Statisticians: Guardians of Democracy

Statisticians: Guardians of Democracy

By Rod Little

For the past two years, I've been working at the U.S. Census Bureau in Washington, D.C., helping director Robert Groves set up a new research directorate. When I tell people what I've been doing, many remark that I must have been very busy counting all those people in the 2010 Census. Some undoubtedly think: "Official statistics, bean counting, so boring and mundane! No doubt worthy, but … rather you than me!"

Maybe they also wonder privately: why is a School of Public Health faculty member wasting his time counting people? Does he have trouble sleeping at night? The Decennial Census is an amazingly complex operation, and important on many dimensions—politically, since it determines the representation of states in the U.S. House of Representatives; fiscally, in that it determines financial allocations to localities for many government programs; and socially, as it provides a localized demographic snapshot of the country every ten years. But I did not work on the 2010 Decennial Census at the Census Bureau.

A pet peeve of Census Bureau employees is that the agency's name—built into the U.S. Constitution and hence not easily changed—fosters the public perception that they do the Decennial Census and then twiddle their collective thumbs for ten years. In fact, the Census Bureau is much more than the Decennial Census.

As the largest government statistical agency, it conducts a myriad of other censuses and surveys that provide crucial information to the public. Data collected by the bureau tell us who we are, what is changing, and what needs to change—from neighborhood to municipality, region, and nation. The knowledge generated from government statistical agencies like the Census Bureau forms the backbone of our information society.

Did you know, for example, that:

  • The principal source of monthly unemployment statistics—a key economic indicator if ever there was one—is a Census Bureau survey, the Current Population Survey?
  • A major national survey of public health, the National Health Interview Survey, is also conducted by the Census Bureau? Objective, nationally representative information about health is pretty vital, given the political controversies over health care reform.
  • An important source of information about health insurance is the American Community Survey, or ACS, another Census Bureau Survey? The ACS replaced the Census Long Form, and hence is often confused with the Decennial Census. It is a massive ongoing survey, and a major source of information about our society.

Survey nonresponse is increasing, and some selected to participate in the ACS are outraged by a mandate to answer questions they find intrusive. Favorite bugaboos are: when do you go to work in the morning, how many toilets do you have in your house, and questions about disability.

Three points need to be made:

  1. First, unlike, say, the Internal Revenue Service, Census Bureau employees have no interest at all in individual responses to these questions, and are subject to jail time or large fines if they breach the confidentiality of their respondents.
  2. Second, it's the answers that really matter: the questions are there because they provide the factual basis for implementation of government programs, passed by Congress.
  3. Third, for the U.S. to maintain its leadership position in the face of global competition, we need to make smart decisions, and this requires good data.

A host of issues is informed by data from the ACS and other government surveys. To name a few: measuring the efficacy of government programs, assessing the impact and insurance costs of climate catastrophes, deciding where to locate a new Target store, planning municipal mass transit, understanding and addressing health disparities, redistricting in a way that meets the requirements of the Voting Rights Act. All these activities use information from surveys like the ACS, where people like you and me take the time to answer questions posed by an interviewer. This information does not grow on trees—you may find it on the Internet, but the underlying source is often a census or survey. Without them, the information is either out of date, distorted, or made up.

It is not a stretch to say that official government statisticians are guardians of our democracy. The Fundamental Principles adopted by the United Nations assert the basic importance of official statistics for "any society that seeks to understand itself and to respect the rights of its members." In the absence of transparent objective data about our society, opinions and whims of the rich and powerful are substituted. Information is power. For example, if you are not interested in addressing health disparities in our society, the best move is not to collect information about them, since without hard evidence, the disparities are invisible.

Why did former University of Michigan faculty member Robert Groves ask me to help form a research directorate? The Census Bureau has a storied history in research. Edwards Deming started his explorations of quality improvement at the Census Bureau. The earliest large-scale digital computers were at the bureau. Morris Hansen and others at the bureau developed landmark practical approaches to probability sampling.

Today, research is more important to maintaining the quality of Census Bureau operations than ever before. There is increased demand for data products, and the questions asked are becoming more complex and hard to measure. Combining information across data sources, protecting confidentiality, asking questions in a way that provides useful information, modeling estimates for small areas, controlling survey and census costs—these require statisticians, social and computer scientists, and others versed in scientific methods.

People at the U.S. Census Bureau are much more than bean counters—they create the information base for our free society.

Rod Little is the Richard D. Remington Collegiate Professor of Biostatistics in the School of Public Health at the University of Michigan, where he is also a research professor in the Institute for Social Research and professor of statistics. In 2010 he became the U.S. Census Bureau's first Associate Director for Research and Methodology and Chief Scientist.

Editor's note: Robert Groves left the Census Bureau in August to become provost of Georgetown University.

Fitting into the Boxes: The U.S. Census and the Multiracial Question

As recently as the 1990 U.S. Census, respondents could choose only one race for themselves and each of their family members. In fact, it wasn't until the 1960 Census that respondents could choose their own race. Prior to that, census takers determined the racial category of the citizens they interviewed. Detailed instructions—such as those listed below, for the 1930 Census—explained how a person of mixed ancestry should be classified:

The 1930 U.S. Census instructions defined the following categories:

  1. Negroes—A person of mixed white and Negro blood should be returned as a Negro, no matter how small the percentage of Negro blood. Both black and mulatto persons are to be returned as Negroes, without distinction. A person of mixed Indian and Negro blood should be returned a Negro, unless the Indian blood predominates and the status of an Indian is generally accepted in the community.
  2. Indians—A person of mixed white and Indian blood should be returned as Indian, except where the percentage of Indian blood is very small, or where he is regarded as a white person by those in the community where he lives.
  3. Mexicans—Practically all Mexican laborers are of a racial mixture difficult to classify, though usually well recognized in the localities where they are found. In order to obtain separate figures for the racial group, it has been decided that all persons born in Mexico, or having parents born in Mexico, who are definitely not white, Negro, Indian, Chinese, or Japanese, should be returned as Mexican ("Mex").
  4. Other mixed races—Any mixture of white and nonwhite should be reported according to the nonwhite parent. Mixtures of colored races should be reported according to the race of the father, except Negro-Indian.

1990 U.S. Census instructions:
Fill ONE circle for the race that the person considers himself/herself to be.

2000 U.S. Census instructions:
The 2000 Census was the first in which people were allowed to officially acknowledge all the sources of their ancestry by selecting more than one racial category. Roughly 6.8 million people chose more than a single race that year.

Source: "RACE: Are We So Different?" a traveling exhibition developed by the American Anthropological Association in collaboration with the Science Museum of Minnesota. The exhibition will be on display at the U-M Museum of Natural History from February 8 through May 27, 2013.