The 2020 Census: The Risks of Getting It Wrong

Thursday, April 18, 2019
Brian Snyder/Reuters
Joseph J. Salvo

Chief Demographer, City of New York

John H. Thompson

Former Director, Census Bureau, U.S. Department of Commerce (20132017)

Marta Tienda

Maurice P. During '22 Professor in Demographic Studies, Professor of Sociology and Public Affairs, and Director, Program in Latino Studies, Woodrow Wilson School of Public and International Affairs, Princeton University

David Westin

Anchor, Bloomberg

Panelists discuss preparations for the 2020 U.S. Census, including the potential influence of disinformation, changes to the nature of Census questions, and the implications of undercounting certain populations.

WESTIN: OK, all right, fine, thanks. OK, so, welcome. It’s great to have you all here. This is, of course a Council on Foreign Relations event. And it is entitled, “The 2020 Census: The Risks of Getting it Wrong.” I’m David Westin. I’m anchor of Bloomberg TV. It’s great to have you all here. I’ll be presiding today. And it’s my honor to introduce a very distinguished panel. It’s a terrific panel that will have a lot to offer all of us.  

So we have Joe Salvo, who is the chief demographer for the city of New York, which is a pretty important job. And then Marta Tienda, Dr. Marta Tienda, who’s the Maurice P. During professor in demographic studies and professor of professor sociology and public affairs, and director of program in Latino studies at the Woodrow Wilson School of International and Public Affairs, Princeton University. Which is a pretty long title. And then we have John Thompson, who’s actually on video conference here. He’s the former director of the Census Bureau, U.S. Department of Commerce. And he has had a fair amount of experience through, I think, twenty-seven years, as I recall, with the Census Bureau. John is joining us from Bend, Oregon. I think we’re all envious that he can be in Bend, Oregon. 

And so we’d like to start, actually, with John. Since he was a responsible and significant part, I think as the associate director, back two Censuses again. And was responsible for part of preparing this Census, the 2020 Census. So, John, why don’t you start with just a few minutes here telling us, you know, what’s involved taking a census and why does it matter so much? 

THOMPSON: Certainly. Can you hear me? 

WESTIN: We can. Beautifully. And we can hear a phone ringing too. 

THOMPSON: I can hear you just fine. 

WESTIN: Great. Yeah, we can hear you, John. Go ahead. 

THOMPSON: Perfect. So let’s start with why the Census is so important. One, it’s constitutionally mandated in Article 1 Section 2 for the reapportionment on the Congress. The data is used to redraw Congressional districts, both state and local. It’s used to allocate over $900 billion of federal funds annually. It supports small and large businesses for making decisions. And importantly, it’s the control for just about every demographic survey that’s taken in the United States. And the Census data carried forward, as post-censual estimates, are used to make those surveys representative. So that’s why there’s so much concern over the accuracy of the Census, because if there’s a bad Census the results will be with us for ten years. 

So let’s talk a little bit about what happens in the Census. So the basic Census process has been in place since 1970. And under this process—it’s a four-step process. You prepare an address list, you mail a paper questionnaire, you capture information from mail returns, and you conduct in person enumeration of the non-responders. And up until through 2010, that operation was a paper and pencil. So the two—the 2020 Census is going to be a little bit different. They’ve changed the way they do address list development. They no longer walk the entire country. They use modern geospatial tools. They allow the internet as a self- response option in addition to using paper and a questionnaire, and also for the first time using a telephone. 

They’re going to use administrative records to some degree to reduce the amount of work they do in the follow-up for those households that don’t self-respond. And very importantly, they’re going to use automation in the form of mobile technology to equip the enumerators and their supervisors to be much more efficient when they do the follow-up for those individuals that don’t self-respond. Importantly, though, something that started in 2000 is still going to take place. And that is what they call an integrated communications and partnership program. 

This program was instituted in the 2000 census and continued in 2010. And the basic idea here is to get out two messages through paid advertising and local partnership. One message is why it’s important to respond to the Census—and that’s very important. That varies from community to community. And the other message that’s critically important is that the Census data is confidential. The Census Bureau doesn’t share information with anyone. And in fact, whenever they public data they make sure that they do not disclose information that would identify an individual. So that’s sort of what’s happening in this Census versus other Censuses. 

WESTIN: OK, John. Thank you very much. It was a great summary. And you have the patience of Job for listening to that phone ringing again, and again, and again. As I understand what we’re going to do now is drop this call and bring you right back, John, because they say they can’t get rid of that ringing. And you’ve put up with enough there. So we’re just going to drop you for a second. We’re going to have you right back. 


WESTIN: In the meantime, Marta, I think a lot of us if we paid attention to the 2020 Census it’s because of one issue, which is the citizenship question. That’s gotten a fair amount of publicity. It’s up on the Supreme Court right now. Explain what that issue is, and why it matters? 

TIENDA: Well, the issue is that in this current climate—which is the backdrop that will aggravate everything else if—under the best of circumstances, but these are not the best of circumstances—this is a question that has not been asked in the Census—the long form—I mean, the short form that everybody gets—since 1950. At that time, 7 percent of the U.S. population, just short of that, was foreign born. That went down. And by 1970, which is the modern census that we’re improving on, that was under 5 percent, the foreign-born for the whole—for the whole nation.  

So that was asked in a way that—the conventional question was: Where does—where was this—in what state was this person born? But, what state, parenthesis, or foreign country, close parenthesis. And if foreign born, you would indicate what country. And then later questions in the current population survey and other surveys that are based on the decennial Census would use that form. And they would ask questions like, when did you first come to the States, et cetera. 

This Census—this question—and this is really important—is a totally different question that has not been tested. And it begins: Is this person a citizen of the United States? That presupposes that everybody understands the difference between a legal permanent resident and a citizen who’s been naturalized or who has been born in the United States. That is not the case. People use the word “citizen” to represent any non-foreign—anybody who wasn’t born in the United States as a non-citizen. So that dichotomy is not generally understood by the total population.  

