By Kenneth Prewitt
In 1998, when I first found my way to census headquarters in Suitland – as its new and naïve Director – I was soon impressed by the Bureau’s relentless attention to errors and mistakes in its data products, and the satisfaction when new techniques for correcting these errors were designed and implemented. I also learned, to my surprise, that while the 2000 census schedule had four months of field operations, five months were dedicated to in-house quality control for the stripped-down apportionment count, then additional months for the more detailed data file needed for redistricting, and even more months for gradually more specialized data products. I thought I had a reasonable understanding of the census, had even published a few pieces in SSRC and Russell-Sage books, but was wholly unprepared for the magnitude of the Bureau’s quality control operations.
How could that be? Because the decennial census has phases which, though not secret, are generally invisible to the public — the master address file is one and then the quality checks that follow enumeration is another. The public is made fully aware of the field enumeration phase – through paid advertising, mailed forms, phone calls, complete count committees, school programs, etc. But the Bureau has never promoted the role of the non-enumeration phases in the census process. It should, and especially brag about its many statistically inventive ways that find and fix mistakes, most of which are made by the public itself – the ten-year old who is married with children, the household that returned duplicate forms.
This promotion will be important before the next census. The 2020 census had to contend with an abnormal level of political interference that produced a steady stream of law suits and uncertainties about major issues – whether a citizenship question would be added to the census form itself, whether undocumented residents would be included in the apportionment count (an issue not yet resolved), and, as political appointees were added by the White House there was conflict over who was actually calling the shots. Disruption of 2020 plans and schedules were frequent and unpredictable. And in recent weeks there has been the forced rush to completion, which has substantially cut into the quality control operations. All of this was extensively covered by the press. The public was confused, left to wonder if the census is more political than scientific. And if so, should it be trusted?
Of course, the census has always and always will be political. However, it has always and always will be scientific, the latter dependent on quality indicators. What I see as a silver lining in 2020 is that what had been largely invisible is now very visible. As the field operations wind down, the centrality of quality indicators in the census process is being prominently highlighted. Going forward, this centrality will be protected by census stakeholders, will be demanded by census data users, will be improved by academics, will be discussed and debated in the press, will be legally protected and will find its way into legislation. In time, the results of the quality checks will be viewed as no less important than enumeration itself. We will not again have a census where the public takes no notice of errors and their correction, where quality indicators appear to be an afterthought, or where insufficient time translates into insufficient application. The Bureau will make publicly clear that a census without ample time and staff to execute quality checks is a flawed census. Although no census is perfect, the difference between one where quality controls are fully applied and one where they are not is the difference between fit for purpose, or not. Publicizing this is a step toward re-building the public trust damaged in 2020.
- Professor Ken Prewitt is the Former Director of the Census Bureau for the 2000 Census, past Executive VP for the Rockefeller Foundation, and currently Carnegie Professor of Public Affairs at Columbia University. He is also a member of The Census Project’s Advisory Committee.
This article is an excerpted version of a more detailed treatment in Public Seminar.
All true. But what is not published ENOUGH is how the Census Bureau moved to online responses, and though other forms of responses were available, most people responded online and there were no issues or hiccups. No hacks, no downtime. Sometimes the news is that things work the way they are supposed to.