Census begins stretch run vowing not to make same mistakes

Listen to Jason's story on the Federal Drive

wfedstaff |

The Census Bureau will release an operational plan for the 2020 population count in the next five months.

Government Accountability Office experts warned this is a key time for the bureau as it needs to get that strategy right in order to avoid many of the same mistakes that impacted the 2000 and 2010 decennial counts.

The operational plan will lay out how the Census thinks the 2020 count will work,...


The Census Bureau will release an operational plan for the 2020 population count in the next five months.

Government Accountability Office experts warned this is a key time for the bureau as it needs to get that strategy right in order to avoid many of the same mistakes that impacted the 2000 and 2010 decennial counts.

The operational plan will lay out how the Census thinks the 2020 count will work, including plans for using new technologies to the hiring of enumerators in the field to the innovations in the works that could save the bureau $5 billion.

John Thompson, the director of the Census Bureau, said his agency is better prepared for the 2020 count and it will be reflected in the September document.

“We are putting some things in place to try to minimize the risk because you can never be sure,” Thompson told lawmakers during a Senate Homeland Security and Governmental Affairs Committee hearing Monday. “For example, the document we are producing this fall, we are calling the 2020 Census operational plan. This will be three years earlier than we issued an operational plan for the 2010 census. So we are getting an early start on it. We have in place an active risk management framework and a risk management process where we constantly identify major risks. We constantly look at how we can find contingencies for those risks. We’d be happy to share that with the committee so you can feel comfortable that we are looking at the right risks.”

Before becoming the Census director in 2011, Thompson was the lead career official for the 2000 count, and watched the 2010 effort from an industry perspective.

Between now and September, Census will be testing out several concepts that will help make up that operational plan.

Innovations to reduce costs

Thompson said there are four broad categories of innovations that could make a huge difference in how the 2020 count is executed.

The first innovation is around using the Postal Service and commercially available data such as geospatial imagery to better determine addresses, and whether buildings are vacant or if they were torn down. In previous population counts, the Census would have sent field staff to check out every address.

The second innovation is to use the Internet to increase the response rate. Thompson said the Census still will offer phone and paper response possibilities, and field staff will go out to hard to reach areas, but they think offering a Web response option will save about $550 million.

The third innovation is around making better use of existing data. Thompson said that is one area where Congress could help them.

“There is a dataset, it’s called the National Database of new hires and it’s maintained at the Department of Health and Human Services,” he said. “It would help us greatly in our program and we would need some legislative changes to have access to that file.”

Sen. James Lankford (R-Okla.) asked Thompson what other datasets does the Census need access to and whether agencies are charging the bureau for access to the information.

Thompson said Census is in the process of acquiring two databases from the Agriculture Department, the Supplemental Nutrition Assistance Program (SNAP) records that are state-by-state, and the Women Infants and Children (WIC) programs records.

“We will put these together in a very private way and we are protecting privacy, and research them as we proceed with the census,” he said. “We are not being charged by the other agencies for these datasets. In fact, the Secretary of Commerce has great authority to ask for these records in conducting the job of procuring at the Census.”

The fourth innovation could save more than $2 billion by bringing better technology to the field. Thompson said this includes using mobile devicesand collecting and transmitting data in real time.

But that innovation also is the cause of the most concern by GAO.

Carol Cha, GAO’s director of IT Acquisition Management Issues, said there are several major concerns about a main piece of the operational plan, the agencywide IT initiative called the Census Enterprise Data Collection and Processing program (CED-CAP).

“Recent estimates put the program’s costs at about $548 million through 2020. Given the bureau’s prior and existing challenges, we highlighted CEDCAP as part of a new entry into this year’s GAO High Risk List as one of a handful of major IT investments in need of the most attention,” she said. “September’s decision is expected to drive the business requirements for CEDCAP’s systems and infrastructure. This milestone, which has already been delayed by a year, cannot afford to slip further.”

GAO reported earlier this year that Census has yet to determine the requirements around the Internet and the IT infrastructure security and scalability.

“If they are not adequately addressed by September, it could lead to system rework downstream eating into an already narrowed schedule margin,” Cha said.

Consistent oversight of IT

CEDCAP is a system of systems focusing on 14 different initiatives ranging from a dynamic operational control system to track and manage field work to developing and testing online response options to creating the back-end infrastructure all of these efforts will run on.

Cha said GAO will initiate a review of Census’ progress in planning for CEDCAP later this year.

She said the concern is that if Census plans on full scale testing starting in October 2018 that’s only three years from today and it’s not a lot of time to get a several complicated pieces in place.

Instead, she recommended Census to simplify CEDCAP and depend on current and proved technologies.

Thompson agreed with Cha, saying CEDCAP is a high-risk program, but he said the bureau is taking steps not to repeat the mistakes of 2010 when it wasted years and millions of dollars on a failed handheld device.

He said Census is innovating not inventing technologies.

“I can assure you this is getting the full attention from the Department of Commerce. The Department of Commerce CIO is fully engaged with this program and with our CIO on this program. In fact, he’s running an independent verification and validation (IV&V) process on the program. He’s taking it very seriously too,” Thompson said. The whole Department of Commerce has this as one of its priorities.”

The second area of concern is cybersecurity where GAO recommended 115 actions for Census to take in January 2013.

Cha said the bureau so far has only addressed about 19 of them.

“Access controls in particular are of vital importance in strengthening because those regulate who and what is accessing the bureau’s systems,” Cha said. “As the bureau is exploring these methods to collect data out in the field, including using personally owned devices or government issued devices, it’s going to be critical for the bureau to secure those back-end systems to make sure the information is properly secure. In addition, if they are expanding the use of administrative records, they also are going to have to sure up the controls there to ensure that information is adequately protected. So addressing our 115 recommendations is the top priority.”

Term ends in 2016

Thompson and Cha agreed that the bureau gave GAO an update on its progress in closing these recommendations on April 17. GAO now will review that update and decide just how many cyber recommendations remain open.

Thompson added the Census is shoring up its network by encrypting all data at rest or in motion, and scanning Internet traffic to ensure no personal information is traveling out of the Census Bureau networks.

Cha said running through both of these challenges is an undertrained workforce. She said the IT security and workforce gaps present the biggest challenge to the 2020 count.

One other non-technology related point to consider around the 2020 count that poses a potential challenge is Thompson’s term as Census director ends in December 2016. At that time, there will be a new President-elect, and potentially a changeover in Congress. It’s unclear what all of this means for him, so over the next 18 months, Thompson said he has to get as much in place as possible before he may have to leave office.

Robert Goldenkoff, GAO’s director of Strategic Issues, said he Census reached the half way point earlier this month in preparing for the 2020 count so there is a greater sense of urgency going forward.


Census decennial planning builds on its telework success

Census nominee promises to fundamentally change the decennial count

Census Bureau offers primer on enterprise risk management

Carol Cha, director of IT acquisition management issues, GAO