PROVIDENCE — Last year, Rhode Island defied expectations in the new census count and managed to hold onto its two seats in the US House of Representatives.
On Thursday, a US Census Bureau report provided an explanation for those results: Rhode Island had the third highest “overcount” in the 2020 census, at 5.05 percent.
“This explains why it was such a surprise when we retained the second congressional seat,” said John M. Marion, executive director of Common Cause Rhode Island. “All other indicators pointed toward Rhode Island’s population being stagnant. This confirms that the census figures we got last year were indeed on the high side because of an overcount.”
Advertisement
The Census Bureau said it had conducted a follow-up survey to gauge the accuracy of the 2020 count, and it found significant overcounts in eight states — Hawaii at 6.79 percent, Delaware at 5.45 percent, and Rhode Island at 5.05 percent. Massachusetts had an overcount of 2.24 percent.
The overcount won’t change Rhode Island’s representation in the House, thanks to a 1999 US Supreme Court ruling that said statistical sampling in the follow-up survey can’t replace the census data for reapportioning Congress.
“The Census Bureau is not allowed to adjust the data based on statistical samples,” Marion said. “In every census, with these estimates, some states have overcounts and some have undercounts. The census is not a perfect process.”
But the overcount is a clear sign that Rhode Island’s second House seat could be in jeopardy in the 2030 census. “Unless we happen to win the statistical anomaly lottery in 2030, we are going to lose our extra congressional seat,” Marion said.
In three years leading up to the 2020 census, population projections had shown that Rhode Island could lose a seat and go from being the most over-represented state, with one House member for about 530,000 residents, to the most underrepresented, with one representative for more than 1 million people.
Advertisement
But the 2020 census said that Rhode Island’s population grew by 4.3 percent over the past decade, rising to 1,097,379.
And Rhode Island ended up being one of 37 states to keep the same number of House seats. That avoided a potential showdown between Representative David N. Cicilline, and Representative James R. Langevin, who has since announced he won’t seek re-election this year.
The Census Bureau, which is part of the US Commerce Department, delivered that good news after an April 2021 presentation by US Secretary of Commerce Gina M. Raimondo, who had been Rhode Island’s governor until March 2 of that year.
But Marion noted that the census count was completed in October 2020, long before Raimondo became commerce secretary. “The Census Bureau is widely respected as a nonpartisan part of the bureaucracy,” he said.
In a statement, Rhode Island Republican Party National Committeeman Steve Frias said, “This is terrible. Rhode Island’s aggressive census counting tactics led to the equivalent of double-counting more than the entire population of the City of East Providence.”
Democracy only works when people trust the system, Frias said. “Double counting 55,000 people in order to hold onto a congressional seat destroys that trust,” he said. “To help restore public trust, perhaps the federal government should investigate exactly how Rhode Island census takers could have double-counted 55,000 people.”
Advertisement
In announcing the overcount and undercount rates, Census Bureau Director Robert L. Santos said that getting an accurate count for all 50 states is always difficult.
“These results suggest it was difficult again in 2020, particularly given the unprecedented challenges we faced,” Santos said. “It is important to remember that the quality of the 2020 Census total population count is robust and consistent with that of recent censuses. However, we know there is still more work to do in planning future censuses to ensure equitable coverage across the United States and we are working to overcome any and all obstacles to achieve that goal.”
The survey found that states with the largest undercounts were Arkansas at 5.04 percent, Tennessee at 4.78 percent, and Mississippi at 4.11 percent.
Marion said Rhode Island’s overcount resulted in part from people being counted twice. “That can happen for a variety of reasons,” he said. “For example, a person moved and was counted in two locations, or their household submitted two responses. The Census Bureau is supposed to weed that out, but it’s not perfect.”
The Census Bureau provided an explanation for “duplicates.”
“We often say our goal is to count everyone once, only once, and in the right place,” the bureau said. “Sometimes in an effort to count everyone in a census, we end up counting some people more than once.”
The bureau said duplications can also occur because of “potentially complex living situations,” the “misdelivery” of census materials, or “an issue with the address” such as a housing unit on a census address list more than once.
Advertisement
A “complex living situation” could include college students counted at both their college residence and at their parents’ home, children counted by both divorced parents who share custody, or people with more than one residence, such as a seasonal or vacation home.
One of the nation’s top census experts, Terri Ann Lowenthal, a former staff director of the US House of Representatives census oversight subcommittee, told the Globe that the post-enumeration survey is “an important tool in the Census Bureau’s arsenal.”
“It is a statistically representative sample of US households,” she said, “and is part of a scientifically complex methodology, developed over many decades, to measure how well the census counted the population.” She said some of the nation’s leading statisticians were involved in developing the methodology, including the late John Tukey, the father of Bayesian statistics.
The Census Bureau used the post-enumeration survey program because the census “is not and never will be perfect,” Lowenthal said. “The final results may look precise, but that precision is an illusion. There are many reasons the census misses people and counts other people twice or in the wrong location.”
If all of the mistakes in a census were spread evenly across the country and population groups, it wouldn’t matter as much, Lowenthal said. “But, as we know from post-enumeration surveys, the census does not count all communities equally well,” she said. “It misses and double-counts some population groups at higher rates than others. That means that some communities receive more than their fair share of political representation and government resources, while other communities receive less than they deserve and need.”
Advertisement
Lowenthal said she is not suggesting that Rhode Island received a congressional district for the next decade to which it is not entitled.
“We cannot know that based on the state-level estimates of accuracy the Census Bureau released today,” she said. “The (post-enumeration survey) only covers the household population, not people living in ‘group quarters,’ such as college dorms, military barracks, prisons, homeless shelters, and nursing homes, or the unsheltered population.”
It is possible, for example, that RI’s college student population was undercounted since universities were locked down just as the census began and most students went “home,” which could have been out-of-state, even though they should be counted at their college-based residence, she said.
The survey produces estimates of undercounts and overcounts, and federal law prohibits the use of sampling methods to derive the numbers used for congressional apportionment, Lowenthal said.
“But that aside, without knowing the extent of undercounting and overcounting in the entire population (household and group quarters), there’s no way to recalculate apportionment, even as a mathematical exercise,” she said. “We only have estimates, and incomplete ones, at that, and the apportionment formula is sensitive to small shifts in population between the states.”
Edward Fitzpatrick can be reached at edward.fitzpatrick@globe.com. Follow him @FitzProv.