The first census (under the direction of Secretary of State Thomas Jefferson) was taken in 1790 by 16 federal marshals (one for each colony and the three territories). Field marshals of the U.S. judicial districts were the field supervisors for the first nine censuses (through 1870), while deputy marshals served as the enumerators.

As 95% of our population at that time lived in rural areas (often with little or no access), collecting the information was quite difficult for the marshals. Often they had to use word-of-mouth to learn of households out in the country. Additional problems included lack of cooperation due to suspicion of the questions, for various reasons. Some of the marshals were tarred and feathered. This nine-month task took 18 months. The final tabulation in March 1792 counted 3,929,326 in the United States. As seen in the Constitution excerpt at the top of the article, this count excluded Indians not taxed, and counted slaves as three-fifths of a free person. When Jefferson sent a copy of this final count to President Washington (the original census lists can be seen on microfilm at the National Archives), they both agreed that the count was incomplete. Jefferson's estimate of the actual population was close to 4,100,000. Blame was placed on citizens for lack of cooperation (many thought the purpose of the census was for taxation), and also on some of the enumerators (census-takers) for laziness. On the whole, they felt that this census was a success, and proof to the European countries of the importance of this new nation.

This first census of 1790 included more than just the number of free people and slaves. It also included the sex of free persons, as well as who was the head of the household, and the age of white males over 16. To some in Congress, this first census was felt to be a great opportunity that was wasted. They wanted to be able to collect much more information on this new nation to help determine how best to govern the people. James Madison (father of our Constitution) proposed to include the occupation of all those employed, as well as the age of all free people. However, some others in Congress felt this to be ineffective on a cost basis. And still others felt all this information gathering to be an invasion of privacy to its citizens. This debate being unresolved, the second census, taken in 1800, asked the same questions as the first.

Soon, however, others came around to Madison's point of view. The next census in 1810 (when Madison was president) began to collect additional information. A census of manufactures began with the 1810 census. In 1850, a change was made from taking the name of just the head of the household, to recording every person's name, along with their age, martial status, race or nationality. As the country was in the midst of the first great wave of immigration, these statistics were later used to determine immigration quotas, something our founding fathers had never envisioned.Beginning in 1880, census supervisors were appointed to each census district to be responsible for the enumeration. The temporary field offices that had been used were tossed in favor of a census office. This was the genesis of the census practices that are used today. In 1890, a new machine was used to help in the taking of the census. The year before, Herman Hollerith, a Census Bureau employee, developed the Hollerith Electric Tabulating Machine, using punch holes in cards to record the information. Hollerith later used this invention to start his own company, which later became known as International Business Machines (IBM). In 1902, Congress enacted the Permanent Census Act, which created the Bureau of the Census.

In addition to the Hollerith invention, the Census Bureau has implemented many other improvements to conducting the task of the various censuses. The development of sampling techniques early in the 20th century made it possible, beginning with the 1940 census, to ask some of the census questions of 5 percent of the population to yield reliable estimates for most of the 1940 census areas. Thereafter, the proportion of the households included in the sample varied from one census to the next. In 1970, several questionnaires were used, resulting in three separate samples - 5%, 15% and 20%. In 1990, and again for Census 2000, the sample questions were asked of every sixth household. In places with less than 2,500 inhabitants, every second household was sampled, while more heavily populated areas were sampled at either a 1-in-6 or a 1-in-8 rate.

A major change in how the census is conducted was made in 1960. For the first time, questionnaires were mailed out to households, with the instructions to fill them out and held until an enumerator picked them up. This was called self-enumeration. The thinking was that with more time to fill out the survey, greater accuracy would be achieved. This would also save time and money, as the enumerator would be taking the time to help only those who needed it. This process worked so well that the Census Bureau changed to a mail-out/mail-back system in 1970. Enumerators would now only be needed for those who did not fill out the questionnaire, or those who needed help. For Census 2000, approximately 95 percent of the households received and were asked to return questionnaires in the mail. For households that received a questionnaire in the mail, enumerators telephoned or visited only in those cases where the questionnaires were not returned, the data were incomplete, or where information was needed for people living in institutions, dormitories, or other such quarters. There also were special procedures for enumerating the homeless, people on maritime vessels, and members of the Armed Forces. In a few remote or sparsely populated areas, enumerators visited each housing unit and completed a short-form or long-form census questionnaire.

