The U.S. Census Is Wrong on Purpose

This is a story about data manipulation. But it begins in a small Nebraska town called Monowi that has only one resident, 90 year old Elsie Eiler.

There used to be more people in Monowi. But little by little, the other residents of Monowi left or died. That’s what happened to Elsie’s own family — her children grew up and moved out and her husband passed away in 2004, leaving her as the sole resident. Now she votes for herself for Mayor, and pays herself taxes. Her husband Rudy’s old book collection became the town library, with Elsie as librarian.

But despite what you might imagine, Elsie is far from lonely. She runs a tavern that’s been in her family for 50 years, and has plenty of regulars from the town next door who come by every day to dine and chat.

I first read about Elsie more than 10 years ago. At the time, it wasn’t as well known a story but Elsie has since gotten a lot of coverage and become a bit of a minor celebrity. Now and then I still come across a new article, including a lovely photo essay in the New York Times and a short video on the BBC Travel site.

A Google search reveals many, many similar articles that all tell more or less the same story.

But then suddenly in 2021, there was a new wrinkle: According to the just-published 2020 U.S. Census data, Monowi now had 2 residents, doubling its population.

This came as a surprise to Elsie, who told a local newspaper, “Then someone’s been hiding from me, and there’s nowhere to live but my house.”

It turns out that nobody new had actually moved to Monowi without Elsie realizing. And the census bureau didn’t make a mistake. They intentionally changed the census data, adding one resident.

Why would they do that? Well, it turns out the census bureau sometimes moves residents around on paper in order to protect people’s privacy.

Full census data is only made available 72 years after the census takes place, in accordance with the creatively-named “72 year rule.” Until then, it is only available as aggregated data with individual identifiers removed. Still, if the population of a town is small enough, and census data for that town indicates, for example, that there is just one 90 year old woman and she lives alone, someone could conceivably figure out who that individual is.

So the census bureau sometimes moves people around to create noise in the data that makes that sort of identification a little bit harder.

A representative from the census bureau explained to the Lincoln Journal Star:

Elsie isn’t the only person this affects so directly. About 50 miles north of Monowi is a town called Gross that in reality has a population of 2 people. But there, too, the census added a person, increasing the population by 50%.

The term for this is differential privacy and I knew that this sort of thing happened with other types of data. For example, if you’ve enabled sending analytics info from your iPhone so Apple can improve things like predictive text, Apple uses differential privacy techniques to obscure your personal usage before it even gets to Apple.

They add noise to the information they gather from you, but it’s done in such a way that with enough data collected from enough people, they can still see the patterns they’re looking for emerge. This way, Apple can improve its services based on what many users do, without knowing what any individual person is doing.

But I was surprised to find out that this sort of thing is done by the census bureau, since I’ve always assumed that the census data was immutable.

In 2092, when the full data from the 2020 census becomes publicly available, will people doing genealogical research on their families be misled by what they find? I assume the individual records are kept intact, but I can still see it leading to confusion when the grandkids of that couple in Gross are wondering who the mysterious unrecorded third person was that must have been living with grandma and grandpa.

Here’s a fun thing I learned while reading up on differential privacy: You can dig down into settings on your iPhone and actually see the data being shared with Apple that is protected using differential privacy.

Go to Settings > Privacy & Security > Analytics & Improvements > Analytics Data and look for entries that begin with “DifferentialPrivacy.” I don’t have my phone set to send diagnostic data, so I don’t have any such entries. (For more on this, you can read the Differential Privacy Overview [PDF] straight from Apple).

And that’s it for another newsletter! Thanks for reading. I hope you’ll consider upgrading to become a paid subscriber and support this newsletter. Or maybe buy me a coffee. Or just share the newsletter with someone you think would enjoy it.

And of course, you can reply or click through to leave a comment if you have any thoughts on differential privacy, or small towns in Nebraska.

See you next time!

David