PRB Library

World Population Data Sheet

Population Bulletin

Population Today

The American People Series

Population Handbook

Guidelines for Effective Data Presentations

World Population: More Than Just Numbers

Graphics Bank

Glossary

Data Sheets

Reports on America

First Glimpses From the 2000 U.S. Census

by Mary M. Kent, Kelvin M. Pollard, John Haaga, and Mark Mather

Population Bulletin, Vol. 56, No. 2, June 2001

Table of Contents
Introduction
The U.S. Census Tradition
Population Change
Increase in Hispanics
Racial and Ethnic Diversity
Metropolitan Growth
Political Implications
More from the 2000 Census
The Next Census
References
Appendix Table
Suggested Resources

Introduction

The U.S. population stood at 281,421,906 on April 1, 2000, according to the decennial census. The new total represented an addition of 32.7 million Americans since the 1990 Census � the largest numerical increase ever between censuses. The 2000 Census recorded a population gain in every state during the 1990s � the only decade in the 20th century with such widespread growth.

The 2000 Census was much more than an enumeration of the population on a specific spring day. It repeated a national event carried out every 10 years since 1790, and ushered in a third century of census-taking in the United States. The census is required by the U.S. Constitution to allocate congressional representation, but its significance extends far beyond this. Nearly $200 billion in federal funds ($185 billion in 1998) are distributed to the states each year based to some extent on census counts. Geographic boundaries of districts for members of Congress, state legislators, and other political leaders are redrawn using census data. Census results also provide information to thousands of people in the public and private sector who make decisions about health, education, transportation, protection of natural areas, pollution abatement, community services, housing, consumer marketing, economic planning, and many other issues. Census results measure progress and give direction for future actions. Experts in demography, economics, and many other fields will spend years examining the 2000 Census data for clues about how the U.S. population is doing and how it has changed.

The latest census was full of surprises even for demographers who carefully track records of births and deaths, and use sophisticated techniques to estimate migration from abroad and within the country. Among the surprises were:

• The census counted nearly 7 million more people than the U.S. Census Bureau had estimated for April 2000 � and it still may have missed as many as 3 million.
• The Census Bureau recommended against adjusting the census for an undercount, contrary to expectations.
• The U.S. Hispanic population apparently grew much faster than anticipated � and edged past African Americans to become the nation's largest racial or ethnic minority.
• New York, Chicago, and several other major cities gained residents, in some cases reversing a decades-old trend of population decline.
• The shift in congressional apportionment was greater than expected: 12 of the 435 seats in the House of Representatives changed from one state to another.
• The census cost less than anticipated.
• U.S. residents were more cooperative than expected about returning census questionnaires.

The 2000 Census broke new ground by allowing Americans to identify with more than one race. It also asked a new question about the role of grandparents as caregivers for dependent children. It was the first census effort to use paid advertising to boost response rates. And the results of this census will be the most accessible ever to Americans because of new computer technologies and the Internet.

But the 2000 Census also raises some questions that may defy solution any time soon. Why was the count so much larger than expected? Was this number larger or smaller than the actual number of U.S. residents on April 1, 2000? Who was missed or counted more than once? How will the new racial categories play out in the coming decade? Is this the beginning of the end of the statistical category "race" in this country?

Several Western and Southern states gained congressional seats at the expense of states in the Northeast and Midwest. Some metro areas saw their populations shift toward outer suburbs � and some suburbs turned from majority white to predominantly minority. How will these demographic changes affect the U.S. political scene? Will Census 2000 escape the legal controversies of the 1980 and 1990 Censuses?

This Population Bulletin looks at some of the major findings of the 2000 Census as of April 2001, and considers the importance of these trends not only to demographers, journalists, business people, and politicians, but to all Americans.

The U.S. Census Tradition

• Figure 1. U.S. Resident Population, Decennial Censuses, 1790�2000
• How Accurate/How Complete?
• Figure 2. Net Census Undercount by Race, 1940�1990
• Box 1. Choosing More Than One Race
• Box 2. Public and Private Uses of Census Data

At the fractious Constitutional Convention in Philadelphia in 1787, America's Founding Fathers conceived the idea of a national census to determine the number of representatives each state would send to Congress. The initial plan was to ensure no more than one representative for every 30,000 free persons. A slave counted for three-fifths of a free person.¹

The idea of a complete population enumeration was not new � the ancient Romans conducted censuses to assess taxes, for example, and William the Conqueror ordered a census of people and property in England and Wales in 1086, which was recorded in a "Domesday Book" � but the United States was just the second Western country (after Sweden) to conduct a complete census. Most previous censuses were conducted for military and taxation purposes, while the U.S. census was initiated as part of a revolutionary system of representative government.²

Just as the form of government hammered out at the Philadelphia convention has withstood more than two centuries of tumultuous change, the tradition of the decennial population census has also endured. The United States has conducted a census every 10th year beginning with 1790. Except during the 1920s, the results were used to reapportion state representation in the U.S. Congress once a decade, and eventually for assessing taxes, gauging potential military strength, and a myriad of other purposes.

The first U.S. census was conducted by 16 U.S. marshals and their 650 assistants. It took them 18 months to visit households and compile the final tally of 3.9 million people, including nearly 700,000 slaves. The census questionnaires have changed every decade � in most cases the changes involved requesting more detail, but sometimes the changes simply reflected the prevailing social and political currents. The 1830 Census collected information on whether people were deaf and blind; the 1840 Census added columns to identify "insane" or "feeble-minded" people (later dropped).³ It was also the first to ask whether people could read or write � to gauge the literacy levels of the population. The 1860 Census was the last to mention slaves, but later questionnaires requested separate information on household servants.

Questions about color or race have been different in every census. The words and categories chosen each time paint a fascinating and revealing picture of how the concept of race has evolved over two centuries. In the past, race was understood as a biological concept. Today, most social scientists agree that race and ethnicity are social constructions and that humans cannot be classified by race according to biological factors. Instead, certain physical characteristics, such as skin color, are used to separate people into racial categories defined by society.

The number of racial categories used in the census has fluctuated considerably over the years. Groups identified by geography (for example, Asians, Pacific Islanders, and Aleutian Islanders) have been listed as races, together with racial groups defined by skin color (blacks and whites). Data on people with varying degrees of white and black ancestry have also been collected by previous censuses. And the racial categorization of nationality groups is changeable. Asian Indians were included in the white race in the 1970 Census but were counted in the Asian and Pacific Islander category starting with the 1980 Census.⁴

Enumerators for earlier censuses were instructed to report a person's race based on observation. Since 1960, people have identified their own race, and that of others in their household, on census forms they fill out themselves. In 2000, for the first time, Americans were given the additional choice of marking all "race" categories with which they identified (see Box 1).

Only a few items on the census forms are required by law. The first census recorded limited information about the age, sex, and race of household members. Slaves were counted separately and were assumed to be black.

Census forms got longer in subsequent censuses as more questions were added. The Census Bureau eventually introduced the use of more than one questionnaire � one for the majority of Americans that asked just the handful of questions required for congressional reapportionment, and longer forms sent to a sample of households that asked additional questions about housing characteristics, birthplace, education, occupation, recent change of residence, mother tongue, and other items.

The 2000 Census asked just six questions on the basic "short form" that went to about 83 percent of U.S. households: age, sex, race, Hispanic origin, household relationship, and owner/renter status. The "long form," in 2000, with these six questions and 46 additional items, was the shortest since the 1940 Census. It was mailed to a sample of about 17 percent of households nationwide.

The way the government conducts the census has evolved from nonstandard forms filled out by a handful of federal marshals and their assistants in each state to a computer-readable questionnaire mailed to a painstakingly prepared address list of every household in the United States. The early censuses were not highly precise and took months to administer. Special enumerators were not used until the 1880 Census, and the Census Bureau did not become a permanent government department until 1902.

Over the years, the U.S. census has both benefited from and spurred technological advances in statistical sampling, operational methods, tabulation technologies, and information systems. Punch cards that could be counted mechanically � an innovation that played a role in the development of electronic computers � were first used in tabulating the 1890 Census. The system was created by R. Herman Hollerith, a former employee of the Census Bureau who cofounded the company International Business Machines (later known as IBM) in 1911. The Census Bureau pioneered the use of computers for mass data processing.⁵

The changes in the procedures for collecting and disseminating census data reflect the country's growth and westward expansion as well as technological changes. During the first century of census-taking, the country's population soared from 3.9 million clustered along the eastern seaboard in 1790 to 63 million spread across the continent in 1890 (see Figure 1). Census-takers realized they were chroniclers of American history. The official report from the 1890 Census begins: "This census completes the history of a century of progress and achievement unequaled in the world's history. The century has witnessed our development into a great and powerful nation." The report called the Atlantic states of the early United States the "sources of supply of a great westward migration. Their children have peopled the great interior valley and the mountains of the west. They have swarmed from the Atlantic coast to the prairies, plains, mountains, and deserts by the millions during the last century."⁶

That first century of census-taking also saw the growth of cities and urban life. In 1790, just 5 percent of the U.S. population lived in cities and the overwhelming majority worked in agriculture. New York was the nation's largest city with 33,131 residents; Philadelphia was close behind with 28,522. Only four other cities had as many as 8,000 people. By 1890, more than one-third of the U.S. population lived in urban areas; nearly one-half worked in manufacturing, trade, transportation, mining, or service industries. New York was still the largest city, with 1.5 million people, but scores of small and medium-sized cities had grown up along major transportation routes throughout the country. The census also documented a decline in the average household size during the 19th century, from about six to about five people, and an increase in the median age of the population from roughly 16 years to 22 years.

The second century of census-taking recorded the phenomenal increases and diversification of the U.S population during the 20th century. Advances in modes of transport, communications, and industrial production helped transform where and how people lived and worked, and massive immigration at the beginning and end of the century infused new ethnic variety into the resident population. The 1920 Census was the first to document the shift from a rural to an urban majority. Politicians from predominantly rural states feared a loss of power to states with large and growing cities and blocked the reapportionment of electoral votes after the 1920 Census. The conflict between the smaller, rural states and larger, more urban states also reflected unease with the influx of millions of immigrants from southern and eastern Europe in the early 1900s. Congress restricted immigration in the 1920s in an attempt to prevent further shifts in the ethnic makeup of the U.S. population.⁷

As the total population count soared from 76 million in 1900 to nearly 250 million by 1990, the costs of conducting the decennial census rose from about 16 cents per person to more than $10 per person.⁸ The mail-out/mail-back questionnaires, first used extensively in the 1960 Census, drastically cut back the need for enumerators to go door to door. In 2000, questionnaires were mailed to nearly all households, yet the Census Bureau still employed nearly 1 million part-time workers to help carry out the 2000 Census. Enumerators visited some 12 million households that did not mail back a census questionnaire, and other Census Bureau employees scanned more than 1.5 billion pages of questionnaires. The 2000 Census cost about $4.5 billion, which was slightly less than expected, but still about $16 per capita.

Preliminary results of the 2000 Census are being released on a flow basis in 2001 and 2002. Details about education, occupation, ancestry, changes in residence, and other variables will be released later � in 2002 and 2003 � but even the few variables already available for states, counties, and census tracts provide a wealth of data. Because the census is the only source of nationwide data for small geographic areas, it is the primary source of information about population gains or losses, and changes within urban neighborhoods for marketers, politicians, and social scientists. These consumers of census data will be mining the 2000 Census for years to come (see Box 2).

