When U.S. Marshals conducted America’s first census in 1790, they posted the answers in the town square so locals could check for accuracy, as required by law. This practice continued until 1850.
How times have changed. Today, the National Archives and Records Administration (NARA) may release decennial census records after 72 years.
The most recent census available to the public is the 1940 Census.
The U.S. Census Bureau and its employees must protect respondent privacy and confidentiality at every stage of the data lifecycle, from collection through processing, publication and storage.
The privacy law, in Title 13 of the United States Code, mandates that information about specific individuals, households and businesses is not revealed, even indirectly through our published statistics.
A new Census Bureau infographic below charts the path of census safeguards from 1790 to now, from zero privacy to today’s absolute strictures designed to maintain public trust.
The earliest steps to protect the data addressed the most basic privacy challenges – but for businesses only.
Amid fears that competitors could extract sensitive information from the data, participation in the 1820 Census of Manufactures was so low that Congress decided not to collect the data in 1830. Realizing how important this information was, officials in 1840 provided assurances to businesses that their information would be safe.
In 1880, a new law gave those assurances teeth by adding stiff fines for census workers who violated that privacy.
A separate law in 1910 added potential jail time.
Those earlier laws addressed direct disclosure of business information.
The first law to address indirect disclosure wasn’t enacted until 1910.
That law made it incumbent upon the Census Bureau to ensure that census statistics couldn’t be reverse-engineered to reveal business identities.
The Census Bureau used a combination of techniques to comply. In 1930, it stopped publishing small area data altogether because, at the time, there was no way to disguise identities at that level.
In 1940, the indirect disclosure law was officially extended from businesses to data about people, as well.
The 1950s ushered in the age of computers and with it the ability to tabulate and publish more data for more levels of geography than ever before. But the proliferation of data brought with it greater indirect disclosure risks.
In 1970, the Census Bureau again suppressed (stopped publishing) whole tables for small areas and groups deemed too at risk.
In the decades that followed, staff worked to limit table suppression by adding more layers of statistical “noise” by swapping characteristics, “topcoding” values to limit outliers, rounding numbers and using other methods that affected precision but better protected privacy.
These “traditional” statistical disclosure methods worked well for decades.
But new threats — the rise of the internet, exponentially faster computing, new data science and an abundance of personal data readily available online —rendered those methods ineffective.
The 2020 Census will be the first census to use a powerful new privacy protection system designed for the digital age.
It uses a rigorous mathematical process, drawing from cryptography, to guarantee a level of privacy regardless of any new technology the future brings.