It says, yes, born in the United States—so that should take care of the majority. Born in Puerto Rico, Guam, whatever, yes. Yes, born abroad of U.S. citizen parents. So it goes through the logic, yes, U.S. citizen by naturalization, that’ the fourth, and then, no, not a U.S. citizen. So what it’s doing, it’s inverting the general way that the question has been asked. From where—was this person born in what state, which is asked in all international censuses: Where was this person born? Where did you live five years ago, if they don’t ask that? So that already conjures up this fear of you’re trying to target non—immigrant, or foreign born, or the undocumented is not the same as an LPR.  

So that’s the—that’s one of the big challenges that we have. And it’s scaring people. So that item, in all the surveys, has had a high non-response rate. It’s double or triple the rate of most Census items. So we have item nonresponses, generally 2 percent on most items. It’s 6 percent on this item. So already on an item basis, it’s much higher. 

WESTIN: So, Marta, let’s just pause there for a moment because, Joe, you’re, as it were, on the street, at the retail level, actually getting this thing done. If that set of questions is included in the Census, what is it going to do to you in New York? 

SALVO: Well, one of the things that I’ve been talking a lot about lately is about what happens in a Census when self-response plummets. As you just heard, since 1970 the Census has been essentially a self-response instrument. It relies on a big share of the American population responding. The Census Bureau’s own research shows that those responses tend to be better, higher quality, than when the Census Bureau needs to pursue you in what’s called non-response follow up, or NRFU. We have a whole bunch of acronyms. I’ll try to avoid them. But non-response follow up. When you start to engage in high levels of NRFU, you introduce all kinds of errors. And people who are reluctant to response frequently will not—will resist a non-response follow-up.  

So if the workload on the Census Bureau is very high in NRFU, what happens is people go out, thousands of people—many in New York went out in 2010. And they have to talk to the neighbors. They have to knock on doors and talk to people who are reluctant to respond. My favorite is, though, is they’ll talk to a landlord in a building in Queens. And they’ll ask the landlord: How many people are in your basement? And the landlord will say, what basement? (Laughter.) So I’ll talk to a next-door neighbor. And the next-door neighbor says, yeah, I see people coming and going but, you know, I really know how many people are there. What is the enumerator to do, especially under pressure, going back four, five, six times? Very difficult job. Introduces error.  

If they can’t responses but they know there are people behind that door, the Census Bureau will resort to something called statistical imputation, where they will use the neighbors—literally an algorithm based on other people in the neighborhood to assign characteristic and assign a count to that basement. So let me ask you, Bedford-Stuyvesant right now in northern and central Brooklyn is experiencing an interesting pattern. We have the Orthodox Jewish population coming in one way, we have Afro-Caribbean and African American pop that’s been there for a long time. And then we have non-family households from the rest of the country, young people migrating in. So let me ask you a question, would you rather have those people self-respond than impute the cases for those people who didn’t respond based on their neighbors? Depending on where you end up, you can really distort the data. And as you say, I operate on the ground and I see the distortions. I see what happened in 2010.  

WESTIN: Well, put some dollars and cents behind this. Because certainly we’d like an accurate Census, but there’s real money—I mean, I’m talking about capital R real money that could turn on this, right, Joe? 

TIENDA: Yeah, there’s big money. So the cost of the Census has been going up over time. There’s no question about it. We have technology to do a good job. And we can do mixed methods using, as was already pointed out, that the geospatial address list, et cetera, instead of doing pounding the pavement and listing and relisting. There’s a lot of movement, a lot of development, as well as people coming and going. Internal migration is very dynamic in this country, as compared to others. But the problem is, so in trying to reduce these costs by putting a question that’s going to increase non-response, we’re actually adding to the cost of the U.S. Census. 

WESTIN: Well, that’s the cost of the Census. 


WESTIN: I’m talking about what turns on the Census in terms of federal funds being allocated? 

TIENDA: They will be misallocated. 

WESTIN: Joe, what does it mean for New York? 

SALVO: Well, your heard John say 900 billion (dollars) is the federal pot annually. About 73 billion (dollars) comes to New York state every year in programs that have an element of population—in other words, where population is one usually of several elements that makes—determines how that money is allocated. There is a lot of money at risk. But let me say this, we have representatives in Washington that provide us with the programs that bring that money in. If we don’t get the representation to begin with then we don’t get the programs, and the money all of a sudden, you know, is not the issue. It’s the ability to bring the dollars to the state, it’s the ability to pass legislation. And therein lies the real rub. 

WESTIN: So this raises the question of why. I mean, why is the Department of Commerce doing this? And by all accounts, Wilbur Ross, the Secretary of Commerce, has said he wants to do this. And he says he wants to do it, as I understand it, because of the Voting Rights Act. The Voting Rights Act, he says, we need this data. 

TIENDA: That’s false. 

WESTIN Well, explain that. Why is it false? Because that’s the government’s claim. 

TIENDA: It’s sheer—well, he backtracked, because he—in testimony they showed that that was not true. The best way that—the way we use the ACS, because the American Community Survey continues—every year we do an additional community survey. You can pull five of them and get a precise estimate at the block level. And what’s good about that is you drop the last year, you add the next year. So you’re always more current. If you get a count on the Census—you only do it on a decennial basis. You’re already out of date by the end, and it’s going to—these errors accumulate. They’re not just one-time errors. So they’re going to accumulate over time in a dynamic population that is redistributing. So that’s what’s at stake. 

The other thing that’s at stake, and this is—this is worrisome—that it’s going to affect the distribution of representation, that is by some estimates the number of seats that could be at stake, just from the citizenship question, because it’s going to disproportionately affect some of the larger states, and the large cities, is up to ten. So there are some of these estimates that they are currently generating. But then when you add to that the reapportionment—and after the—after the fact, all games on. You know, whatever politics can play with. Where you have a good count. If you have a good count, the rest—politics, and the Census, and gerrymandering, and restricting voter registration, that’s one thing. But not getting the count right in the first place only will compound that over time. And it is missing the fundamental purpose of the Census, which is governance.  