How Accurate/How Complete?

As the costs of and expectations for the census have increased, the completeness of the count and accuracy of the information collected have come under increased scrutiny. For recent censuses, demographers have used two methods � postcensus surveys and demographic analysis � to estimate the number of people missed or overcounted by the census. From 1940 to 1990, the total number missed exceeded the number who were counted more than once.⁹ The demographic analysis method showed that the net percentage undercounted declined from 5.4 percent of the population counted in the 1940 Census to 1.2 percent in the 1980 Census, only to rise to 1.8 percent in the 1990 Census. The postcensus survey had a similar result for the 1990 Census. It showed that 8.4 million people were missed and 4.4 million people were counted twice, yielding a net undercount of about 4 million people (1.6 percent).

The undercount has always been much greater for the black population than for other Americans; this gap has not narrowed substantially since 1940 (see Figure 2). Certain other population groups are also consistently difficult to enumerate. Children, urban residents � especially in low-income central-city areas � people with limited English-language skills, and racial and ethnic minorities all are much more likely to be missed in the census or in surveys than other people. These groups are less likely to have a regular address, may fear cooperating with government authorities, or may face language or cultural barriers to complying with the census. Where these population categories overlap � for example, for minority immigrant children living in poor urban neighborhoods � the undercount tends to be especially high.

Other population groups are frequently overcounted, because they fill out or are included on more than one census form. Retirees with both a summer and winter residence may receive and fill out a census form for both addresses, for example, and college students may be counted at college as well as at their parents' home. Likewise, military personnel, prisoners, nursing home residents, and other people who are temporarily living away from their usual residence are especially subject to overcount.

The 2000 Census undercount and overcount have not been fully investigated, but the Census Bureau's Accuracy and Coverage Evaluation (A.C.E.) Survey conducted to evaluate the 2000 Census indicated that between 1.0 percent and 1.4 percent of the population was missed in the census. A.C.E. results show a 2.85 percent national undercount rate for the Hispanic population, for example, equivalent to another 1 million Hispanics. This evaluation also suggested a slight overcount among Americans ages 50 or older, as well as a large undercount of men ages 18 to 29 and American Indians and Alaska Natives.¹⁰

Using the second method to evaluate census coverage � demographic analysis � the Census Bureau reported evidence of a net overcount of 0.7 percent, that is, the census counted 4 million more Americans than actually live here. If subsequent analysis supports this result, it will be the first ever net overcount for a U.S. census. The discrepancy between the survey and demographic analysis methods for evaluating census coverage also led the Census Bureau to recommend against adjusting the census count pending further analysis.¹¹

Advocacy groups and state and local governments, among others, have sued the federal government to adjust the census numbers for the estimated undercount. They argue that population groups that are under-represented in the census may fail to receive a proportionate share of the federal dollars allocated according to population. Also, the undercounted groups do not factor in the redrawing of political districts, which means they get less than their share of political representation at the local, state, and national level.¹² By April 2001 � one year after the census � the state of Utah and the cities of Chicago, Los Angeles, San Antonio, Stamford, Conn., and Inglewood, Calif., had filed suit against the federal government because they disputed the census results for their area.¹³

Population Change

• Shifting South and West
• Box 3. The 2000 Census Count: Exceeding Expectations?

The 2000 Census recorded an addition of 32.7 million U.S. residents during the 1990s. It was the greatest increase in population ever between two censuses, and the largest percentage increase since the 1960s. The 2000 Census marked the only decade in the 20th century in which every U.S. state gained population. The national count was 6.9 million higher than estimated for April 2000 based on the 1990 Census count and demographic analyses of births, deaths, and migration trends over the decade (see Box 3).

Shifting South and West

The Western and Southern states increased the fastest in population and the Northeastern states grew slowest � continuing demographic trends evident since the 1950s. The South has emerged as the most populous of the four regions defined by the Census Bureau. Its share of the U.S. population expanded from 31 percent in 1950 to 36 percent in 2000. But the westward movement of the population has been the most dramatic shift over the past few decades, and this trend was still evident in the 2000 Census results. In 1950, just 13 percent of Americans lived in the West; in 2000, 22 percent lived in the West � up slightly from 21 percent in 1990. Although the Midwest and Northeast have gained population in the past five decades, their growth has been overshadowed by the rapid gains in the West and South. The share of the U.S. population living in the Northeast fell from 26 percent in 1950 to 19 percent in 2000, while the Midwest's share declined from 29 percent to 23 percent. The population living in the Midwest, which includes such big states as Ohio, Michigan, and Illinois, just barely outnumbered those living in Western states in 2000. With its consistently faster growth, the West is likely to overtake the Midwest before the next census, just as it overtook the Northeast after the 1990 Census.

Nevada, Arizona, Colorado, Utah, and Idaho � all in the West � were the five fastest growing states over the decade (see Appendix Table). Nevada, which had just 1.2 million people in 1990, surged 66 percent over the decade to reach nearly 2 million. Arizona grew 40 percent to 5.1 million, for a much larger numerical gain. Hawaii, Montana, and Wyoming were the only Western states with relatively slow growth. Montana's increase was just under the national growth rate of 13.2 percent, while Hawaii and Wyoming grew just 9 percent. Georgia, Florida, Texas, and North Carolina were the fastest growing Southern states, and among the top 10 gainers nationwide. The populations of all four of these Southern states increased more than 20 percent between 1990 and 2000.

Population growth was much slower in the Midwestern and Northeastern states. New York grew by just 5.5 percent � although it added nearly 1 million people � and Pennsylvania increased by just 3.4 percent. North Dakota barely grew at all over the decade, and had the slowest rate and smallest numerical growth of any state. It added just 3,400 people to bring its population total to 642,000.

Although California is not growing as rapidly as in the past, it still logged the largest numerical increase over the decade � 4.1 million people � to reach a population of 33.9 million. The volume of new residents was nearly as high in Texas, which added 3.9 million and pushed past New York to become the nation's second most populous state. Florida gained 3.0 million people over the decade, the third largest increase. Wyoming remained the least populous state with 493,782 residents. It gained 40,194 residents over the decade.

Increase in Hispanics

• Table 1. U.S. Population by Race and Hispanic Origin, 1980, 1990, and 2000
• Figure 3. U.S. Hispanic Population by Race, 2000
• Figure 4. Hispanic Population Growth in U.S. Counties, 1990�2000

One of the biggest surprises of the 2000 Census was the phenomenal growth in the U.S. Hispanic population. The number of people who identified themselves as Hispanic increased from 22 million to 35 million between 1990 and 2000; the number of Hispanics edged past the number of non-Hispanic African Americans for the first time (see Table 1). The Hispanic population has grown faster than the U.S. black population because Hispanics have higher birth rates and immigration rates than blacks. Although many blacks immigrated from Africa and the Caribbean, the flow is minor compared with the entry of Hispanic immigrants from Latin America.

Table 1
U.S. Population by Race and Hispanic Origin, 1980, 1990, and 2000

NA: Not applicable
1 The 2000 figures are not comparable to the other years because respondents could mark more than one race.
2 Includes Native Hawaiians and other Pacific Islanders.
3 More than 2 million Hispanics marked two or more races in 2000.
Sources: U.S. Census Bureau, 1980 Census of Population PC80-1-C1 (1983): table 233; 1990 Census of Population 1990-CP-1-1 (1992): table 3; and Census 2000 Redistricting Data (P.L. 94-171), accessed online at www.census.gov/clo/www/redistricting.html, on April 21, 2001.

Other explanations for the rapid growth of the Hispanic population revolve around the census itself. The Census Bureau made special efforts to count undocumented immigrants (many of whom are from Latin America), and it moved the question on Hispanic origin to a more user-friendly location, preceding the race question, to encourage a greater response. Many analysts thought people were confused in past censuses by being asked their race first, then whether they were Hispanic. And many Americans do not distinguish between race and ethnicity as defined by the federal government.¹⁴ Previous analyses have estimated that about 90 percent of Hispanics would be considered white by current Census Bureau definitions, yet 42 percent of U.S. Hispanics said they were "some other race," in the 2000 Census (see Figure 3). Another 5 percent said they were "some other race" and white, black, or some other multiracial combination. Similarly, nearly 40 percent of Hispanics checked "other race" in the 1990 Census.

The U.S. Hispanic population has been highly concentrated in the Southwest and West, and in a few metropolitan areas outside these regions, such as Miami, New York New Jersey, and Chicago. One of the big demographic stories of the decade has been the dispersion of Hispanics out of these areas to smaller cities and even rural areas in the Midwest, South, and Northeast. These areas saw the largest percentage increases in Hispanics between 1990 and 2000, as shown by the darker areas in Figure 4. Long-term residents in many Midwestern towns and Southern cities had little interaction with immigrants, or in some cases with minorities, before the arrival of Hispanic workers and families in the 1990s. This dispersion of Hispanics means that many more Americans are seeing and experiencing the country's new racial and ethnic diversity.

Figure 4
Hispanic Population Growth in U.S. Counties, 1990�2000

Source: Created by the Population Reference Bureau based on data from the 2000 Census.

The 2000 Census documented anecdotal evidence that the Hispanic population is getting more diverse. U.S. Hispanics of Mexican origin � the largest Hispanic group � added more than 7 million people over the decade. They account for nearly 60 percent of all Hispanics. And, while Central and South Americans were the second largest group in 1990, they were superceded by an amorphous "other Spanish/Hispanic/Latino" group in the 2000 Census.¹⁵ This surprising shift may reflect confusion with the questionnaire; it may also be connected to the inability to mark more than one national origin. However, some demographers and social commentators suggest that it may signal the gradual assimilation of Americans from diverse national origins in Latin America to a "pan-Latino" identity. The term Hispanic was created as catchall statistical category for people from Spanish-speaking countries, but was not widely embraced by the people it meant to identify. Many identified with their national origin rather than as "Hispanic." This may be changing for U.S.-born Hispanics, who may feel less affinity with the country of their forebears than they do with other Spanish-speaking Americans, who increasingly include high-profile athletes, politicians, and entertainers. Hispanics whose parents are from different origins (Puerto Rican and Dominican, for example, or Salvadoran and Mexican) might also favor the more general term.¹⁶

After the country of birth and other variables from the 2000 Census are released in 2002 and 2003, demographers will have additional clues about why the Hispanic, Spanish, or Latino category surged over the past decade.

Racial and Ethnic Diversity

• Figure 5. U.S. Population Under Age 18 and Age 18 or Older by Race and Hispanic Origin, 2000
• Figure 6. Minority Share of the Population Under Age 18 in U.S. Counties, 2000
• Figure 7. Americans Who Identified With More Than One Race, 2000

The number of Asian Americans also soared during the 1990s, continuing a trend of recent decades spurred by high levels of immigration from Asian countries. In 1990, 6.9 million non-Hispanic Americans identified themselves as Asian (including Pacific Islanders). By 2000, the number was nearly 10.5 million. Survey data and immigration records indicate that most Asian Americans (about 60 percent) are foreign born, and many settled in the United States after 1980.¹⁷ Asian Americans were also less concentrated geographically in the 1990s than they previously had been. While 36 percent live in California, large communities of Asian Americans are now found in Georgia, Pennsylvania, Minnesota, and several other states. This dispersion reflects the settlement of refugee populations that entered the United States in the 1980s and 1990s, and the arrival of new Asian immigrant groups from, for example, India and Pakistan, which do not have the thriving communities that Chinese or Filipino immigrants already had in the United States.

