group. This measure indicates the number of widowed, divorced, and never married men per 100 women of the same age range who lived in each state and the District of Columbia.
A ratio of 100 means the number of unmarried men and women aged 15 to 44 years are equal in that area. A ratio above 100 means that unmarried men outnumber unmarried women, and a ratio below 100 means unmarried women outnumber unmarried men.
Figure 3 maps the ratio of unmarried men to unmarried women for ages 15 to 44 by state. The ratio for the United States is 108 unmarried men per 100 unmarried women. States with a ratio statistically higher than the ratio for the nation are mainly in the West, but also include Wisconsin, Iowa, and Virginia. States with a ratio lower than the ratio for the nation are in two areas. One group is in the South, from Louisiana to South Carolina. The second is composed of Midwestern and Northeastern states, including Michigan, Ohio, New York, and Rhode Island.
The District of Columbia also had a ratio below the ratio for the nation, indicating 93 unmarried men per 100 unmarried women.
Table 4 shows the ratio of unmarried men to unmarried women for cities with a population of at least 1 million. Of these cities, Dallas (125), Phoenix (120), Los Angeles (115), San Diego (115), and Houston all had ratios higher than 100,
while Philadelphia had one of the lowest ratios, 92.
The data in this working paper are from the 2004 ACS. The population represented (the population universe) in the ACS is limited to the population living in households and excludes the population living in institutions, college dormitories, and other group quarters. According to Census 2000, 7.8 million people, or 2.8 percent of the total population, lived in group quarters. Of this number, 4.1 million were institutionalized–primarily in correctional institutions and nursing homes, 2.1 million were in college dormitories, and 1.7 million were in all other types of group quarters.
Statistics from surveys are subject to sampling and nonsampling error. Data from the ACS are based on a sample and are estimates of the actual figures that would have been obtained by interviewing the entire household population using the same methodology. All comparisons presented in this working paper have taken sampling error into account and are significant at the 90-percent confidence level unless noted otherwise.
This means the 90-percent confidence interval for the difference between the estimates being compared does not include zero.
Nonsampling errors in surveys may be attributed to a variety of sources, such as how the survey is designed, how respondents interpret questions, how able and willing they are to provide correct answers, and how accurately the answers are keyed, coded, edited, and classified. Nonsampling errors in the ACS may affect the data in two ways. Errors that are introduced randomly increase the variability of the estimates. Systematic errors consistent in one direction introduce bias into the results. The Census Bureau protects against systematic errors by conducting extensive research and evaluation programs on sampling techniques, questionnaire design, and data collection and processing procedures.
The final ACS population estimates are adjusted in the weighting procedure for coverage error by controlling specific survey estimates to independent population controls by sex, age, race, and Hispanic origin. The final ACS estimates of housing units are controlled to independent estimates of total housing. This weighting partially corrects for bias due to over or undercoverage, but biases may still be present, for example, when people missed differ from those interviewed in ways other than sex, age, race, and Hispanic origin. How this weighting procedure affects other variables in the survey is not precisely known. All of these considerations affect comparisons across different surveys or data sources.
For further information on the ACS sample, weighting procedures, sampling error, nonsampling error, and quality measures from the ACS, see http://factfinder.census.gov/home/en/datanotes/exp_acs2004.html.
Detailed tables from the 2004 ACS are available on the Internet at the Census Bureau’s web site https://www.census.gov. Once on the site, click on the “American Community Survey,” or contact the Customer Services Center at 301-763-INFO (4636).
For additional information about marital status, visit the Census Bureau's Internet site at https://www.census.gov/population/www/socdemo/ms-la.html or www.census.gov/population/www/socdemo/marr-div.html. You may also contact the author of this working paper, Rose Kreider by e-mail.
 For more information on the American Community Survey, including its sampling design and the reliability of the estimates, go to https://www.census.gov/acs/www/index.html.
 For a comparison of population estimates derived from the marital status item on the ACS and the ASEC for 2004, see Martin O’Connell and Gretchen E. Gooding, “A Comparison of ACS and ASEC Data on Households and Families: 2004.” https://www.census.gov/acs/www/.
 The estimates in this working paper (which may be shown in text, figures, and tables) are based on responses from a sample of the population and may differ from actual values because of sampling variability or other factors. As a result, apparent differences between the estimates for two or more groups may not be statistically significant. All comparative statements have undergone statistical testing and are significant at the 90-percent confidence level unless otherwise noted. The data in this working paper are from the 2004 American Community Survey. The population represented (population universe) is the household population living in the United States.
 A mathematical model used to project the likelihood of divorce and remarriage, based on marital history data from men and women, suggests a higher proportion of men will ever marry after a divorce than will women. See Kreider, Rose M. and Jason M. Fields, “Number, Timing and Duration of Marriages and Divorces: 1996,” Current Population Reports, P70-80, February 2002, [Table 1]1.
 In this working paper, the term non-Hispanic White is used to refer to people who reported being White and no other race, and who are not Hispanic. The term Black is used to refer to people who reported being Black or African American and no other race, and the term Asian is used to refer to people who reported only Asian as their race. The use of single-race populations in this working paper does not imply that it is the preferred method of presenting or analyzing data. The Census Bureau uses a variety of approaches.
Because Hispanics may be any race, data in this working paper for Hispanics overlap slightly with data for the White, Black, American Indian and Alaska Native, Asian, Native Hawaiian and Other Pacific Islander, Some Other Race, and the Two or More Races population. Based on the population 15 and older in ACS 2004, 6 percent of single-race Whites, 2 percent of single-race Blacks, 12 percent of the single-race American Indian and Alaska Native population, 1 percent of single-race Asians, 7 percent of the single-race Native Hawaiian and Other Pacific Islander population, 96 percent of the single-race Some Other Race population, and 25 percent of the Two or More Races population were also Hispanic.
 Detailed marital status tabulations by age, sex, race, and Hispanic origin are available at the national level from the 2004 ASEC. See detailed Table A1 at https://www.census.gov/population/www/socdemo/hh-fam/cps2004.html.
 Although age at first marriage is not asked on the ACS, an indirect method estimates the median age at first marriage based on the proportion of the population who were ever married for 5-year age groups ranging from 15 to 54 years.