WESTIN: So the fundamental importance of the Census, is that the issue in the courts? Because a number of states, including New York, have sued over this, saying you shouldn’t include that question. It’s in the Supreme Court. There’s going to be an oral argument next Tuesday on the Supreme Court. It’s expedited. They jumped over the court of appeals, took it straight from the district court up to the Supreme Court. But is the fact that you have NRFU, is that the new term I’ve learned her, is that a basis for the Supreme Court to say you can’t do it, if your NRFU’s too high? 

SALVO: Going back to what Dr. Tienda said earlier, it takes years to test the Census question. There’s going to be a new question on the 2020 Census on race, for example, that’s going to allow—if you check off “white” you can write in your group. If you check off “black” you can write in your group. That question was under test for eight or nine years before the feel they got it right. The scientific basis for putting a question on the Census is laid out in a whole series of publications that have been put out from the National Academy of Sciences, the Committee on National Statistics. There are statistical practices that are out there, and practices and policies of federal agencies that need to be followed. And those practices have not been followed here.  

Let me say this, to go a step further, in the litigation it came out that the Department of Justice really never asked for the question. 

TIENDA: Right. 

WESTIN: That was the claim from the Department of Commerce, that the Department of Justice asked us to put it. And it turns out, not so much. 

SALVO: Exactly. And it turns out, it wasn’t the case. There was—in December of ’17, that letter went out. But the decision on this—on including a question was made months and months before that. What came out was that essentially the secretary of commerce asked the Department of Justice to do this. And you know, that in and of itself is a major issue, obviously, in this litigation. But what it comes down to is whether the practices as scientists, as people who know the Census, whether those practices were followed. And they were not, clearly were not. The chief scientists at the Census Bureau said—testified, essentially, they were unable to do the proper testing of the question. 

Now, just to give you an on-the-ground example, in the year 2000 the Census Bureau decided they were going to change some of the wording on the Hispanic question. The Hispanic question, it asked you whether you were Puerto Rican, Cuban, Mexican, or other. Under the “other” they had examples that they used in the previous Census. And they would say Ecuadoran, Colombian underneath. In 2000, they changed the other wording to “Other Hispanic, Hispanic or Latino.” They didn’t test it well. We lost in New York City thousands of Dominicans who wrote in “Other Hispanic, Latino,” because they saw the examples underneath and they didn’t write in Dominican. We had to adjust our numbers. And in fact, the Census Bureau went back and adjusted their own numbers and admitted they had inadequately tested. 

Now, think about this. That is a wording in one category of a question, and it screwed things up. Now we’re putting a whole question on without adequate testing. It has not passed the scientific—the scientific basis is not there for doing this. 

TIENDA: And I want to emphasize that it’s inverted from conventional practices, the way the question has appeared all of the surveys, asking where were you born and then asking, if a foreign country where is that country? Now they’re doing it in reverse. Are you a citizen? And that is intimidating because in 1980 that was the last time any question was added to the 100 percent questionnaire. It was the are you Hispanic, Latino, et cetera. And even then, in 1980, we now know that question well—although it continues to add new dimensions when they modify any verb, or any noun, or any example. But it took a long time to get that clear.  

There are people in the central and southern parts of the United States that didn’t know what a Hispanic was. That was the case in ’80 and ’90. And they would look for the word “Amer.” Anyplace you could see “American,” and they reidentify from American because they were from Central America or, if it didn’t appear that way, they’d put “Amer.” So the presumption that what scientists know about questions is understood by the general population? No, the academics know. Certain segments, but not all. 

WESTIN: Well, just on that point, just for a moment, we unfortunately can’t get John Thompson back on the phone. The connection we couldn’t get going, and I’m really sorry not to have him here. But to pick up on that point, scientists versus the general public, let’s talk about the Supreme Court of the United States. You folks are skilled statisticians. You’re demographers. You know this stuff backwards and forwards. You spent your career on it. I am a lawyer. I actually clerked on the Supreme Court. Trust me, they don’t understand statistics. It’s not their job. It’s not why they’re there. And as I think about this argument in front of the Supreme Court, there are things that scientifically you could demonstrate are wrong. But the Supreme Court said, yes, but it’s up to the department of—Congress, the president of the United States to decide that. That’s their right. And we’re not going to second-guess them on that, because we don’t have the expertise. So how does that argument go, Marta? 

TIENDA: Well, having served as expert interest for lawyers, I can tell you there were two kinds of lawyers: Those that did have statistical understanding and those that didn’t. And there was a lot of extra work with the latter. But it is possible to give examples that illustrate what is at stake. And hopefully there will be able to show the driver being a partisan goal to deliberately undercount. And that’s what needs to be shown. The examples that we’ve just heard, Dr. Salvo, showed what the consequences are for minor changes in wording. And that has to be understood, because there is a series of regulations of what is the order for these statistical testing. They have been totally side-skirted. 

WESTIN: Joe, you’re very familiar with this case. And it’s expedited, as I said. You’re not even going to have the benefit of the court of appeals to review it beyond the district court. And as I understand it, time is a-wasting. I mean, it’s being argued on Tuesday, and as I understand it they have to got to get this thing resolved within what period of time—sixty days, ninety days? 

SALVO: June is the— 

TIENDA: By the end of June. 

SALVO: June is the—it’s essentially in June. It takes a long time to print 150 million pieces of something, as it turns out. And they need to commence with all of that in June. So, yeah, that’s why the court—you know, that’s the rule. 

TIENDA: Well, this—you know, politics in Census is not new. In 1970, there was a big driver to try to get the Hispanic question on there. They fought it and fought it, and then it was turned over. So they had to throw away a lot of questionnaires and print it. And they put it on the 5 percent samples. Got a terrible count. And that’s why they led to the ’80. But there was a big drive for that. It didn’t happen just because the Census Bureau though, oh, we have to do a better count of the Hispanic population. It was the fastest growing. They knew something had to be done. So there is a precedent for trying to change things. But there was a very long process. And they did not change the 100 percent questionnaire until then next decade. And they had a whole decade to test and pretest and understand what versions of the question needed to be modified and what kind of adjustments they would have to do after the fact when they get the response error. 