Asian Indians were the fastest growing major Asian American group during the 1990s. Among Asian Americans who identified with one race, Asian Indians are now the third-largest U.S. Asian group, after Chinese and Filipinos. They were the fifth-largest group in 1990. Many Asian Indians are settling where they find jobs rather than where there are existing communities of the same ethnic origin. The result, again, is more diversity in the country's heartland and small cities.

The African American population increased faster than the non-Hispanic white majority, but lacked the additional push from immigration to keep up with the Hispanic or Asian American growth rates. African Americans remain the predominant minority group in the South, however. Blacks made up 19 percent of the population of the South in 2000, while they made up about 12 percent of the total U.S. population.

Although Hispanics outnumbered blacks nationally, Hispanics made up just 12 percent of the South's population in 2000. During the 1990s, record numbers of blacks moved to the South from other regions, which reversed a pattern that prevailed for most of the 20th century. More than 90 percent of African Americans lived in the South in 1900, but the percentage dropped to 53 percent by 1970, reflecting an African American exodus to Northern and Midwestern cities. The 1970s saw a reversal of this trend, and the flow south intensified during the 1990s. Demographers suggest that blacks have been attracted to Southern states by the region's booming economy, attractive life style, improved racial climate, and the historic African American roots.¹⁸

The racial and ethnic diversity of the U.S. population is most evident among children. The 2000 Census found that nearly 40 percent of the population under age 18 was African American, Asian American, Hispanic, American Indian, or another minority, while 61 percent was non-Hispanic white (see Figure 5). One of the big news stories from the 2000 Census was that California is now a "minority majority" state, meaning that non-Hispanic whites make up less than one-half the state population. But minorities already make up more than one-half of the population under age 18 in five states (Arizona, California, Hawaii, New Mexico, and Texas), and in selected counties throughout the country (see Figure 6). In some cases, minority majority counties are clustered around large urban areas; in other cases, these counties identify Indian reservations or nonmetropolitan counties with large black or Hispanic populations.

Figure 6
Minority Share of the Population Under Age 18 in U.S. Counties, 2000

Source: Created by the Population Reference Bureau based on data from the 2000 Census.

Tracking changes among racial and ethnic groups is more difficult with the 2000 Census: This was the first to allow people to mark more than one race. The federal government added this option because of increasing rates of interracial marriage and the growing population that identifies with more than one race, especially among children. Of the 281.4 million people counted in the census, about 6.8 million (2.4 percent) identified with two or more races. About 4 percent of children were identified as multiracial, compared with 2 percent of adults.

The multiracial population included 3.2 million people who reported "some other race" in combination with one or more other races. About 41 percent of these respondents were Hispanic. They often use the "some other race" designation to express their nationalities � for example, Mexican or Salvadoran or Nicaraguan � which for them have more meaning than the category Hispanic. Those who chose some other race along with white, black, or Asian were the most common multiracial combinations in the 2000 Census. Next to these came white and American Indian and Alaska Native (1.1 million), white and Asian (868,000), white and black (785,000), and black and American Indian and Alaska Native (182,000) (see Figure 7).

The new options for answering the race question on the census form have made it difficult to measure the size of racial groups and to track trends over time, especially for groups with high rates of intermarriage. The American Indian and Alaska Native population, for example, could number as low as 2.5 million or as high as 4.1 million, depending on how the multiracial American Indian population is classified. It is also a challenge to measure the growth or decline of racial groups since 1990. Using the single-race definition, the American Indian and Alaska Native population grew by 26 percent, but under the alternative definition, which combines single-race and multiracial American Indian groups, the population grew by 110 percent.

Metropolitan Growth

• Table 2. Ten Fastest Growing and Fastest Declining Metropolitan Areas, 1990�2000
• Figure 8. Population Growth in Metropolitan Area Counties, 1990�2000
• Patterns of Growth
• Three Patterns of Metro Growth
• Figure 9. Population Growth in the Atlanta Metropolitan Area, 1990�2000
• Changing Neighborhoods

The vast majority of Americans live in metropolitan areas � urban counties surrounding a city (or urbanized area) with a population of at least 50,000. The 2000 Census has painted a broad-brush picture of the 276 metro areas that are the center of American society. Eighty percent of the population lives in metropolitan areas, a slight increase over the 1990 share. The metro area population increased by 14 percent between 1990 and 2000, much faster than the population in nonmetropolitan counties, which grew about 10 percent over the decade.

Most of the nation's metropolitan areas saw their populations increase in the 1990s, but growth was much faster among metro areas with populations between 1 million and 5 million: They grew 19 percent during the 1990s, while larger and smaller metro areas grew by 11 percent and 12 percent respectively.

Metro areas in the West and South grew fastest � by about 20 percent on average � while metro areas in the Midwest and Northeast increased by less than 10 percent. Las Vegas, Nev., saw its population soar from 853,000 in 1990 to nearly 1.6 million in 2000 (an 83 percent increase). Such major regional hubs as Phoenix, Atlanta, and Denver grew at least 30 percent during the last decade as did emerging areas like Austin, Tex. (see Table 2). The faster growth in the South and West continues overall regional trends evident for the last 40 years that have shifted the U.S. population away from the Midwest and Northeast. In the 1960 Census, for example, the South and West accounted for 46 percent of the U.S. population; by 2000, these two regions accounted for 58 percent.

Table 2
Ten Fastest Growing and Fastest Declining Metropolitan Areas, 1990�2000

Source: U.S. Census Bureau. Table accessed at www.census.gov/
population/cen2000/phc-t3/tab02.xls on April 17, 2001.

Most of the 24 metropolitan areas that lost population between 1990 and 2000 included smaller, aging cities in the Northeast and Midwest � Pittsburgh, Pa., Buffalo, N.Y., and Youngstown, Ohio, for example � that have been losing population for decades. But some metro areas centered on small manufacturing towns in the South � such as Anniston, Ala. � also lost population. Counties within the same metropolitan areas grew at varying rates (see Figure 8).

Figure 8
Population Growth in Metropolitan Area Counties, 1990�2000

Note: Alaska and Hawaii (not shown) each have one metropolitan area county that increased by 0% to 14% between 1990 and 2000.
Sources: Created by the Population Reference Bureau based on data from the 2000 Census.

What is behind these trends? The 2000 Census data slated for release in 2002 and 2003 will allow more detailed analyses, but the existing data suggest some explanations.

First, fast-growing metropolitan areas tended to be "job magnets," assisted by rapid growth in one or more economic sectors. Las Vegas, for example, has continued to emerge as a major entertainment and tourism center; tourists alone pumped $22.5 billion into the local economy in the late 1990s.¹⁹ Other fast-growing metropolitan areas � such as Austin, Tex., Phoenix, and the Research Triangle area of Raleigh-Durham-Chapel Hill, N.C. � were high-tech boom areas.²⁰ Conversely, many slow-growing and declining metropolitan areas suffered recent or long-term economic downturns. The loss of population in the areas surrounding Buffalo, Pittsburgh, and the eastern Ohio cities of Youngstown and Steubenville, for example, continued a trend initiated by losses of manufacturing jobs decades earlier.²¹

Economic factors are not the only explanation for metro area population change. Several booming metro areas have emerged as retirement Meccas, including Naples and Ocala in Florida; Yuma, Ariz.; Myrtle Beach, S.C.; Wilmington, N.C.; and Las Cruces, N.M.²²

International migration also influences metro area population growth. While 2000 Census data on migration will not be released until 2002, an analysis of demographic trends from 1990 to 1998 by demographers William Frey and Ross DeVol suggests that "gateway" metros such as New York, Los Angeles, Miami, and Chicago were top destination choices for immigrants. Many immigration magnets lost native-born residents to other states and metro areas. Among the 10 metros that Frey and DeVol listed as high immigration magnets, only Dallas, Houston, and Miami grew significantly faster than the national average over the 1990s, partly because eight of those 10 (Dallas and Houston were the exceptions) had a net loss of native-born residents.²³ Frey, among others, suggests that U.S.-born residents left because the growing immigrant population held down wages.

Patterns of Growth

Metropolitan areas usually encompass one or more central city areas surrounded by an inner ring of suburban counties, and often an outer ring of less densely settled suburban counties. In recent decades, central cities have lost population to suburban counties as middle-class families moved out to suburbs under the assumption that the schools were better, neighborhoods safer, and property values more stable. But the 2000 Census documented a surprising population gain in most central cities, especially in newer metropolitan areas, and drew the outlines of at least three general patterns of metropolitan growth: increasing density in the central counties; sprawling, less concentrated urban growth; and a declining urban center, surrounded by slower-growing inner suburbs and faster-growing outer suburbs.

Eight of the nation's 10 largest cities (Detroit and Philadelphia were the exceptions) gained population between 1990 and 2000.²⁴ While many smaller cities in the Northeast and Midwest lost population during the 1990s, the declines generally were less steep than expected and less severe than the declines of the 1980s. Baltimore's population total was down 11.5 percent; Cleveland was down 5.4 percent; New Haven, Conn., lost 5.2 percent, and Philadelphia lost 4.3 percent of its population. Other cities in the Northeast and Midwest actually grew, notably Chicago (up 4.0 percent) and New York City (up 9.4 percent). New York City, as defined by five constituent counties, topped 8 million in 2000, just above its previous high count of 7.9 million in 1970.

Demographers will know more about who moved into these metro areas after census data on place of birth and migration are released in 2002. But it appears likely that central-city population growth was fueled mainly by increasing numbers of international migrants who offset a continuing exodus of the U.S.-born population out of central cities to surrounding suburbs or to other metro areas and states. Chicago, for example, had fewer black and white non-Hispanic residents in 2000 than in 1990. However, an increase in Hispanic residents, many from abroad, more than offset the losses and allowed the city to grow.

Suburbs in many metropolitan areas also gained population from international migration. More international migrants lived in the suburbs than in central cities during the 1990s, according to Census Bureau estimates. Stories about high schools in which dozens of languages are spoken or about hospitals needing emergency-room translators have been familiar fare in newspapers for decades. What changed during the 1990s is that these stories were written about inner suburbs like Wheaton, Md., outside Washington, D.C., or Marietta, Ga., outside Atlanta, rather than traditional immigrant gateways like lower Manhattan or downtown Los Angeles.

Three Patterns of Metro Growth

The patterns of growth across metropolitan areas are difficult to compare using only the census concepts of "central city" and "suburb." Some city boundaries, especially in the West, are large and still encompass areas of low-density settlement, and some cities like Houston and Indianapolis can annex new land as the metropolitan area grows. But maps of population growth linked with aerial photos or remote-sensing images of land use, confirm that "sprawl" is not a uniform nationwide phenomenon. Population densities and growth rates vary across metropolitan areas, and even within suburbs of metropolitan areas, as illustrated by Kansas City, Atlanta, and Los Angeles.

The Kansas City metropolitan area, which includes four counties in Kansas and seven in Missouri, showed a pattern common for slow-growing metropolitan areas in the North and Midwest. The central cities grew little or lost population: Kansas City, Mo., grew 1.5 percent for the decade; Kansas City, Kan., lost almost 2 percent of its population. The inner metropolitan counties grew slowly (less than 10 percent during the decade), while the outlying counties grew rapidly (more than 20 percent). The metropolitan area as a whole grew 12 percent, just below the growth rate for the nation as a whole.

