When a woman told me she couldn’t talk because she had thrown up, I didn’t press the issue. And I didn’t contradict the shirtless, thoroughly buzzed young man who glanced at the official list of racial choices, smiled broadly, and said he was a dog. But I did make note to contact these people again, up to five more times, until they gave me an answer I could put on the form. And when an old white guy wearing a baseball cap with an eagle on it said “there are three adults here and that’s all I’m going to tell you” and then shut the door in my face, that was okay. I took out my number two pencil and wrote “03” in the box for question S5 and “REFUSED” under the other questions. It didn’t bother me if the guy was cursing through his closed door while Rush Limbaugh was grinding away in the background. I had a legal right to ask these questions, and I had done my job. He had been counted.
The 23rd version of the U.S. census cost about $14.5 billion. Over the spring and summer of 2010, 635,000 employees (most of them temporary) counted more than 300 million Americans in 134 million housing units. They also reached about 8.3 million more in prisons, nursing homes, college dorms, and other group quarters, as well as in more than 60,000 shelters for the homeless, under bridge abutments, and even in caves. It took less than $2.8 billion to count 72 percent of the initial housing units—those whose inhabitants simply received a form in the mail, filled it out, and sent it back. More than $5.2 billion was spent chasing people who either didn’t get a form or ignored it. The “nonrespondents” were mostly young, poor, rural, or transient. They were pursued by a legion of badge-toting men and women who hoisted shoulder bags emblazoned with a large census logo. For a few weeks, I was in that army. I was curious to meet some of those slackers and ask them why they had neglected their civic duty.
Censuses form the political infrastructure of representative government. They are ancient—Moses took one in the Book of Numbers, and Mary and Joseph were on their way to a census taker when they stopped at the manger—but the older ones were usually used to collect taxes or form armies. The United States was the first country to use a census to determine political representation for its citizens. That stipulation is in article 1, section 2 of the Constitution. The last several U.S. censuses have also determined the distribution of federal aid to localities: about $450 billion a year in 2008, or $1,470 for every man, woman, and child, according to the Brookings Institution.
An old joke says that demographers are people who wanted to be accountants but lacked the personality for the job. As a reporter who is inordinately fond of demographers, I understand what the people-counters are up against. For the 2010 count, $338 million was spent on advertising and marketing, yet the shoulder bag I wore was emblazoned with the motto “Helping You Make Informed Decisions.” How’s that for a hook? Most of the people I talked to already knew that government workers would be coming around asking questions, but they didn’t really know why. So I wasn’t surprised when a woman sitting on her porch said, “I got that form and started filling it out, but I threw it away when they asked me what my race was. What business does the government have asking me that?” Well, I tried to explain, the race question is used for a lot of good things, like enforcing civil rights laws, distributing aid to public schools, and providing basic facts for public health programs.
“That’s not right,” she replied. “I’m white and I’m poor. Why don’t they care about me?” They actually do, I said; there are a lot of programs out there to help people in poverty, but government data on income is based on the American Community Survey, which is different from the census. All I want to know is your name, address, and phone number; whether you own or rent your apartment; whether you ever stay at another address; and the names, birthdays, sex, race, and family relationships of everyone who lives with you. I don’t think I changed the woman’s mind. But I helped her pass the time while she was taking a smoke break, and she was appreciative enough to answer the questions.
Mind you, I cannot reveal that woman’s name or where she lives. I raised my right hand and took an oath. I swore to uphold the Constitution, to not strike against the government or take bribes, and to not “disclose any information contained in the schedules, lists, or statements obtained for or prepared by the Bureau of the Census.” Violate that last oath and the fine is up to $250,000 and five years in a federal prison. So I can tell you about the what, when, and why of taking a census, but not the exact who or where.
The bureau is fond of saying that the public’s trust in the confidentiality of census data is their most valuable asset. What they don’t tell you is that they have a troubling habit of cracking under pressure during national emergencies. During World War I the bureau gave the War Department personal information about several hundred young men suspected of being draft dodgers. And in 1942 it helped persuade Congress to pass a law temporarily allowing it to share data with other agencies “for use in connection with the conduct of the war.”
Historian Margo Anderson and researcher William Seltzer found that, before the law expired in 1947, the bureau gave the names of 79 Japanese Americans in the Washington, D.C., area to the Secret Service, which was investigating possible threats to the president. It also loaned a staffer to the War Department who helped locate citizens of Japanese ancestry on the West Coast and send them to detention camps, and it considered imposing a national identity registration.