The principal topics listed below indicate how the range of the population censuses grew during the past two centuries: In the 20th century, interest focused as well on people's economic characteristics - their jobs and how they traveled to work, their income, and how well they were housed. Most of these questions have continued to be asked on a sample basis.

The concept of color or race in the censuses has never denoted any scientific definition of biological stock. "White "and "Black "persons have been identified in every decennial census since 1790. American Indians were first enumerated as a separate group in the 1870 census; however, until 1890, those in the Indian Territory or on reservations were not included in the official U.S. population count used for congressional apportionment. Data have been collected on the Chinese population since the 1870 census, and on the Japanese beginning in 1890. Until recently, the census taker determined a person 's color or race according to the Census Bureau 's guidelines. Beginning with the 1960 census, however, respondents who completed their own census questionnaires were able to classify themselves and other household members. Census 2000 made provision for each person 's race to be marked as White, Black, African American, or Negro, American Indian or Alaska Native, Asian Indian, Japanese, Chinese, Filipino, Hawaiian, Korean, Vietnamese, Guamanian or Chamorro, Samoan, or to be written in if none of these applied. In addition, American Indians were asked to report their tribe. The Hispanic population, which the Census Bureau recognizes as an ethnic group, now is identified primarily by a question on Hispanic origin or descent (Mexican, Mexican American, or Chicano, Puerto Rican, Cuban, or other Spanish/ Hispanic/Latino), although other measurements are also available in terms of language spoken at home. Prior to Census 2000, respondents were asked to select only one category when identifying their racial group. With the Office of Management and Budget 's (OMB) decision in July 1998 to revise its standards for the classification of Federal data on race and ethnicity (according to Directive 15), however, the Census 2000 questionnaire allowed the respondent to mark one or more races depending on what that person considered himself/herself to be.

To learn more about the history of the United States Census, view the document put out by the U.S. Census Bureau entitled "Measuring America: The Decennial Censuses from 1790 to 2000"

The law requires that the Census Bureau release the final population counts by State by the end of the calendar year in which the census is conducted. This release allows for the determination of the number of Congressional seats that will be allocated to each State. The next legal deadline is April 1, one year after the Census is conducted, when population counts by age (18 and over), race and ethnicity must be released to the census block level for states to begin the process of redistricting their allotted congressional districts to comply with the one person/one vote requirement.

Once the deadlines are met, the Census Bureau begins to release the full range of decennial producets. The majority of scheduled products are released over the following three years. The following sections of the census users guide detail some of the changes from previous censuses and highlight the key data categories.

Many data subjects found in the 2000 Census can also be found in the 1990 and previous censuses. However, caution must always be used when examining temporal changes. One should familiarize himself/herself with the census questionnaires to make sure that the data one is comparing are collected in a similar manner. The Census Bureau often changes the collection categories or universe surveyed, thus resulting in changes in the data reported. Numerical changes over time (i.e., increase in persons of Hispanic origin; changes in ancestry counts, etc.) may be due to a change in procedures (question placement or wording), rather than a true demographic change. The ancestry question provides a case in point. Always asked on a sample basis, the question asks respondents to write in their ancestry or ethnic origin on a blank line. While the census includes an information booklet to assist respondents understand the intent of the questions, there is additional assistance in this question with examples of possible answers provided below the line. In 1990 the Census Bureau included Cajun as one of these examples. 1990 results showed a ten-fold increase in the country’s Cajun population. As there was no adequate demographic reason for this increase, the next best explanation was that the example provided led to a response bias as respondents, unsure of how to answer, found this to be an acceptable alternative.

In order to assist you in understanding the potential biases created by the 2000 Census, we have created this section as a way of highlighting the most important changes to the census between 1990 and 2000.

The Census Bureau used two surveys to conduct the 2000 Census. The first survey, which is known as the “short form,” was sent to 83% of households in the United States. It consisted of 7 questions covering age, gender, race, and ethnicity (Hispanic origin), household relationship (used to categorize households by type), and housing tenure (owner v. renter). These same 7 questions also lead off the “long form,” or sample questionnaire, and thus cover all persons and households in the United States. Much of the data collected through the short form are required by the Constitution for the purpose of reapportionment and political redistricting. Click here to view the short form (Acrobat Reader required).

