The first twelve U.S decennial censuses were taken by temporary organizations created specifically to carry out the provisions of the corresponding census act. During this long period of time in which the censuses were taken by the temporary organizations, there had been demands for a permanent census organization. Such demands must have been triggered by the recognition that it is highly inefficient to assemble and break down a complete statistical organization every ten years(Eckler, 6).
The demand for a permanent Census Bureau was further strengthened by dissatisfaction with the 1890 census results and the time required to publish them. Two major professional associations, the American Economic Association and the American Statistical Association, asked the Congress for consideration of plans for the 1900 census and the possibility of a permanent census office. Hearings on the proposed legislation took place in both houses of Congress in 1897, but in the absence of any conclusion in either house, the whole matter was held over until the next Congressional meeting(Eckler, 9).
After all of the recommendations and efforts over many years toward the establishment of a permanent census organization, the Permanent Census Act was finally passed in March, 1902. Under this Act, Census became a permanent rather than a temporary office and was given a rather diverse program of censuses and surveys to keep it busy between the decennial population censuses(Eckler, 10).
With more and more immigrants arriving in the United States every year, each decennial count was a slower task than the count before. Even with the help of Superintendent Charles Seaton's simple device which has rollers allowing clerks to enter tallies in compact columns, the office force could scarcely keep up with the work to be done. Census figures were so obsolete when they were finally published that they were of little use except for historical purposes. Statisticians feared that the 1890 census might not be completed until it was time to take another in 1900(Scott, 55).
Dr. John Billings, a distinguished army surgeon, had worked on the 1870 statistics of mortality and continued his relationship with the census by supervising vital statistics. Young Herman Hollerith had just graduated from the School of Mines of Columbia University at the age of 19 and had come to Washington to work on statistics of power and machinery. Together they created punch-card tabulation and set in motion the development of automated data processing(Scott, 56).
Hollerith method marked the first stage in insertion of automation of census. The method proved to be most efficient in a census test tabulation in St. Louis in 1888, and his method was later adopted for the census taken two years later. In the Hollerith method various characteristics are represented by holes punched in a card in a specified location. The card is then placed in the tabulating machine in position so that every hole in the card is exactly over a mercury cup. Each hole allows a needle-like connecting wire to come down and make an electrical contact with the mercury. As the contact is made, the circuit is completed and the corresponding counter on the machine advances one. At the same time a lid opens in a sorting box and the operator places the card inside the proper compartment for use in the next tabulation. Mr. Hollerith's method proved a success, saving some five million dollars and two years' time in processing the 1890 census(Scott, 57).
As the years went by, the Census machines' engineers made improvements and modifications in every part of the system. Between 1907 and 1911 the "unit counter machine" was developed to tabulate characteristics and print results, an improvement over the earlier equipment which required a clerk's copying the totals from the counting dials. In 1909, James Powers engineered an automatic card-punching machine. He later established the Powers Tabulating Company, principal competitor in the commercial field to Hollerith's Tabulating Machine Company, which developed into the International Business Machines Corporation(Scott, 58).
The design of the cards changed to include more information, the number of columns gradually increased from 24 to 80, and the shape of the holes condensed from the original round punch to a narrow vertical slot. The speed of the unit counter was improved, devices were engineered for "gang punching"(punching a whole set of holes in one operation), and a special multi-column sorter was invented by Anthony Berlinsky in time for the 1950 census. Mr. Berlinsky, known as the "Thomas Edison of the Census Bureau," has invented dozens of mechanical improvements for economical and efficient census processing, with savings to the government by millions(Scott, 58).
Although more than a thousand general and special punch-card processing machines were being used in census work, the Bureau of the Census was caught in a breathless race with the people of the United States by the mid-twentieth century. As the population climbed from 63 million to 92 million, and then to 132 million, even the most modern punch-card equipment had to struggle to keep up with the tabulation for exponentially increasing growth of population. It was becoming clear that mechanical processing equipment was no longer adequate for tabulating the facts about the 151,325,798 Americans in the mid-century(Scott, 59).
The second stage of insertion started when in 1946 Census Bureau contracted with the recently created Eckert-Mauchly Computer Corporation to design a machine for statistical purposes. In the following year Eckert-Mauchly produced the plans for UNIVAC, the Universal Automatic Computer. In June 1948, the Bureau of Standards and the Census Bureau ordered a machine for processing the 1950 census, but the machine was delivered in March 1951. UNIVAC was a system of machines; it included a card reader, which converted the information on the traditional census punch cards to magnetic tape. The tapes then became the input to the central processing unit, which was made up of some eighteen thousand vacuum tubes. This was programmed to tabulate the data. The results were printed out on a Uniprinter, initially a rather slow device. The machine arrived too late for much of the decennial census processing, but it did fulfill the hopes of the statisticians. It was the first nonmilitary computer and it demonstrated efficiency in tabulating and analyzing substantial amounts of data(Anderson, 197).