WESTIN: We’re going to open this up to questions in just a few minutes, so you can be thinking about your questions. But before that, I want to turn away from the citizenship question to a broader question. If this issue had never come up, Joe, what concerns would you have about the methodology that’s being applied in this Census? 

SALVO: First off, it’s the first time that the internet is being used as the primary method of data capture. The Census Bureau expects that there will be obvious advantages to real-time intake of data. I like the idea, you know, of having real-time intake. There were problems in the 2010 Census in New York in specific neighborhoods that could have been averted if the regional director was able to monitor the enumeration as it was occurring, which should be possible now. The flipside of this is that there’s differential access to the internet. We have internet deserts in the city. And if you consider—every day we reach about breaches. People are very nervous about putting their data out on—you know, on that—on their computer or on their phone, actually. There will be a phone app for the Census. And we don’t know—the Census Bureau has been chronically underfunded this decade. And I want to say it goes back a number of years, actually, where the testing has in some cases been inadequate and of concern. That’s one point. 

The other is that the Census Bureau is trying to reduce their workforce in NRFU, in non-response follow up, by using administrative records. That is, everything from tax returns to other types of federal information, food stamps and the like, to try to provide evidence that someone is in a unit at a particular address that—and, if need be, assign characteristics—all in an effort to cut the workload. And they have made assumptions as to how much that workload will be cut. And we’re very concerned about that because if that workload is not cut—in other words, if that workload booms because people don’t respond on their own, the Census Bureau is going to be short-handed. We have a very much lower unemployment rate this time around—much, much lower than in 2010. And the Census Bureau essentially has cut the number of enumerators from six hundred thousand nationwide to about three hundred and fifty thousand, on the premise that they’re going to be able to do these things—these things that have not really been tested adequately. 

WESTIN: How much more money do we need than what we’re spending, Marta? How underfunded would you say it is? 

TIENDA: Well, they’re cutting—there’s so many parts about the labor market. Yes, it is very tight. They’re also—in the past, you know, normally you have to be a citizen in order to participate in the Census, but they usually provide a waiver for that because it’s a one-time really intense operation. They haven’t done that yet. So that’s an issue. But in answer to the question that you had asked him, what would I worry about? Well, here, we’re the government and we’re here to help you. Please tell us all the information we’re asking about you. You should trust us. And if there is a misinformation campaign comparable to what was done—I mean, if there are errors or omissions, if the statisticians know how to fix it, it won’t be the best count but can be—it can be a usable Census. If there is a breach so that—so that there are information campaigns, falsehoods as we saw during the elections, saying: Don’t answer. They’re going to really try to do raids. They’re going to try to get your grandmother, et cetera. That can do a lot of damage. So technology can be your friend and it can be your worst enemy. And I don’t think that they have thought through that, especially when you think about how much security one has to maintain. 

You know, when I listen—on the internet—I listened—I was—a Lyft driver recently. And he said he gave up his job at Best Buy. He was tired of people not understanding that if they came in and bought a router, that that wasn’t sufficient. That they actually had to have an internet connection in order for the router to work for Wi-Fi. And that level of misinformation and lack of understanding—you know, those of who are connected, we just take it for granted. Your cellphone works. You don’t know why it works. You know it works in some places and not in others. But what does that mean for you to have access? Not everybody has access to a library. 

And here’s where the rural-urban divide is going to come up, and possibly have unintended consequences. So that division, the regional divisions, even though large cities are going to be affected by one question, rural areas may be affected by the lack of internet connectivity, Native Americans, et cetera. So there are so many ways that we could do a better job. Shame on us if we don’t. 

WESTIN: Joe, is there any solution to the sorts of issues that Marta has just raised, that you have raised, a practical solution? When it comes to the citizenship question, Supreme Court can just say yes or no. That’s easy. But the things you’re talking about, that would require really active supervision of the methodology, of the budget, and things like that. the Supreme Court’s not going to be able to do that. And this administration has a somewhat different attitude toward immigration than prior administrations, I think it’s fair to say. Can Congress do anything? 


SALVO: I want to just talk a little bit about the two-pronged approach that we have here in New York City, if I may, and then I’ll get to the question about Congress. We have a two-pronged approach. One is to get out there and to try to inspire people to stand up for who they are. I was in Queens at a Complete Count Committee meeting a few weeks ago. And I put the question up that’s going to be on the 2020 Census. And I said, this question will define you. This question is going to determine whether Queens is the most diverse place—according to people in Queens—on Earth. I’ll take the U.S. as a point of comparison. (Laughs.) People responded to that. Essentially if the response is low, and the data do not come in on this question, the statistical reality will not confirm what you all believe. And it’s messages like that that we’re trying to bring across. That’s the messaging side. 

Then there’s another side to this, which where I—the effort that I lead. We’re working hand-in-hand with the Census Bureau in trying to develop training, how to identify those basements, make sure you walk around the block, not just on the block face. We did two and a half years of field work. You heard John talk earlier—John Thompson—about the address list. We worked two and a half years to get the address list right, OK? That’s the first step. That’s the foundation. If you don’t have a recognized address in the Census Bureau’s master address file, you don’t get enumerated, right? You don’t. You have to be placed in a particular location that gets recognized. So we work very hard to do that. But where now we’re working is on training, helping the bureau train their own people. So it’s these linkages. 

Now, you asked about Congress. A few weeks ago I was up on Capitol Hill talking to our congressional delegation and other people, trying to educate them about this. The appropriation for partnerships, the appropriation for what you heard referred to as the integrated communications campaign and I’m happy to say the appropriation for questionnaire assistance—the original plan had no questionnaire assistance centers on the ground in the neighborhoods, none, zero. You call on the phone. Telephone these days is not exactly the best way to get messages across, all right? But the stakeholders all got together, and we pressed the issue, and there’s an allocation now for physical, on the ground, questionnaire assistance. The Congress has provided for that, OK? 

