What’s America’s largest ethnic group, and why did we get it wrong for so long?
The identity of America’s largest ethnic group has long seemed cut and dried. But new data is challenging the conventional wisdom.
It makes intuitive sense. In the 1800s, Germany (or the bits of Europe that would more or less become Germany) sent the United States more immigrants than any other country by a fairly wide margin. If you account for the nation’s population at the time, it’s the largest wave of immigrants America has received from any country since we started keeping track in 1820.
But what if our nerdly intuition failed us yet again, and the normies had it pegged all along?
Germans have long reigned supreme thanks to this American Community Survey question: “What is this person’s ancestry or ethnic origin?” But in 2020, the census debuted a more detailed race question addressing almost the same topic. It encouraged folks not just to mark a box about their race (labeled “White” or “Black or African American”) but also to record the details of their “origins” (“for example, German, Irish, English, Italian, Lebanese, Egyptian, etc.”).
By that measure, the world has turned upside down — and this time the English are on top.
According to the results of the 2020 Census, 46.6 million of us claim at least some English background while only 45 million claim some German. (If you look only at people who claim a single ethnic background, the English thump the Germans, 25.5 million to 15.4 million.)
The English now appear to outnumber Germans and qualify as the largest ethnic group in much of the country, particularly if we focus only on European or other non-Hispanic White origins. (For more on how the Census Bureau measures Hispanic Americans, Native Americans or Pacific Islanders, see our previous columns.)
English predominate across the South, the West and northern New England. The Germans hold sway in the northern Great Plains and the Rust Belt.
Should we trust this apparent reversal? Well, to answer that, we should probably explain why it happened. Only we couldn’t. Sure, the Census Bureau measures the two questions differently. It looks at just two responses for its ancestry question, but at least six for its revamped race question, and it goes to greater lengths to push folks to answer the race question. But that didn’t seem to account for the differences. So, we kept digging.
We soon hatched a cockamamie theory: In a country still super-smudged by the fingerprints of British colonialism, it seems possible folks think of “English” as a default rather than as a specific ethnic group. Sort of like how the Amish refer to all their non-Amish compatriots as “English,” be they Slovak or Salvadoran.
Even outside Lancaster County, Pennsylvania, “English” in America resembles a trademark so omnipresent no one remembers it belongs to a specific brand. It’s the Kleenex, dumpster or Speedo of race and ethnicity. However, by listing it as a potential response in 2020, the new census race question may have dragged English out of the background.
And as we dug into the data for the related but separate ancestry question, we found what looks an awful lot like a smoking gun: The first and only time the census listed English as a suggested ancestry was in 1980 — which also just happens to be the only year that the English beat out the Germans as the top U.S. ancestry!
Does that mean it’s more accurate to have English on top? Perhaps. English was the top answer for the “origins” portion of the revamped race question, but not the Census Bureau’s long-standing ancestry question, and the Census Bureau may eventually shift to race as its preferred measure, writing that it may “recommend the removal of the ancestry question.”
But we were reluctant to question the survey designs of the absurdly talented folks at the Census Bureau, who can spend years and careers testing, debating and refining a single question.
So, we called in the big guns.
With guidance from Washington Post pollster Scott Clement, and the megaphone wielded by Jackie Weisman at the American Association for Public Opinion Research, we sent an APB (All Pollsters Bulletin) to AAPOR membership. Almost instantly, we hit the mother lode of absurdly smart people who care so much about survey design that they’ll put Thanksgiving on hold to dissect a single survey question.
We first heard from Dina Smeltz, self-proclaimed “rogue pollster” and vice president at the Chicago Council on Global Affairs. She mentioned that folks might have seen “English” and assumed they were answering a question about language and assimilation with American anglophone culture.
Her theory quickly received backup from Stanford University’s Jon Krosnick, a jazz drummer whose work on the psychology of surveys has been cited 60,000 ways to Sunday. He said that given the way the fill-in-the-blank origin component of the race question was designed, it’s nigh on impossible to divine what respondents meant.
Ideally, he said, you would write a question that doesn’t require examples. But that might not have been possible in 1980, an era before the genealogy boom, when the word “ancestry” wasn’t in common use. The 1980 Census data release came as the use of the word in books indexed in Google was around its lowest point in a century, and it coincided with the beginning of a long and steady rise.
