Democracy’s Data: The Hidden Stories in the US Census 
by Dan Bouk.
Picador, 362 pp., $20, August, 978 1 250 87217 3
Show More
Show More

William Faulkner​ ’s story ‘Go Down, Moses’ begins with a lengthy description of a man. But it’s hard to get a handle on him. If the drapes, pleats and price of his suit are all ‘too much’, his face is ‘black, smooth, impenetrable’. The story elaborates on these details but also draws attention to the fact that they fail to interest his interlocutor, a ‘spectacled white man sitting with a broad census-taker’s portfolio’. The census-taker quickly runs through his list of questions and then departs the story as, in a rather different way, does the man whose name, age, occupation and family background he records: Samuel Worsham Beauchamp. Beauchamp has been convicted of killing a Chicago policeman, and the next day he will face the electric chair. The rest of the story is about his grandmother’s efforts to bring his body home to Mississippi.

Early drafts of the story don’t include this opening scene, and I’ve often wondered why Faulkner chose a census-taker to introduce Beauchamp and his situation. It must partly have been a matter of timing. ‘Go Down, Moses’ was written in July 1940, a few months after the United States launched its 16th decennial count, and stories of the census interview were everywhere: in cartoons, movies, advertisements and endless newspaper sketches. On 27 April, for example, the cover of the Saturday Evening Post was Norman Rockwell’s depiction of a census-taker making notes as a large woman counts on her fingers the names of the red-headed, freckled children who peek out all around her. Their irrepressible liveliness threatens to overwhelm the little man in the too-big raincoat but his authority is ultimately asserted by the emphatic black rectangle of his portfolio. That portfolio, bigger than a broadsheet newspaper, is also the key prop in No Census, No Feeling, a slapstick short featuring the Three Stooges. Not that the Stooges were after information; their portfolios were used to hide from the police. The movie is, as the poster promises, ‘Bangful of Laughs!’ –but, like Faulkner’s story, it also taps into anxieties about the whole process: a feeling that census-taking is at once invasive and oddly inattentive to the lives it records.

‘Getting at the Facts’, a skit published in a Kentucky newspaper in 1905, deals with this directly. ‘How many people live here?’ the census-taker asks the girl at the farmhouse door. ‘Nobody lives here,’ she replies. ‘We are only staying through the hop season.’ So he tries again, rephrasing the question: ‘How many of you are there here?’ Again this fails to get the required response: ‘I’m here,’ she says, ‘Father’s in the woodshed, and Bill is …’ At this point, the census-taker loses his patience: ‘See here, my girl, I want to know how many inmates there are in this house, how many people slept here last night?’ ‘Nobody slept here,’ she replies. ‘I had the toothache dreadful, and my little brothers had the stomach ache, and the new hand that’s helping us got sunburned so on his back that he has blisters the size of eggs; and we all took on so that nobody slept a wink all night long.’ Looked at one way, the joke is on her: she’s too naive to understand the nature of his query. But it’s also on him and, by extension, on a government that asks such unimaginative questions. The official wants to ‘get at the facts’; the farm girl suggests how complicated they are.

In 2020 only a minority of Americans met a census-taker and those who did had already been given the chance to answer questions on the phone, by post or, for the first time, online. In 1940, however, the count was entirely reliant on an ‘army’ of what the government called ‘enumerators’ who went from door to door collecting ‘democracy’s data’. Dan Bouk confesses to ‘romantic ideas’ about the doorstep encounter as a symbolic ‘moment of co-operative civic action’ – the representative of the nation-state carefully attending to each of its members, one by one. That’s the ideal; his book, however, is often about what the encounter fails to achieve, and the negotiations and ‘tiny subversive manoeuvres’ it demands.