Many fast-growing metropolitan areas, especially in the South, followed a more haphazard growth pattern. In this prototypical "sprawl," rapid development occurs at the outer fringes of metropolitan areas, often leaping over low-density areas and following major highways. The Atlanta metropolitan area exemplifies this pattern as its population surged by 39 percent during the 1990s. Forsyth County, Ga., on the northeast border of the Atlanta metropolitan area, and Henry County, southeast of central Atlanta, more than doubled in population during the 1990s (see Figure 9). Major highways run through both counties. Portions of several of the Atlanta metro area's western and southern counties also saw their populations increase by 50 percent or more, while adjacent areas both closer and further from the center had relatively slow growth.

Atlanta's central-city population, in portions of Fulton and DeKalb Counties, grew less than 6 percent. Even this low growth rate was higher than expected � the city had lost population during the 1970s and 1980s.

Fast-growing metropolitan areas in the West followed another pattern as their populations became more concentrated during the 1990s. Suburbs in Western metro areas gained population, but did not expand in land area as rapidly as in the South. Physical geography was one reason these areas became more densely settled � many Western metropolitan areas are physically limited by coastlines, desert, or mountain ranges. Their outlying areas are seeing more development than in past years, but still cannot support the densities of settlement that "exurbs" in other parts of the country can.²⁵

Changing Neighborhoods

Population shifts within metropolitan areas are altering the racial and ethnic makeup of cities and suburbs. In previous decades, metro areas have experienced "white flight": White families would move from central-city areas to the suburbs, which concentrated minorities and urban poverty in the inner cities. Neighborhoods and schools were highly segregated; central cities were predominantly minority, suburbs were predominantly white. Immigrant groups typically settled in central cities and created unique ethnic communities. Some of this same phenomenon occurred during the 1990s. The non-Hispanic white percentage of the population in the 100 largest cities dipped from 52 percent to 44 percent, according to a recent report from the Brookings Institution.²⁶ But some cities, including Washington, D.C., and Atlanta, saw the non-Hispanic white share of their populations increase. In Washington, this reflected a larger exodus of blacks than whites; but in the city of Atlanta, the increase captured an influx of non-Hispanic whites during the decade. Atlanta was one of a few rapidly growing cities, including Austin, Tex., and Las Vegas, Nev., that saw an increase in their non-Hispanic white population, according to census results and the Brookings Institution study.

The 2000 Census data show high levels of racial segregation in residential areas of metropolitan America, and only a slight decline in segregation since 1990. White Americans tend to live in neighborhoods that are overwhelmingly white; minorities live in neighborhoods with other minorities.

A recent study shows that the average white American in a metropolitan area lives in a neighborhood that is about 83 percent white and about 7 percent black, 6 percent Hispanic, and 3 percent Asian. In 1990, whites lived in neighborhoods that were 86 percent white. The 2000 Census showed the average black person lives in a neighborhood that is 33 percent white and 51 percent black. Compared with 1990, blacks were more likely to have Hispanic and Asian neighbors in 2000, but they were no more likely to have white neighbors.²⁷ Asian and Hispanic populations � which include large numbers of recent immigrants � were somewhat more isolated from other racial and ethnic groups in 2000 than they were in 1990.

Older, large metro areas in the Northeast and Midwest, like Newark, N.J., and Detroit, tend to be the most segregated, while newer, rapidly growing areas in the West and South are least segregated. Metropolitan areas with large military populations, such as Norfolk, Va., and San Diego, also tend to be less segregated.

One of the distinctive patterns of the 1990s was the movement of middle-class minority families from cities to suburbs, and in some areas, the emergence of immigrant communities in the suburbs rather than city neighborhoods. This has introduced ethnic and racial diversity into what once were all-white communities. But minorities who move to the suburbs do not necessarily live in integrated neighborhoods. The decade saw an increase in "minority suburbs" in such places as Atlanta and Washington, D.C.

In some suburban areas, neighborhoods evolve over time from predominantly white, to mixed white/minority to predominantly minority � similar to the progression that took place in central city areas in the last half of the 20th century. Washington, D.C., for example, was two-thirds white and about one-third black in 1950. In 2000, the city was two-thirds black and one-fourth non-Hispanic white. In suburban Prince George's County, Md., just across the district line, the population was about 85 percent white in 1970. In 2000, the county's population was 24 percent non-Hispanic white and 63 percent black, reflecting the movement of blacks from the District of Columbia into the county.

Herndon, Va., a suburban community in the Washington, D.C. metro area, attracted many immigrants from Asia and South and Central America during the 1990s. Hispanics increased from nearly 10 percent to 26 percent of Herndon's population; and Asian Americans increased from about 8 percent to 14 percent of the population.²⁸

In Atlanta's metro area, minority suburbs have emerged in newly developing areas. Middle-class black families moved from the central city to suburban areas in Clayton County. Between 1990 and 2000, the county population rose by 30 percent, and the percentage of residents who were black more than doubled, from 24 percent to 51 percent. The percentage who were Hispanic rose from 2 percent to more than 7 percent in Clayton county over the decade.²⁹ In contrast, the explosive suburban growth in neighboring Henry County primarily involved white families, although the county also saw an increase in blacks and Hispanics. About 80 percent of Henry County's population was white in 2000, down from 88 percent in 1990.

Political Implications

• Figure 10. Electoral Votes by State, 2000
• Box 4. Could We/Should We Replace the Electoral College?

The legal requirement to conduct a census is rooted in the American system of representative government. The law requires that the total population for each state, as determined by the census, be given to the president by December 31 of the census year and that this population be used to reallocate the number of seats held by the states in the U.S. House of Representatives. The 2000 Census reapportionment will take effect when the 108th Congress is elected in November 2002 and convenes in 2003.

The new census numbers also determine the number of electoral votes each state will wield in the 2004 and 2008 presidential elections. Because the U.S. Constitution grants all states at least one representative and two senators, each state has at least three electoral votes (equal to its number of representatives plus the two senators). The number of electoral votes was fixed at 535 during the apportionment following the 1910 Census; three more were added in 1961 when the District of Columbia was granted three electoral votes (although the District has no voting representation in Congress). The number of votes held by an individual state increases or decreases depending on its relative share of the total U.S. population at each census (see Box 4).

The census figures reported to the president in December 2000 added congressional seats to eight states, and subtracted seats from 10 states (see Figure 10). Arizona, Florida, Georgia, and Texas will each gain two seats in the 108th Congress, while California, Colorado, Nevada, and North Carolina will each pick up one seat. New York and Pennsylvania will lose two seats each, while Connecticut, Illinois, Indiana, Michigan, Mississippi, Ohio, Oklahoma, and Wisconsin will lose one seat each.

Figure 10
Electoral Votes by State, 2000

Source: U. S. Census Bureau. Accessed online www.census.gov/population/www/censusdata/apportionment.html on April 21, 2001.

The population totals used to reapportion congressional seats include the resident population of each state plus military and U.S. government civilian employees from that state (and their dependents) who are posted overseas. Overseas populations were included in the apportionment population after the 2000, 1990, and 1970 Censuses, but were not included in other census years. If only the current state resident population totals had been used in 2000, North Carolina would not have gained an additional seat � that seat would have gone to Utah. North Carolina's substantial overseas military population pushed up the state total just enough so it could claim another seat. Utah has contested the result, arguing that if the Census Bureau had included the 11,000 Utah residents serving temporary tours as missionaries overseas, Utah's apportionment population would have been enough so that Utah rather than North Carolina would have gained a seat.³⁰

Even with the expected gains for the South and West, the new apportionment numbers contained some surprises. Apportionment projections based on 1999 Census Bureau estimates had indicated that Florida and Georgia would each gain only one seat, while Indiana, Michigan, and North Carolina would remain the same. Those projections also mistakenly predicted that Montana would regain the second House seat that it had lost in the 1990 apportionment.

Because the states vary tremendously in population size � from just under 500,000 in Wyoming to nearly 34 million in California � and the total number of House seats stays at 435, this system guarantees a wide disparity among states in how many people each state delegation represents. Montana has one member in the U.S. House of Representatives, who represents 902,000 state residents, for example, while Wyoming's one member represents 494,000 people. Under the new apportionment, New York has 29 members, or one for every 654,000 New York residents. This disparity among states will be reflected in the 108th Congress when it convenes in 2003.

How will the 2000 Census affect the U.S. political scene in the next decade, and what are some of the political implications of the population changes captured by the census? In general, states like New York and Pennsylvania that have gone for Democrats in recent presidential elections lost ground, while states like Texas and Arizona that have tended to support Republican presidential candidates in recent elections gained electoral votes. This shift reflects the general population movement to Southern, Western, and Mountain states � from abroad and from Midwestern and Northeastern states. The 2000 Census results reaffirm these trends.

The increasing percentage of the U.S. population in states that were Republican-leaning in the 1990s, however, does not guarantee a long-term dominance of the Republicans or any other party. The population moving into these high-growth states may not have the same political leanings as the population already living there.

The political future is especially uncertain in states that are gaining large immigrant populations. Immigrants tend not to vote: Most recent immigrants are not citizens and many are young, poor, and have little formal education � population groups that have low rates of voter turnout.³¹ But over time, some become politically involved: They may naturalize, vote, and even run for office. Recent legal changes have accelerated the transition in some areas. The number of immigrants seeking U.S. citizenship surged in the late 1990s, in part because of changes in the procedures for obtaining immigrant visas and because welfare and other public services for noncitizens were limited by the Personal Responsibility and Work Opportunity Reconciliation Act of 1996.³²

In New York City, this increase in the number of naturalized citizens coincided with the imposition of term limits that forced out two-thirds of the incumbent city council members, which created opportunities for political newcomers. At least a dozen foreign-born New Yorkers ran for city council in 2001.³³ The U.S.-born children of immigrants � citizens by birth � are much more likely than their parents to participate in elections, which could shift the political balance in some areas.

The major political parties are well aware of the potential for political support or opposition from the growing Hispanic population. Hispanics are underrepresented in the U.S. Congress: They are 12 percent of the U.S. population, yet hold just 4 percent (19 voting seats) in Congress in 2001. African Americans, also about 12 percent of the national population, hold 36 seats. Asian and Pacific Islanders hold 6 voting seats, while non-Hispanic whites occupy 472 seats.³⁴ Hispanics have long been prominent in the political parties in states like California, Florida, New Mexico, and Texas, however, and their influence is expected to increase elsewhere. During the 1990s, the Hispanic population doubled in many states, including Iowa, North Carolina, and Oregon, and rose sharply in fast-growing states like Georgia. Political candidates in Hispanic districts may increasingly champion issues of special concern to Hispanics, such as combating high school dropout rates and providing health care for uninsured workers.³⁵

African Americans make up nearly the same percentage of the U.S. population as Hispanics, but they make up a larger percentage of the population that is eligible to vote. And African Americans are bucking the national decline in voter turnout. In the last two national elections, the percentage of blacks who voted has increased or held steady, while the percentage has slipped for whites.³⁶ Blacks have not been particularly successful at winning elected office in majority white districts, but they wield considerable power in many states and metropolitan areas. In the November 2000 presidential election, the black share of the vote was greater than their share of the voting-age population in five states: Florida, Mississippi, Missouri, Tennessee, and Texas.

