2022 Franz Edelman Competition: U.S. Census Bureau

Ashley Kilgore:

This podcast is part of a special series featuring the 2022 finalist teams for the INFORMS Franz Edelman Award for achievement in advanced analytics, operations research, and management science, the most prestigious award for achievement in the practice of O.R. and advanced analytics. For more than four decades, the Edelman Award has recognized contributions that are transforming how we approach some of the world’s most complex problems. Finalists for the Edelman Award have contributed to a cumulative impact, more than $336 billion since the award’s inception, as well as countless other non-monetary benefits.

Ashley Kilgore:

The winner of this year’s award will be announced to the 2022 INFORMS Business Analytics Conference held April 3rd to the fifth in Houston, Texas. Joining me for this episode are Vicki McIntire, assistant regional director for the Denver regional office and Tammy Adams, senior advisor for IT and operations to discuss the finalist entry from the team representing the US Census Bureau. The US Census Bureau conducts a decennial census every 10 years as mandated in the US constitution. Prior to the 2020 census, this was done with manual assignments.

Ashley Kilgore:

In 2020 optimization and machine learning techniques, automated the scheduling, workload assignments, and management of field data collection. MOJO, an operational control system based on these techniques provided optimization of caseloads handled by enumerators through a geographic information system. The 2020 census result, 99.9% of all addresses in the nation and MOJO via assignment optimization provided a productivity increase of over 80%. The system was developed in collaboration with Princeton consultants, as well as others in the private sector and academia. Vicki, Tammy, thank you for joining me. I’m excited to share a sneak peek of the US Census Bureau’s project in the lead up to the 2022 Edelman competition.

Vicki McIntire:

Thank you.

Tammy Adams:

Thank you.

Ashley Kilgore:

To start us off, could you share some background on the US Census Bureau and the types of services it provides?

Tammy Adams:

Absolutely. The Census Bureau is the federal government’s largest statistical agency. It is dedicated to providing current facts and figures about America’s people, places, and economy. The Census Bureau honors privacy, protects confidentiality, shares our expertise globally and conducts our work openly. The Census Bureau is guided on this mission by scientific objectivity, its strong and capable workforce, its devotion to research based innovation, and its abiding commitment to its customers. The Census Bureau conducts more than 130 surveys and programs each year in addition to its constitutionally mandated decennial census.

Ashley Kilgore:

And now your project was conducted to support this decennial census. Could you share a bit about this and why it is important?

Vicki McIntire:

Yes, I sure will. As required by the US constitution, the Census Bureau is to count every person living in the United States every 10 years. We attempt to count everyone once, only once, and in the right place. This means the population in all 50 states, the District of Columbia, and five US territories. Our primary, our constitutional obligation is collecting the data that’s used to determine the number of seats for each state in the House of Representatives, but the data are also used by decision makers at every level of the government to plan and provide services for their communities over the next 10 years, until we take the next census and deliver those numbers again. The cycle then starts all over. So the census state is used for redistricting purposes, such as defining the representative boundaries for congressional districts, state legislative districts, school districts, and voting precincts.

Vicki McIntire:

Governments also rely on this data to ensure voting rights and civil rights legislation. Decennial census data along with other Census Bureau collected data are also used to support important government functions, such as calculating monthly unemployment, crime, and poverty rates as well as health and education data. Our data collection efforts results in the annual appropriation of hundreds of billions of dollars in federal funding for important community based initiatives like infrastructure and critical public service programs. Things like road improvements, school locations, hospitals and healthcare services, the allocation of funds for neighborhoods, fire and emergency response services and hundreds of other programs.

Ashley Kilgore:

And what were some of the problems or challenges associated with the decennial census that you worked to help address?

Tammy Adams:

We looked back at how the 2010 census was conducted. Anticipating the challenges and issues for the 2020 census, we needed to rethink our approach. And we targeted the decennial operation that by far incurs the highest costs, our non-response follow up operation. This is our field team of door knockers, whose job it is to visit the 60 million households that did not self respond. We only have one chance every 10 years to get this right. There was no doubt in our minds, we had to modernize our operations or we jeopardize everything in our constitutional mandate. We needed to go from the stone age to the space age.

Ashley Kilgore:

Can you walk us through how your solution came together?