These were bad moves, to be sure, and several of the people involved knew it. Forrest Linder, a young statistician who worked on the relocation project, wrote that Americans might see such a register as “the beginning of an American Gestapo.” They were also bad moves because once the word got out, the Justice Department and other agencies started asking for confidential data, and it took an act of Congress in 1962 to stop the pestering.
Still, suspicion and politics have always dogged the census. In 1920 southern congressmen blocked reapportionment for 10 years when the numbers showed rapid urban growth at the expense of rural areas. Over the past 30 years, various plans to make the census more accurate by statistically adjusting it have been kicked through the halls of Congress; the Supreme Court said in 1999 that adjusted numbers could not be used for legislative reapportionment without the approval of Congress.
This time around, most of the hyperventilating has come from the anti-government movement. In April a conservative radio talk-show host railed against the American Community Survey by telling his audience, “I’m not filling out this form. I dare them to try and come throw me in jail. I dare them to.” That comment was not helpful. A few weeks later, an attorney in Leander, Texas, fired five shots over an enumerator’s head while demanding that he get off her property. It isn’t unusual for enumerators to be threatened for all kinds of reasons. The bureau recorded around 700 incidents in 2010, including 132 where someone showed a weapon or threatened to use one, and 13 incidents when shots were fired.
Whenever a large number of Americans feel that something is going fundamentally wrong in their county—as they do today and as they did during the run-up to the Civil War, the immigration wave of 1900–1910, and the civil rights movement—they take it out on the Census Bureau. And as the hue and cry goes on around them, the bureau’s employees slog on like infantry troops, stoically and inefficiently, until the job gets done. Except for me. I was really excited by all of this. I thought it was beautiful and couldn’t understand why so few people saw it as I did.
The rules and procedures for enumerators are marvelously complex—refined by decades of experimentation and testing—and nearly airtight. Like the Talmud and the tax code, they attempt to place a logical grid over a world that keeps trying to wriggle away. The main point of the exercise is to capture the names, personal information, and addresses of every resident of the United States on National Census Day, April 1, even though the forms are circulating from mid-March until the end of June. So when I would come upon someone unpacking boxes in his new apartment on May 15, I had to treat him as if he didn’t exist—at least, not in that location.
It’s important for enumerators to be thorough and follow procedures exactly, because making the tiniest mistake in filling out a form means that a computer will reject it and someone will have to fill it out again. The quality checks and data processing that happen after the enumeration is finished are also complex and time-consuming, and everything has to be done by December 31, the date the office of the president demands to know the total population of each state so it can get started with reapportionment.
Enumerators are the weakest link in this chain. I certainly was. My mind wandered often during the four-day training session. In a drafty room, a friendly retired man read us the procedures in detail, from a prepared text. I yawned, fidgeted, and doodled. My classmates weren’t doing much better. A woman asked, What if my son had a party on April 1 and four of his guests were homeless? You count the four friends and make a note on the form, said the friendly man. Then another student said that he had heard about a house that was jacked up into a trailer and moved from one location to another on April 1. Count it in the second location, he was told, in a slightly less friendly tone of voice. As the trainer struggled to make himself understood, the class dissolved into a free-for-all and a woman behind me said loudly that she really needed a beer. We quickly broke for lunch.
The Census Bureau would love to do the count without enumerators. It planned to conduct the 1970 census largely without them, but then a wave of Great Society laws increased the pressure from cities and advocacy groups to get African Americans and low-income people on the books. It tried to replace many enumerators with statistical adjustment procedures in 1990 and 2000, but Republican lawmakers who feared that the adjustments would boost the numbers in legislative districts that usually vote for Democrats blocked the plans. The inflation-adjusted cost of the census went from $14 per household in 1970 to $39 in 1990, and will probably reach $112 per household in 2010.
As any war veteran will tell you, large, time-critical missions funded by the federal government often include a lot of waste and some fraud. This time around, two local managers in Brooklyn were fired, and about 10,000 households were re-counted after the managers fell behind and started making stuff up. In early July, census director Robert Groves estimated that his fact-checkers would find about 1,000 incidents of fabricated information. And a $600 million plan to equip enumerators with handheld computers had to be scrapped in 2008, which is the last minute in an operation of this size. However, the handhelds were used to refine the address list before the count began, so the manufacturer, Harris Corporation, got its money. The bureau doesn’t plan to use them again.
More than 1,000 house fires happen in America on an average April day. But if the house was intact and inhabited one moment after midnight on April 1, it counts. About 7,000 people die on National Census Day—but if their souls stay in their bodies until 12:01 a.m., they, too, get in. And what about the guy who told me he was a dog? If he had insisted, I eventually would have checked “some other race” under question 6 and written “dog” in the box next to it, because people are allowed to choose whatever racial categories they want. Door-to-door census taking comes down to several hundred thousand fallible humans, being paid about what people earn for picking up garbage, going out alone to encounter wildly unpredictable events, armed with a dizzying list of procedures. No wonder it makes the demographers so nervous.