So we are working closely with our representatives to get the dollars that we are going to need to mobilize. That is a critical feature. And Congress, you know, is—I don’t know how many of you know this—but Congress is basically in charge of the Census, as per the Constitution. Congress delegates it to the Commerce secretary. So Congress can actually push back on this. 

TIENDA: They’ve had this before about who are we counting? Do we count inhabitants? Do we count persons? Do we count citizens? Do we count voters? We’ve had these debates and they’ve gone to the courts in the past. And in the past, there has always been a support to count everybody once and in one place, not twice or, you know, those people who have multiple homes. But this has become an issue. And Congress is aware, but Congress is also polarized. So the polarization that we see is manifesting itself in the way the operations should—they should do the best count possible, and then fight out the politics after the fact, if they want—if use the data for political purposes, that’s different. But to get—not to get the right count has monumental consequences for wellbeing and inequality.  

And not all cities have the kind of Joe Salvo leading their Census effort. You go to some of the smaller cities, or go to south Texas, you won’t see the kind of orchestrated, concerted effort to try to capitalize on diversity. That’s not going to happen in many parts of the country. 

WESTIN: Not that we have anything against south Texas, mind you. (Laughter.) 

TIENDA: I think it’s pretty good. I’m from there, so. (Laughter.) 

WESTIN: It’s time to open this up to questions. And I suspect all of you know the rules, but I’ll just go through them again so everybody knows. First of all, this is on the record. So everything that’s being said is being recorded and streamed, and things like that. We’d like you to wait for the microphone, use the microphone, stand up, identify yourself and your affiliation, ask a question. It would be better if it was a question and not a speech. It would be nice to have a question mark at the end of it. (Laughter.) And make it not too long. But we’d love to hear from you, because we want to know what you’re interested in. So questions, please.  

There’s a mic right there, coming to you. 

Q: Maryum Saifee, CFR international affairs fellow. 

So historically Arabs and Iranians have been classified as white. And there was a push to sort of reclassify them as, you know, Middle East or North Africa. And what—you know, first question, what is the rationale behind that decision? And second, how does this affect resource allocation and data collection? 

SALVO: Part of that eight or nine years of tested resulted in the Census Bureau recommending the inclusion of MENA, as we used to refer to it, Middle East North African, because, as you well know, we’ve had this frustration of people feeling like they didn’t have a place for themselves on the questionnaire. That went to the Federal Office of Management and Budget who has the ultimate—the executive branch—who is the ultimate decider of this. And it turned out that, you know, the Federal Office of Management and Budget takes thousands of comments. There were a lot of comments on this. You know, there were federal register notices and there were comments that people provide. And there were several thousand comments that they had to take into account. And their ultimate decision was not to include that. Their decision, though, was to go with these modified categories that I described earlier. But the ultimate decision, based upon all the feedback that they got, was not to go with that.  

To those of us who were here in New York, I can tell you that was a disappointment, as it is to—obviously to members of the Arab community. I don’t want to call it good news, but you can this time around check off “other” and write in, you know, Egypt—write in what you—your category. And the question that we’re asking the Census Bureau right now, though—everybody should know this—is: What are you going to tabulate? What are those tables going to look like? Because the Census Bureau has a—is trying to move into the 21st century. The world of cryptography is influencing this decision. They are instituting more heavy-duty, let’s put it this way, privacy measures to protect the privacy of individual respondents. And in the process, we want to make sure we get data. You know, you can up the noise in the data to the point where everybody’s protected, but we don’t get any data. So there’s going to be a debate about how detailed those race and ethnic tabulations are going to be. 

TIENDA: And there’s another reason—another challenge there. The Census Bureau does not like people to say, “other race.” It makes them crazy. They have spent an enormous amount of time, and effort, and money trying to get the Hispanics to not check “other race,” because you have to fit in one of these boxes. We have this pentagon and it’s sacred and check off one of these boxes. Well, that didn’t work that way. And they spent so much time and effort to try to change the share of the Hispanic population as some other race. And so that—they’ve been trying to deal with this.  

And instead of having two questions, and even though they always declare Latinos are not a race, it’s an ethnic group. It’s always their little proviso. In reality, it’s become a racial group, because it’s Hispanics, and blacks, and whites, and Asians, and others. So that’s been a really big challenge for the Census Bureau, to try to figure out how to classify race and ethnicity. They don’t like the subjective classification because peoples ancestries and their identities change. Once they go to college they suddenly realize who they are or who they’re not. And they change. So it’s a fungible category and it makes it very difficult for purposes. 

But people are what they are. You know, gender also. So now we’re getting to the point where we are allowing this diversity. The real question is—but it takes a lot of time to hand-code that kind of material. And even with machine learning, when you’re writing things in, that’s still a formidable task. And there are big decision rules.  

WESTIN: OK. Other questions? Right here, sir. There we go. Microphone right behind you. 

Q: Thank you for this presentation. I’m Joel Cohen from the Rockefeller University and Columbia. 

A number of major corporations are reported to be supporting a complete count effort because their marketing and planning for installations depends on the accuracy of the Census. I didn’t hear any discussion—would you talk about the possible role of big business as an ally in the struggle to get a Census that actually has correct information? 

WESTIN: Joe, are you enlisting— 

SALVO: Where’s John? (Laughter.) 

WESTIN: I know. We miss John. We miss John. 

SALVO: I will tell you what John would say, OK? John, when he was director, was joined at the hip with Google and with other big operations in an effort to try to get the real-time intake of data through the internet, get it right, and to protect the Census against all kinds of cyberthreat. That has been their role in trying to get the enumeration right. You know, think about what UPS and FedEx do, OK? All that technology is going to be brought to bear on the Census. And John actually during his tenure tried to make sure that that would happen. But that’s one side of it. 

The other side of it has to do with the application of the data. And we know all those marketing studies are driven heavily by what comes out of the American Community Survey and the decennial Census. And I can’t emphasize enough something John said: You take the decennial Census, that’s the definition of your universe. All of the sample data that gets collected for a whole decade is driven off of that image of that description of that universe. So if something gets messed up—if you don’t have enough children or the black population’s undercounted, it gets carried over into everything that happens thereafter. So it is truly the case that you have to live with it for ten years, which is why obviously many of us are very concerned. 

