Around 1 in 20 residents in Arkansas and Tennessee were missed during the 2020 census, and four other U.S. states had significant undercounts of their populations which could short-change them of federal funding in the current decade, according to figures from a survey the U.S. Census Bureau released today.

In Florida, and Texas, undercounts appear to have cost them congressional seats too.

On the flip side, residents in eight states were overcounted during the once-a-decade head count that is used to allocate political power and federal funding. In Minnesota and Rhode Island, overcounts appear to have helped save them from losing congressional seats.

In the remaining 36 states and the District of Columbia, the overcounts and undercounts were not statistically significant. Undercounts signal people were missed. Overcounts suggest they were counted more than once, as for example, children of divorced parents who share custody or people with vacation homes.

The figures released today from the Post-Enumeration Survey serve as a report card on how well residents in the 50 states and District of Columbia were counted during a census that faced unprecedented obstacles from a pandemic, hurricanes and wildfires, social unrest and political interference by the Trump administration.

States that did a better job of getting residents counted scored greater Electoral College and congressional representation, or did not lose expected seats in the House of Representatives. They also are now better positioned for the annual distribution of $1.5 trillion in federal funding in the coming decade.

Nothing can be done at this point to change how many congressional seats are allocated among the states, and neither can the data used for redrawing congressional districts be adjusted.

Today’s release did not break down by demographic traits how good a job the 2020 census did at the state level, but a national report card released in March showed significant undercounts for the Black and Hispanic populations, as well as for those identifying as some other race and American Indians and Native Alaskans living on reservations.

Academics and civil rights leaders are pressing the Census Bureau to tweak yearly population estimates that traditionally have used census numbers as their foundation and incorporate other data sources to produce a more accurate portrait of the undercounted racial and ethnic communities for the numbers that help determine the distribution of federal funding. The Census Bureau has set up a team to explore this.

Arkansas, Tennessee, Mississippi and Illinois respectively had undercounts of 5%, 4.8%, 4.1% and 1.9%, while Florida and Texas respectively had undercounts of almost 3.5% and 1.9%.

Arkansas, Florida, Tennessee and Texas did not direct as many resources as other states in encouraging residents to fill out census forms. Mississippi spent around $400,000 and Illinois allocated $29 million toward those efforts. Historically, groups that have undercounts are racial and ethnic minorities, renters and young children.

In a statement, Arkansas Gov. Asa Hutchison said he was interested in getting more details on his state’s undercount, especially since Arkansas grew by more than 95,000 residents over the decade and surpassed 3 million residents for the first time.

Demographer Allison Plyer also observed that Arkansas, Mississippi and Tennessee have among the highest rates of households without a computer or internet subscription. The 2020 census was the first head count in which most participants were encouraged to fill out the form online.

“Get-out-the-count efforts can make a big difference, even when your community has poor internet access and is less likely to answer the census,” said Plyer, chief demographer of The Data Center in New Orleans.

Florida’s undercount translates into around 750,600 missed residents, and an analysis by Election Data Services shows the Sunshine State needed only around 171,500 more residents to gain an extra seat. The undercount in Texas translates into around 560,000 residents, while the analysis put Texas as needing only 189,000 more residents to gain another congressional seat.

Hispanics make up more than a quarter of Florida’s population and almost 40% of Texas residents, and critics say the Trump administration’s failed efforts to add a citizenship question to the census form may have had a chilling effect on the participation of Hispanics, immigrants and others.

Arturo Vargas, CEO of NALEO Educational Fund, said there was a “desperate need” for information about undercounts and overcounts of racial and ethnic groups at geographies smaller than states, especially in places like Texas where the undercount most likely was in the Hispanic population.

Given the inaccuracies in the count, there is a real risk of an unfair distribution of congressional seats among the states, he said.

“Without knowing below the state level, we aren’t able to understand the extent of that error,” Vargas said.

Minnesota was allocated the 435th and final congressional seat in the House of Representatives; if Minnesota had counted 26 fewer people, that seat would have gone to New York. Minnesota’s 3.8% overcount amounted to around 219,000 residents.

Other states with overcounts were Hawaii, at almost 6.8%; Delaware, at 5.4%; New York, at 3.4%; Utah, at almost 2.6%; Massachusetts, at 2.2%; and Ohio, at almost 1.5%.

Eugene Tian, Hawaii’s chief state economist, said the state’s large number of vacation homes likely contributed to people being counted twice.

“People were counted as Hawaii residents while vacationing in Hawaii or waiting out the pandemic in their vacation homes between April and October 2020, but their usual residences were on the U.S. mainland,” Tian said in an email.

In Rhode Island, the 5% overcount translates into more than 55,000 residents. It would have lost a seat if 19,000 fewer residents had been counted, according to Election Data Services.

John Marion, executive director of the government watchdog group Common Cause Rhode Island, said it was difficult to pinpoint exactly why Rhode Island had such a large overcount. There were significant outreach efforts and the state has a large summer home population, but the same applied to other states, he said.

“We’re essentially the lucky beneficiary of a statistical anomaly,” Marion said. “And as a result, we’ll have more representation in Congress for 10 years.”