The Census Bureau fine tunes almost everything it does

The Census Bureau is still assessing what it learned from a national count conducted during a pandemic. It’s also looking ahead to how it can best operate as a statistical agency, perhaps the premier statistical agency, 25 years into the 21st century. For an update, the Federal Drive with Tom Temin with Census Director Robert Santos.

Interview transcript:

Tom Temin
And you have written quite an extensive blog here, it looks like you thought about this a lot. A lot of bullet points in there. And I guess, let’s start with the first point that I caught, if you don’t mind. The 2020 derived products still to come out in 2023. There’s a lot of work to be done there, too. You are telling the bureau.

Robert Santos
That is absolutely correct. The pandemic led to delays in the census products. We focused much of our resources to get out the, absolutely constitutionally, required counts for apportionment, those are 50 numbers. We got those out as quickly as we could followed by the redistricting files. And since then, we’ve been working to develop the remainder of the releases, the first of which will be coming in May of 2023, called the demographic and housing characteristics files. And that has the basic demographic population counts for sex, age, race and ethnicity, down to a small geographic levels. Followed by in August, there will be a release of the Detailed Demographic Housing Characteristics File A. And all it means is that they were split up, the detailed tables were split up into three products, the first of which will be released in August. And that contains of very highly detailed and granular geography and race and ethnicity types of categories. So there’ll be over 370 separate race and ethnicity categories on that file, including over 1200 individual American, Indian and Alaska native village tribes. So it will be really nicely, highly detailed, very rich data that’s coming out in August, as well as a standard product that we usually put out, which is the congressional district profiles for their governance. That’s always a hit with Congress.

        Join us Apr. 23 and 24 at 1 p.m. EST for Federal News Network's CX Exchange where we'll be hosting leaders and industry experts to get insight on how agencies are embedding CX into their missions. | Register today!

Tom Temin
It strikes me that if the nation was a pointillist painting, the big 2020 count of 360 million, etc., is looking at it from a block away. But the products coming out next, we can walk right up to it with a magnifying glass and see what’s really going on at a very localized level. I think something just came out the other day, that town I live in Maryland is like one of the top 10 diverse places, who knew. I guess you can walk around and see it, but now we have that evidence. Fair way to characterize these forthcoming products?

Robert Santos
Yes, they’re much more pixelated so that you could do deep dives into them. And you’d have to watch out, you can’t go too low, because you might end up with a little bit of noisy data. But it’s really nicely rich data that will tell us who we are as a nation. Paint a very bright, colorful portrait of the diversity that we are.

Tom Temin
And this is really useful for policymakers, but also for people in business that really want to tailor what they do, to whom they expect to serve.

Robert Santos
The uses of data that is that detailed, are only limited by one’s creativity and imagination. There are many applications, both in the business sector, in the government sector, as well as for local communities. The school boards, or community economic development groups, chambers of commerce and businesses love these data. There’s going to be a lot of utility.

Tom Temin
We’re speaking with the U.S. Census Bureau director Robert Santos. And you have done some reengineering now, for 2030 already. I guess to the public, it seems like, wow, that’s really soon. But to census, each 10 years is almost like a heartbeat away.

Robert Santos
Actually, that is correct. And we throw in some extra heartbeats as well. We actually started on the 2030 census in the 2018-2019, part of the decade. So we are now going full steam ahead. There is much to be planned. And what’s interesting is that we are approaching this in a couple of different ways. One of which, is that the Census Bureau itself, is in the process of transforming itself into a 21st century federal statistical agency enterprise. Where we’re doing a paradigm shift, from being a transactional data collection operation where we have surveys, we have censuses, we knock on doors, we solicit individual entities and ask them to provide information to one that leverages existing data. Administrative data from IRS records. We have MOU’s , memoranda of understanding to share data with IRS, with the Social Security Administration, with Medicare, with Medicaid and so forth. These data are very rich and can provide a lot of information, often information that is asked in the decennial census or in a survey, but folks fail to include it. Either, they’re in a rush and they skip over, or they don’t know the answer.