TIENDA: Yes, it’s even longer because by the time they get all their new sampling frames and try to do all the correct, it’s a problem. And there’s another—well, I would say it would be important to see who—if there were any amicus briefs filed that support the Census Bureau in this—in this effort. So I don’t know whether the big corporations have done that. They should have—and universities should have done it, because that’s— 

WESTIN: They did? Is that right? Because I don’t know the answer to the question. I should. 

Q: (Off mic.) 

TIENDA: Yeah, because they did in several other cases, like affirmative action and the like. So that’s an important piece. Even if they’re not statisticians, they may at least understand the business side of this—of this case, and what implications that has. 

The other issue is that there’s differential impact of the undercount. So the share of households that have at least one foreign-born person is about 14 percent. They’re not evenly distributed, as we know. So the real question is, it’s not just the percent or the number of people, but it’s also the number of people in households where there’s one or more foreign-born people. And if they feel—and in the environment where there’s been a threat, and raids and the like, and we’re going to trust—this is going to undermine the trust for the Census Bureau more generally, which is a real problem going forward. And that’s going to be a very difficult process to correct. But the impact is going to be quite differential. And I would hope that we understand that. That’s key for marketing. Anybody who wants to sell something. General Motors wants to know where people are for the kinds of people who buy Chevys. 

WESTIN: Yes, ma’am. 

Q: Hi. My name is Allison Silver. I’m from 4Context, also NBC THINK. 

You both have mentioned several times that in talking about what a terrible count is. I mean, you’ve said it several times. And I think it might be helpful if you would actually give some idea about what you’re saying is a terrible count, and how—and what the ramifications are. Are there a certain number of schools that would be closed? Are there a certain amount of federal programs that would be cut off? I mean, I would like to have some idea, since you guys have studied this so completely, well, what it is we’re talking about. 

WESTIN: I—well, you can start with New York. You know New York terribly well, and then you can broaden it out, Marta. 

SALVO: In 2010, Astoria, Queens and parts of Jackson Heights were heavily undercounted because the Census Bureau declared thousands of housing units vacant. Anybody who knows those parts of Queens knows that housing is under tremendous pressure. The level of—the rising vacancy rates in those neighborhoods and a bunch of neighborhoods in southern Brooklyn would be indicative of abandonment. It was totally implausible that these areas experienced big vacant surges. It was an enumeration problem that occurred in two Census offices that’s been well-documented now. When we go to the health department, and the health department says we have to calculate a vital rate, a rate of illness for children, let’s say, they use in the bottom of their formula, in the denominator, they use the number of people from the Census, frequently by age. And in the numerator they have their disease incidents. They came to us and said: We can’t use that data. You need to adjust that data for us, because we don’t have an accurate measure of whether—of the rate of illness in a particular area.  

This is just one example of what can happen when children—for example, when the number of children is undercounted. And I want to emphasize, it’s differential by neighborhood. In 2010, according to the official Census Bureau calculations, there was no undercount in New York City. It was net zero. But you know why? Because there were neighborhoods in Staten Island which were overcounted. We have duplicates. And there were other neighborhoods, like Bedford-Stuyvesant and like Astoria, that were undercounted. The omissions and the duplicates don’t occur in the same places, see? So we have undercount, even though citywide the picture doesn’t look bad it is very bad at the neighborhood level. 

One other example. We helped schools decide how to draw their zones when populations grow and they have to make a determination as to how to draw a zone around a school, or how to draw a zone around a new school. They come to us and they say: What is going on in this neighborhood? We have to anticipate what we think might happen. So you look at data on births. You look at data from the American Community Survey. You look at data from the Census in order to come up with that definition. And I don’t have to tell you what the school budget in New York City looks like, OK? We want to make good decisions. 

Those are just two examples. I’ll throw in one more. We are projecting by neighborhood the aging of New York City’s population right now. And why? Because we have to anticipate where services have to go. If the base is bad for that projection, the projection’s going to be lousy. Fortunately, though, people sixty-five and over really get involved in the Census. (Laughter.) Sometimes really involved, to the point where they—we get extra people. So in that case, I’m not so concerned. But where I’m concerned is the number of children below poverty in a place like Bushwick—not Bushwick. In a place like—I’m thinking Ocean Hill-Brownsville, for example, and Bedford-Stuyvesant, where you have a lot of children. If those children are missed, we have a real problem because the absolute numbers are not there, and we can’t anticipate what the priorities are regarding services. 

WESTIN: Well, that’s New York. Take it national. 

TIENDA: Well, at the national level the Population Reference Bureau has done some analyses of differential undercount. And the Latino undercount was very low in 2010. They’re projecting that to one of the largest ones if this goes forward as currently planned, and the citizenship question is added. They’ve also focused on age, but at the other end of the distribution. And they’re showing at the national level that the differential representation of poor kids and minority kids is going to be quite egregious. So these are the area that have the worst schools, oftentimes. And they’re the areas that are going to be even more underserved. So compound underrepresentation times underfunding, and we have a disaster. Because guess who’s going to support the elderly population? The young people. But not if we don’t educate them. So this is going to be compounding in a very—you know, if we’re present-oriented we’re not thinking that way. But if we’re thinking about the future of the country and its global competitiveness, this is not a—this is not a trivial issue. It’s a fundamental issue for the future of the country. We cannot afford to—at a time of declining fertility and population aging, to undercount the future resources for this country, which is human capital. 

SALVO: I want to look for a silver lining here. 

WESTIN: Oh, good. Please do. 

