How Mapping And Data Could Make The Census Cheaper

A new data analytics and mapping tool could save the Census Bureau from spending scarce funds tracking down the people who don’t respond to the 2020 count.

The Response Outreach Area Mapper analyzes a wide array of demographic data to predict the percentage of people in a given census tract who will self-respond to the decennial. The bureau could use that information to help local officials tailor outreach efforts to specific communities and proactively direct money and manpower to the areas that need it most.

Roughly 74 percent of U.S. households filled out and submitted the questionnaire in the 2010 census, leaving the bureau to send employees to visit the homes of more than one-quarter of Americans.

The 2020 census will be the first to accept responses online and over the telephone, but the bureau expects only about 60 percent of households to self-respond, said Deirdre Dalpiaz Bishop, chief of the geography division at Census. While there are a slew of factors that might prevent someone from responding to the Census, these follow-up visits remain the most costly aspect of the decennial count, she told Nextgov.

When the count is underway, officials using the ROAM tool could help encourage potential non-responders to submit their information, Bishop said, and for every person who replies, that’s someone whose door “we will not have to knock on.”

In addition to estimating response rates, ROAM also includes granular demographic data on every census tract in the country, such as age, ethnicity, language, income and resident mobility. Census and local officials can use this information to determine how to best target a given community.

“You don’t want to reach out to all those hard-to-count populations in the same way,” said Suzanne McArdle, the Census cartographer who developed ROAM, in a conversation with Nextgov. Different messages appeal to different groups, she said, and the bureau can use the tool to show local leaders how to hone in on their specific constituents.

Bishop also noted the tool will inform the hiring process at Census, as the bureau prefers its employees to “look like, talk like and be like the people in the community” they’re counting.

McArdle said Census has relied on some form of response prediction for the past few decades, but the agency only recently began developing an interactive digital model. She and her team began working on ROAM in April 2016 and launched it publicly last month.

The tool was intentionally designed for users who don’t have a technical background, she said, but the program also allows for tech-savvy public servants to pull Census data and build their own applications with additional information.

ROAM stands as one of the few silver linings of the 2020 count—budgetary and cybersecurity concerns led GAO to designate the decennial as a “high-risk” project, and a recent inspector general report found a number of shortcomings in the bureau’s process for vetting new hires.