One of the major innovations in the beginning of the second stage was the introduction of FOSDIC processing of the schedule data. FOSDIC, which stood for Film Optical Sensing Device for Input to Computers, eliminated the punch cards that had been used to transfer the data from the schedule to machine-readable form. The enumerators transferred the answers on the census form to the appropriate circles on a paper form. The forms were then microfilmed. FOSDIC then read the data according to the position of the marked circles and transformed the answers to computer-readable magnetic tape(Anderson, 202).
Unlike the first stage of insertion, the second stage brought great changes in the work force; dramatic evidence of the efficiency of new methods were demonstrated through the change of size of the census clerical and office staff. In 1950 an office staff of 9,233 worked on processing the census; in 1960 that number dropped to 2,960. In 1950, 3,000 key punchers transferred the data from the schedules to the machine-readable punch cards. In 1960, 100 operators accomplished the same task running the FOSDIC equipment. In 1950, all the data was ready for tabulation by July 1951; in 1960, it was ready two months earlier, in May 1961. All these gains in processing efficiency freed up funds for additional tabulations and publications. The bureau published 58,400 pages of published reports in 1940, 61,700 in 1950, and 103,000 in 1960(Anderson, 203).
The second stage also had direct and tremendous impact on the American society. As more and more public officials demanded better and larger amounts of information, the census data was upgraded constantly by the newly invented computers. As the bureau developed more and better census statistics in the late 1940s, 1950s, and 1960s, legislators and officials in other government agencies learned to make more use of them. Census data was being used for increasingly detailed analyses of the social and economic condition of American society. Furthermore, Congress expanded federal aid to state and local governments through the grant-in-aid system. Congress enacted programs that used statistically based funding formulas to cover federal support to states and local regions for school lunch programs, airport construction, and hospital construction in 1946, and for water pollution control in 1948. In the 1950s and 1960s, the grant-in-aid was used to supply federal funds for the interstate highway system, housing assistance, antipoverty programs, employment and training programs, urban redevelopment, and water and sewer projects(Anderson, 203).
With the help of census data, it was possible to determine the sections of the city where many old people were living and the sections where young couples were starting their own families. It was also possible to distinguish between high, middle, and low-income areas, and to locate the neighborhoods of stable and declining housing. For example:
Businessmen could pinpoint sections of the city that offered the best market for their products.
The director of the city's libraries could use census regional statistics to learn more about the people within walking distance of each branch library. He could learn about their educational background, their income, and their professional interests or skills.
The Board of Education's planning specialists could also use regional statistics to help them evaluate possible school sites, examining each area of the city to predict future school populations. In considering sites for future elementary, junior high, and high schools, they could estimate the number of students within convenient distance and the problems of traffic flow, public transportation, recreational facilities, and so on. City and county planners could also use regional statistics to help them develop new facilities like parks and recreation centers; investors could also rely on them as they choose the sites for a multimillion-dollar suburban shopping center or a downtown office building(Scott, 195).
Before the completed census schedules could be tabulated by the computers, more work was needed. The work of processing the forms of short questionnaires was almost entirely automated and proceeded with extraordinary speed. However, for some of the sample questions, human beings must first perform the tedious job of converting the answers from handwriting into machine-readable specified numerical values(Scott, 136).
Experience with self-enumeration indicated that most Americans would do a satisfactory job in filling out their own census questionnaires. The use of FOSDIC forms for these schedules greatly simplified the preparation of data for tabulation because the original forms could go directly to the huge microfilm cameras to photographed(Scott, 138).
Some census schedules were returned with errors and omissions that must be corrected before the data could be tabulated. The increase of automation in the second stage helped greatly in this correction process. For example, a small store owner reported that he had 33 employees instead of 3, or the owner of an 130-acre farm reported 300 acres of wheat under cultivation instead of 30. Consequently, much of the computer's time was devoted to editing raw data, a job that was once performed by whole groups of clerks backed up by groups of analysts. The computer was then programmed to review the data and to make automatic corrections in habitual errors. For example, when a baker reported his total bread production in pounds instead of in thousands of pounds as requested in the economic censuses, the computer could first uncover the error from the baker's report of the total value of bread produced, and then corrected the error by going through a carefully defined procedure. In the end the computer printed out a list of all the corrections it had made in a given set of data(Scott, 139).
