The census has always been a weapon of political power

Census volunteer Gail Alexander places a sign in front of the home of Sarah McCloud as they visit homes during the "March to the Mailbox" effort on April 10, 2010 in Miami, Florida. (Photo by Joe Raedle/Getty Images)

(CNN)Months after President Donald Trump's Justice Department filed its request, his Commerce Department announced late Monday that it would restore a citizenship question to the 2020 Census.

Though we don't yet know the precise language, the prospect of reviving an inquiry last featured in 1950 has sparked charges that the administration is trying to depress the response rate in parts of the country with larger immigrant populations. That, in turn, could scramble congressional delegations -- shifting power away from traditionally Democratic bases -- and cause reductions in federal funding to major urban centers
The administration has argued that the data is needed to better enforce the 1965 Voting Rights Act. But the timing -- the decision came late in the process -- means there won't be time for the kind of field testing that experts like census historian Margo Anderson, a distinguished professor at the University of Wisconsin-Milwaukee, say is required for getting the most accurate possible count.
Still, as Anderson made clear during our conversation, the politicization of the census process is nothing new. Partisans have been using it to further their narrow interests since the early days of the republic. Now, as a group of Democratic senators introduce legislation to make sure, as Hawaii Sen. Brian Schatz put it, the 2020 edition is "on the level," and California Attorney General Xavier Becerra prepares to sue the administration, this old fight is poised for a return to the headlines.
    I spoke to Anderson on Tuesday afternoon. Here's how she explained what's happening now, the history behind it, and what we should expect to hear next.
    This interview has been edited and condensed for clarity
    Krieg: Why do questions, generally speaking, come and go off the Census?
    Anderson: The Census reflects whatever the social and political and economic issues of the days are, so questions come and go as those issues change. The Trump administration's proposal to put citizenship on is connected, as they indicate, to their concerns about voting rights enforcement.
    Krieg: So you're saying this has always been a kind of political football?
    Anderson: It is a provision of the 1787 Constitution, designed to apportion political power and tax capacity among the states. So it's never been an academic enterprise -- it's always been a function of the instrumentality of the state.
    Krieg: Focusing now on the citizenship question and the issues around it -- why did it initially come out, decades ago?
    Anderson: The main run of questions on citizenship was from 1790 through 1950, when it was on the main Census form. It originally went on that form as a sub-question of the "place of birth" question.
    So, we start asking, "Where were you born?" in 1850. And then, in the late 19th century, because of mass immigration, we say, "OK, if you were born abroad, are you a citizen? Are you naturalized? Did you take out first papers?" That was salient when we had a period, especially from 1890 to 1924 of very high levels of immigration.
    It remained on the form in 1930, '40 and '50 as a continued monitoring of the processes of migration and citizenship and naturalization. It went off in 1960 and, this is one of the great mysteries -- I don't think there was much fuss about it at that point. We were in a period where a fairly low proportion of the population being foreign-born.
    Krieg: Do you view this, adding this question about citizenship, as an attempt to restrict funds and representation to regions that have higher populations of non-citizens?
    Anderson: Certainly, that's the fear. Now, here's the issue: usually when a change like this is to take place, and this one of the reasons why the stakeholders and people who are close to this are jumping up and down, there is a period of testing. In other words, you propose a question be put on the form, the Census Bureau goes out and tests it, they decide whether it works or not, and then there's a public comment period, hearings in Congress, etc.
    None of that has happened.
    Krieg: We heard some talk about this in January, right? But then nothing, and then this last night -- it felt like it came out of nowhere.
    Anderson: Yes, and that's one of the great concerns.
    I looked at the secretary's memo this morning, but I have yet to see what the question actually looks like. On the form. Are we asking place of birth, the way we used to, and then this (citizenship question) becomes a sub-question if you're not born in the United States?
    Is it just a bold question, are you a citizen in general, asked of everybody?
    At the bottom of the secretary's memo, there is a statement that says: "To minimize any impact on decennial census response rates, I am directing the Census Bureau to place the citizenship question last on the decennial Census form."
    Krieg: That sounds like they're acknowledging, implicitly, that the question will lower response rates and scare people off.
    Anderson: Well, maybe, but even more confusing to me, as somebody who does survey research, the person who confronts this question hasn't yet actually hit "send" if they're doing it on the internet, or stuffed the form into an envelope and put it back in the mail.
    So the problem is that, if somebody gets to that question, they might toss the whole thing.
    This (deciding to put the question last) is a hypothesis about survey methodology. As a technician, what I would say is, "Hmm, that's an interesting idea. What we'll need is a split design where we put the question in the middle of the form, and at the end of the form and then we go out and ask 500 people to fill out the form and we see what they do. Because we don't know.
