By Terri Ann Lowenthal, Funders Census Initiative Consultant

November 2015

On October 6th, the U.S Census Bureau released its baseline 2020 Census Operational Plan. Already a year behind schedule due to budget constraints, but well ahead of the same milestone for the 2010 Census, the preliminary plan puts “meat on the bones” of sweeping design reforms the Bureau has been researching and testing since the beginning of the decade.

The Census Bureau estimates that full implementation of the plan will save $5 billion over the cost of repeating the paper and pencil, brick and mortar, and boots-on-the-ground traditional census design. That’s an important goal, because Congress has said the 2020 Census must cost less than the 2010 count. The price tag for the new plan would be $88 per housing unit, the Bureau says, compared to $124 per housing unit in 2010, for a total lifecycle cost of $12.5 billion.

In the June 26, 2015, FCCP newsletter, I outlined the major reforms in store for 2020. Based on the results of census field tests in 2014 and 2015, the formal operational plan nails down key components of each major phase of the census; additional tests in 2016 and 2017 will help refine the methods. Meanwhile, the Census Bureau must begin to develop IT systems to support a largely automated enumeration, as well as a massive communications plan to boost participation, especially in historically undercounted communities. The infrastructure and operations must be ready to deploy in time for an end-to-end readiness test in early 2018.

The following summary highlights key operations in each phase of the count and notes significant changes from the 2010 Census. (Note: Most numbers and cost figures cited are estimates at this point in time.)

Building the address list: Address canvassing is the first major census operation, and it takes place well before the count starts. An accurate address list is critical because it determines where to mail census materials or send a census taker. In 2010, 150,000 canvassers walked every street and road in the country to verify and update the Master Address File. For 2020, most work to update the address file will take place “in-office” and on a continuous basis, using aerial imagery, government (e.g. Postal Service, state, and local records) and commercial databases, and GIS tools. Targeted “in-field” address canvassing will cover roughly 25 percent of housing units, in areas with high growth or significant change in housing stock. The Bureau will open 30 Area Census Offices to oversee this operation, compared to 151 in the last census.

Self-response phase: For the first time, the six-week self-response phase (formerly and appropriately known as “mail-out/mail-back) will focus on boosting response via the Internet, which is the least costly way of collecting data. People also can respond by paper questionnaire or telephone. Every housing unit will receive an initial letter, inviting them to respond on-line using a unique ID code; on-line responses will be accepted without a unique identifier, as well, and matched to the address list. Paper forms will be sent in the first mailing only to 20 percent of housing units, primarily targeting low-income (urban and rural) and elderly households that are less likely to have or use the Internet. After two weeks, the Census Bureau will send a paper questionnaire to all housing units that have not yet responded. All households will receive a thank-you/reminder postcard later in this operation. The Bureau estimates that 47% of households will respond on-line, 5% by telephone, and 11% by paper form.

Non-response Follow-up (NRFU): The Census Bureau hopes to substantially reduce census costs by reengineering this major field operation. In 2010, 550,000 temporary census takers (“enumerators”) knocked on every unresponsive door up to six times in an effort to gather information. For 2020, the Bureau estimates that 56 million addresses will fall into the NRFU universe. It will first try to identify vacant homes, using Postal Service and other databases, and remove those addresses from the follow-up workload. Census takers will then visit every non-responding household once.

The next proposed step (which is generating concern among advocates for hard-to-count population groups) is to use data gathered for other government programs (“administrative records,” such as IRS, SSA, SNAP, TANF, WIC, Indian Health Service) to “count” some of the remaining households. (The Bureau estimates it can enumerate about six million households using administrative records.) Census takers will then go back into the field to visit still unresponsive households several more times, or ask neighbors and landlords for information. As a last resort, the Census Bureau uses statistical methods to “impute” data for households it has failed to reach in person. (That’s why an accurate address list is so important. If a housing unit isn’t on the list, the Census Bureau doesn’t know that no one has responded.)

Automating the field work: Using technology, the Census Bureau plans to streamline field operations considerably, reducing its “footprint” across the country. For 2020, there will be six Regional Census Centers, compared to 12 in 2010, and 250 Area Census Offices, compared to nearly 500 Local Census Offices in 2010. The Bureau will automate recruitment, training (using videos, for example), human resource functions (e.g. payroll), and, most importantly, data collection. Census takers will be equipped with GPS-enabled devices (smartphones or tablets); the Bureau will use a range of data sources to optimize enumerator routes, determine best times to knock on doors, and track enumerator progress in real-time. The ratio of enumerators will be 15:1 (or higher), compared to 8:1 in 2010.

The 2020 Census Operational Plan represents a marked departure from past enumerations — a modern census in the age of technology and information. The plan is not final, and the Census Bureau’s ability to demonstrate that proposed new methods will work depends, in significant part, on adequate funding levels from Congress. The 2016 field tests, supplemented by ongoing research, will be critical opportunities to implement and evaluate the plan in diverse communities. Understanding the basic structure of the census plan will help funders and grantees alike target resources effectively — and at the right time during the census process — to maximize participation and improve accuracy in hard-to-count areas.