“A good city street neighborhood achieves a marvel of balance between its people’s determination to have essential privacy and their simultaneous wishes for differing degrees of contact, enjoyment, or help from the people around,” wrote Jane Jacobs, the legendary observer of urban life. “This balance is largely made up of small, sensitively managed details, practiced and accepted so casually that they are normally taken for granted.” The border between public and private is the enumerator’s workplace. Our bags and badges gave us glimpses of things we would otherwise never see.
Every day I went out with a list of addresses whose census forms hadn’t been returned. This address list is several years in the making, and building it is like taking a shadow census. It starts with every postal delivery route in the nation and is enhanced over several go-rounds by local officials and bureau employees. Much of the work of enumerators boils down to discovering the flaws in the list. I saw apartments that were vacant and others that had been turned into offices. On top of that, many people told me they had never gotten a form in the mail, and they were probably telling the truth. A person who uses a post office box won’t receive a census form there, for example, because everyone in the count has to be assigned a physical location.
People who don’t have mailboxes are supposed to be visited by a worker who hands them a form or hangs it on their door. Some folks also told me that they had already mailed back their forms. I was instructed to apologize for our error, a standard customer-service tactic, and then talk them into filling out another form on the spot.
But what if those people really had filled out a form? It isn’t unusual to count someone twice, such as a college student who answers the census in her apartment but is also claimed by her parents, a child whose divorced parents share custody, or a person who owns a weekend home. During the data-processing stage, the bureau’s computers flag forms that contain exact or near matches for names and addresses. Then they use an algorithm to assign scores to similar responses for each question on the suspicious forms. Staffers have the final say on which ones should be thrown out as duplicates. In 2000, 3.6 million suspected duplicate forms were deleted. Despite this, post-census research estimated that the final numbers double-counted 5.8 million people.
I also met people who just hadn’t responded. Most of them were happy to cooperate and seemed slightly embarrassed; they understood that they had ignored a small civic responsibility. But I worked in neighborhoods that were shabby yet functional. No meth labs were nearby, street violence was rare, and few residents were undocumented. It’s a different story in places like the near west side of Cleveland, where newly arrived Latinos pack themselves into rundown houses and try hard not to attract attention. Local officials and neighborhood leaders understand how valuable a complete count would be there, but they also know that some residents don’t even claim their food stamps out of fear of being deported. “You can almost tell when a person is not going to come to the door,” says Angela Woodson, a city employee who worked with enumerators there. “We told them that it was important and confidential, but they still resisted. They didn’t believe us.”
People who don’t feel secure are also unlikely to see much value in a census form. I remember a duplex with broken toys and junk all over the yard. I rang the bell and a boy opened the front door but not the screen; he looked like he was in the upper grades of elementary school, and I could tell by his eyes that there wasn’t anyone else home. He said there was another apartment around the back and someone might be in there. “Go up some stairs and through a door that looks like a shed door,” he said. “Then climb some more stairs. He might be in there.”
I lifted an old iron latch and opened the door into a two-story back porch. Someone had enclosed it a long time ago to keep out the weather. It had very dirty windows and a stairway running up one side and was packed with stuff that looked scavenged and secondhand: bow-hunting equipment, a bag of stale bagels, broken bits of wood that might once have been toys, kitchen and garden tools mixed together, and books moldering in soggy boxes. The door had a glass panel that was covered with an old towel, and from inside I could hear music—the Grateful Dead, “Fire on the Mountain”—so I figured someone was probably inside. I knocked, paused, and knocked again, louder. A youngish fellow with long, dirty hair yanked the door open. He smiled when he saw my bag and jumped out onto the landing, followed by a cloud of marijuana smoke. He was uncomfortably close. “Hey, a census guy!” he said. “This is gonna be fun.”
The kid said he was a dog and a resident of Earth. Then, to my relief, he added that he had filled out a form already at the shelter downtown, or maybe in another town, because he had no permanent address. Maybe he’d go out to California next, or maybe he’d stay here for a while. I kept asking questions. Is someone home who lives here most of the time?
“Hey, bro,” he yelled, “come talk to the census guy!”
Then another man came to the door. He was older with gray hair and bad teeth and was not smiling. “It’s just me here,” he said, and gave me his age by holding up six fingers, then two. I took him through the questions and the young guy started fidgeting and moving his hands in time to the music. After I finished, the older man asked if he could ask me a question, which was, “How much are they paying you?”