The decennial census is the United States’ fundamental form of big data, its defining act of symbolic mediation and the cornerstone of functional government. Its existence is mandated in the US constitution as the means to apportion congressional seats every ten years. Who was to be counted, and how, is more complex. ‘Indians’ were excluded completely from the constitution’s formula, on the grounds that their sovereignty was legally separate; ‘free Persons’ and the indentured were each attributed a ‘whole Number’; and ‘all other persons’ were counted as ‘three-fifths’. Those ‘other persons’ were slaves (a word the constitution strenuously avoided) and the odd figure came about as a compromise between the sectional interests of the Northern and Southern states. In 1783 delegates of the recently formed confederation of thirteen states met to discuss the tax burden of each. After real estate valuations were dismissed as unreliable, it was agreed that population would offer a reasonable index of each state’s wealth. The next question was how slaves might be included in that measure, if at all. Northern delegates made no distinction for these purposes between slaves and free men, while Southern delegates began with the proposition that since slaves were property, they should not be taxed at all. As negotiations proceeded, various arithmetical compromises were proposed, with an enslaved person weighted, variously, at a quarter, a third, a half or three-quarters of a free person. Finally, with ‘despair on both sides’, the ratio of three-fifths was agreed on.

In the end this tax plan wasn’t implemented. But the formula wasn’t forgotten; four years later, the Constitutional Convention suggested it should be used to allocate congressional seats. Everything changed once political power rather than taxation was at stake. Now it was the Southern delegates who wanted to count slaves as full persons although they did not, as Gouverneur Morris of Pennsylvania pointed out, wish to ‘make them citizens’ with voting rights. Rather the three-fifths rule meant that

the inhabitant of Georgia and South Carolina who goes to the Coast of Africa, and in defiance of the most sacred laws of humanity tears away his fellow creatures from their dearest connections and damns them to the most cruel bondages, shall have more votes in a Government instituted for protection of the rights of mankind, than the Citizen of Pennsylvania or New Jersey who views with a laudable horror, so nefarious a practice.

In short, the formula handed huge political leverage to Southern whites; leverage that continued after Reconstruction with the effective disenfranchisement of Black voters. Only with the passage of the Voting Rights Act in 1965, Bouk argues, did the census change its purpose from a mechanism for reinforcing white supremacy to one for dismantling it.

Who is counted, how, and for what purpose, has changed a lot since 1790. No census has exactly matched its predecessor in method or design: each time, some questions are dropped and others added, while classificatory categories – especially unstable and byzantine where race is concerned – have been regularly rethought. The only constant, as one former census bureau director, Kenneth Prewitt, admitted, is that they are ‘never politically neutral’. For example, the category of ‘Free Coloured Persons’ was introduced in 1820 largely because pro-slavery apologists were determined to prove that freedom made former slaves crazy, dissolute and suicidal. The matter came to a head after the 1840 census seemed to offer evidence to this effect. Although the American Statistical Association quickly identified ‘various and gross errors’ in the count (‘insane’ Blacks were recorded in some Massachusetts towns that had no Black population), the figures continued to be wielded by pro-slavery congressmen. Harriet Beecher Stowe was one of those who expressed outrage that ‘the most important statistical document of the United States has been boldly, grossly and perseveringly falsified’ and appended a discussion of ‘Fact v. Figures’ to her ‘key’ to Uncle Tom’s Cabin. Stowe’s novel itself can be read as a refutation of the ‘false returns’ of 1840, but it also engaged with the new racial science which informed the 1850 census, in particular the adoption of ‘mulatto’ as a new ‘statistical race’ (in Prewitt’s phrase). The term remained on the schedule until 1920, even though many enumerators reported ‘considerable uncertainty’ in its application.

It was not only the categories of Jim Crow America that were absurd, however, but the expectation that census-takers should use visual or contextual ‘clues’ to decide a person’s race, rather than just asking. Looked at from this point of view, the doorstep encounter was less a moment of civic co-operation than, in Margo Anderson’s words, ‘the Achilles heel of census accuracy’. It was only after 1970, when a postal census was widely implemented, that people were finally able to represent themselves, albeit within the limits of the categories offered.

Back in 1939, however, self-identification was not on the agenda. Instead, the bureau created scenarios – each, Bouk says, ‘amounting almost to a short story’ – for would-be enumerators to disentangle. What was the race of a child born to a white man and a Japanese woman? Or the marital status of a woman who had just filed for divorce? And what birth country should be recorded for a man born in Warsaw, then part of Russia but now under German control? The answers, respectively, were Japanese, married and Poland, a country that didn’t exist when the man was born or when he was enumerated. Bouk is particularly interested in the way ‘partner’ came to be used as a blanket term for all kinds of living arrangements and relationships (from lodger to lover) that the system found anomalous. The term had first been used in 1880 to record immigrant labourers in shared accommodation, but by 1940 it also applied to one of a pair of men or women who were ‘not related by blood or marriage’ but had a ‘common dwelling’. The ‘household’ was the census’s primary organising unit, and enumerators were also given detailed instructions about how to identify one – in short, a shared cooker rather than bed. No one asked what the relationship between the cohabiting couple was, or whether it made sense to designate one of them as the household’s ‘head’.