As the work of census data processing gradually shifted from human being to machine automation, more consistent and accurate results were generated. The accuracy and reliability of census information are very important because they guide the President, federal executive agencies, the governors of the states and their legislatures, and the administrators of city and county governments as they develop policies and programs for the people. Thus the statistical information is often an important factor in the decision-making process(Scott, 187).
As computers demonstrated their computing power in the second stage of automation insertion, many knew that the processing of completed census forms would become a completely automated affair, literally untouched by human hands from start to finish. Many kinds of information would not be included in the census schedules of the future because the facts would be collected directly from the computer tapes of other government agencies(Scott, 201).
"With more widespread use of microfilm and magnetic tape to record information, many of the thousands of printed census reports may be eliminated entirely. Users will either purchase computer tapes directly or buy print-outs of the particular information they need. As computers learn to 'talk' more easily to each other, systems will develop making it possible for a user in Salt Lake City, Utah, or Augusta, Maine, to ask a question of his own computer and, through a special paid service, receive a direct answer from a census computer in less time than it takes to address the envelope for a letter(Scott, 202)."
The above description was one's view of the future census processing in the 60s, which is already a phenomenon in today's information technology. With the existing computer technology which enables communication between different machines, the current stage of insertion falls right between the second and the third ones.
The automation process of census work can still be improved in many ways. For example, telephone technology has made great advances since 1990, and related technologies such as touch-tone data entry, voice recognition, and automated voice recording will make it possible to provide a wide range of services to callers without human intervention, including the conduct of a computer-administered interview without assistance from a live operator(Steffy, 63).
The process can also be improved by making use of computer-assisted telephone interviewing(CATI) to complete enumeration with mail nonrespondents. The cost of a census telephone interview is estimated at $3.55 compared with $10 for an interview completed by a field enumerator. These estimates are compatible with the research which has consistently found telephone interviews to be less expensive than comparable face-to-face interviews. Also, the research clearly indicates that CATI provides enhanced data quality when judged against similar paper-and-pencil surveys. In the framework of the census, using telephone follow-up when possible would also reduce the number of enumerators required and thus reducing the cost of census(Steffy, 63).
Currently the Census Bureau's goal for the 2000 census is to offer the public a questionnaire assistance call-in program through a single integrated telephone network so that anyone can access the system from any telephone in the United States by calling a single 800 number. Services will be accessed through a menu system, either by touch tone or voice recognition; for example, the caller will be instructed by the system to either press the "1" key or else to say the word "one." The Census Bureau seeks to offer as many services as possible without human intervention. Nevertheless, there will always be an option for the caller to exit the menu system to speak to an operator(Steffy, 64).
Various methods have been evaluated and experimented by the Census Bureau. For example, the technology assessment of interactive cable television commissioned by the Bureau concluded that this technology will not be mature enough by the year 2000 to be adopted as a data collection method. However, interactive cable television is changing rapidly, and the Census Bureau plans to continue to monitor the progress of this technology. The technology assessment of personal computers is also not encouraging. It is estimated that less than 40 percent of households will have PCs by the year 2000, and only 12 percent will have modems. Consequently, the Census Bureau does not plan to experiment with PCs as a response option in the 1995 census test. However, this response mode remains on the long-range research program. One possible project would be to establish a computer bulletin board through an existing on-line service(Steffy, 73).
In general, the Census Bureau have tried hard looking for the best method that not only increases automation, productivity, and efficiency, but also reduces the enormous costs in various aspects of census work. Even though the current census automation falls right between the second and third stage, it can be predicted that the census automation will soon enter the third stage when the majority of households either own or have access to communication devices such as touch-tone telephones or modems. The later are preferred since they are cheaper and no operator assistants will be needed. Without human intervention, the data collection process can be performed at speed, and increases of efficiency and automation are also to be expected. We encourage the Census Bureau continue looking for new method that will give us better and more accurate census results, and the government agencies can make better use of them to improve our communities and living environments.
Anderson, Margo J.. The American Census: A Social History. New Haven: Yale University Press, 1988
Duncan, Joseph W.. Revolution in United States Government Statistics. U.S. Department of Commerce, 1978
Eckler, A. Ross. The Bureau of the Census. New York: Praeger Publishers, 1972
Scott, Ann Herbert. Census, U.S.A.. New York: Seabury Press, 1968
Steffey, Duane L.. Counting People in the Information Age. Washington D.C.: National Academy Press, 1994