Conducting the 2020 Census under Risk of Contagion

The US Constitution mandates that a decennial census take place in 2020. It is standard procedure for households that do not mail their response (or complete it online) to be surveyed in person by enumerators hired for this purpose. But the highly contagious SARS-Cov-2 (novel coronavirus) makes door-to-door enumeration a risky endeavor. To follow through with its important duties while maintaining public health, the Census Bureau should proceed with pilot projects to test and verify virus-safe practices in a few locales and conduct the remainder of the survey with constant reevaluation of its own safety practices. If successful, the 2020 census will provide a biosafety model for in-person interaction with customers that businesses can emulate.

Nonresponse Follow-Up

Online and mail-back operations have counted 57.7 percent of US households through May 7, and many more will likely be counted via these safe methods. The remainder will be counted using door-to-door “nonresponse follow-up.” 

Under normal procedures, a census taker knocks on many doors, conversing face-to-face with a person in each household for several minutes. But census workers who become infected with the coronavirus could become super-spreaders who transmit the virus to dozens of homes a day. To proceed, the Census Bureau must develop, test, and replicate operational policies that will significantly mitigate the risk of contagion.

Pilot Program

We recommend that the Census Bureau first implement a pilot nonresponse follow-up program with virus-safe practices in a few regions. Unlike pre-census test surveys, the pilot would yield final data. This program must be closely monitored to verify that virus-safe practices have succeeded in preventing the spread of coronavirus among census takers and, by implication, enumerated households.

The Census Bureau should choose a few regions (states, territories, or counties) for a pilot survey conducted according the practices detailed below. Hawaii, for example, could be among the pilot regions: it has a low infection rate at present, and its remoteness makes sudden outbreaks less likely. 

Census managers from other regions should monitor and, where possible, take active (remote) roles in the pilot survey in order to disseminate experiential knowledge widely. 

The Census Bureau should work closely with state and local authorities, throughout the pilot survey and all following operations, to monitor community-wide infection rates. The Census Bureau should set thresholds for the level and growth of infections above which it will presumptively pause operations in an area. This presumption could be overridden at the discretion of local health officials and the Census Bureau.

First-Stage Pilot Program

Given the high risk of a serious outbreak, the first pilot should be conducted in only a few small, relatively isolated areas within the pilot regions, following the personnel policies laid out in the following section. This pilot test could be stage-gated with a first pause after a week to test enumerators, grant leave to those infected, and assess the success of operations. High rates of infection should lead to a provisional suspension of the pilot and a redesign of personal protective equipment (PPE). Low rates would allow the pilot to continue. 

Given that training and management will be remote, institutional learning will be slower; the Census Bureau must intentionally induce remote “water-cooler talk” to facilitate learning. Managers and census takers should discuss the unforeseen difficulties that arise during the pilot. For example, supervisors from other regions could be required to interview pilot census takers for firsthand accounts of the survey experience in order to inform their own management choices. 

Second-Stage Pilot Program

If the results from the first week indicate that biosafety measures are effective, a second-stage pilot program should expand the survey to the remaining areas of the pilot regions. Census Bureau officials should require all supervisors for regions participating in the second stage to submit a written action plan of what measures they intend to adopt.

A week after starting second-stage operations, enumeration should be paused in some selected local areas for enumerator testing and evaluation of practices. 

National Expansion

If ongoing testing suggests that the pilot program was successful, the Census Bureau should gradually extend the survey nationwide. There are three good reasons to stagger initiation of the national survey: (1) to allow the Census Bureau to remotely reassign experienced personnel, (2) to continue the learning and adjustment process, and (3) to delay the survey in the hardest-hit places until they stabilize.

As the survey expands, the Census Bureau should constantly evaluate the practices and results of its enumerators, using testing, observation, and discussion. 