The fact that Bouk is able to reveal the names of some of those who were classified partners in 1940 – such as Emily H. Brand, a 29-year-old secretary, and Katrina S. Grant, a 27-year-old social worker – is the result of legislation which ensures the release of all personal data after 72 years. The rule represents another balancing of ideals: on the one hand, the requirement for absolute confidentiality; on the other, the belief that, as a 1930 bureau official put it, every person has ‘a permanent place in the history of the country’. If one kind of knowledge requires the ‘translation’ of people into data, another strives to reverse the process. But unlike novelists who can make up stories for those they count (in a 1936 map of his apocryphal Yoknapatawpha county, Faulkner recorded ‘whites, 6298; Negroes, 9313’), historians are confined to the archive. Bouk is left wondering of one intriguing family: ‘What further stories might they have told if asked different questions than those asked by this census?’

It has always been difficult to get people to believe that their personal details won’t be handed on to the authorities. When Johnson Jones Hooper wrote about his experience of ‘Taking the Census in Alabama’, he signed it ‘by a “Chicken-Man” of 1840’ – a reference to the common belief that his job was to identify (for tax purposes) every chicken, cow and loom. A hundred years later, almost every newspaper ran stories to reassure readers that, as the Knoxville News Sentinel put it, the census-taker ‘could be jailed for telling Mrs Jones, next door, how old you are or how much rent you pay’. (That didn’t stop the radio comedian Gracie Allen asking a census man about another Mrs Jones ‘across the street’: ‘Is she a Republican? Or is that her natural hair?’)

In 1940 the anxiety was not confined to the prying eyes of the Joneses. For some Republicans, the enemy was ‘big government’ itself, especially because a question about income had been introduced. Conducted as Roosevelt campaigned for his third term in office, the 16th census became a tool to assess the consequences of a decade of depression and the effectiveness of particular New Deal programmes, and to determine where new resources should be deployed. If, as a 1938 report by the National Resources Committee claimed, the aim was to ‘progressively make available to all groups what we assume to be American standards of life’, much more data, and more accurate data, was needed. For once, the bureau poster of Uncle Sam – portfolio in hand, accompanied by the emphatic slogan ‘It’s Your America!’ – seemed to make sense.

Only a few months later, however, that promise of universal embrace rang hollow. Bouk tells the story of how, as early as September 1939, ‘backroom conversations’ began to take place about the ways in which the census bureau might contribute to war preparations. The first issue concerned workers for the factories that Roosevelt promised would turn the United States into ‘the great arsenal of democracy’. Fearing sabotage, the government required that all defence workers should be US citizens, but few of those citizens had documents to prove it. Thousands of people wrote to the census bureau asking for certified records of their census responses. The Division of Vital Statistics was already under pressure following the 1935 Social Security Act, since few elderly or middle-aged people had the birth certificates required to apply for aid. The bureau struggled with this new demand, while, at the same, processing the census returns and, Bouk argues, doing its best to appear like a ‘statistical standing army’. Bureau officials ‘feared the United States could lose the war’ but they also worried that ‘the United States might win without the bureau’s help.’

This anxiety explains why, within hours of the Japanese bombing of Pearl Harbor on 7 December 1941, it set to work looking for cards marked ‘Jp’. Even before the Second War Powers Act repealed the confidentiality of the census, the bureau helped the military authorities by providing population tables for particular neighbourhoods. By August 1942, more than 120,000 Japanese Americans, two-thirds of whom were US citizens, had been ‘evacuated’ and interned. The story of the census bureau’s involvement in what the White House later called ‘one of the most shameful instances of the government misusing its own data’ has been painstakingly documented over the last twenty years by Margo Anderson and William Seltzer. Bouk approaches it through the story of one family, the Moriyamas, whose son Iwao had moved to Washington in 1940 to work in the Division of Vital Statistics. In the census that year, probably because he was living in a boarding house full of white people, the enumerator classified Iwao Moriyama’s race as ‘white’ and he escaped internment.

