Holding the Government Accountable

The Citizenship Question: A Lot Riding on a Single Change in 2020 Census

Citizenship status may seem like a simple question, but in the context of the larger ongoing immigration debate, it could be more ominous to some communities. (Photo: Shutterstock)

The decennial census is right around the corner. It is an extremely important function, which may be why it is one of the few specific governmental responsibilities spelled out in the Constitution. Every ten years the Census Bureau must accurately and fairly count all people living in the country—not just citizens, but everyone. The goal is to count everyone once, only once, and in the right place.

Yet on March 26, Department of Commerce Secretary Wilbur Ross announced that a question about citizenship would be included in the 2020 census. The question hasn’t been included in the decennial census since 1950 but the Department of Justice requested in January that the question be added, claiming that it was to help them enforce the Voting Rights Act.

The announcement set off major criticism from census experts and civil rights leaders arguing that the addition of the question will lower response rates, increase the cost, and produce a less accurate census. Flawed population numbers could have a significant impact since we rely on the census to apportion seats in Congress and allocate billions of dollars in federal funds.

There has also been quick criticism of the Ross decision from numerous Democratic politicians, and conservative voices have noted that the effects of a less accurate census could harm red states as much as blue states. The opposition has culminated in legislation being introduced to prevent the change and a massive multi-state lawsuit.

Decennial Census

Let’s step back and first talk about the importance of the decennial census.

This total population information helps directly shape our democracy and the balance of power because it is used to determine the number of Congressional representatives each state sends to DC. It is worth emphasizing again that the assignment of Congressional representation is based on total population, not just citizens. It’s estimated that over 40 million immigrants lived in the United States in 2016, the majority of whom are here legally. The United States admits more than one million lawful permanent residents each year to work in tech firms and as nannies, conduct advanced research and teach at universities, and more. There are also tens of millions of immigrants who come here on legal temporary visas. Most are tourists, but others come for months at a time, spending money to attend college, performing temporary or seasonal work that American companies need, or for other legitimate reasons. These non-citizens don’t vote, although many of them pay taxes, but if they have concerns or problems with the federal government they go to their representative just like the rest of us.

Each decennial census has an immediate impact on Congress. After the 2010 census, eight states gained Congressional seats and 10 states lost them. The 2020 census will decide the next set of states to gain or lose seats in the House of Representatives. Additionally, most states use census data to draw legislative district lines so that each roughly includes the same number of people.

In addition to political power, the census data shapes how we spend money—a lot of it. Census information helps direct more than $800 billion in government spending every year. The population data helps determine where to locate schools, hospitals, police units, and other local services; where new roads need to be built; where companies should expand to and hire more people and build new shopping centers; and so much more. Most social and economic research in this country is based on the census. Again, it is not about being a citizen because non-citizens drive on roads, use public transportation, go to hospitals when sick, and send their kids to schools.

The census count is used with other information to allocate money and resources accordingly. If we get the count wrong, then we will be giving some states and localities too little money while others get more than they need. Considering we will rely on this information for a decade, the numbers will impact trillions of dollars in government spending and an enormous number of policy decisions.

How will the question affect the census?

No one knows exactly how the question will affect the census. It is untested, so the exact impact on responses is impossible to accurately project. It is highly unusual to include an untested question in the decennial census and the idea of doing so in the 2020 census has been the subject of strong criticism. Despite the uncertainty about the impact, census experts seem unanimous that the question will place the accuracy of the census at risk and make the survey cost more money.

Two former Department of Commerce Secretaries who oversaw the census—Carlos Gutierrez, a Republican, and Penny Pritzker, a Democrat—wrote, “this question will put in jeopardy the accuracy of the data that the census collects, and increase costs.” They described a “multi-year process for suggesting and testing new questions. Questions should be added only after the Census Bureau has adequately tested the potential effect of the question on response and accuracy rates.” They say that was not done in this case, and that it put the accuracy of the census and the reputation of the Census Bureau in jeopardy.

All six living former directors of the Census Bureau agree. They wrote to Secretary Ross in January and urged him leave the citizenship question out of the upcoming 2020 census. The former directors, with a collective 25 years of leading the bureau under both Republican and Democratic administrations, stated plainly that “adding an untested question on citizenship…would put the accuracy of enumeration and success of the census in all communities at grave risk.”

Census question 8
Question #8 on the 2020 Census. (Photo: U.S. Census Bureau)

The directors emphasized the importance of careful survey design and question selection by noting that “There is a great deal of evidence that even small changes in survey question order, wording, and instructions can have significant, and often unexpected consequences for the rate, quality, and truthfulness of response.”

The Census Bureau’s own Science Advisory Committee met one week after the announcement and discussed the added question in detail. Committee member D. Sunshine Hillygus, a Duke University political science professor and author of a book about the 2000 census, offered a sharp rebuke of the move in her presentation, which started with a slide simply saying, “W.T.H.” (what the hell?). The full committee offered their official opposition to the citizenship question because of concerns about the lack of adequate testing. The committee noted that “there is a hierarchy of needs for the decennial census, with an accurate count of foremost importance” and felt that the addition of the untested question ran contrary to that need.

Unseen by most of us, the Census Bureau conducts extensive research and testing over the course of years to design the census questions. The purpose of the testing is to ensure as full and accurate participation as possible. Questions, format, and instructions are tested so the Bureau fully understands the impact of any changes. Currently the Census Bureau is conducting a field test of the 2020 census in Providence County, Rhode Island, which does not include the citizenship question.

