Some college towns plan to challenge the results of the 2020 census, claiming they were shortchanged because the pandemic forced students to leave campuses and complaining that the undercount could cost them federal money.

College communities such as Bloomington, Indiana; Tuscaloosa, Alabama; and State College, Pennsylvania, are exploring options for contesting the population counts, which they say do not accurately reflect how many people live there.

When the pandemic struck the U.S. around spring break of 2020, it set off an exodus in college towns as classrooms went virtual almost overnight. The sudden departure of tens of thousands of students from these communities made it difficult to count them in the census, which began at almost the same time.

Because universities were able to provide the Census Bureau with records for students living in dorms and other on-campus housing, off-campus students “ran the risk of being missed,” said Dudley Poston, a sociology professor at Texas A&M University.

An Associated Press review of 75 metro areas with the largest share of residents between 20 and 24 showed that the census results fell well below population estimates in some cases but also exceeded them significantly in others.

Officials in college towns are not sure why there was such variation, and they are reviewing whether it was due to the timing of spring breaks, outreach efforts or the percentage of students living on campus versus off. Another variable is whether schools cooperated when the Census Bureau asked for records on off-campus students. Only about half of schools did so since many had privacy concerns or did not have the requested information.

“You can kind of go crazy thinking about the variations,” said Douglas Shontz, a spokesman for the Borough of State College, home of Penn State University, where officials believe the census missed 4,000 to 5,800 residents.

The AP review showed that the population counts were below estimates by about 5% to 7% in the Mount Pleasant, Michigan; Greenville, North Carolina; and Bloomington, Indiana, metro areas, which are home to Central Michigan University, East Carolina University and Indiana University, respectively.

Counting university students has always been a difficult task, even before the pandemic. The Census Bureau’s rule of thumb was that students should be counted at their college addresses, even if the coronavirus temporarily sent them elsewhere on the April 1 date that provides a benchmark for the census.

In State College, home to Penn State’s 39,000 students, the bureau’s message prior to the pandemic was that people should be counted “where they sleep most of the time,” which was confusing to students after they went home. As a result, neighborhoods dominated by students had the lowest census response rates in the borough, said State College Borough Manager Tom Fountaine.

Cities, states and tribal nations can start contesting their numbers in January through the bureau’s Count Question Resolution program, but it looks only at number-crunching errors, such as an overlooked housing unit or incorrect boundaries. The program only revises figures used for population estimates over the next decade that help determine federal funding.