On Friday, the U.S. National Archives will unlock the complete records of the 1950 Census. These files describe more than 150 million people, one by one. They bear the handwriting of a small army of Americans (around 150,000 in all) who scoured the nation, city and country, and even the recently sprouted suburbs, asking more than 20 questions of every person.
Today, we can see the 3 billion answers to those questions as “big data,” possibly yielding new discoveries about the past. Scholars will probe the data set to study social mobility, residential segregation or the everyday lives of Americans at mid-century. Countless genealogical investigators will mine the records for information about their kin or communities.
These records are of particular significance because they are the last of their era. In subsequent decades, the government embraced more fully a statistical technique that allowed it to ask far fewer questions of most individuals. Censuses from 1960 on would still record nearly every person, but with far less depth or detail. The 1950 census therefore stands as a testament to the value of an expansive, exhaustive census and all that it saves for posterity.
The Constitution instituted the U.S. census as part of the machinery of democratic governance, tying representation to population. It called for an “actual Enumeration” to determine what proportional share of the House of Representatives each state would be due. In meeting this goal, the first census of 1790 did little more than count heads.
That first enumeration only tallied the numbers of White men (split into two age groups), White women, those the census referred to as “all other free persons” and enslaved people. It excluded Native Americans and credited enslaved people as only three-fifths of a person for apportionment. Only heads of household were named in the ledgers, and so only their details were recorded.
Through the 19th century and into the 20th, the scope of the census expanded. Congress added new questions, hoping the answers would inform its deliberations. In 1850, as the census became more curious, it also began recording separate answers for each individual person in a household, though only free people were named.
After the Civil War and with the passage of the 14th Amendment, the three-fifths compromise was struck out, making all Black Americans full statistical people who were asked the full battery of questions. Beginning in 1870, the census was supposed to count every Native American as well — though only some would count for representation purposes.
Congress began to add new questions to be asked of all those counted, as it would for decades to come. By the turn of the century, new questions focused particular attention on the place of birth and first language (called “mother tongue” in the questionnaires) of each person and also that of each of their parents. In those first decades of the 20th century, enumerators filled in the printed census sheets, which were then transformed into statistics. The resulting tables of numbers informed debates about immigration that culminated in restriction laws passed in the 1910s and 1920s.
It had become plausible by the 1930s for a Census Bureau official to claim: “Every person in the United States, however insignificant he may be, has a permanent place in the history of the country.” Yet not long after those words were written, a technical revolution swept through the census that made it possible for the government to ask many fewer questions of most individuals.
The 1940 Census introduced more questions than its predecessors, including a controversial set of questions related to Americans’ incomes. Yet the growing list of questions did not satisfy the people and interests who shaped the census: The New Deal’s proliferating government agencies, increasingly influential labor unions, business leaders marketing to the masses and data-hungry social scientists.
They all wanted to know more but were running up against the physical limits of the census sheet. They were out of space.
But a new method that applied probability theory to the quest for national self-knowledge offered a solution. By asking questions of a randomly selected “sample” of the entire population, the Census Bureau could predict what the answers were for the whole. The census would stay nosy, or even become nosier, but it could do so while bothering only a tiny fraction of the population.
All Americans in 1940 were quizzed with more than 30 questions, the most that could be fit on a newspaper-sized sheet of paper. Five percent of the population, evenly distributed across the country with individual respondents selected at random, answered even more.
Then in 1950, the census slimmed down to around 20 questions, with a large set of sample questions asked to a bigger subgroup: 20 percent of all residents. As the nation expanded, the census took on new significance, evolving into one of the most democratic forms of historical preservation ever imagined.
But a new commitment to the aggressive use of sampling led to the decision in 1960 to slash the number of required questions for all people to seven. That’s the same number people answered in 2020. To make up for all the questions no longer being asked of us and to make space for new questions that meet our changing times, we have been, since 2005, each represented by a tiny fraction of the entire population in a Census Bureau survey conducted every year.
The growth of the census questionnaire had not been designed with historians in mind. The data collected was meant to guide legislators. Yet the census nonetheless had developed into an unparalleled archive of ordinary lives. Then, from 1960 on, much of that archive’s richness evaporated. Census records would continue to record nearly every person, but without individual details.
A consolation for ancestor-seekers and other researchers may be that the pared-down 1960 Census also introduced widespread self-response. The bureau had attempted to remove census enumerators from the process wherever possible that decade, in the hopes of limiting certain kinds of error. In the process, it gave greater control to individuals to determine how their identity would be recorded, a control that has since grown as activists have successfully advocated for more inclusive and more precise options for defining a person’s characteristics, most prominently race and household role.
Today, politicians and civil society groups champion the census as a tool for accurate planning, for winning visibility and representation and for channeling trillions of dollars in federal funds.
Some reformers are also pushing to make 2020, like 1950, into a last-of-its kind census by replacing most self-response or enumeration with records generated by other government agencies — such as birth registrations, immigration records, drivers’ licenses or tax documents — and possibly supplemented by private data sources, too. Such a count would probably be cheaper. It is possible, though far from certain, that it could provide more accurate data. It might reduce, but could also amplify, persistent undercounts of Hispanic, Black and Native American residents, and of children generally. We might also consider, though, a future census that gives more prominence to the responsibility of democratic representation in the nation’s archives.
As we celebrate the release of the 1950 Census records, it is an opportune moment to think again about the role the census has played — and may still play — in preserving the nation’s past by preserving a substantial accounting of each of us.