How we count, not who we count, matters most
In October 2009, Sen. David Vitter (R-La.) and Sen. Robert Bennett (R-Utah) sparked a debate when they introduced an amendment to an appropriations bill that would have required the 2010 census to collect data on citizenship status. This kind of controversy, revolving around precisely what data the census should collect, is nothing new.
But there is another census controversy that focuses on how best to collect data, rather than on what data to collect. Since the 1960s, debate has steadily simmered over how, for instance, to correct for the systematic undercounting of minority populations.
Debates limited to census methodology, such as the undercounting issue, impact the census’s long-term value as a public good much more substantially than do disputes like the one over the Vitter-Bennett proposal. Such controversial, high-profile tussles tend to resolve on their own accord after a few years, but questions about the statistical validity of the census itself have lingered for decades. Maintaining the public’s trust in the census, which ultimately helps distribute political power around the country, requires resolution of the methodological concerns.
The Census as a Public Good
The census, conducted every ten years since 1790 by constitutional mandate, collects basic demographic data about the composition of the American populace. The explicit purpose of the census is to apportion congressional representatives, but more recently, America has started using census data to ensure the proportional allocation of funds across all levels of government. An accurate census is therefore vital for ensuring fair representation and distribution of public goods. As Margo Anderson, a census historian at the University of Wisconsin, told the HPR, “Any threat to the credibility of the census will make the political uses of it difficult.”
But the census’s value is not only narrowly political. Kenneth Prewitt, a professor of public affairs at Columbia University and the former director of the Census Bureau, told the HPR that the census provides the “basic data used to calibrate every piece of survey data,” both public and private. The census is the baseline that helps determine the accuracy of all other social surveys in America, and it therefore underlies most meaningful research on American public opinion.
Who to Count?
As the most comprehensive study of the American populace, the census often becomes entangled with the sociopolitical concerns of the time. Throughout America’s history, disputes over the status of immigrants and minorities have spilled over into the question of who to count in the census. This began at the Constitutional Convention, when delegates debated the enumeration of slaves.
In 1920, for the first time, the census showed that Americans living in urban areas outnumbered those in rural regions. Political elites, who still mostly represented rural districts, recognized the profound political implications of this demographic shift, and Congress refused to redistrict itself until the Reapportionment Act of 1929. The 1920s also saw a rise in nativist and anti-immigrant sentiment, with groups like the Ku Klux Klan gaining popularity. These groups attacked the census in an attempt to forestall recognition of the increasing diversity of the nation. Terri Ann Lowenthal, a consultant for the nonpartisan Census Project, told the HPR, “It’s really the census that shows us how the racial and ethnic composition has been changing dramatically.”
A similar controversy about immigrants recurred in the 1980s, and the Vitter-Bennett amendment of 2009 shows that this dispute is resurfacing today. As Lowenthal noted, these controversies “tend to come and go in cycles”; they emerge for reasons unrelated to the census, and after a few years, they lose prominence for reasons, again, unrelated to the census itself. These debates about immigrants and minorities may give Americans reason to lose trust in their policymakers, but they ultimately have little to do with the underlying validity of the census.
How to Count?
Debates over census methodology ultimately have a much greater effect on the census’s ultimate value than do these more public controversies. One of the most significant methodological concerns is the undercounting of minority groups in urban areas. As Anderson wrote in her book Who Counts?, this issue first became apparent to census officials during World War II, when the number of young men registering for the draft—especially minority men—far exceeded the number then thought to exist.
For nearly two decades, most statisticians have argued that a procedure known as dual-systems estimation—which compares two surveys that together give a more accurate count than either survey alone—is the best method to adjust for this undercounting. The debate over statistical adjustment was most prominent in the 1990s and continues today. Lowenthal, who worked for congressional Democrats in the ’90s, noted that it was “front and center for Republicans” to forbid statistical adjustment and today it is part of “Republican dogma.” Anderson added that, for 2010, there is “no possibility of adjusting the census for undercount.” Official census data thus does not reflect what most statisticians consider the best estimate of true American demographics.
Another problem is raised by the nation’s large prison population. The census counts prisoners only at their places of incarceration, rather than at their homes, which generally resulted in the overrepresentation of rural prison districts, and the underrepresentation of urban districts, in state legislatures and in Congress. As Peter Wagner, executive director of the Prison Policy Initiative, told the HPR, “Nobody knew about this ten years ago, and twenty years ago it didn’t matter because the prisoner count was not nearly as high as it is today.”
Maintaining the Public Trust
The census is thus much more complicated than a simple tally. Can it ever be perfected? According to Anderson, “The census will always be carefully scrutinized, so one can expect the public and politicians to hunt for—and usually find—problems with the quality and accuracy of the count.” This scrutiny does not necessarily preclude public trust in the census. The degree to which the census implements the best available methods for accurately counting the population will determine the extent of that trust.
Historically, the census has been able to implement these with little delay. Wagner added that he is confident in the census’s ability to adapt. “The census does a very good job, and each census is better than the one before,” he said. Anderson added that the Census Bureau is “an incredibly innovative agency”; it leads the world in machine tabulation and probability sampling.
While debates over issues that do not affect the reliability of the census will come and go, the future of the census will depend on the proper resolution of more methodological concerns. As long as the census continues to innovate and rely on the most advanced methodologies and techniques, it will maintain its credibility as a basis for scientific research and political decision-making.
Jeffrey Kalmus ’12 is the Webmaster and Rajiv Tarigopula ’14 is a Contributing Writer.