Although Asian Americans remain a relatively small proportion of the U.S. population, they have become increasingly active politically. Even more than blacks and Hispanics, Asian Americans must gain the support of other racial and ethnic groups to win elected office, but several have achieved this.³⁷ Political analysts have noted that coalitions among minority voters can sway an election even when the size of a specific population is relatively small. In 1998, black, Hispanic, and Asian coalitions were instrumental to the U.S. Senate race in New York and the governor's race in the California, and a coalition may be emerging among Hispanics and blacks in New York City.³⁸

The growing share of minorities in the U.S. population � and their increasing involvement in politics � all but guarantees to raise their visibility on the political landscape. The major political parties are positioning themselves to woo more of the minority vote. Recently, one analyst calculated that if George W. Bush wins the same percentage of the minority vote in 2004 as he did in 2000, he would lose the popular vote by 3 million votes because minorities will account for a larger share of the voting-age population in 2004.³⁹ The rate of minority participation in elections will be increasingly important for candidates' success.

More From the 2000 Census

• Table 3. Selected 2000 Census Population Data to be Released
• Box 5. Media Coverage of the 2000 Census
• Box 6. The Legacy of the 2000 Census

The flurry of news stories generated by each new release of data from the 2000 Census is unparalleled. The news value stems from the surprises and mysteries associated with the census, but it also reflects a more sophisticated approach by the news media and the easier availability of information through the Internet and other technologies (see Box 5). A wealth of additional and even more detailed information about the American people � some of it down to the census-tract level � will be released over the next few years (see Table 3).

Table 3
Selected 2000 Census Population Data to be Released

* Urban/rural data are on the final national file. This is the only difference from the advance national file.

What will the legacy of the 2000 Census be? Kenneth Prewitt, director of the Census Bureau during the 2000 enumeration, cites the new approach to classifying race as a watershed event in 2000 (see Box 6). Legal battles that are brewing over adjustment, coverage, and redistricting may lead to precedent-setting court rulings. And the press attention and easy accessibility of census data may well expand the use of these data far beyond the traditional business, academic, and public consumers.

The Next Census

• Box 7. The American Community Survey

The 2000 Census followed in the 200-year tradition of enumerating each American to comply with the Constitutional requirement for political apportionment. It may have been the most complete and accurate census ever administered, and it may have been the last "traditional" U.S. census. The rising costs of conducting a census and and growing concerns about personal privacy have prompted a search for alternative ways of getting the same information. One strategy is the American Community Survey (ACS), which is slated to replace the long census form that went to about 17 percent of the population in 2000. Assuming it is fully implemented in 2003, the ACS would be conducted nationwide for the rest of the decade and would track changes over time and measure relative differences among population groups and communities (see Box 7).

New technologies for storing and transferring data and for making maps are likely to make more information available to the public. But aside from the technological issues, a number of ethical and governance issues will remain important in the third century of census-taking. Ken Prewitt highlights � and distinguishes between � the issues of privacy and confidentiality: "Privacy really has to do with the question 'what does anyone have the right to know about me?' irrespective of whether that information is held confidentially or not." Many Americans are concerned that so much detailed personal information is available for marketers and others, and view the census as another government invasion of privacy.

The Census Bureau is required to ensure the confidentiality of the data it collects. It does not release information that can identify specific individuals or households enumerated in recent censuses. The Census Bureau will release household information for genealogical or other research only 72 years after the census was taken (records from the 1920 Census are the most recent available in 2001, for example).

Because the census and related surveys involve collecting information on individuals, Prewitt says that the Census Bureau, which "has a practically impeccable record on confidentiality can still be vulnerable to charges of 'you're invading my privacy,'" and concedes, "There's no easy solution to that one."

The census will probably continue to be a political lightning rod, just as it has been for the last two centuries. The advances in methodology and technology will probably encourage more lawsuits because they expand the number of possible interpretations and expose possible errors or inconsistencies. Based on his experience with the 2000 Census, Prewitt urges an early agreement among the contending political interests about what methodology to use so the professional staff has time to gear up. Noting that he testified before Congress 17 times in less than two years, Prewitt commented that "You would never run a military campaign that way. �[General Norman] Schwartzkopf, when he was running Desert Storm, wasn't going back and forth to Capitol Hill to get permission for every step.

I understand the politics, and I understand they are intense, but resolve it if at all possible." The census was created by and for politics, and is not likely to be able to remain aloof from it. At the same time, the census remains a gold standard for social science research and a valuable venue for developing new communication and data technologies. It also provides a fascinating portrait of Americans at the end of each decade.

References

1. Margo J. Anderson, The American Census: A Social History (New Haven, CT: Yale University Press, 1988): 7-18.
2. United Kingdom Office of National Statistics, "Census Taking Through the Ages," Factsheet 8. Accessed online at www.statistics.gov.uk/
   census2001/publicity/factsheet8.pdf on April 5, 2001.
3. U.S. Census Bureau, "Population and Housing Inquiries in the U.S. Decennial Census, 1790-1970," Working Paper 39 (Washington, DC: U.S. Census Bureau, 1973): 5-39.
4. Sharon M. Lee, Using the New Racial Categories in the 2000 Census, A KIDS COUNT/PRB Report on Census 2000 (Washington, DC: The Annie E. Casey Foundation and the Population Reference Bureau, March 2001).
5. Harvey Choldin, Looking for the Last Percent (New Brunswick, NJ: Rutgers University Press, 1994): 11-13.
6. U.S. Census Bureau, 1890 Census (Washington, DC: Government Printing Office, 1897).
7. Anderson, The American Census: 131-59.
8. Bryant Robey, "Two Hundred Years and Counting: The 1990 Census," Population Bulletin 44, no. 1 (Washington, DC: Population Reference Bureau, 1989): 4-6.
9. Barry Edmonston and Charles Schultze, eds., Modernizing the U.S. Census (Washington, DC: National Academy Press, 1995): 30-37; and U.S. Census Bureau, "Population and Housing Inquiries in the U.S. Decennial Census, 1790-1970," Working Paper 39: 1-3.
10. U.S. Census Bureau, "Preliminary Estimates Show Improvement in Census 2000 Coverage," Press Release CB01-CN.03, Feb. 14, 2001.
11. U.S. Census Bureau, Report of the Executive Steering Committee for Accuracy and Coverage Evaluation Policy (March 1, 2001). Accessed online at www.census.gov/dmd/www/EscapRep.html on April 16, 2001.
12. Terri Ann Lowenthal, The Road to Census 2000: A Chronology of Key Issues, A KIDS COUNT/ PRB Report on Census 2000 (Washington, DC: The Annie E. Casey Foundation and the Population Reference Bureau, March 2001).
13. August Gribbin, "Census Toll Challenged by 5 Cities and Utah," The Washington Times, April 23, 2001.
14. Hispanic origin is an ethnic � not racial � designation in federal statistics. Hispanics may be of any race. See discussion in U.S. Office of Management and Budget, Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity. Accessed online at www.whitehouse.gov/omb/fedreg/ombdir15.html; Albert M. Camarillo and Frank Bonilla, "Hispanics in a Multicultural Society: A New American Dilemma?" in America Becoming: Racial Trends and Their Consequences, vol. 1, ed. Neil J. Smelser, William Julius Wilson, and Faith Mitchell (Washington, DC: National Academy Press, 2001): 119-21.
15. Betsy Guzman, "The Hispanic Population," Census 2000 Brief C2DBR/01-3 (U.S. Census Bureau, May 2001).
16. D'Vera Cohn, "Shifting Portrait of U.S. Hispanics," The Washington Post, May 10, 2001; and Mireya Navarro, "The Latino Candidate: Yours, Mine or Ours?" The New York Times, May 6, 2001.
17. Sharon M. Lee, "Asian Americans: Diverse and Growing," Population Bulletin 53, no. 2 (Washington, DC: Population Reference Bureau, June 1998): Table 1.
18. William Frey, "Migration to the South Brings U.S. Blacks Full Circle," Population Today 29, no. 4 (Washington, DC: Population Reference Bureau, May/June 2001): 1, 4-5; and Kelvin M. Pollard and William P. O'Hare, "America's Racial and Ethnic Minorities," Population Bulletin 54, no. 3 (Washington, DC: Population Reference Bureau, 1999).
19. Michael Barone and Grant Ujifusa, The Almanac of American Politics 1998 (Washington, DC: National Journal, 1997): 875.
20. Ibid.: 86, 1066, and 1361.
21. Ibid.: 1146 and 1235.
22. William H. Frey and Ross C. DeVol, "America's Demography in the Next Century: Aging Baby Boomers and New Immigrants as Major Players," Milken Institute Policy Brief 9 (Mar. 8, 2000): 9-10.
23. Ibid.: 18-19.
24. U.S. Census Bureau, "Incorporated Places of 100,000 or More, Ranked by Percent Population Change: 1990 to 2000." Accessed online at www.census.gov/population/cen2000/phc-t5/tab04.xls on April 25, 2001.
25. David Firestone, "90's Suburbs of West and South: Denser in One, Sprawling in Other," The New York Times, April 18, 2001.
26. Brookings Institution Center on Urban and Metropolitan Policy, Racial Change in the Nation's Largest Cities: Evidence From the 2000 Census (Washington, DC: The Brookings Institution, 2001). Accessed online at www.brookings.edu/es/urban/census/citygrowth.htm on May 3, 2001.
27. John Logan, et al., Ethnic Diversity Grows, Neighborhood Integration Is at a Standstill (Albany, NY: Mumford Center, State University of New York, Albany, 2001). Accessed online at www.albany.edu/
    mumford/census, on April 15, 2001.
28. Audrey Singer, Samantha Friedman, Ivan Cheung, and Marie Price, The World in a Zip Code: Greater Washington, D.C. as a New Region of Immigration (Washington, DC: The Brookings Institution, April 2001); and U.S. Census Bureau, American FactFinder. Accessed online at www.census.gov on May 3, 2001.
29. David Firestone, "Census Shows Growth in Atlanta's Population," The New York Times on the Web, March 23, 2001. Accessed online at www.nytimes.com on April 18, 2001; and U.S. Census Bureau, Census 2000 Redistricting Data (Public Law 94-171) Summary File. Accessed online at www.census.gov in April 2001.
30. Joe Baird and Dan Harrie, "Judges Reject Utah Census Claims," The Salt Lake Tribune, April 18, 2001. Accessed online at www.sltrib.com/2001/apr/04182001/nation_w/89820.htm on April 19, 2001.
31. Jennifer C. Day and Avalaura L. Gaither, "Voting and Registration in the Election of November 1998," Current Population Reports P20-523 (Washington, DC: U.S. Census Bureau, May 2000).
32. Philip Martin and Elizabeth Midgley, "Immigration to the United States," Population Bulletin 54, no. 2 (Washington, DC: Population Reference Bureau, 1999): 33-35.
33. Susan Sachs, "Give Me Your Tired, Your Poor, Your Vote," The New York Times on the Web, April 8, 2001. Accessed online at www.nytimes.com on April 9, 2001.
34. Infoplease.com, "Minorities and Women in the 10th Congress (table)." Accessed online at www.infoplease.com/ipa/A0878575.html on May 24, 2001.
35. Eric Schmitt, "Hispanic Voter is Vivid in Parties' Crystal Ball," The New York Times on the Web, April 9, 2001. Accessed online at www.nytimes.com on April 9, 2001.
36. David A. Bostis, The Black Vote in 2000: A Preliminary Analysis (Washington, DC: The Joint Center for Political and Economic Studies, December 2000): 1. See also, Avalaura L. Gaither and Eric C. Newburger, "The Emerging American Voter: An Examination of the Increase in the Black Vote in November 1998," Population Division Working Paper, no. 44 (U.S. Census Bureau, June 2000).
37. Lee, "Asian Americans: Diverse and Growing": 29-32.
38. David C. Ruffin, "Black Voters Assert Their Power," Focus 26, no. 12 (December 1998): 3-4; and Mireya Navarro, "The Latino Candidate: Yours, Mine or Ours?" The New York Times, May 6, 2001.
39. Schmitt, "Hispanic Voter is Vivid in Parties' Crystal Ball."

