The 2020 census has endured many challenges that have delayed operations, from wildfires to hurricanes to the coronavirus pandemic, leading U.S. Census Bureau officials to suggest the agency may miss the Dec. 31 deadline to provide state population counts.
The agency insists that despite the challenges, and possibly being a week or two late with final counts, it will provide an accurate census of the nation's population, which is mandated every 10 years by the U.S. Constitution.
Census officials said they were working to provide the state-by-state population count — apportionment data used to determine how many members in the House of Representatives each state gets — as close to Dec. 31 as possible.
However, Al Fontenot, the census bureau's associate director for decennial programs, told reporters on a call Wednesday, that the apportionment data might not be available until the "first or second week of January." He added it was too soon to say whether there would be a delay in providing state redistricting data, which is normally provided three months after the state population counts.
Tim Olson, the bureau's associate director for field operations, laid out the "multiple challenges" affecting the 2020 census: storms and hurricanes, particularly in the South, wildfires in the West, and "major civil unrest that gripped our nation in every state … while we were conducting field operations. It goes without saying the biggest challenge of this census is the coronavirus pandemic," emerging just as census field operations started in March, which then were postponed due to the nationwide lockdown.
"In spite of these unplanned and unprecedented challenges, we completed the count operation October 15," Olson said, adding that "based on early indicators, I believe we did a really, really good job."
Those early indicators include "resolving" approximately 152 million addresses, or 99.9%, meaning addresses were verified against address files from local governments and post offices, for example, Fontenot said. That compared with 132 million resolved at a similar stage in 2010.
Fontonot said 67% of housing units in the nation self-responded to the census, or 99 million households, which, he added, exceeded expectations. Of the 99 million who self-responded, more than 79 million did so online, 18 million mailed in the paper form and another 1.8 million responded by phone.
Outside observers have raised concerns about the accuracy of the data amid the shifting deadlines of the nonresponse follow up phase, when census takers knock on the doors of addresses that did not respond to the census. The U.S. Supreme Court allowed the Trump Administration to end the count two weeks earlier than the bureau had announced originally.
The census has been the focus of legal wrangling with advocates and local governments suing the Trump Administration over other measures, such as the president's memo to the bureau to exclude undocumented immigrants from the state population totals for apportionment.
Bureau officials would not comment on the legal cases.
Fontenot stressed the bureau's focus on accuracy.
"Technology has advanced significantly" since 2010, he said. "A lot of processes can happen faster."
For example, Fontenot said computers determined the most efficient routes for census takers and the best time of day to knock on doors, based on other statistical information the bureau had about addresses.
Olson said the nonresponse follow-up phase involved 64 million addresses, with census takers "resolving 99.93% of all of those addresses over a 68-day period.
Olson said 55.5% were completed through an in-person interview and 20.4% through "high quality administrative records," such as Medicare or Medicaid records.
In 2020, 24.1% of nonresponse follow-up of occupied addresses were enumerated through interviews with proxies, such as a neighbor or landlord. In 2010, Olson said 23.8% were enumerated through proxies.
"All of these operations are not the end of the census," Fontenot said. "We will convert, check and validate and turn them into final apportionment numbers," he added, "as close to the statutory deadline as possible."
CORRECTION: A previous version of this story inaccurately attributed a statement that apportionment data might not be available until the "first or second week of January." In addition, it contained an incorrect percentage for addresses collected using high quality administrative records.