Because of quality concerns caused by the pandemic, data from an annual Census Bureau survey that provides a wide-ranging picture of the U.S. on everything from commute times to education levels won’t be released in its usual format this year, officials said today.

Answers for the 2020 American Community Survey one-year estimates were collected only from two-thirds of the people needed in the survey because of difficulties posed by the pandemic. Those who were missed skewed toward people with lower incomes, less education and who didn’t own their homes, raising the risk that survey results would be biased, said Donna Daily, chief of the American Community Survey Office.

“The COVID-19 pandemic posed numerous challenges to collecting data,” Daily said. “While this is a difficult decision for us, it shows our commitment to providing high quality data.”

Mailings sent out by the Census Bureau to survey participants were canceled for several months in spring 2020, and field workers who follow up at the homes of people who didn’t respond also didn’t make visits during that time, census officials said.

The survey polls 3.5 million people each year, asking about their jobs, income, housing costs, disabilities, marital status, Internet access, health insurance, number of vehicles owned and types of appliances they have in their homes.

The American Community Survey launched in 2005 as a substitute for the census’ long-form questionnaire which had been sent out to about a sixth of all U.S. households during the once-a-decade head count of every U.S. resident. The long-form census questionnaire asked dozens of detailed questions about residents’ social, economic and housing situations. The short form, which was basically what every U.S. household got during last year’s census, asks fewer than 10 questions.

The idea behind the American Community Survey one-year estimates was to give researchers and policy makers up-to-date yearly information about every place with a population of 65,000 people or more without having to wait every 10 years for the next census.

Like census data, the American Community Survey numbers help determine the distribution of hundreds of billions of dollars in federal spending. The census figures, however, also are used to determine how many congressional seats each state gets in a process known as apportionment and for drawing congressional and legislative districts.

The apportionment figures from the census were released in April, and the numbers used for redistricting are expected to be made public next month.

Because of the concerns over its quality, the American Community Survey one-year estimates won’t be released in their usual form in the fall. Instead, an “experimental” version with limited data in tables at only the national and state levels will be released, but they won’t even be placed on the Census Bureau’s usual public website for downloading data sets. Census Bureau officials are urging researchers and policy makers to use the experimental figures with caution.

“These experimental estimates do not meet our data quality standards,” Daily said.