SALVO: And let’s try to find one or two. I’ll give you one, OK? I’ve never seen the level of interest and mobilization that I’m seeing now towards 2020. This is my—I always joke I’m a hundred years old. This is my fourth Census, OK? And I’ve never seen the level of interest and mobilization that I’m seeing now. We are just—in New York. I’m also seeing it in other places, but especially in New York. We have the private sector coming to us, the association for better New York is totally mobilized. We have a formal office. In 2010, we did not have a real budget. We have a real budget now for a Census office, a director of outreach for the city. People are being hired to go out. We are just so active in trying to promote the Census. And people are interested. I get calls: What can I do? I realize this is important. What can I do?  

So there’s my silver lining, because maybe through this mobilization—and what I always keep thinking of: Stand up for who you are. Don’t let anybody stop you. It’s a similar message that you might get, you know, going to the voting booth. But here it’s so elementary to our democracy. 

TIENDA: Yes, but I—you know, I don’t want to be a naysayer, and I totally agree, you get on one of these ethnic radio stations, like Chapulin (ph) in California. And if you get on that radio station, then he could create all kinds of fire. You have to registered. You have to do this. And he can start that kind of a mobilization because they have a huge following, and it will outdo any Twitter that you can imagine. But at the same time, I think that we’re—if people don’t get out—if they’re afraid they’re not going to—it’s going to be—it’s going to be a real issue. And what you see in New York is what we could be as a nation. 

And after 9/11, there was a great ad by the Ad Campaign, I American. I am an American. But it was done by all kinds of voices. And it is an amazing—an amazing documentary that—it brought the country together. We need a campaign like that to revitalize the Census, because time is really short. The fact that it is now April, we’re not going to have a decision till June. And what it takes to actually—that last push to get the Census out and to do it on April 1 is not—is a formidable operation. And we’re walking with one of our hands tied behind our back. But there’s still one that’s working. It’s the one that writes. And I hope that that is what’s going to overcome all of the—all of the counter-pressures that we’re currently experiencing. But it will take a village. And we have one hundred thousand villages in this country. And they all have to be mobilized and trust the Census. If we lose trust in the Census, it’s going to affect many of our surveys and data collection systems. 

WESTIN: Other questions? Oh, I’m sorry. I missed you over here. 

Q: (Off mic)—sorry—(comes on mic)—Princeton University. 

We learned a lot tonight about the Census that at least I, and I think many of us, did not know, about the statistics, about the internet. But, Dr. Salvo, you emphasized that it’s important to have an address, that you look at buildings as how many people may live in them. How do you count the homeless, who it’s particularly important, I think, will be counted in San Francisco, in Chicago, in New York? How do you count the people who are roaming around the country in RVs, or a family that might be out of the country for six months, or an expat lease where nobody’s there? I mean, how—if you focus on an address aren’t there categories you’re going to systematically undercount? 

SALVO: Well, the reason why we focus on an address is because essentially apportionment and redistricting are, you know, the essence of the Census. And where you reside, what state you reside in, obviously has a big impact on the distribution for the—of the House, right? So I mean, I can say this, that the residence rule in the Census is where you live most of the time, most of the year, OK, essentially. And if you are in an RV, this is a very interesting—I can’t resist but go to that one first, because it’s the most challenging in some ways. I guess you’ll have to talk about what state you were in most of the year, or something like that. I—you know, the truth is, I really don’t know. We don’t have too many RV people in New York, in the city. So I guess, no, this is better left to the experts that are doing— 

WESTIN: But more broadly, it raises an interesting question. I mean, I know this just from exit polling. That changed fundamentally when people started getting cellphones. Because we used to do it all by calling up landlines. And then when they didn’t have landlines, it really undermined it. Are there changes in people’s lifestyles—whether it’s homeless, whether it’s Airbnb, whether it’s—or not? I mean, are we seeing a change in the population overall that might undermine the validity of the address approach? 

TIENDA: Well, the homeless, they have tried—I think that’s a really important question, ma’am. The homeless population, they would go out before the day of the Census. And they’d do random samples. So they’d get a little mini army here in Los Angeles, and they’d do another one in Santa Monica. That’s very replete now with a lot of homeless people. Austin, Texas. I just came from there, and there were homeless people—I hadn’t ever seen it the way it is now. So they do samples in certain cities. If they have the architecture, the—to actually do some counting estimates. But it’s never with the same kind of precision that we’re going to get an address from. 

So—and then we do the non—the institutionalized population. There are estimates of the institutionalized population. And there are also estimates of the military and the people abroad. So they ask how many people are living here most of the year, which is another issue. Because this was a problem for migrant farmworkers. If you think about the timing, the Census is on April 1. Well, many people from south Texas had already gone on the migrant streams. So their houses were empty there, but they were in an RV. And that’s where they started to do some of the imputation and hot decking, where they would create the household based on the neighbors who didn’t go into the migrant stream.  

But this other issue of how lifestyles have changed I think is going to be a challenge going forward, but I don’t think it’s a serious as some of these other challenges to the undercount. But it’s a part. And where it manifests itself is whether there’s a differential by group—by demographic group, if it’s affecting seniors, if it’s affecting the young people, or African Americans, or Latinos. Who is living in RVs and roaming around the country? That’s one issue. Who is likely to be moving during that data, Census date April 1? So if you’re moving and you want to fill it out in advance, you can do that. Say, where are you going to be on April 1, your habitual residence. So for some of that we can anchor even if the people aren’t there at the time, and the internet allows you to do that. I can fill out—and college students, et cetera, there’s a lot of—we haven’t let them out yet. So all of that affects where people are counted. 

SALVO: The group quarters enumeration is a fixed operation at the bureau. Part of—and we gave the bureau already about two thousand addresses in New York City of facilities, essentially institutions, and what we call non-institutional group quarters, such as dorms. They go to them special—especially the dorms—they go to them early, because students leave, right? So in February-March, they start this, of 2020. The homeless enumeration, there’s a special night—usually referred to as S night. By the way, it’s never called the homeless enumeration. It’s called service-based enumeration. They go to known locations of persons who don’t have—who go to those facilities. So they would go to shelters, they would go to mobile food vans and these places on a selected night in March, OK?  