    Krieg: The act of counting people, in any country, is such an important thing for so many reasons. But it seems to be uniquely contentious here. Why is that?
    Anderson: The main point is that the American census is an instrument of the American state embedded in the Constitution of 1787. "Representatives and direct taxes shall be apportioned among the states within this union according to their respective numbers." That's the language in the Constitution.
    So in addition to the mechanics and technical questions of counting, the US was the first country in the world to take this sort of method, mechanism, and then tie it so closely to its political system.
    You take a Census and then Congress reapportions and states redistrict. That was a real innovation in representative government in the 18th century. But the fallout from that was that, therefore, Americans figured out very, very quickly that the census had profound political implications.
    One thing I always point out to people is that the first gerrymander -- we're talking about gerrymandering a lot these days -- (began in) 1810. So the question of the manipulation, if you don't like it, or the enhancement if you do, of the process is embedded in the history of the country. The census gets tangled up in the Civil War, it gets tangled up in immigrations restrictions, it gets tangled up in the measure of unemployment during the Depression.
    Krieg: What was the problem then, during the 1930s?
    Anderson: In general, the problem with (measuring) unemployment is that it's a rapidly changing phenomenon. It goes up and down. The traditional method of counting the unemployed population, which at that point was done at that point every 10 years, was inadequate to actually use it for policy purposes. So there was a question in the 1930 census about unemployment and there was actually a special unemployment census in 1931.
    But there needed to be, essentially, a technical re-understanding -- namely the development of probability sampling -- that allowed for what we now do, which is a monthly unemployment report. But it took about 10 years to work out how to do that, to define unemployment, and there were big questions.
    Franklin Roosevelt in his press conferences would repeatedly obfuscate the question by saying, "Well, you know, if the husband is unemployed and his wife and his son go to work, you know, how do we deal with this?" So, you had to get political and academic agreement on what we mean by unemployment and of course we now have several definitions. That took about a decade and by 1940, we'd worked it out and life has been fine, in terms of the measurement process, ever since.
    Krieg: But there were more fights -- on the Census itself -- after that, right? It seems like this is a quiet war that carries on parallel to the big debates of each era?
    Anderson: Oh, it gets tangled up in the Civil Rights movement and the measurement of the undercount of minorities in urban areas. It has been tangled up more recently in family and gender relations. Do we count same-sex marriage. All sorts of things.
    The two other other pieces are -- the country has a very diverse and dynamic population. So the population grows very rapidly by world-historical standards, which means that the apportionment and redistricting functions are very disruptive. If you have 10% to 30% more people and they're in different places, you're going to take political power away from some people and give it to somebody else. Just like when you win or lose and election.
    Then there is, of course, the history of the technical issues of the Census -- it was the pioneer of what was first called machine tabulation, which became IBM and computerization. There's very big and often unnoticed political and technical implications, which most people aren't supposed to pay much attention to until the year or two before the count and then the year or two after.
    That's where we are now. What triggered this is, there's a statute that requires the commerce secretary to report to the oversight committees in Congress, now, what's going to be on the 2020 census.
    Krieg: We've seen some attorneys general from blue states and other opponents of the administration, and this decision, already saying they're going to sue. How have efforts like that played out in the past?
    Anderson: There have been many different types of lawsuits around the census. There are, classically, one or two apportionment lawsuits over the last seat (in Congress). North Carolina and Utah have sued over who got the last seat in Congress in the last 15 years. From the 1970 Census through the 2000 census, there were undercount lawsuits. I think after 1980 there was over 50 of them, which were consolidated mostly into two suits in New York and Detroit.
    The last big one was (former House Speaker) New Gingrich suing Bill Clinton over the design of the 2000 Census and the Supreme Court deciding that case in 1999. The book (I wrote) with my former collaborator Steve Fienberg, "Who Counts? The Politics of Census-Taking in Contemporary America," recounts in gory detail the technical and political issues that went into that.
    So what you are going to see is the revival, using a lot of those precedents, of the decisions in both of those cases -- on both sides -- over the legitimacy of what Ross is proposing.
    Krieg: Which precedents -- what are the fundamental arguments we're going to hear?
    Anderson: Some of this turns on the language of the Legislative Procedures Act, which is a relative obscure piece of federal legislation over the authority of the secretary of commerce and the census director to make certain kinds of technical decisions. So I'm sure that Secretary Ross will use the Administrative Procedures Act.
    The urban areas and attorneys general are going to be arguing on the basis of the 14th Amendment and Civil Rights issues.