It should also be noted that should response problems arise, then the 2020 census process will wind up costing more. The initial self-response is the least expensive way for the Bureau to collect the information. But if people in some communities don’t respond or provide incomplete responses, then the Bureau must use more expensive options such as paying canvassers to go door to door for responses.

Why would people not fill it out?

Citizenship status may seem like a simple question, but in the context of the larger ongoing immigration debate, it could be more ominous to some communities. President Trump’s Administration has made increased immigration enforcement and deportation activities a cornerstone of his first term. That creates heightened fears among immigrant communities that any information provided to the federal government will be used to target them.

As conservative commentator David Brooks stated, “In normal times, frankly, it doesn’t strike me as an odd question to ask, are you a citizen? …. [but] given the climate, it strikes me as a menacing question and probably a counterproductive one.”

Even prior to Secretary Ross adding the citizenship question, the political climate spelled trouble for the 2020 census. According to the Census Bureau’s own research, they found an “unprecedented ground swell in confidentiality and data-sharing concerns among immigrants or those who live with immigrants.” As one interviewee said, “The possibility that the Census could give my information to internal security and immigration could come and arrest me for not having documents terrifies me.” Even if the census data is protected under law from such disclosure, the perception that this could happen existed before the citizenship question was added.

Do we need citizenship information?

The Department of Justice requested that the citizenship status question be added, claiming that it needed the data to implement the Voting Rights Act of 1965. That explanation doesn’t make much sense, though, because the Department has been enforcing the Voting Rights Act robustly since the law was enacted more than 50 years ago without citizenship information from the decennial census.

That is not to say that we don’t collect citizenship information. In fact, the Census Bureau has continued collecting citizenship data since the question was removed from the decennial census after 1950 through what is now called the American Community Survey. The ACE collects information from representative samples of the population and that data is used to enforce the Voting Rights Act. It isn’t clear that we urgently need citizenship information from the decennial census.

What would an undercount mean for states?

Each census faces a problem of some categories of people being overcounted and some undercounted. Whites tend to be overcounted as they have a larger percentage of families with multiple homes that mistakenly file more than one census response. Blacks, Latinos, Native Americans, Asians, and young children under the age of 5 tend to be undercounted as larger percentages live in circumstances that make them harder to count, such as renters, those who move often, or live in group settings. When there is an undercount it distorts political power, wrongly distributes federal funds, and can result in poor decision-making by policymakers because they have faulty information.

The full and exact impact of how adding a citizenship question will affect the count is nearly impossible to project, but it will almost certainly not be uniform across the country. Instead, it will likely have a greater adverse impact on states with larger, and possibly less responsive, immigrant populations. That imbalance in the count could translate into states unfairly losing or not gaining deserved seats in Congress, and not receiving billions of dollars in needed federal funds over the next decade.

Some have claimed that Democrats, who have been the most vocal opposing the citizenship question, are worried that an undercount resulting from the citizenship question would affect them more because immigrants reside in greater numbers in urban centers, which are often more Democratic-leaning.

However, others have noted that there are plenty of Republican-controlled states that could just as easily lose out. An analysis of Texas indicates that an undercount of the state’s almost 5 million immigrant residents could jeopardize the three Congressional seats the state has been expecting to gain from the 2020 census.

A recent George Washington University study was able to project some of the financial impact on states from a census undercount. The study examined five federal assistance programs that rely directly on the total population count to calculate how much the federal government reimburses the state. Reimbursement levels are tied to a state’s income per capita. States with higher income per capita only get the minimum 50 percent reimbursed. States with lower income per capita can qualify for up to 83 percent reimbursement.

So if the census undercounts the number of people in a state but the total state income remains the same (it is derived from a different source), then the per capita income would rise. If that state was getting more than the minimum 50 percent reimbursement, the higher per capita income calculation would cause it to get a lower reimbursement rate than it should be getting.

The study calculated what a 1 percent undercount in state population would cost states for these five programs in 2015 and found that 37 states were at risk of losing money. Texas would lose the most, with $291 million at risk from a 1 percent undercount. Four other states would also lose more than $100 million annually—Pennsylvania, Florida, Ohio, and Illinois. The study also noted that 13 states, including major population centers like California and New York, would not lose any money from these programs as a result of a census undercount because they are already at the minimum reimbursement level.

What opposition has there been?

Some politicians have engaged in direct action to oppose the citizenship question inclusion. Representative Carolyn Maloney (D-NY) and Senator Brian Schatz (D-HI) focused on the untested aspect of the question with new legislation titled 2020 Census Improving Data and Enhanced Accuracy (IDEA) Act. The IDEA Act would prevent any major changes, including new questions, that have not been “researched, studied, and tested for a period of not less than 3 years.” The legislation, which does not have any Republican co-sponsors yet, is unlikely to move forward in the Republican-controlled House and Senate.

Additionally, 18 states and 6 cities have joined together to file a lawsuit seeking to block the federal government from adding the citizenship question. They claim that including the question is unconstitutional and arbitrary. The suit argues that since the federal government has a constitutional requirement to “fairly and accurately” count all people in the country, inclusion of a question they know would reduce the response rate and lead to inaccurate numbers would violate the government’s constitutional responsibility.

What should be done?

Given the untested nature of the question and the real possibility of it affecting the critical results of the census, the Project On Government Oversight recommends it be removed from the 2020 decennial census. The Department of Justice can continue to use citizenship information collected through the American Community Survey to properly enforce the Voting Rights Act. If more information on citizenship is needed, new questions for the decennial census or changes to other surveys can and should go through full and proper testing before being implemented. Trillions of dollars over the next decade and Congressional representation is too much to gamble on an untested question that the most knowledgeable census experts say will compromise the validity of the census.