The actual Enumeration shall be made within three Years after the first Meeting of the Congress of the United States, and with- in every subsequent Term of ten Years, in such Manner as they shall by Law Direct. (The United States Constitution, article 1, section 2)
The United States Census is one of the great statistical studies carried out anywhere. Originally, the census began with a simple goal. After the American War of Independence the American states formed a loose association called a confederation. The structure of the new country was described in a document called the Articles of Confederation, and from 1781 until 1789 the new nation struggled
to function with the system of government defined by the articles. The experiment was a failure. Under the Articles of Confederation the central government was too weak to govern.
In 1787 delegates from the various states met to correct the prob- lem. One of the more contentious issues they faced was the prob- lem of representation. States with large populations wanted representatives apportioned on the basis of population. Not sur- prisingly, states with small populations expressed concern that they would be ignored under this system of government. The compro- mise to which the delegates finally agreed included one legislative body, the House of Representatives, in which each state would be represented by a number of delegates roughly proportional to its population, and a second legislative body, the Senate, in which each state would be represented by two delegates regardless of popula- tion. This compromise required fairly detailed information about the size and distribution of the U.S. population, so the men who drafted the Constitution included a section mandating that a
It is difficult to appreciate the scale of contemporary United States Census Bureau efforts. (Courtesy of the United States Census Bureau)
census, or count, of the U.S. population be made every 10 years, beginning in 1790. The U.S. government has fulfilled this consti- tutional requirement every 10 years since 1790 during years of war and peace, prosperity and hardship.
The first census, the 1790 census, set the pattern for succeeding surveys. The census was taken very seriously. Census administra- tors divided the nation into districts and arranged for a count to be made in each district. All reports were made part of the public record. Hefty fines were established for any census worker involved in filing a false report, and any citizen involved in sup- plying false information to a census worker faced a substantial fine as well. Information was to be taken from each person at his or her place of residence. Any individual without a fixed place of resi- dence was counted at the location he or she occupied on the first Monday of August. The final tally was 3,929,214.
The United States has always been geographically large, cultur- ally diverse, and economically complex. Even in the 1790s, many thought that a simple head count did not provide enough infor- mation about the nation to help establish policies. Policy makers wanted to know more, and a few questions about the characteris- tics of the U.S. population were added to the survey. Most of this additional information was not required to decide how to appor- tion representatives; it was extra information that was designed to shed more light on who lived in the United States. For the 1790 census workers counted (1) the number of free white males older than 15 years of age, (2) the number of free white females, (3) the number of slaves, (4) the number of free white males younger than age 16, (5) the number of all other free persons, and (6) the names of heads of households. This was just the beginning. Every 10 years the demand for information about the U.S. population increased. Remarkably, this information was all collected by a small staff and the legions of census takers who were briefly hired every 10 years to collect the information.
The old volumes make for interesting and sometimes uncomfort- able reading. Information about race uses the language of slavery decades after slavery was eradicated as an institution. Nor was race the only equal-rights issue. There was an enumeration of
“imbeciles” and the “insane.” These labels were sometimes used for purposes of repression. Large percentages of racial minorities in some districts were routinely described as insane. To their credit, census officials acknowledged at the time that these data were inac- curate and revealed more about the census taker than the population that was surveyed. Despite their flaws the early censuses reveal a great deal about the characteristics of a rapidly growing nation.
As the nation became larger and more complex, the task of the statisticians became increasingly difficult. This difficulty was not due only to the fact that the nation was getting larger. There was also an ever-increasing demand for raw data and statistical analy- ses of the state of the nation. Policy makers wanted information about the manufacturing sector, the agricultural sector of the economy, and demographics of the U.S. population so that they could better formulate policy.
Even in the late 19th century and early 20th century the amount of data collected was enormous. The United States Census was one of the first institutions to make systematic use of data pro- cessing equipment, the most famous of which is the Hollerith tab- ulator. For the census of 1890, the superintendent of the Census contracted for six Hollerith tabulators. Individual information, such as age, sex, color, and place of birth, was punched into a card. The cards were then run through the tabulator, which read the card and recorded the information. Data processing equipment increased accuracy and efficiency. It also enabled statisticians to identify correlations between various characteristics of the popu- lation more easily. With new technology and better statistical techniques the amount of information that could be gleaned from the raw data collected by the census takers continued to increase. Nevertheless, it barely kept pace with the amount of raw data to be analyzed. For more than 100 years all of this was accomplished by a Census staff that was largely organized for a particular census and then disbanded once the report was written.
By 1900 the job of constructing a statistical description of the United States had turned into an enormous task, and a permanent Census Office was established in 1902. In 1903 its name was changed to the Census Bureau. The demand for information
required that the bureau work continually, taking surveys, analyz- ing data, and publishing the findings.
By the 1920 census the volume of data that was being collected and analyzed was impressive even by today’s standards. It is partic- ularly impressive when one remembers that the computer had not yet been invented. The 1920 census involved a workforce of 90,000 and questionnaires for 107.5 million people, 6.5 million farms, 450,000 manufacturers, and 22,000 mining and quarrying companies. There were 300 million punched cards. The report entailed the calculation of 500,000 “percentages, averages and other rates,” and the publication of 12 large, or quarto, volumes of 1,000 pages apiece. The Information Age began earlier than many of us realize.
The Hollerith tabulator. Designed for the United States Census, it was a breakthrough in electric data processing equipment. (Courtesy of IBM Corporate Archives)
The decennial census is now only one part of the mission of the U.S. Census Bureau, but it is still an important one. Each question that appears on the decennial census, the one census that is mandated by the Constitution, is required as a matter of law. This information is used to manage and evaluate federal programs and to draw state and federal legislative districts. Some of the infor- mation is used to monitor or satisfy legal requirements that have been imposed by U.S. court decisions. It is also routinely used for planning and decision-making purposes by many companies.
Because of the size and importance of the census, the bureau goes to great lengths to collect and analyze information in a time- ly manner. The 2000 census, for example, was published in 49 lan- guages. Some of the languages, including German and French, were predictable. Some of them, such as Chamarro, Dinka, and Ilocano, are less well known. The volume of data has required the bureau to automate the process as much as possible. During the decennial census the bureau makes an electronic image of every questionnaire that is returned. All returned envelopes are auto- matically sorted, and households that did not comply are auto- matically identified so that a census worker can interview the residents. The information collected is fed through data-analysis software that provides a statistical snapshot of the United States that is as complete as possible.
In addition to the decennial census, the Census Bureau man- ages a number of surveys and performs numerous statistical analyses. The bureau publishes a Census of Manufacturers, an American Housing Survey, a Consumer Expenditure Survey, a Survey of Income and Program Participation, and numerous other surveys of interest to economic planners. It is important to realize that each such activity is an application of statistics. The Census Bureau collects massive amounts of data and pro- vides statistical analyses of the data it obtains. The bureau does not use this information; it is not a policy arm of the govern- ment. Instead, it provides information to the many interested governmental and private institutions that need it as an aid in their decision-making processes. This is one of the main reasons statistics are valuable.