The homeless who are more transient, we work with the regional director to give them known locations—another acronym—TNSOLs, Targeted, Non-Sheltered Outdoor Locations. We give a list to the bureau, and they send people to these places. Our Department of Homeless Services has a list. We do a census every year of the homeless population in New York City. So we have the known locations. And all of this is put together as the group quarters population, which in New York City is about one hundred eighty-four thousand people out of 8.4 million, so. 

WESTIN: I think here, and then here. 

Q: Hi. My name is Muriel Chase. 

And so you were talking about that 2020 is going to be the first time that internet is really being implemented in this—in this Census. And I was curious, you talked about pros and cons, pros being real-time, which I think is going to be a huge gamechanger. But cons, do you see any concerns on privacy? And could you expand on that? 

SALVO: Yeah, I think people are concerned with the daily reports of our data being hacked and, you know, infringements on our privacy, yeah. And people are very reluctant and nervous about this. The general—you know, you asked a good question about—forget the citizenship question, let’s just talk about the environment we’re in, OK? Put the citizenship question aside. We’re in an anti-immigrant environment. We’re in an environment where our data get compromised. We’re in an environment where privacy and government intrusion is a big concern. All of these things, I’m not saying they didn’t exist before, but everything’s been— 

TIENDA: On steroids. 

SALVO: Yeah. Everything has been amped up in a way that’s got us all really concerned. 

WESTIN: But, Joe, go back to your silver lining for a second. There must be some ways in which that technology is going to help us as well. There’s always risks with technology, privacy and things like that, but there must be ways in this day in age it’s really going to help us. 

SALVO: I’ll tell you how. We’re going to get daily reports of self-response—daily this time. Every day, we’re going to get a report as to which—how neighborhoods are doing. Every one of the 2,168 Census tracts in New York City is going to have a self-response rate. And we are going to have an army of people that will go out and knock on doors if need be, to say: Did you fill it out. You go to this community health center, we understand. It may disappear if you don’t answer the Census. Got to really make it personal. That’s the point we’re at. So, yeah. 

TIENDA: But you think about this. We’ve had a lot of government intrusion in many ways. We want your—we want your email address. We want your cellphone number, your regular number. We want your name. We want to know everybody who’s in your household. We’re not going to use this against you. And we’re going to protect your data because we’ve protected—you know, I was compromised because I was—I got some security clearance at the federal government. So then they said, oh my God, that security breach? Well, I was on the list. And I think, I didn’t even do anything. But because they don’t—the government is notoriously bad for maintaining the up to date systems. 

WESTIN: But there’s your solution to get Amazon to do it. (Laughter.) We give Amazon all that data without any hesitation, right? 

TIENDA: No, I don’t. I’m not a teenager. (Laughter.) The teenagers do that. 

WESTIN: OK. One last question. Here, please. 

Q: You were talking earlier about how the exit poll has been interfered with because of the phones and the cellphones— 

WESTIN: Undermined, yeah. 

Q: Right, exactly. And earlier you guys were saying that the advancements involve using cellphones as opposed to just personally going out and man-to-man, woman-to-woman. And I have been wondering, because fewer and fewer people pick up their telephones when it’s a number that they’re unfamiliar with. I got a call from California from a state agency there, and they began the conversation by profusely thanking me for picking up the phone several times. Thank you so much for picking it up. Thank you so much for picking it up. Is there any plan to handle things if these cellphones aren’t the answer? 

WESTIN: It’s a great point with spam—I mean, you get so much spam. 

TIENDA: It’s not—there’s just so much spam right now that that’s a real challenge, because now it’s your own area code that is being used and hijacked. And then there’s a lot of spam on email that’s saying—somebody you know that just, you know, gets all the context. So those are challenges. We are up to fixing it, if we so resolve. If we want to do it. But that requires resources and keeping the systems—the encryption systems at the highest level, not negotiable. That’s an empirical question.  

Q: So are there plans to do this? 

SALVO: For what specifically? 

WESTIN: To get people to answer their phones. 

SALVO: Well, no, answering the phone is not the—OK, let me say this. You get the request by mail, a letter. Eighty percent of the country gets a letter to go to the internet. Twenty percent will get a paper questionnaire, depending on the level of internet access and a number of other variables. All right, you get that letter. You can go to the internet and answer. If you do not want to answer on the internet, there is a phone number that you can call and you can give your response over the phone, OK? It’s not that the bureau’s going to call—they’ve already given up on the idea that, you know, massive callouts to people. I mean, I get twelve messages on my machine every day. It’s like—we all do. It’s crazy, right? But it’s capture by telephone that they’re trying to institute this time to bump up the self-response rate. OK, so—yeah. 

TIENDA: Right. Right. But you get a confidential number. I was in the ACS when—I did one on the internet and I did one in paper a few years before. And you get this little code. I thought, if they’re going to call me send me a paper, I don’t want paper. I don’t want paper. So I just—you can do it online. You get all the codes, just like you do when you’re doing a survey or whatever. This is your code. You get online and you fill it out. And then you send it. And then it’s gone. 

Q: What I was asking about is the follow-up. You were saying the follow up is going to be by phone. Is that— 

TIENDA: They tried that. They do that sometimes. And sometimes they just go and knock on your door. 

SALVO: No, it’s principally going to be knocking on your door. They’re not going to call you, yeah. No, they’re given up on that idea. 

WESTIN: OK. The time has come to thank everybody. First of all, thank you all for coming and being so attentive and asking some terrific questions. I also want to thank our panel, particularly John Thompson in absentia. (Laughter.) 

SALVO: We tried to— 

WESTIN: I know. We missed John. We can all tell him that. And also Joe Salvo, he’s the chief demographer for the City of New York. And Marta Tienda, she’s a professor at Princeton. Really terrific. To remind everybody, this was all on the record. And thank you very much. And good evening. (Applause.) 


Top Stories on CFR


During Kenya’s state visit, the United States should work toward building a more resilient model of U.S.-Africa partnerships.



Ebrahim Raisi was more loyal to hard-line Supreme Leader Ali Khamenei than previous presidents, and whoever succeeds him is likely to be just as conservative.