How local governments ensure an accurate census

An accurate census count is critical for state and local governments. It affects the number of seats a state has in the U.S. House of Representatives and helps determine how $590 billion of federal funding is spent on infrastructure, hospitals and schools. But getting an accurate count can be tricky, especially in the wake of a disaster that relocates residents, rent increases that foster more creative housing arrangements or a housing boom that fuels new developments.

To ensure it has an accurate list of addresses to work from, the Census Bureau enlists the help of state and local governments to update the addresses it has on file.  Its Local Update of Census Addresses program is open to all 39,139 tribal, state and local governments that can use LUCA to compare their local housing data with Census’ and make additions, corrections or deletions to the addresses on the lists and maps the bureau uses to conduct the decennial census.

Before comparing files, participants can use other Census tools to standardize and collect data. The Census Geocoder is an address lookup tool that converts addresses to an approximate latitude and longitude based on the ranges in the Master Address File/Topologically Integrated Geographic Encoding and Referencing (MAF/TIGER) shapefiles. Governments can enter up to 10,000 addresses at once and receive information about state, county, tract and block codes for those addresses in a matter of minutes.

For files of more than 10,000 addresses, participants may use the LUCA Geocoding Service, which geocodes the area in which the addresses are found and the next day returns a total count of each matched, unmatched or tied address and an address count -- data the Geocoder does not provide.

“Our goal is for them to be able to go and take a look and try to use that list to identify areas where we may be missing housing units,” said Brian Timko, the 2020 census coordinator for the Census Bureau's geography division.  "If there was a new subdivision that was built that we don’t have the addresses for, they can use these block count tools to determine, outside of the Title 13 environment, where we may be missing addresses,” he said . Title 13 requires the bureau to ensure confidentiality of census information, including individual addresses and structure coordinates that identify the location of living quarters.

About 10,950 governments registered to use LUCA for the 2020 census -- almost 28 percent of the invited governments. “We actually have 96.9 percent of the population and 98.6 percent of the housing covered by at least one LUCA participant. This includes 42 states that are registered to participate,  and that’s up from 28 states in the 2010 LUCA,” Timko said.

LUCA was developed after concerns about undercounts in the 1990 census led to the Census Address List Improvement Act of 1994, which authorized the bureau to implement LUCA. The technology was part of the 2000 and 2010 censuses, but a new element for 2020 is Census Bureau Geographic Update Partnership Software (GUPS), a self-contained geographic information system tool that lets participants load, display and edit geospatial and address data.

“We’re seeing better-quality data come in because the GUPS software dictates the format that the data comes back to us in,” Timko said. GUPS allows state and local governments "to standardize the data so that when we get the data back, we know it’s in the right format and it passes all our validity checks.”

The bureau sends disks with the software to participants who load it onto a computer. The software will prompt users for the Title 13 address data disk and the MAF/TIGER shapefile disk. Next, participants enter a password to be able to work with the encrypted data on the disks. GUPS, whose core function is Python code built on the open source QGIS platform, decrypts it so users can edit the data, which is displayed in a list, map and address count list.

“The GUPS software maps their data and so they can just select addresses from their address list and easily add them to our address list for the LUCA updates,” Timko said.

GUPS is a standalone system that doesn’t connect to any bureau networks or servers, added Monica Mardel, GUPS project manager. “The participant submits [data] via what we call the Secure Web Incoming Module, so it’s a separate system and it’s a secure way for the participants to deliver their files to the Census Bureau,” Mardel said.

Other governments are taking accuracy into their own hands. New York is using the New York Block Browser LUCA Evaluation System (NYBBLES) from Cornell University’s Program on Applied Demographics, for example. It uses address lists based on real property data to get estimates of the number of residential units per parcel.

“The Census Bureau was willing to generate a count of housing units for each block in preparation for LUCA, and with my experience of the real property data, I can do a similar count based on the real property data,” said Jan Vink, extension associate at the program.  Comparing those two could "find blocks [where] the Census Bureau might be missing some addresses,” he said.

Local assessors populate a local database with the information that they submit to the state once a year. NYBBLES also relies on parcel maps, which show places independent of the database and which the state Office of Taxation collects from counties. NYBBLES gets the data from that office and analyzes it to come up with estimates for each parcel and block.

“I put all the data in an online database and created a web interface to display and make the data publicly available,” Vink said.

NYBBLES contains information on all areas of New York state except New York City, which does its own count.

Other states could benefit from a NYBBLES-like system, he added. The real property data fields and other aspects would have to be customized, but the basic framework is shareable.

“The local knowledge is very important for accuracy,” Vink said. “There’s only so much you can do at the state level.”