ORLANDO, Fla. — A new automated system designed to create efficient routes for census takers as they knocked on the doors of homes that hadn’t responded to the 2020 census instead often sent them on journeys that made no sense and distributed cases unevenly among the workers, according to a watchdog report released Tuesday.
The routes created by the system known as “the optimizer” frustrated census takers with illogical assignments, sending them on confusing paths that may have slowed efforts to reach every uncounted household before the door-knocking phase ended last October, according to the report from the U.S. Government Accountability Office.
When asked by GAO investigators, only 21.6 percent of area census office managers were satisfied with the accuracy and efficiency of how cases were assigned, according to the report, a broad review of what worked and didn’t work during data collection for the release 2020 census. The report was released hours ahead of Senate testimony Tuesday by acting Census Bureau director Ron Jarmin.
The complaints from census managers and supervisors outlined in the GAO report echoed what census takers were telling The Associated Press during the middle of the door-knocking phase of the 2020 census last summer and fall. These census takers were assigned daily routes via bureau-issued iPhones, which they also used to record the race, Hispanic origin and relationship of everyone living in each household.
“They basically have us going to the same unit over and over again, based on a computer program,” census taker Linda Rothfield in San Francisco told the AP in September. “The program is atrocious.”
Bureau officials told the watchdog agency that the automated system worked effectively and increased productivity, though they said they could have done a better job of explaining it during training.
Census takers also had difficulty entering large apartment or condo buildings to interview households since many building managers were working remotely during the pandemic. Census supervisors wished they had been given more control over reassigning such cases, the report said.
The report also noted that nearly 8,000 census takers who had failed their training tests were given assignments anyway, and that almost 1.2 million households were reached by census takers by telephone, a new option given to census managers.
The report also said that the bureau spent $98.4 million on financial rewards to census takers for working additional hours or for traveling to other locations where the door-knocking was lagging.
The bureau is currently in the data processing phase of the 2020 census. Numbers that will decide how many congressional seats and Electoral College votes each state gets are scheduled to be released next month. The census figures also determine how $1.5 trillion in federal funding is distributed each year.