Personnel Policies, Equipment, and Practices

Taking a census during a pandemic requires some important changes in what is expected of census takers: they must agree to novel coronavirus testing, be willing to abide by public health directives, and accept greater scrutiny of their on-the-job public health practices. The Census Bureau should dismiss enumerators who disregard public health mandates.

The Census Bureau should screen, hire, and train census takers remotely. All prospective census takers should be batch-tested for novel coronavirus and test negative as a condition of employment.

Census takers should commit in writing that they will comply with all existing public health regulations, including assembly restrictions, for the duration of their fieldwork. Their training should include public health practices in daily life as well as on the job. The Census Bureau could work with local governments to provide centralized, isolated residential arrangements for census workers on a voluntary basis.

Before beginning door-to-door operations, each enumerator should be equipped with PPE deemed sufficient by medical experts. This equipment may include face shields or other items not currently in visible everyday use.

Supervisors should shadow census takers during work to correct imperfect practices. The Census Bureau should also set up a hotline for reporting unsafe practices by any census taker.

Pause Buttons Everywhere

A single day’s delay in pausing contagious activities can result in an exponentially more costly epidemic. It is imperative that census operations be able to stop instantaneously.

Every individual in the door-to-door operations hierarchy should have the authority to immediately pause face-to-face operations in his or her area of responsibility. Individual census takers who believe they may have been exposed to COVID-19 should be able to pause their own work without prior approval. Supervisors who suspect their teams are not being careful should be able to pause their work without the sign-off of a superior.

In all cases, a pause should immediately be reported two levels up the hierarchy. Overruling a pause should require consultation; no individual manager should be allowed to overrule a pause on his or her own.

In most cases, a pause will be brief or limited. In other cases, it may be indefinite, such as in a city that experiences a major COVID-19 outbreak.

Funding Virus-Safe Census Operations

A virus-safe census will cost more than a typical census. Frequent pauses and extended training quarantines will add labor costs. The census will also require large supplies of PPE. The 2010 census employed 564,000 people at its peak; all of these would need to be equipped with face shields, masks, gloves, and other necessities.

To properly fund the 2020 census, the Census Bureau should publish a detailed plan, complete with cost estimates. Congress should scrutinize and, when satisfied, fully fund the operation, recognizing that the spending will directly (though temporarily) alleviate unemployment and will generally comport with efforts to limit the spread of COVID-19 by pioneering safe face-to-face work practices.

In addition, Congress should pre-commit to purchase census PPE at a minimum per-item price. As Caleb Watney and Alec Stapp have argued, purchase guarantees combine the fiscal strength of the federal government with the efficiency and innovation of the free market. 

Innovation in the Census

The 2020 census is an opportunity to develop best practices for conducting service-sector business amid the threat of a pandemic. As Americans grope toward a “new normal,” the census can provide a rubric for carrying out interpersonal business with an acceptably low level of risk to employees and customers. 

This is not the first time that the census has spurred innovation in business practices: the Census Bureau invented the “spreadsheet,” and leading tech companies such as IBM were first seeded by Census Bureau prize money awarded for data processing equipment. Thus, beyond the benefits of an enumeration of population, the 2020 census can have a productive role in modeling customer interaction practices for the aftermath of the crisis.

As employers reopen workplaces, there will be an understandable tension between efficiency and safety. The census model can offer employers and employees a coordinating point—a set of equipment and practices that many people have seen on their streets. 

Will the 2020 Census Be Accurate?

The ongoing pandemic is likely to raise questions about the accuracy of the census. After all, COVID-19 has caused changes for many households, including people moving in with parents and moving out of homes with elderly relatives. 

People who move frequently—snowbirds, individuals with multiple homes, students, and seasonal workers—have always posed a challenge to census-taking, and existing census policies for counting them are sufficient.

While the crisis may have caused some residential dislocation, the overwhelming result has been residential stasis. Cell phone tracking data show a sharp decline in mobility since early March. It is almost certain that more people are at home during the day than ever before, so door-to-door operations will be more efficient and precise.