So we’re finding and leveraging those data to create this different paradigm where we start with all the information from different sources that we compiled together, and then we identify gaps, and then fill in those gaps through targeted solicitation. We’re looking, as an enterprise, for this combination of blended administrative data and solicited data to actually be much more powerful and useful, have higher utility for social problems and governance, then the individual data programs that generate them. And so we’re really excited about that. It does have implications to the 2030 census, because as we’re moving forward and testing and such, we’re adopting as much as possible, because we’re in the midst of transformation. We’re adopting, as much as possible, the ability to leverage this different enterprise, this different approach, so that we can create a more efficient and more highly accurate decennial process that capitalizes on information we have. And then looks to fill in for, basically hard to reach populations and historically undercounted folks.

Tom Temin
What about information sharing, say, at the state level? Where there might be much more detailed records about, say, businesses, locations, and ownership and that kind of thing, as the state’s trying to grapple with this issue themselves?

        Read more: Agency Oversight

Robert Santos
What an amazing commentary, you read our minds. We are looking to supplement the federal national data that we have available, with whatever additional sources of information we can find. And we have, for a long time already, been soliciting and working with states to see if we can identify and leverage their administrative data sources from things like, motor vehicle registration, voter registration, things of that sort. And we have had limited success, we respect the sovereignty of each state, and recognize that not everyone is going to want to join the bandwagon to create a larger data product that benefits society. Just because, states or states and they have their own vision of what they want to do with their data. We are accumulating state by state as much information as we can. If we have it, and it’s determined to be useful, we will use it. It does require a bit of processing, because states are free to, and they should, design their data systems and their collections according to what meets their needs for their population. And that does not always necessarily sync with the way that we need it or use it. And so there’s this processing and exploration that needs to be conducted before we can make it fully usable.

Tom Temin
And what about commercial data sources? A lot of agencies are using that for purposes of identity verification in cybersecurity in this kind of thing. But there’s also a lot of information. Again, economic, population, demographic that is in those commercial databases, is that suitable? And is it proper for census to look at those?

Robert Santos
The answer is yes, and yes. The opportunity to leverage information does not lie solely with other federal agencies or departments and the states, it also includes the private sector. We are exploring, and actually use for 2020, some commercial data to help inform us as to things like, which addresses had occupied housing units and things of that sort. So there have been some uses of it, already. We’re looking to expand that, but careful to make sure that we don’t overly rely on the data, because you’re only one CEO or board decision away from suddenly not having it. So we tread carefully, but we are deliberate and we want to seize whatever opportunities we have out there.

Tom Temin
It strikes me that a lot of these sources that augment what Census Bureau gathers itself, even if it doesn’t change numbers necessarily, provides a really good error checking mechanism that makes you more accurate.

Robert Santos
Yes it does. Although, you always have to question, which one is the more accurate?

Tom Temin
Well, you’re the statistician, not me.

Robert Santos
Our bigger approach is that more information is better than less information. And so we’re willing to bring it all in, and line it all up and match it and see to what extent we can exploit it, to create more accurate data and fulfill our mission of providing the nation with quality data on our our population and our economy.

        Want to stay up to date with the latest federal news and information from all your devices? Download the revamped Federal News Network app

Tom Temin
And census operates. I think as we indicated at the beginning, as part of a population of statistical agencies in the government, there’s a whole community. But I always think of census as kind of a big one in the room there. What is going on in the federal statistical community, on some of these data sources and data usage, gambits that are shareable and how is that all changing?

Robert Santos
Oh, well, we do have an interagency committee on statistical policy that is composed of, not only the 13 federal statistical agencies, but also because of the evidence based policymaking act, representatives, statistical officers from all the other departments. And we coordinate and share information about how we are leveraging our data, and working with, and under the guidance of, the chief statistician of the U.S., Karen Orvis and her wonderful staff at [Office of Management and Budget (OMB)], to make sure that we provide consistently and report data that fulfills the OMB regulations and standards, as to things like race and ethnicity reporting and such.

Tom Temin
And you can’t have a discussion about statistics or gathering or data nowadays, without talking about artificial intelligence and machine learning. And you have stated again, going back to your blog, that you are trying to modernize economic statistics. And that AI and ML might have a role here, maybe just briefly explain what that would be.

