• 2002 March CPS (Current Population Survey)
• 2000 1% Decennial Census PUMS (Public Use Microdata Sample)
• 1990 5% Decennial Census PUMS (Public Use Microdata Sample)

In the decennial census, the census bureau collects data in two forms - the long form and the short form. The short form data is available in STF (Summary Tape File) format. While it has the advantages of representing 100% of the population and finely divided geography, it has relatively few data items for each person and household, and it is already summarized by the census bureau.

The long form data is available in an unsummarized form, called the PUMS 5% sample, for "Public Use Microdata Sample". There are also PUMS 1% and 3% samples available.

We are using the PUMS 5% sample because it is unsummarized (we can summarize it according to your needs), it contains the much richer long form data (as opposed to the sparse short form data), and for most analyses, a 5% sample should be sufficient. It is the largest data sample available for long form data. Also, the census bureau weights each person and household, in order to simulate 100%. On the down side, the long form data is not available in finely divided geographies. The census bureau will not divide it into geographies containing less than 100,000 people, in order to protect privacy.

For a list of available fields, the values that each field can contain, and an in-depth description of most fields, see data. More information, such as data collection procedures, sample census forms, discussions of sampling, etc. are available at this university site.