Fear​ of the misuse of personal data remains a constant threat to census accuracy. A recent example resulted from the the Trump administration’s attempt to add a question about citizenship to the 2020 census (a question not mandated by the constitution and last asked in 1950). As later became clear, the aim was explicitly to discourage participation from undocumented immigrants, who Trump believed had cost him the popular vote in 2016. The census bureau’s chief scientist, John Abowd, estimated that at least 630,000 households (mainly Asian Americans and Latinos) were likely to be deterred, resulting in fewer congressional seats (and electoral college votes) as well as reduced federal spending in Democrat-leaning states. The question was eventually dropped after the US Supreme Court ruled that the rationale for its inclusion was ‘contrived’.

The census had narrowly avoided not merely a significant undercount but, more important, what statisticians call a differential undercount. ‘None shall escape me and none shall wish to escape me,’ Walt Whitman declared in ‘Carol of Occupations’, one of many poems in which he presents himself as a kind of ideal census-taker. In reality the poorest and most precarious parts of the population often escape the purview of the census, whether they wish to or not. At the turn of the 20th century, a deliberate undercount of African Americans was motivated by white supremacists hoping to find evidence of the population’s decline. Kelly Miller, a mathematician at Howard University, demonstrated this on several occasions, but few listened until 1947, when Daniel O’Haver Price published an article comparing two data sets: men who registered for the draft in October 1940 and those counted six months earlier in the census. Price concluded that the census had missed 3 per cent of men of draft age, while the shortfall for Black men was 13 per cent. Was Ralph Ellison thinking of this while writing Invisible Man? ‘I am invisible, understand, simply because people refuse to see me.’

Langston Hughes’s story ‘Data for Simple’s Census’, which first appeared in the Chicago Defender in November 1961, addresses the issue head on. Unlike the protagonists of most census interview stories, Simple is not trying to hide from the government; the problem is rather that the government has failed to notice him:

I was born young, black, voteless and hungry, in a state where white folks never put Negroes on the census. My daddy said he were never counted in this life by the US government. And no one could ever find a birth certificate for me nowhere in Virginia. It were not until I came to Harlem that one day a census-taker dropped around to my house and asked me where I was born and why, also my age and if I was still living.

Simple does not regard these questions as reductive. ‘Put me on your census now,’ he urges the man, ‘because I may not be here when the next census comes around.’

Since the 1960s, when racial statistics became important for monitoring and enforcing civil rights legislation, estimates of differential undercounts (and strenuous follow-ups and statistical readjustments) have become a regular part of the census. Although most people fill in their forms themselves, the number and diversity of enumerators has therefore grown exponentially. The census has come to be understood as a crucial tool in the politics of recognition, a means to highlight the problems and interests of particular groups: that’s why the catch-all ‘Hispanic’ was introduced in 1970, and why, after 2000, when respondents were allowed to self-identify with more than one racial category, many opted simply for Black. Barack Obama was one of them. By publicly completing his form in the Oval Office in 2010, Melissa Harris-Perry argues, Obama ‘embraced Blackness, with all its disprivilege, tumultuous history and disquieting symbolism. He did not deny his white parentage, but he acknowledged that in America, for those who also have African heritage, having a white parent has never meant becoming white.’ Obama was engaging in what Bouk calls ‘strategic self-enumeration’, picking the part of one’s identity that could most use some representation. The old 19th-century racial categories remain, in other words, but they are now understood to refer to political rather than biological entities. And since political and cultural loyalties change, it’s not unusual for people to describe themselves differently from one decade to the next.

But what about the original purpose of the census? Although Americans can now, within limits, determine how they appear on the form, it is less clear how that personal representation translates into the political representation that the constitution promised. Bouk traces the source of the problem back to 1929 when the House of Representatives decided to cap its membership at 435. Nearly a century later, while the population of the country has tripled, that number remains frozen, with all sorts of undemocratic consequences. As a recent report from the American Academy of Arts and Sciences notes, the current system means that when the electoral college comes to choose a president, a vote cast in sparsely populated Wyoming is worth 3.6 times that of a vote cast in California. If Bouk’s book has a message, it’s that we should all pay attention to the messy stories and politics behind the seemingly precise numbers that govern our lives.

Send Letters To:

The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN

Please include name, address, and a telephone number.

Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.

Sign up to our newsletter

For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.

Newsletter Preferences