Robert Santos
The notion is that AI and machine learning are tools. And the tools are only as good as where you apply them appropriately. So our initial forays, and it’s going to expand, our initial forays had been on things like the Economic Census, which is ongoing, we’re collecting data from 4 million businesses right now. And if you look at the application, the survey online that you go through, there are areas where you have to put in things like, what types of services do you provide? Do you put it into text? Or what is the industry that you’re in? And as a respondent records text of what their business is all about, a drop down menu will automatically pop up that takes the text that was written, and uses an AI algorithm to sub select the most likely categories that fit into whatever text that was. And so it eases the burden of having to scroll and scroll and scroll through countless. We produce so many things, and we have so many industries in our country. So it really cuts down that burden, so that they can quickly and easily identify the category for that particular question that’s being asked. And if you think about it, that does two things. One, is it eases the burden, and so the individual is more likely to continue until completion. And the second, is it creates more accurate data, because it basically does what previously had been a laborious effort to have humanize among the millions of text responses, and then having to go through the mental cognitive challenge of taking that and trying to identify what code it belongs to.

Tom Temin
All right, and then all of these things together, when you talk about the data centric operation, use of ML and AI and everything else. There must be a pretty strong technology component to support this new approach. And just maybe, give us a quick survey of what’s going on, on the IT front there to enable all of this.

Robert Santos
Certainly. We are, as I mentioned as an enterprise, the Census Bureau is transforming into a 21st century data centric type of single enterprise operation. And that requires that we take our platforms that historically have been stove piped into individual data collection systems, like the decennial had their own system for collecting data. Current Population study had their own, crime victimization survey had their own, etc. And creating a single platform where it doesn’t matter which survey you’re doing, you can use the common platform to, not only create all the questions in whatever survey that are being done, but also manage the sample, however that’s defined. And it doesn’t matter whether it’s an economic study, annual business study, or a national health interview study of persons. It’ll all be one common platform. Then the data from that platform, because it’s now a single enterprise platform, gets ingested into a large data lake, where all the data’s exist and can be linked together to create more powerful databases to, maybe, answer questions that couldn’t otherwise be answered, as well as to provide the historical data products that are done. And then there’s the data dissemination aspect, where we’re creating a single enterprise type of operation where we can easily disseminate data and have people access that very easily. And it leads to better more efficient data products, including data visualizations and things of that sort. Really facilitates that whole operation.

Tom Temin
And just a final question, with all of these activities going on, and the establishment of memoranda of understanding, and so on. Are people at the bureau, what’s the percentage of people teleworking remote, and coming back into the bureau, these days?

Robert Santos
These days, we are in a unique situation. The Census Bureau actually has upwards of 13,000 plus individuals who work with us across a large Suitland based headquarter facilities, as well as six regional offices, and a national processing center and then contact center, basically a telephone center in Tucson. So what the Suitland operation, which has the bulk of our full time staff, is actually in a high flexible telework situation out of necessity. We’re in the process of taking our 900,000 square feet and remodeling it. The contractors are in place, work has begun. And that will take a little bit of a while. So in the near future, there will be a continued, very flexible telework. In the regional offices, naturally, they’ll be a combination of in office work, as well as some liberal telework type of thing. So we’re in a really good place, I think right now, as a holding area, we develop the new renovated space, at which time we’ll have a full telework policy.

Tom Temin
So you’re remodeling for the purpose of supporting telework better.

Robert Santos
In fact, one of the things if you look backwards, the pandemic and the remodeling, for the last two plus years, required that all of the headquarter staff be in maximum flexible telework, because we had very limited space in the headquarters. That’s continued. So we’ve had a long experience with it now. And we found that we can, in fact, continue to work and thrive. So we’re not afraid of having to be in that situation. And in fact, there are people in family situations who embrace that and find that those types of situations work. So we’re looking to the future to see how we can take our revised new facility, and match the benefits of having people come together and meet each other and have that in person synergy that can happen only in person, mixed without some flexible total work as well.

Tom Temin
Yeah, so no traditional cubicle farm for all 10,000 people.

Robert Santos
Well, yes and no. No, in that there will be a telework situation. But yes, in the sense that we are doing a hoteling type of operation, where we take a space, essentially be large cubicles, as well as offices, and then folks will be able to say, I’m going to be here. I take this space for these days, and I’ll be working at home otherwise.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.