The other questionnaire, known as the “long form,” was sent to approximately 17% of all households in the United States. The long form contained the original 7 questions asked on the short form plus 26 additional population questions and 20 additional housing questions. These questions go into more detail about demographics, education, income, employment and housing characteristics. Because only 1 in 6 households receive the long form, the data collected are regarded as sample data because only a sample of the US population completes the form. While not required for redistricting purposes, there are a host of other federal programs that rely on the sample data to determine funding levels and program eligibility. Click here to view the long form (Acrobat Reader required).

The Census Bureau releases the vast majority of its census results in the form of summary data. These represent responses aggregated across people, or households, living in a specific geographic area – be it state, county, city or census block. In other words, summary data will provide you with the number of households in a particular city that have 3 or more vehicles, or the number of persons, 25 years or over, in your county with a bachelor’s degree. The structure and content of the summary data tables are predetermined and cannot be altered. If the Census Bureau has not tabulated the data in a particular way (for instance, poverty rates for persons 25-34 years of age), there is nothing that can be done to change that.

Microdata, on the other hand, are the individual records that contain information collected about each person and housing unit. The Census Bureau uses these confidential microdata in order to produce the summary data that go into the various reports, summary files, and special tabulations. Public use microdata samples (PUMS) are extracts from the confidential microdata taken in a manner that avoids disclosure of information about households or individuals. These files allow the user to produce tabulations that were not available in the summary tables. There is a limitation, however, in that the geographic areas for which PUMS data are available must have a minimum population of 100,000, so as to ensure confidentiality.

The information gleaned from the 2000 Census short and long forms are released in a series of Census Bureau data products over a 3 to 4 year period. The method of release is to first produce products from the 100% questions, and then to follow with products containing the sample items. The following is a partial list of these products, containing brief descriptions and release dates for Michigan.

For more information on the above listed data products and others, please see http://www.census.gov/population/www/censusdata/c2kproducts.html. For information on how to obtain the above listed data, please see the section: How to Obtain Data.

The following is a list of widely used data categories from the 2000 Census. Included in the description is their definition, how to use and interpret the information and how to calculate percentages (if applicable).

Census geography provides the framework for interpreting, analyzing, and understanding census data. The Census Bureau classifies all geographic entities into two broad categories:  1) legal and/or administrative entities, and 2) statistical entities. The geographic areas may work in a hierarchical fashion, with smaller areas nesting in larger ones (e.g., blocks in block groups; block groups in census tracts; census tracts in counties, etc.).  Some areas, like ZIP Code Tabulation Areas (ZCTAs), are not structured to nest (though at times they may nest within a city or county) and are thus given only as subsets of the nation (See Standard Hierarchy).

Legal/Administrative entities generally originate from charters, laws, treaties, resolutions or court decisions. They include:

Statistical entities usually evolve from practice, custom, usage, or need, and generally the Census Bureau develops criteria and guidelines for their identification and delineation. They include:

Data from the 2000 Census are released in three main formats: Internet, File Transfer Protocol (FTP) and CD/DVD.The Census Bureau is moving away from providing printed data tables in publication format as was done in previous censuses. Data users must now rely on digital formats for their data needs. The following covers the three formats and how to access them.

There are a number of other helpful resources available to guide you through the Census maze. Among the best that we have found are the following:

Introduction to 2000 Census Data Products (US Census Bureau)
http://www.census.gov/prod/2001pubs/mso-01icdp.pdf

Census Guide 2000 (University of Michigan)
http://www.lib.umich.edu/govdocs/cen2000.html

Census 2000 at ICPSR (Inter-University Consortium for Political and Social Research)
http://www.icpsr.umich.edu/CENSUS2000/

US Census Resources on the Web (University of Wisconsin-Madison)
www.library.wisc.edu/guides/govdocs/census/2000int.htm

Census Tutorial (Alaska Library Association)
http://daniel.cornwall.home.att.net/census_tutorial.html