“Thirteen bucks an hour,” I said. “Shit,” said the older man. “I’ve been working at the restaurant for three years, working my ass off, and I get ten. SHIT!” I shook both of their hands and started down the stairs. “Have a grateful day,” said the kid.
The census of 2000 missed about 4.5 million people. But since it double-counted 5.8 million, it was actually less than half of one percentage point away from the correct total number. That census also marked the first time since the bureau started measuring accuracy (in 1940) that the final count came in too high. It missed 1.8 percent of black Americans and 0.7 percent of Hispanics, while it double-counted 1.1 percent of non-Hispanic whites. Yet it was considered a big success, because the net undercount in 1990 had been a lot higher: 4.6 percent for blacks, and 5 percent for Hispanics.
We know all this because of two things that happen after the enumeration ends. The first is demographic analysis, where staffers find out how far the head count falls short of a number derived through official records of birth, death, and immigration. The second is the Census Coverage Management (CCM) survey. The bureau’s best enumerators return to the field and take another big survey. They interview about 187,000 households in 6,400 very small geographic areas selected to represent specific demographic segments, such as Hispanics in Cleveland. Forms from the CCM survey are matched to forms from the census, and the difference indicates the size of the undercount or overcount.
I can tell you that teasing information out of some of the households on my list required a lot of patience, persistence, and tact. Once a woman came to the door of an apartment that took up half of a small duplex. She was holding a baby, a slightly older child was standing next to her, and a teenage boy stood in the distance. She told me she was too busy to talk and asked me to come back later; I did, several times, with no luck. On my last attempt, the teenage boy came to the door just to say his mother wasn’t home. I asked him how many people lived in the apartment. He hesitated a moment before saying nine, and that’s all I could get out of him.
When enumerators cannot get a response from a household after visiting in person three times, they are allowed to get the information from a proxy—such as a neighbor, a friend, or, in most cases, the landlord. This is how I ran into one of the two ideologues I met during a month of legwork. When I called one landlord and told him as gently as possible that the law required him to tell me the names of his tenants, he launched into a tirade. “I am outraged that the government expects me to give out information about other people,” he said. “Go ahead, send in the prosecutors! You people are going way too far!”
My patriotic friend might have been a fan of Ron Paul of Texas or Michele Bachmann of Minnesota, two Republican members of the U.S. House who publicly discouraged Americans from co-operating with the Census of 2010 as a protest against big government. There’s little evidence that their statements had much effect, though. The mail participation rates in Paul’s district were about average, and Bachmann’s district had some of the highest participation rates in the country. Paul and Bachmann join a long list of Congress members who, back through the decades, have yelled and screamed about the census until someone explained to them how much money it brings to their district—and that their district could even be eliminated if enough people who live there aren’t counted.
If the Census Bureau knows someone resides at a particular address but doesn’t have other information about him—for example, if a person mailed back a form that was blank except for the number of people, as Paul and Bachmann urged us to do—then information from one of his cooperative neighbors is duplicated onto the blank form, a technique known as imputation. In 2000, 2.3 million people were wholly imputed. In addition, 13 percent of census forms had items missing for one or more questions, and the missing items were imputed. In 2000 imputation cost Utah an additional congressional seat that eventually went to North Carolina. This spurred Utah to challenge the constitutionality of imputation before the Supreme Court. When the Court found for the bureau, demographers everywhere heaved an immense sigh of relief.
At the end of my month of enumeration, the only people I met whom I would still describe as slackers were the ones who just couldn’t be bothered. There was the man with an expensive haircut who sighed deeply and sent text messages while I was talking to him, and the woman who sent me away because she was putting her kids to bed, and then because she was about to serve dinner, and then because she was watching a movie. Eventually she stopped taking my phone calls.
Census coverage isn’t perfect, but it is amazingly close, given the size and difficulty of the task.
And after 220 years, the census still goes right to the heart of America. It celebrates our technological prowess while it exposes our willful ignorance of poverty. It also shows just how many people in this country are still struggling mightily to find a secure home and a better future. I found inspiration in these people, one of whom turned out to be a woman who had eight children. I caught her at her office, when she had a moment to talk. She was intelligent, serene, quietly commanding—just the kind of person you’d want for a mom. I found out that she had taken four of the children from a friend who couldn’t care for them anymore, and she had filed for adoption so they could stay together. I’m glad I tracked her down. It’s clear that she wants those kids to count.
Excerpted from the Autumn 2010 issue of The American Scholar, a publication of the Phi Beta Kappa Society since 1932 and a four-time National Magazine Award recipient. www.theamericanscholar.org
This article first appeared in the January-February 2011 issue of Utne Reader.