Appendix Table: Population and Electoral Votes for U.S. States in 2000 & Population Change Since 1990

Note: Population change 1990-2000 and VAP per electoral vote were based on unrounded numbers.
Sources: U.S. Census Bureau. Tables accessed online at www.census.gov/population/cen2000/tab01.txt, www.census.gov/population/www/cen2000/phc-t2/tab03.xls, and www.census.gov/Press-Release/www/2001/tables, various files, on April 30, 2001.

Suggested Resources

Anderson, Margo J. The American Census: A Social History, New Haven, CT: Yale University Press, 1988.

Anderson, Margo J., and Stephen E. Fienberg. Who Counts? The Politics of Census-Taking in Contemporary America, New York: Russell Sage Foundation, 1999.

Darga, Kenneth. Sampling and the Census: A Case Against the Proposed Adjustments for Undercount, Washington, DC: AEI Press, 1999.

Edmonston, Barry, and Charles Schultze, eds. Modernizing the U.S. Census. Washington, DC: National Academy Press, 1995.

Lowenthal, Terri Ann, The Road to Census 2000: A Chronology of Key Issues. KIDS COUNT/PRB Reports on Census 2000. Washington, DC: The Annie E. Casey Foundation and Population Reference Bureau, May 2001.

Prewitt, Kenneth, "The U.S. Decennial Census: Political Questions, Scientific Answers," Population and Development Review 26, 1 March 2000: 1-16.

Selected websites with reports based on the 2000 Census

U.S. Census Bureau www.census.gov and http://factfinder.census.gov

Population Reference Bureau and Social Science Data Analysis Network www.AmeriStat.org

Lewis Mumford Center for Comparative Urban and Regional Research, State University of New York, Albany www.albany.edu/mumford/census

Center on Urban and Metropolitan Policy, Brookings Institution, Washington DC www.brookings.edu/es/urban/census/censusmatters.htm

Mary M. Kent is the editor of the Population Bulletin at the Population Reference Bureau (PRB) and has written and edited publications on a variety of U.S. and international population issues. She holds a master's degree in demography from Georgetown University.

Kelvin M. Pollard is a research demographer at PRB. He is the author of numerous publications on U.S. population-related topics, including two previous Population Bulletins, PRB's United States Population Data Sheet, and a chapter in a revised edition of a textbook, The Methods and Materials of Demography (forthcoming). He holds a master's degree in applied social research from the University of Michigan.

John Haaga is director of PRB's domestic programs. He is co-editor, with Reynolds Farley, of the forthcoming series of PRB/Russell Sage Census 2000 Bulletins. His publications deal with social policy and demographic change. His doctorate in public policy was awarded by the RAND Graduate School.

Mark Mather is a policy analyst at PRB, where he coordinates several projects that communicate population research to advocacy groups, educators, the media, and the public. He holds a doctor's degree in sociology from the University of Maryland and has written on the U.S. census and demographic issues.

Other PRB staff who prepared text or graphics for the Bulletin are Bingham Kennedy, Jr., associate editor; Cheryl Lynn Stauffer, education and outreach specialist; and Kerri L. Rivers, research associate. Others who contributed to the Bulletin include Steve Doig, Arizona State University; Reynolds Farley, University of Michigan; Ken Hodges, Claritas Corporation; Barron Holmes, South Carolina KIDS COUNT Project; Sharon M. Lee, Portland State University; and Kenneth Prewitt, New School for Social Research, New York. PRB thanks Arthur Norton for reviewing the manuscript.

Box 1 Return to text
Choosing More Than One Race

In the 1990 Census, half a million people disobeyed instructions to mark only one race and checked two or more races instead.¹ This mild act of civil disobedience (or, for some, misunderstanding) signaled a growing social acceptance of multiracial identities and dissatisfaction with the current categories; it also coincided with a marked increase in multiracial marriages and births.

The 2000 Census allowed multiple racial responses for the first time (see figure). Nearly seven million people were marked in more than one racial category. The option of choosing more than one race provides a more accurate, if complex, portrait of diversity in America. Although the people identifying themselves as multiracial were just 2.4 percent of the U.S. population, their numbers may grow faster than the total population as interracial marriages increase and more people acknowledge their multiracial backgrounds. Already, children are much more likely to identify themselves as multiracial than adults. Four percent of the population under age 18 were identified in more than one racial category in the 2000 Census, twice the percentage for adults.

The multiracial population is larger in certain geographic areas and population groups. People in urban areas, for example, are more likely to be multiracial, as are residents of racially diverse states such as Hawaii and California. The percent of people reporting more than one race was relatively high in Honolulu (15 percent), and the Bronx, N.Y. (6 percent), and relatively low (less than 1 percent) in Madison County, Miss., and Luzerne County, Pa.² In general, large urban areas in the West and Northeast are likely to have a higher percentage of multiracial people (because of racially diverse populations and higher intermarriage rates), than smaller cities or rural areas in the South or Midwest, which have less racial diversity and lower rates of intermarriage.

Racial groups with relatively small populations tend to include a higher percentage of multiracial people partly because they have fewer potential marriage partners within their own group and higher rates of interracial marriage.³ American Indians, for example, make up only about 1 percent of the U.S. population, and they have a long history of intermarriage with non-Indians, especially whites. About 40 percent of American Indians reported at least one other race (usually white) in the 2000 Census. Asian Americans, who are about 4 percent of the U.S. population, also have a substantial multiracial population. In 2000, 14 percent of Asian Americans reported at least one other race (usually white), compared with 5 percent of blacks and 3 percent of whites.

Children are generally more likely to be multiracial, especially among some racial groups. A majority of Asian American adults were born abroad, for example, and immigrated to the United States after 1965, when restrictions on immigration from Asia were relaxed. Intermarriage among Asians and whites (and other groups) has been quite high in recent years, especially among native-born Asian Americans.⁴ The children of these fairly recent unions make up the bulk of the multiracial Asian Americans counted in the 2000 Census. About 24 percent of Asian American children were identified as multiracial in the 2000 Census, compared with 10 percent of Asian Americans age 18 or older. This age difference is not as common among multiracial American Indians and blacks, however, reflecting the long history of intermarriage between American Indians and other races and low intermarriage rates among blacks.

The relatively large number of Asian Americans and American Indians who marked more than one race in the 2000 census makes it difficult to calculate the exact size of these groups and to measure their growth or decline since 1990. The American Indian and Alaska Native population could be as low as 2.5 million or as high as 4.1 million, depending on how the multiracial American Indian population is classified. In some geographic areas with high rates of intermarriage, the single-race counts of the American Indian and Asian American populations may show declines because of the new multiracial categories.

Although the new categories and combinations may cause confusion among data users, the multiracial reporting did not substantially affect the relative distribution of the main racial populations. These differences may increase as the multiracial population grows, however, creating additional challenges for demographers, journalists, and anyone else using the census data on race. The Office of Management and Budget has issued provisional guidelines for federal agencies reporting racial and ethnic population data.⁵ Other consumers of census data will probably use these guidelines as well, but they will also be able to display the data most relevant for an individual task: The total number of people who chose African American (alone or with another race), for example, vs. people who identify themselves as African American and nothing else. The mission and goals of organizations using race data, and the organization's clientele also determine the way data are tabulated.

The new options for reporting race are a challenge for data users, and it is not clear now what standard categories will be used for race and ethnicity in coming decades. But if the "check all that apply" option requires U.S. society to reconsider what race is, this is an appropriate result. After all, race is best understood as a social construction, subject to changing social and political influences for its meaning and measurement.

Adapted from Sharon M. Lee, Using the New Racial Categories in the 2000 Census, KIDS COUNT/PRB Report on Census 2000 (PDF: 320KB), a joint publication of the Annie E. Casey Foundation and the Population Reference Bureau (March 2001).

References

1. Katherine K. Wallman, Suzann Evinger, and Susan Schechter, "Measuring Our Nation's Diversity: Developing a Common Language for Data on Race/Ethnicity," American Journal of Public Health 90, no. 11 (2000): 1704-08.
2. Reynolds Farley, "Racial Identities in the Census of 2000: What Did Respondents Do When They Had the Opportunity to Identify With Multiple Races?" Paper presented at the Jerome Levy Economics Institute, Bard College, Conference Multiraciality: How Will the New Census Data be Used? Sept. 22-23, 2000; and Jorge Del Pinal, et al., "Reporting of Two or More Races in the 1999 American Community Survey." Paper presented at the Jerome Levy Economics Institute, Bard College, Conference Multiraciality: How Will the New Census Data be Used? Sept. 22-23, 2000.
3. Barry Edmonston and Sharon M. Lee, "The U.S. Multiracial Population: Theoretical and Research Issues," Working Paper 00-05 (Portland, OR: Population Research Center, Portland State University, 2000); and Joshua R. Goldstein and Ann J. Morning, "The Multiple-Race Population of the United States: Issues and Estimates," Proceedings of the National Academy of Sciences 97, no. 11 (2000): 6230-35.
4. Sharon M. Lee and Marilyn Fernandez, "Trends in Asian American Racial/Ethnic Intermarriage: A Comparison of 1980 and 1990 Census Data," Sociological Perspectives 41, no. 2 (1998): 323-42.
5. Office of Management and Budget, Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity (December 2000). Accessed online at www.whitehouse.gov/omb/inforeg/index.html#SP and at www.whitehouse.gov/omb/bulletins/b00-02.html on May 12, 2001.

Box 2 Return to text
Public and Private Uses of Census Data

The census provides more detail for smaller geographic areas than just about any other data source. People use these data for a variety of activities in the public and private spheres � often in conjunction with data from administrative records or surveys. In interviews with the Population Reference Bureau's Bingham Kennedy, Jr., April 2001, Barron Holmes of the South Carolina Kids Count Project, and Ken Hodges, Claritas Inc., offered their perspectives and experiences using census data.

Barron Holmes, Kids Count Project Director, South Carolina State Budget Control Board, Columbia, S.C.

My work involves understanding and reporting the condition of children in South Carolina. The census helps us do that in several ways. The first thing is that when you're working with census data, you don't have to worry about sampling error. So that's an advantage that it has over surveys. In addition, the census data gives you information about the subcounty level. That helps us understand things like poverty and family situations � the number of one-parent families, for example � in a very local kind of way.

It's less helpful when we're trying to understand things like the educational levels of parents. The census will tell you how many people between the ages of 25 and 29 have less than a high school degree, but it won't tell you whether or not they have kids. And � there's no way to match up kids with fathers � .

�The information that comes out of the census on things like disability and cohabitation rates tends to be less accurate than what you would get from a professional survey. That's simply because the census relies on people to fill out forms, rather than responding to questions from a professional survey-taker.

Ken Hodges, Director of Demography, Claritas Inc., Ithaca, N.Y.