Moreover, conducting a census during a pandemic is of particular importance. COVID-19 is a world-historic event, leading to death spikes in many areas on a scale not seen since the 1918 influenza pandemic. This event will inform policy for generations to come. A full-population census conducted near the peak of the pandemic, while many people are still sheltering in place, will yield extraordinarily valuable data for future researchers and policymakers. 

Thus, to the extent that is possible and reasonably safe, the Census Bureau should resume operations in order to capture a portrait of American society during the pandemic.

Alternative Courses of Action

PPE shortages or a surge in COVID-19 infections could render a 2020 census infeasible. If so, Congress has two options: it can delay the entire survey until 2021 or beyond, or it can cancel the unfinished parts of the 2020 census, potentially filling in the gaps using population estimates and data from the American Community Survey, which is conducted every year for more expansive purposes than the census. Both of these options pose practical and constitutional problems.

Delay is the worse option. Census-taking without protection is very likely to spread SARS-Cov-2, a risk that could remain active for years even if the virus is contained by herd immunity or widespread vaccination. Delaying until 2021 or 2022 would force households that have already responded to complete census forms a second time, likely resulting in some fatigue and a lower response rate, and also requiring a costly second round of census advertising and hiring. With fewer remote responses, door-to-door operations would be even more expensive as well. 

Canceling the 2020 census or replacing it with a population estimate would likely require a constitutional amendment. If states trust that the estimation process would be fair to all, it might be relatively easy to pass such an amendment. In 2030, the decennial census would resume as normal. Cancellation would be easier, cheaper, and more certain than delay. However, this strategy is risky: it is not clear what constitutional or statutory implications might arise from fiddling with the census clause. 

Furthermore, the decennial census has many uses besides apportionment of congressional seats. Along with the American Community Survey, it is a reliable baseline against which other data are calibrated. When a private company runs a survey, it reweights responses based on census-estimated population shares, on the assumption that these figures are reliable. The very population estimates that the Census Bureau would use to substitute for a census are themselves based on prior censusesLacking a 2020 census would lead to gradually accumulating error until 2030. That would make the 2030 census a wild card, the 2030 reapportionment even more volatile, and thus the political debates around it even more toxic. 


The United States has never canceled a census before. Counting the population is one of the most basic functions of government, dating back, quite literally, to the biblical book of Numbers. From the ancient censuses of the Han Dynasty to William the Conqueror’s Domesday Book to early colonial censuses to the decennial US census, census-taking has always served to exhibit state capacity. National hiring, mass contact, scalable logistics, and large-scale coordination are a display of the state flexing its muscles. The COVID crisis has revealed that these very areas are deficits in US state capacity. Policymakers should not respond to those deficits by abandoning one of the few remaining tools for building capacity: rather, the census should be used as a laboratory of epidemic response. The United States government should not accept the conclusion that it simply lacks the state capacity to count the population safely; rather, it should, through a strenuous exertion of political will, find a way to forge ahead.

About the Authors

Salim Furth is a senior research fellow and director of the Urbanity Project at the Mercatus Center at George Mason University. He studies regional, urban, and macroeconomic trends and policies, and he has testified before the US Senate and House of Representatives. Previously, he worked at the Heritage Foundation and Amherst College. His writing has been featured in National Affairs, American AffairsThe City, and Public Discourse, and he wrote regularly for the Wall Street Journal’s Think Tank blog. He earned his PhD in economics from the University of Rochester in 2011.

Lyman Stone is the chief information officer of the population consulting firm Demographic Intelligence, a research fellow at the Institute for Family Studies, and an adjunct fellow at the American Enterprise Institute. He is a former international economist at the US Department of Agriculture, where he forecasted cotton market conditions. His work has been featured in the New York TimesWashington PostWall Street JournalBoston GlobeSouth China Morning Post, and many other media outlets. He and his family live in Hong Kong.