Using the U.S. Decennial Census
Why have it? Why use it?
The U.S. Census Bureau has been taking a census of the whole population every ten years since 1790. Its original purpose was to determine congressional representation and to apportion taxes to pay for the expensive Revolutionary War. Congressional apportionment is still the core of its mission, but the Census also helps determine the annual distribution of over 300 billion dollars, and it is an essential tool in monitoring and enforcing civil-rights legislation in such areas as employment, voting rights, housing and mortgage lending, health-care services, and educational opportunities. The Census includes (or attempts to include) all those living in the United States on Census Day (legal citizens, as well as undocumented). It provides an unmatched wealth of socioeconomic data on individuals, families, and households available down to detailed geographic areas -- towns, census tracts, and even blocks.
The questions asked vary each decade (table of all questions asked, 1790-2000), reflecting the concerns of the time. But the the core data include:
Census 2000 and Race/Hispanic identity
Census 2000 made a radical departure from previous decades by allowing individuals to identify as more than one race. This was a tremendously controversial decision, and it makes the 2000 data non-comparable with earlier years and very complex to use (since each person can be up to six races). In addition, each person was asked whether or not s/he were Hispanic -- since the Census defines Hispanic as an ethnicity, not a race. Hispanics can, for example, be Black, White, American Indian, Asian or Some Other Race (about 50% of Hispanics chose Some Other Race, accounting for about 98% of that race category). It is also easy to add up to more than 100% of the population, since a person who is Asian and Black is counted in both categories. There are various ways to work with this data, though the simplest is to use the race groups "alone" rather than in combination, and then have a category of Two or More.
Six major race groups
Again, these can be combined -- any person can have up to six combinations -- or can be summarized as Two or More Races.
Not just six
In addition to the six major race groups (illustration on right) which are used in Summary Files 1 and 3; Summary Files 2 and 4 break the data down further into hundreds of race/ethnicity categories. This is useful, for example, if you want Asians not as a group, but as Cambodians, Koreans, Chinese, etc; or Hispanics broken into Puerto Rican, Mexican, Cuban, etc.
Census 2000 and Geography
Because of the need to have Congress proportionally represent the population, and for civil-rights enforcement against illegal segregation, geography is key to the Census. Each person is counted at one address. The Census then provides many geographic slices of the data. Block is the smallest unit, average 100 people all within one census tract or county. Sample data is not available to the block level. Block Group is a subdivision of a census tract--optimally 1,500 people--and is the lowest level geography for which sample data is tabulated (confidentiality issues arise more often with smaller population groups). Census tracts have an optimum size of 4,000 people. They are supposed to be stable over decades, so they usually follow visible features, but they can also include the boundary of a state or county. Places are either legally incorporated cities or Census Designated Places (downtown Amherst, for example) which are requested by the city for analysis of an area of concentrated population (downtown Amherst, for example.) Metropolitan Areas are core areas containing a large population nucleus together with adjacent communities having a high degree of economic and social integration with the core. Counties are used even if there is no county government. County Subdivisions are used in New England (and a few other states) for towns. Geographic Terms and Concepts provides a more detailed explanation.
Which Dataset to use? How to use them?
Census 2000, though rapidly becoming "old" data, is still the most detailed data available for small geographic areas (towns, small cities, census tracts, and blocks). Two questionnaires were sent out in 2000; 100% of the population received the short form, with an additional sample (approximately 1 in 6) filling out the long form. The short form asks: age, sex, household composition, own or rent home, race, and Hispanic identity. It does not collect economic information. This data is available in Summary Files 1 and 2 (SF1 summarizes race data, while SF2 has more detail). SF3 (summarized race data combined with socioeconomic) and SF4 (socioeconomic and detailed race). See census data choices for more about each dataset, and How to Use Detailed Tables for a cheat sheet.
Census and Poverty
The Census determines poverty for individuals living alone and for households. It also provides poverty information about non-family households, but it isn't meaningful -- the poverty status of the entire household is determined by that of the householder. If the householder is in poverty, then each individual in that unit is in poverty. For families and individuals, the Census Bureau uses a complex formula that takes into account all of the income of the family and the number and age of its members. The poverty threshold for 1999 (used by Census 2000) is here. Note that the poverty threshold does not vary by state, or between urban and rural areas. There is another very useful table, "Ratio of Income to Poverty," which summarizes those living at up to 200% of poverty (eligible for many needs-based social services) and the depth of poverty (those at 50% of poverty).
