Home » Featured

ASA and Census … A Long and Productive Relationship

1 December 2016 One Comment
Margo Anderson

    In 1839, when a small group of men got together in Boston and founded the American Statistical Association for “the purpose of collecting, preserving, and diffusing statistical information,” they may not have realized what a momentous event it would be. The ASA was among a number of what were then called “learned societies” being established in the still relatively new United States, and most had modest goals: to bring together likeminded men of some intellectual interest to read and critique papers in a seminar setting and discuss their common interests. Most of these societies remained local affairs, often supported by the energetic personal interests of a few members, and many disappeared, merged with other societies, or functioned more as social clubs than intellectual venues.

    Not the ASA. Though, in practical terms, the association remained firmly anchored in Boston (it didn’t hold its first meeting outside the city until the 1890s), these founding members made their way onto the national policymaking stage within a decade and joined in an international debate about statistical theory, statistical policy, and what was one of the key matters of controversy at the time: the census.

    By 1840, the United States was taking its sixth census, mandated by the 1787 Constitution to apportion seats in the House of Representatives and Electoral College. Previous censuses were modest affairs—simple counts of the number of people by age, sex, civil condition (free or slave), and color for each household. At each census, Congress asked for a bit more information, experimenting, for example, with questions about occupation, immigration, and disability. By the sixth census, the questions had ballooned from their original six to more than 70, all on large unwieldy sheets of paper carried around by assistants to the United States marshals who went door to door to collect the information.

    “Tabulation” of the results of this data collection was done in the field by the assistant marshal, who added up the totals for his local area and shipped the schedules and summaries to the U.S. marshal, who in turn added up the results for his area and shipped the lot to Washington. There, a “clerk” in the Secretary of State’s office, aided sometimes by a few assistants, compiled the national results, made minor corrections of arithmetic, and sent them to Congress so reapportionment could proceed.

    Congress printed the results, basically as they came from the field. It wasn’t until 1830 (the fifth census) that the secretary printed a uniform schedule and sent that for use in the field. The whole effort took several years to collect and compile. Nevertheless, at each census, the effort got a bit more complex.

    By 1840, the schedule was unwieldy at 70+ questions on large, two-sided pieces of paper. One of the questions in 1840 asked how many “white” and “colored” “insane and idiots” there were in the household. When the results were published, they seemed to show a dramatically higher rate for insanity for “colored” in the North (free states) than in the South (slave states). Debates about the abolition of slavery were intense at the time and defenders of slavery used the data to argue that African Americans were benignly held in “slavery” and would become insane if freed.

    Absurd as this sounds today, the controversy exploded in the press, and members of the ASA entered the fray, trying to determine what had produced such results. Edward Jarvis, who would serve as ASA president for 30 years, examined census reports in detail and found that the column for “colored” insane was larger than the total number of “colored” in a local area in parts of New England. Since what fed the controversy were the comparative rates of insanity in North and South, small errors in the counts of insanity would be masked in the large slave populations in the South, but would stand out against the small populations of free colored in the North. There had to have been some mistake.

    The ASA memorialized to ask that the results be corrected. They were not, but the controversy led ASA members (including Lemuel Shattuck, another ASA founder) and other officials to propose reform of 1850 census procedures, most notably to the use of the individual level rather than the household schedule, and thus the establishment of a large central office in Washington to tabulate the census. Congress adopted those reforms in May 1850 and the Census Office became one of the largest administrative offices in government during the census period.

    The Census Office was technically a “temporary” agency until 1902, though it often remained open for years before and after the actual ‘census period.’ ASA members became increasingly influential in the technical and administrative development of statistics and census taking, and census leadership frequently came from ASA membership. In turn, census officials from other professional venues came to join the ASA.

    The ASA grew as the discipline of statistics expanded in the second half of the 19th century. Francis Walker—census superintendent in 1870 and 1880, ASA president from 1883 to 1896, and president of MIT from 1881 to 1897—led the push for the ASA to become a truly national organization in the 1890s.

    At the end of World War I, the Census Bureau set up an official advisory committee, initially composed of members of the American Statistical Association and American Economic Association. The ASA continues today as part of the official advisory committee system for the census.

    All because, I’d suggest, a few men in Boston in the early 1840s recognized they could help address technical problems of taking the census in a large diverse country, thus establishing the tradition in the ASA of contributing to both the development of statistics and the hard questions of American public policy.

    Want to Know More?

    There are a number of historical articles in JASA and TAS about the founding of statistical societies in the 19th century and the early relationships between the ASA and statistical innovation. See the following in particular:

    Fitzpatrick, Paul J. “Leading American statisticians in the nineteenth century.” Journal of the American Statistical Association 52(279):301–321.

    ____________. “Statistical societies in the United States in the nineteenth century.” The American Statistician 11(5):13–21.

    __________. “Leading American statisticians of the nineteenth century II.” Journal of the American Statistical Association 53(283):689–701.

    Regan, Opal G. “Statistical reforms accelerated by sixth census error.” Journal of the American Statistical Association 68(343):540–546.

    For background on the development of the field of statistics, statistical thinking, and census taking in the United States in the period, see the following:

    Anderson, Margo. The American census: A social history. 2nd ed. New Haven: Yale University Press, 2015.

    Cohen, Patricia Cline. A calculating people: The spread of numeracy in early America. Routledge; Revised edition (1999).

    1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)

    One Comment »