Claritas provides marketing information resources for business applications. The company was founded after the 1970 Census, which was the first to make data available to the public in computer-readable form. The census provides the small-area data needed for business applications, but the census tape files were huge and contained far more information than most companies needed. The initial role of the private suppliers was to provide value-added access to census data. Only from a private supplier could a company acquire selected variables for selected areas � such as income for selected census tracts. We added further value by aggregating small-area data to areas relevant to business applications, such as ZIP Codes or a three-mile radius around a store location.

Since those early years, the value-added component has grown immensely. We supplement the census with small-area demographic estimates, estimates of consumer demand, and industry specific data resources. Lifestyle cluster systems link small-area demographic data with consumer behavior and media usage surveys to provide powerful consumer segmentation applications.

Our customers include businesses in a wide range of industries, including retail, financial services, media, advertising, automotive, telecommunications and health care. Increasingly, our products are tailored to the specific needs of these industries, and are disseminated through sophisticated desktop and online retrieval systems. But whether the applications involve site evaluation or consumer segmentation, small-area data remain a critical element, and the census remains an indispensable source of small-area data for these applications.

Box 3 Return to text
The 2000 Census Count: Exceeding Expectations?

Editor's note:
Demographic analysis conducted after the 2000 Census suggested a net overcount of about 1.8 million, or 0.7 percent. Yet a report of the Census Bureau's Accuracy and Coverage Evaluation Survey (A.C.E.) released in March 2001 suggested that the 2000 Census count was too low and had missed a net total of 3.3 million people (1.2 percent). As of May 2001, the Census Bureau had not settled the issue of whether the count was higher or lower than the actual population. Another report on the census undercount is expected in late 2001.

In the text below, University of Michigan demographer Reynolds Farley discusses the possible reasons why the census count may have exceeded the estimates.

On December 28, 2000, the Census Bureau announced an unexpectedly large count from the 2000 Census. The 281,422,000 residents enumerated were 6.9 million more than expected for April 1, 2000. The census count exceeded estimates in every state and the District of Columbia. In Washington, D.C., the count was 12 percent greater than the estimate, in Nevada, 7 percent greater, and in Rhode Island, 6 percent greater. In an additional 26 states, the count was more than 2 percent above the estimated population (see map).

In contrast, the population counted in the 1990 Census exceeded the Census Bureau's estimate in just 15 states and it fell at least 1 percent short of the estimates in 26 states.¹ Subsequent analyses showed that the 1990 Census had a net undercount (the number missed minus the number counted more than once) of about 5.25 million people � or 1.8 percent.²

Did the 2000 Census eliminate the net undercount that has plagued previous U.S. censuses?³ The Census of 2000 appears to have done an excellent job enumerating the population. The Census Bureau conducted an effective advertising campaign and especially targeted populations known to be difficult to count, including people in inner cities, the poor, minorities, and people with limited English-speaking ability. The percentage of households that mailed back their census forms rose from 65 percent in 1990 to 67 percent 2000 � again, higher than expected.⁴ The Bureau also had sufficient funds allocated by Congress to hire enough enumerators to visit the households that did not send back forms. When recruitment lagged, the Census Bureau was able to quickly raise the enumerator wage rate. In the San Francisco area, it eventually went above $20 per hour.

The Census Bureau also may have been more effective in 2000 than previously in reaching immigrants and people with limited English skills. Thousands of immigrant and community organizations developed partnerships with the Bureau, in part to demonstrate their large numbers to local government officials. The census form was distributed in six languages in 2000 in contrast to only Spanish and English in 1990.

But the better response rates and coverage do not explain why the count was so much larger than estimated. The 2000 Census count was 2.6 million higher than the estimated population even after adjusting for the undercount in 1990 and projecting the adjusted count forward to April 1, 2000 based on recorded births and deaths and estimated net immigration. There are several possible explanations. First, the Census Bureau may have underestimated population growth during the decade. While births and deaths are almost completely enumerated, it is possible that current demographic procedures overestimated emigration from the United States and underestimated undocumented immigration.

A second possible explanation is that the Census Bureau estimate of the net undercount in 1990 was too low. This seems unlikely. There are two methods to determine how many people were missed or counted more than once in the census: demographic analysis and the Post Enumeration Survey (PES). Demographic analysis takes into account births, deaths, and net migration in the decades before the census to estimate how many "should" be counted at each age and in each race. Using this procedure, 1.85 percent of U.S. residents were missed in 1990.⁵ The second method, the PES, is based on the careful and thorough revisiting of a sample of housing units included in the census to ascertain who was missed and who was counted at two or more locations. For 1990, this procedure estimated a net undercount of about 1.6 percent (plus or minus 0.2), about the same as the demographic procedure.⁶

Third, it is possible the population was overcounted in the Census of 2000. Preliminary results of demographic analysis suggest the census count was 1.8 million too high � yielding a net overcount of 0.7 percent.⁷ Evaluation of the 1990 Census found that certain groups were overcounted, even though there was a net undercount nationwide. The white population was overcounted in five states, for example.⁸ At the same time African Americans consistently were undercounted. The count for African Americans was most complete in Alabama and Michigan, which still had net undercounts of 3.3 percent in 1990. At the other extreme, 8 percent of blacks living in Arizona, California, Colorado, and Nevada were missed by the 1990 enumeration.

Whether the census overcounted or undercounted the total population, the issue of the differential inclusion of specific population groups must still be addressed. Minorities, children, the urban poor, and other population groups that are consistently undercounted may plausibly claim that their share of representation in elected bodies or their fair share of federal spending allocated according to population was diminished by the undercount.

This key constitutional issue was discussed by Justice Sandra O'Connor in her majority decision in a Supreme Court case, Department of Commerce v. U.S. House of Representatives in 1999.⁹ Writing for a 5 to 4 majority, Justice O'Connor observed that throughout recent history, minorities, children, renters, and other specific groups were undercounted in the census. She then described sampling and demographic procedures that would likely produce more complete counts.

Justice O'Connor concluded that the constitution requires an actual enumeration of the population for determining congressional representation. But, the Supreme Court decision neither called for nor prohibited the use of adjusted census data for drawing congressional or other legislative districts within states, or for allocating federal spending to local governments. This allows legislatures to use either actual census counts or adjusted counts to draw congressional and other legislative districts. States could use adjusted data effectively to level the playing field � local areas that were overcounted would lose representation and allocations while those that were undercounted would gain. And when appropriating federal funds, Congress could use adjusted census data to make sure that net over and undercount do not send excess dollars to some localities while shortchanging others.

There is continuing controversy about the quality of this census. To rectify long-standing social inequities that spring from a differential undercount, state legislatures need census data adjusted for net census over- or undercount. However, the Department of Commerce and the Census Bureau will not release adjusted data until they complete further studies of the quality of the enumeration. The City of Los Angeles and other local governments have gone to federal courts demanding immediate release of the best estimates of adjusted census counts but, thus far, have not obtained those data. Quite likely the federal courts and, perhaps, the Supreme Court, will adjudicate this issue in Summer 2001.

Adapted from Reynolds Farley, "The Unexpectedly Large Census Count in 2000 and Its Implications," Research Report 01-467 (Ann Arbor, MI: University of Michigan Population Studies Center, 2001).

References

1. Margo J. Anderson and Stephen E. Fienberg, Who Counts? The Politics of Census-Taking in Contemporary America (New York: Russell Sage Foundation, 1999): table 6.1; and U.S. Census Bureau. Accessed online at www.census.gov/population/cen2000/tab04.txt on Dec. 28, 2000.
2. U.S. Census Bureau. Table accessed online at www.census.gov/dmd/www/90census.html on May 22, 2001.
3. Harvey Choldin, Looking for the Last Percent (New Brunswick, NJ: Rutgers University Press, 1994): 11-13.
4. U.S. Census Bureau. Tables accessed online at www.census.gov/main/www/cen2000.htm, on April 21, 2001. Barry Edmonston, "The 2000 Census Challenge," PRB Reports on America, no. 1 (Washington, DC: Population Reference Bureau, 1999): 4-5, cites a predicted mail-back response of 55 percent.
5. Gregory J. Robinson, Bashir Ahmed, Prithwis Das Gupta, and Karen A. Woodrow, "Estimation of Population Coverage in the 1990 United States Census Based on Demographic Analysis," Journal of the American Statistical Association 88 (September 1993): 1061-71.
6. Howard Hogan, "The 1990 Post-Enumeration Survey: Operations and Results," Journal of the American Statistical Association 88, no. 423 (September 1993): 1054.
7. U.S. Census Bureau, "Report of the Executive Steering Committee for Accuracy and Coverage Evaluation Policy" (March 1, 2001): 15. Accessed online at www.census.gov/dmd/www/EscapRep.html on April 30, 2001.
8. Ibid.: Table 3.
9. Department of Commerce et al. v. United States House of Representatives et al. 1999, No. 98-404.

Box 4 Return to text
Could We/Should We Replace the Electoral College?

The framers of the U.S. Constitution were visionaries who created a legal foundation for a new form of government. Incredibly, the document they created has stood the test of time exceedingly well for more than two centuries. The system for electing a national president was a compromise they hammered out to limit federal power and maintain the power of states. The U.S. president would be selected by electors from each state, not by direct vote of the people. The system was meant to ensure that small states would always have a say in national elections, and many argue that presidential candidates today would bypass small states altogether if they did not need their electoral votes.¹

But many Americans now seem to feel either that the Electoral College was a mistake or that it has outlived its usefulness. A Gallup poll conducted after the 2000 presidential election found about three in five Americans, 61 percent, in favor of abolishing the Electoral College and replacing it with a direct popular vote.

Why do so many people want to scrap the Electoral College? One reason is that it challenges our notions of fairness. It allots each state a number of electors equal to the number of its representatives plus its senators, creating a bias in favor of states with small populations. Although this bias is well-known, the extent of bias introduced by this system is probably not widely appreciated. Each of Wyoming's three electors represent some 122,000 members of the voting-age population, according to the 2000 Census. By contrast, New York's 31 electors each will represent a voting-age population of some 461,000 people. Thus, the Electoral College gives each voter in Wyoming nearly four times as much weight as a voter in New York.

The problem is broader than the extreme cases of Wyoming and New York: In six small states and in the District of Columbia, each elector represents fewer than 200,000 potential voters, while in 14 large states each elector represents more than 400,000 using the new apportionment figures. This disparity, according to Lawrence Longley, co-author of The Electoral College Primer 2000, departs from the one-person, one-vote rulings of the U.S. Supreme Court in the 1960s. These rulings mandated that each congressional district represent essentially the same number of citizens.

If the Electoral College is unfair and obsolete, what system should replace it? Seven Gallup surveys dating back to 1966 show that most Americans believe the president should be elected by popular vote. France, Mexico, and many other countries elect their leaders by direct popular elections, although Canada, the United Kingdom, and several other parliamentary systems do not.

It would be nearly impossible to get rid of the Electoral College, however, because doing so would require amending the Constitution. Traditionally, amending the Constitution requires a two-thirds majority of both houses of Congress plus ratification by three-quarters of the states. Obtaining approval of three-quarters of the states would be the harder of the two, because small states are clearly advantaged by the current system. And there are enough small states to insist on retaining the status quo.

Twelve small states plus the District of Columbia have either four or fewer electoral votes. Five other states have just five electoral votes each. These states had only 7 percent of the national voting-age population in 2000, yet any 13 of them could block any change in the Electoral College.

With most potential amendments � for example, a flag burning amendment or a balanced budget amendment � small states may share the same interests as large states. But in the case of the Electoral College, the small states have a clear interest in protecting their advantage in electing the president.

