COVID-19 Threatens 2020 Decennial Census

The 2020 Census is facing an existential crisis stemming from the COVID-19 pandemic. As of Sunday morning, March 29, the national response rate was 33.1 percent of all households, which might appear to be a good number, since April 1, the official Census Day, was days away, with many months left before the count is finalized.

However, there are towns, like East Hampton and Southampton, where the census hasn’t even started. Southampton’s percentage of households that had completed their census forms stood, as of the end of day March 28 at 13.1, while East Hampton was even further behind the national average, at only 9.8 percent. Shelter Island is even worse, with a response rate below 2 percent.

The census numbers are the foundation of our representative democracy, and the key to the distribution of hundreds of billions of dollars across our nation. The U.S. Census Bureau has been conducted it as required by the U. S. Constitution every 10 years since 1790.

The cost of counting every person who resides in America has grown exponentially over the past several decades. The operation cost $1,078,488,000 in 1980, $2,492,830,000 in 1990, and came in at $4.5 billion in 2000. The 2010 Census operation was the largest peace time mobilization in U.S. history, employing 500,000 people above and beyond the fulltime staff already working for the Census Bureau, and cost American taxpayers almost $13 billion, which equates to $42 for each person counted.

The Government Accountability Office estimated that if 2010 were used as a model for the next census, the price tag could rise as high as $30 billion. The bureau’s own estimate is $22 billion.

The main cost-cutting measure for 2020 is the introduction of an online census form. Instead of an initial massive mailing out of forms, the Census Bureau sent out invitations to most American households, asking them to go online and participate. Americans began receiving them on March 12.

Prepared For COVID-19?

In the years leading up to each decennial, the Census Bureau does test runs, and while it may have been prepared for tornadoes, hurricanes, or other natural disasters, was it prepared for a national pandemic? Large sections of East Hampton and Southampton towns not receiving forms by Census Day answers that question.

Because a percentage of the residents of those two towns do not get mail directly, but rather use post office boxes, the Census Bureau employs a system called Update Leave, where a census taker hand delivers invitations to participate, as well as the form itself, if needed, directly to the residence. That program was supposed to start on March 15.

There are about five million U.S. residences that the bureau reaches with this system.

In the days leading up to March 15, the Census Bureau pushed back the Update Leave start date two weeks, to March 29. On March 28, the bureau announced it was pushing that start date back two more weeks, to April 15. The following day, President Donald Trump told the country that the social distancing protocol will remain in place until at least the end of April, which will further delay the implementation of Update Leave until the beginning of May, at the earliest.

East Hampton’s response rate in 2010 by Census Day was 66.7 percent. If census takers cannot get out into the field until May or June or July, how will that impact the Census Bureau’s ability to meet its legally-required deadline of December 31 to present the numbers to the president, which will be used to allocate seats in the House of Representatives?

Nonresponse Follow-Up

Between East Hampton and Southampton, there are tens of thousands of residents. In Queens County, there are millions.

Queens County is likely the most ethnically-diverse county in the country. It is also the epicenter of the pandemic. Currently, the response percentage to the Census Bureau’s initial mailing of invitations to participate stands at 22.4 percent. Not good, but better than its sister New York City borough on its western border, Brooklyn, also known as Kings County. The Census Bureau has recorded a response rate of only 21.6 percent there.

The operation after the initial outreach to residents is called nonresponse follow-up. That is when census takers go knocking on doors at residences that haven’t been accounted for by the bureau. That was supposed to start on May 13 and conclude by July 31. Those dates have also been pushed back two weeks.

There is a solution that the Census Bureau has explored before.

It does not just count people. The Census Bureau also counts a slew of things, such as how many people work in different professions, and where they work in those professions. It looks at where they live in the country, or how many cars a family owns. This is done through the American Community Survey.

The data collected is extrapolated through statistical sampling, to form a picture of the nation as a whole.

In 1998, Kenneth Prewitt was approached by the Clinton administration to become the director of the Census Bureau. He spoke about that experience with The Independent last August.

“By the time I got the job, a great deal of work had been done,” he said in preparation for the 2000 decennial. “It is too late to rearrange the furniture.”

There was one cost-cutting measure he believed the Census Bureau should adapt in its approach to the 2000 Census: Dual-system estimation, a form of statistical sampling. It would be a way to get a more accurate count of population groups that are hard to count. Traditionally, those groups include the immigrant community, particularly if they are undocumented residents of the country, as well as members of communities of color.

It is the stuff that statisticians’ dreams are made of. But not, Prewitt learned, politicians, particularly Republican ones. Republicans were suspicious that the approach would increase the numbers of population groups that tend to vote Democratic, pushing power from red states to blue.

“Sampling became a toxic word,” he said. Congressional Republicans sued the Census Bureau. In January of 1999, the Supreme Court, in a 5-4 vote that fell along ideological grounds, sided with Republicans.

“There is a legitimate difference of opinion either way,” Prewitt said about the use of sampling for the entire census. To this day, he regrets not narrowing the focus of the proposed use of sampling. Prewitt believes that if he had done so, applying the use of sampling to only the nonresponse follow-up phase of the operation at the tail end of the Census might have passed muster.

Come this August, when the 2020 version of Nonresponse Follow-Up is currently scheduled to conclude, that might look like a pretty good idea.

“The Census, Then and Now,” is an ongoing series in The Independent.

T. E. McMorrow has worked in three decennial censuses and was a field operations supervisor covering a large swath of Manhattan during the 2010 decennial.

If you are part of the 2020 Decennial operation, and wish to comment on the current operation itself, excluding any personal information gathered, which must and is protected by law, you can reach the author at t.e@indyeastend.com.