Tammy Adams:

Sure. In 2010 temporarily hired enumerators or the door knockers we just talked about were given their own cases and chose when and in what order to work those cases. This led to obvious inefficiencies. For example, to get more cases, the enumerator would have to meet with their supervisor in person. The census wanted to modernize this approach. MOJO assembled information on what cases had to be in a given day and which enumerators at what times were able to work them. It then provided efficient routes to maximize the number of cases completed while minimizing cost to the taxpayer. The solution was remarkably successful. In 2010, enumerators drove an average of 5.05 miles per completed case. In 2020, this was 3.6 miles per case. With over 60 million cases to visit, these savings alone resulted in approximately 85 million fewer miles driven, almost the distance between the earth and the sun.

Ashley Kilgore:

And now could you share what was unique about your approach?

Tammy Adams:

Sure. Commonly, the problem we are trying to solve is called the vehicle routing problem. The goal is often to maximize revenue, in our case number of completed cases while minimizing distance for vehicles to travel with relatively few stops. In the academic literature, this is usually about six vehicles and 300 stops. The most unique thing about our approach as with much of the census is sheer scale. Each one of our problems had over 300 enumerators, that’s the vehicles in our case and 43,000 stops or the cases, the houses we have to visit. And we had to do that 1400 times a night every night for 12 weeks. By breaking the massive problem down into smaller chunks, we were able to overcome this massive challenge.

Ashley Kilgore:

So in my intro, I briefly mentioned what the impact has been for the US Census Bureau since implementing MOJO. Could you show share some more detail on this?

Vicki McIntire:

I’d be happy to do that. So MOJO has taken field operations to an entirely new level. It used to be an interviewer which we call enumerators, would get a paper case list, paper questionnaires, and some paper maps all assembled in a binder. They would have to meet their supervisor in person to turn in their payroll, exchange information, turn in completed work and get new work assignments. Enumerators were also left to plan their day and routed travel using the list of cases and paper maps. MOJO took this era of operations and transformed it into the space age. One where an enumerator received their cases on a mobile device each day, already planned and routed for them. Supervisors could monitor that the enumerator was working and could easily communicate with them via a text or phone.

Vicki McIntire:

No more daily in-person meetings were needed to exchange completed work or get new cases. And payroll became automated and was submitted each day by the enumerator electronically. MOJO also captured the work intent of enumerators, which days and block of time were available or planning to work. A tremendous tool for supervisors who could proactively anticipate staff shortages and take actions to ensure the demanding deadlines for completing the census could be met.

Ashley Kilgore:

Tammy, Vicki, I want to thank you both again for joining me and wish you and the rest of your team good luck in the 2022 Franz Edelman competition. Are there any final thoughts you’d like to share regarding the US Census Bureau’s finalist project before we wrap up?

Tammy Adams:

Sure. Findings from the National Academy of Sciences propelled the Census Bureau forward and how they could re-engineer their field operations for the 2020 census. The bureau agreed, but took an approach that would shun the creation of multiple large resource intensive teams for a more streamlined, targeted approach of fashioning a small group of 12 surgically selected highly talented and experienced individuals in their fields from across the entire organization to build a proof of concept for how to reengineer the 2020 field operations. The original proof of concept encompassed, considerably more concepts and approaches for field operational recasting in addition to the MOJO optimization and routing component. However, once the proof of concept was delivered the team of 12 laser focused solely on optimization and routing to deliver the highest yield and operational efficiency and cost savings. This small yet powerful federal team comprised of members who specialized in research, engineering, algorithmic design, and data science developed, implemented, and operated the core solution that ultimately powered the nation’s largest civilian mobilized operation with efficiency and effectiveness.

Tammy Adams:

Census only had one chance to get this right and get it right we did. There was not the ability to roll this out in phases, nor is there the luxury of any redos. In the end, we reduced the number of enumerators needed previously by half. We moved our hours per case from 1.05 cases per hour to 1.9 two cases per hour, over the entire length of the workload. That’s over 80 hours of work for each of our 300,000 enumerators. Over 27 million working hours saved even with half of the staffing levels and local offices as 2010. These impacts will not be limited to the 2020 census. And the Census Bureau is looking to incorporate the success of the optimizer to all of its survey operations, ensuring it will remain the elite institution for providing the nation with timely, accurate statistics about the nation, its economy, and its people.

Ashley Kilgore:

Want to learn more? Visit resoundinglyhuman.com for additional information on this week’s episode and guest. The podcast is also available for download or streaming from Apple Podcasts, Google Play, Stitcher, and Spotify, wherever you listen, if you enjoy Resoundingly Human, please be sure to leave a review, to help spread the word about the podcast. Until next time, I’m Ashley Kilgore. And this is Resoundingly Human.