Adapted from Theodore D. Fuller, "A Demographic Perspective on Replacing the Electoral College," Population Today, February/March 2001.

References

1. Ken Gormley, "Election 2000 Forum: Electoral College-Keep It," Post-Gazette. Accessed online at www.postgazette.com/election2000/
20001112edgorm7.asp, on April 20, 2001; and "Editorial: Keep the College," Waco-Tribune Herald. Accessed online at www.wacoaccess.com/
auto/feed/news/opinion/, on April 20, 2001.

Box 5 Return to text
Media Coverage of the 2000 Census

Steve Doig, Professor and Knight Chair in Journalism, Walter Cronkite School of Journalism and Telecommunication, Arizona State University, was interviewed about 2000 Census media coverage by the Population Reference Bureau's Bingham Kennedy, Jr., on April 17, 2001.

PRB: How does the media coverage of the 2000 Census compare with coverage of previous ones?

Doig: There's been much wider interest among the media in the 2000 Census. I was very familiar with the 1990 Census, which I covered while working at the Miami Herald. Back then, I thought there would be widespread competitive rush to get the good stories coming out of the census, but there wasn't in 1990. The only news organization besides Knight-Ridder papers like the Miami Herald that was aggressive about using the 1990 Census data was USA Today, with its interest in demographics and so on. I finally realized that part of the reason was that the computer revolution was just getting underway in news organizations. There weren't that many people who were proficient at using computers, the equipment was much harder to work with, the software required more programming skills, and things like that. So there were maybe half a dozen reporters who were actually ready to do big stories on the census back then.

�There are now hundreds of news organizations � not just newspapers, but TV and radio stations, too � that are doing major census stories. Several things happened to spark such widespread interest. One, computer-assisted reporting has really spread across the industry: The idea of reporters using social science methods to do much more precise stories is much more of a reality these days. Also, the equipment is much easier to work with � back in 1990, I had to use a mainframe computer to work with census data.

�The Census Bureau has really gone out of its way to make its data more accessible this time. Not only have they put it on the Internet with American FactFinder, which makes it easy for even very small papers to work with the data, but you can also get the bulk data on CD-ROM.

At the same time, we reached a critical mass of computer-assisted reporting in newsrooms around the country. Given the competitive nature of reporters, once it became apparent that the census was a story that some of us were interested in, nobody else dared ignore it. And so everyone said, "Oh my god, we have to be ready!" There was also the realization that, because of the heightened skills, that everybody would have to do it on deadline, too. So there's really been an incredible improvement in the quality of the stories that can be done with each wave of data � .

The Los Angeles Times had an eight-page special section that was out the morning after the California data came out. You have to be in the news business to appreciate what a testament that is to planning, flexibility, and commitment by the higher management that this is an important story. Not every news organization can match that kind of production, but papers of all kinds of sizes have made really wonderful use of the information.

PRB: Where do you think there is room for improvement in the media's coverage of the census?

Doig: It will be interesting to see how well they learn to go back to the data when other stories come along. Right now, the data is coming out, and we're treating that as an event � pulling stories out of it and telling those stories.

I think the best use of the data will be when other kinds of stories need to be told, and the reporters hopefully will be familiar enough with the demographic data that they will see it as another source, another way of adding precision to all those other stories.

Another thing that journalists certainly can do better is that we can learn more about some of the good statistical tools that academic researchers use. We won't turn journalists into academics, but greater use of these statistical tools can help keep the media from drawing conclusions that aren't there and other pitfalls that journalists sometimes fall prey to when you need to tell a story right away.

PRB: What kinds of stories do you think can really benefit from greater use of census data?

Doig: Any of the issues that revolve around change, whether its growth, ethnic tensions, education, whatever. Journalists have known forever how to tell those stories anecdotally � but it really adds credibility to the story to show that the specific examples are not isolated instances but part of a broader pattern. If we can show that the pattern is there using census data, it really adds credibility to the journalism.

PRB: What are the main difficulties that remain for journalists in reporting census data?

Doig: The Census Bureau in releasing the data isn't really thinking of journalists as their primary audience � and no one expects them to think of journalists that way � so it is often difficult to do a story on very short notice on data that is coming out on any given day.

There are also a variety of technical problems in dealing with census data. Once you get to the sample data, for example, you have to make sure you understand things like error margins. You don't want to make a big thing out of a small thing that is actually within the margin of error.

One other thing to keep in mind is that there are very few journalists today who make use of the full-range of Census Bureau products. There are a lot of information products that come out between censuses, such as the Current Population Survey or Census of Agriculture. These are every bit as rich in information as the census about the communities that we cover, but most journalists are only dimly aware that these things exist, if they know about them at all. Maybe one of the good things about all this coverage of the 2000 Census, and the interest that readers have shown in it, is that smart reporters will start turning their skills in computer-assisted reporting to making better use of these other products and more complicated data.

Box 6 Return to text
The Legacy of the 2000 Census

Kenneth Prewitt was director of the U.S. Census Bureau, October 1998 to January 2001, and oversaw the execution of Census 2000. He is currently dean of graduate faculty of political and social science at the New School for Social Research in New York. Prewitt was interviewed by the Population Reference Bureau's Bingham Kennedy, Jr., on April 28, 2001.

PRB: What are the implications for social policy of the opportunity to register more than one race on the census? What about the implications for how we think about race in America?

Prewitt: I think that's the most important thing that happened in the 2000 Census. I think when the historians write about Census 2000 in 70 or 80 years, the sampling debate will be a footnote, the improved coverage will be noticed, but the books will be about the multiple race item.

From 1790 to 1990, the Census Bureau collected race data in terms of a small number of discrete categories. Throughout that entire 200-year history, a lot of social policy was based upon those discrete categories. From 1790 to roughly 1960, the social policies were essentially detrimental to racial minorities, that is they were first about slavery and sustaining the slave system, and then in the late 19th century, they became about trying to establish the principles of racial superiority and inferiority.

Then in 1960, those policies shifted 180 degrees. �You get equal opportunity programs, quota programs, affirmative action programs, the Voting Rights Act � a whole series of public policies that are trying to right some wrongs that have been in place for the entire nation's history. The classification system was still important because you did this with the basic principle of statistical proportionality. If there is 12.5 percent of the population that is African American, but they make up only 2 percent of the college population, then something is wrong.

That's not going to work anymore. First, once you go to 63 race categories (or 126 if you include Hispanic origin), there is no upper limit. You cannot design public policy that depends upon a racial classification system when the number of categories is just exploding like that.

The other thing that is going to happen is that the scientific community � which already knows it's hard to measure race and that it doesn't mean anything biologically or anthropologically � is going to get more restless with even the attempt to measure race. And I think the public will get more restless.

I think the multiple race issue is a tremor in 2000 for what is going to become a political earthquake that is coming down the road. It will take a while, but if this works its way through the system, the number of categories will continue to expand. �The pressure to expand will be intense, and the statistical agencies will accommodate that pressure.

PRB: What do you see as the upside and the downside to this development?

Prewitt: The downside is that there is still discrimination in this country. We have now weakened � and I think will eventually eliminate � one of the tools that our political process has to correct for discrimination � .

The upside is that it is very healthy for a country to be beyond placing people into discrete categories. The fact that we're going to have to dismantle and construct other ways to deal with discrimination and lack of opportunity for certain individuals is no easy task, but it was not easy to do the affirmative action and civil rights work in the first place � .

In the long run, if we quit thinking in racial categories, that is nothing but good, nothing but a positive development for our country. We got multiple race responses from 2.4 percent of the population this time, which is a reasonable level. But I think it will grow enormously as older people who never [identify themselves as multiracial] die off and younger people who are more comfortable with it take their places � . It will also grow because it has now been legitimated. �I don't know where it will level out, but in 15 to 20 years, I would expect to see about a quarter of the population identifying themselves as multiracial.

PRB: Because the implications are so big, do you think some people with a stake in the old system will push to go back to the old system or push to make multiracial a discrete category?

Prewitt: Absolutely. I think there will be a push in that direction and I think it will be unsuccessful. I just don't think you can put this genie back in the bottle. It will now be thought of as racist. Trying to put all multiracial people into one category will cause people to say things like I'm Indian-Black and this person is native Hawaiian-Japanese and we belong in different groups.

Box 7 Return to text
The American Community Survey

The census provides a snapshot of the U.S. population once every 10 years. But the few questions that are asked of all Americans � age, sex, race, and a handful of other items � can only draw the outlines of the picture. The rich details emerge from the "long form" questionnaire that goes to between 15 percent and 25 percent of households, depending on the community. While some people complain that the long form is too long and intrusive, those who use these data maintain that the data are collected too infrequently, especially considering the nation's rapidly changing demographics. The U.S. Census Bureau has been developing an alternative to the long form that will collect information more frequently, with a better response rate. Assuming continued funding from Congress, the Census Bureau plans to update the picture that emerges from the 2000 Census with an ongoing American Community Survey "video."

The American Community Survey (ACS) is slotted to replace the decennial census long form by collecting essentially the same information throughout the decade rather than once every 10 years. With updated information, local officials and residents will be able to track changes in their communities. In addition, the ACS will allow researchers to compare differences among communities and over time.

The Census Bureau has been developing the ACS since 1996, when it was tested in just four sites. The ACS has since expanded to 31 diverse sites, collecting data that can be compared with Census 2000 results. These comparisons are laying the groundwork for the transition from the long form to the ACS. The plan is to fully implement the survey in 2003 and continue data collection every year thereafter.

The basic design of the survey is self-enumeration through mail-out and mail-back operations in every county to a representative sample of about 3 million addresses (households and group quarters) across the country. The questions are essentially the same as those on the decennial census long form. Follow-up of those who do not mail back their forms is conducted first by telephone and then through face-to-face interviews with one-third of the remaining nonrespondents.

The results have been encouraging. Effective final response rates for the ACS have been about 97 percent even during the Census 2000 period when response rates for the census long form dropped. ACS field staff work full-time in their areas and can explain to local communities how the ACS benefits them. Because of this, the ACS has not encountered the kind of resistance from respondents that the census long form has engendered.

Since 1996, Census Bureau staff have been meeting with users of the decennial census long form to ensure that the new survey will meet their needs. Among these is a need for reliable data on small areas. The ACS accumulates samples for small areas such as rural areas and census tracts (statistical subdivisions of counties) over multiple years to provide data quality similar to that obtained from the long form. For areas with fewer than 20,000 people, collecting a sufficient sample to provide reliable data will take five years. The first such data release is planned for 2008, and the five-year "moving averages" will be updated every year thereafter to provide, for the first time, the ability to track trends for small areas. Data users have testified to Congress that such updated multiyear averages will show general trends that are more valuable than out-of-date long-form data.

Information updated every year also will open new possibilities for using data. Researchers are considering how the updated trends provided by the ACS can be used to improve needs assessment, predictive models, and estimates of characteristics such as disability and poverty. Geographic Information Systems (GIS) can use the current population and housing information to help community officials monitor and evaluate programs.

As an ongoing survey, the ACS is a flexible vehicle, capable of adapting to changing customer needs. Once it is fully implemented, it can be expanded to accommodate questions of national policy interest and even specialized supplements to help identify the characteristics of special population groups.

Adapted from an article by Cynthia Taeuber, U.S. Census Bureau, in Population Today (Washington, DC: Population Reference Bureau, November/December 2000). Detailed information about the American Community Survey is available on the Census Bureau's website at www.census.gov/acs/www/.