Poverty as defined by the Census is an absolute, not relative, measure. It is based on an equation from the mid-1960s determining that a family must spend 1/3 of its income to provide minimal nutrition. This basic equation has not been changed much in the intervening years (even though costs such as childcare, housing, and fuel have risen significantly), but it is multiplied by an inflation indicator.
The historic census is an invaluable research tool. Amherst College has the bound volumes for all years in the stacks, and the print volumes provide rich detail; much of the data has also been made available as files online. If you don't need geographic areas smaller than counties, and don't need specialized tables, the Historic Census Browser can be very helpful. Historical Statistics of the United States, Millennial Edition Online is a great index to the subjects covered in each census, as well as non-census historical statistics. Selected Historical Decennial Census Population and Housing Counts are useful reports, looking at particular aspects of the Census over time. And Demographic Trends in the 20th Century outlines the major issues that each census identified...a very useful overview of what you can learn from the census. Social Explorer has some historical data in a useful map format in its freely available version. Finally, if you are looking for information about individuals, it is available 70 years after the census is taken. Ancestry Library provides full images of U.S. Census records up to the 1930 Census, and other resources such as selected newspaper obituaries, immigration records, and military records. Finding enumeration districts for cities can link census data to specific streets or intersections for 1880-1940.
Historical Racial Classification and the Census
|1890||White, Black, Mulatto, Quadroon, Octaroon, Chinese, Japanese, and Indian|
|1900||White, Black, Chinese, Japanese, and Indian|
|1910||White, Black, Mulatto, Chinese, Japanese, Indian, Other (plus write-in)|
|1920||White, Black, Mulatto, Indian, Chinese, Japanese, Filipino, Hindu, Korean, and Other (plus write-in)|
|1930||White, Negro, Mexican, Indian, Chinese, Japanese, Filipino, Hindu, Korean (Other races, spell out in full)|
|1940||White, Negro, Indian, Chinese, Japanese, Filipino, Hindu, Korean (Other races, spell out in full)|
|1950||White, Negro, Indian, Japanese, Chinese, Filipino (Other races, spell out)|
|1960||White, Negro, American Indian, Japanese, Chinese, Filipino, Hawaiian, Part Hawaiian, Aleut, Eskimo|
|1970||White, Negro or Black, Indian (American), Japanese, Chinese, Filipino, Hawaiian, Korean, Other (print race)|
|1980||White, Negro, Japanese, Chinese, Filipino, Korean, Vietnamese, Indian (American), Asian Indian, Hawaiian, Guamanian, Samoan, Eskimo, Aleut, Other (specify); each person can also be Hispanic or Latino.|
|1990||White, Black, Indian (American), Eskimo, Aleut, Chinese, Filipino, Hawaiian, Korean, Vietnamese, Japanese, Asian Indian, Samoan, Guamanian, Other Asian Pacific Islander, Other race; and Hispanic or Latino.|
|2000||White; Black, African American, or Negro; American Indian or Alaska Native (specify tribe); Asian Indian; Chinese; Filipino; Other Asian (print race); Japanese; Korean; Vietnamese; Hawaiian; Guamanian or Chamorro; Samoan; Other Pacific Islander (print race); Some other race. Two or More Races -- any combination of up to six races (57 possibilities); each person can also be Hispanic or Latino (of any race).|
1790-1990 data adapted from Anderson and Fienberg (2000: Tables 3 and 4) and 2000 data from U.S. Census Bureau (2001a). From: Measuring Racial Discrimination, 2004. Also online at: http://www.nap.edu/books/0309091268/html/205.html.
American Community Survey (ACS)
ACS is an ongoing survey conducted by the Census Bureau to provide more timely information about communities between the decennials. It has a different methodology and a smaller sample size -- and it isn't yet available for smaller cities (like Amherst). It is available for counties, and for larger cities like Springfield. By 2010, the "Long Form" will not be included in the census. Some of the detailed socio-economic data that was gathered from the long form and included in Summary Files 3 and 4 will be available through ACS. Some questions may be dropped, and there will also be some new questions. The ACS General Handbook (pdf) is a helpful guide.