“Now in 2024, the word Ancestry is the name of a company! It’s out there in public discussion constantly. But in 1980, it wasn’t a highly used word. It wasn’t a word used in every community every week,” Krosnick said. “So imagine someone misinterprets this question as ‘What language do you speak?’”
Federal researchers also identified this as an issue, but another possible explanation lurked in the background, he said: “frame of reference.” In a world where Americans often use “ethnic” to mean “not Anglo,” it may not occur to people to think of English as an ethnic group with specific roots in the United Kingdom.
Consider an analogy. If you asked us to list the best things on earth, we might start with friendship or French dips, and we almost certainly wouldn’t mention oxygen. But of course we consider breathable air to be a wonderful feature of our planet — among the most important and perhaps even the best. But it’s just not in our frame of reference until you nudge us in that direction.
So with English, we are left wondering whether we were dealing with a language miscue or a refreshed frame of reference. We didn’t know whom to call next. But it turns out the poll people were on the case. Our friends at the Pew Research Center got wind of our request and tracked down some shockingly suitable specialists.
The first was a friendly face, Pew demographer Jeffrey Passel, who you may recall as one of the “demographic demigods” who used a census mistake to show that most Brazilian Americans identify as Latino, no matter what the official stats say.
He’d seen the first signs of this phenomenon after the 1990 Census, when German first overtook English on the ancestry question. That year, in an shake-up of ancestry examples, the Census Bureau added about 10 new choices, and each of them — from Taiwanese to Cape Verdean — increased in popularity. At the same time, the Census Bureau dropped five others, most of which saw their populations fall. (The exceptions were Honduran and Venezuelan, both of which saw broader population increases that decade.)
That would suggest the surge in English identity stems from a joggled frame of reference — and may also rely on the painfully simple notion that we’re more likely to pick from the suggested examples, since they’re right in front of us and we know they’re “right” answers.
But from 2000 onward, the Census Bureau offered neither English nor German as examples. And when left to their own devices, folks were more likely to produce German and less likely to produce English. Why?
Our Pew pals soon put us in contact with another decorated survey methodology expert and past AAPOR president, Roger Tourangeau, retired vice president at Westat, a sizable survey outfit based in D.C.’s Maryland suburbs. In a pair of analyses, Tourangeau and his collaborators found that examples can help survey purveyors such as the Census Bureau guide us toward answers they can compare on an apples-to-apples, dust-to-dust basis.
Consider how broad a question about ancestry is — do we answer “Black,” “sub-Saharan African,” “Angolan” or “Well, my dad’s from Duluth, Minnesota?” If all of the examples are races or regions, we might respond with Asian or Latin American. If they’re countries, we’ll respond with Cambodian or Bolivian.
But we’re also learning how to deal with borderline cases. In particular, Tourangeau and his friends found that including “African American” in the list caused more folks to provide that answer. Similarly, they saw a sharp increase in Irish when that got tossed into the example lineup. (Like the English, the Irish have spread far and wide throughout America and tend to be subsumed into mainstream culture.)
“Left to their devices, people may try to highlight what is distinctive about themselves — and German ancestry may seem more distinctive than English,” Tourangeau said, adding, “I think people with multiple possible answers may glom onto the one they see among the examples.”
“What all of this shows,” Krosnick told us, is that “words matter. Professionals who work in this area have published hundreds of papers on this topic. When you change the words of a question, you often get different results.”
So, in the battle for top spot between the English and the Germans, to us, the winner seems clear: When folks are reminded to keep an eye out for England as they mentally shake their family tree, they find it. Consider that, when the Census Bureau added English as an example to the race question on the annual American Community Survey, and thus made that background more salient in our minds, we also suddenly became more likely to answer “English” to the separate ancestry question farther down the form.
Then again, “English” may simply provide an easy answer for someone who may not be an amateur genealogist but knows that they speak English, that they have an English surname (such as Smith, Johnson or Williams) and that they learned in fourth grade about the Mayflower, which dumped a load of English families on these shores in 1620.
Hi there! The Department of Data seeks your quantifiable queries. What are you curious about: Who’s most likely to commit a crime? Why are more and more Americans declining to report any ancestry at all? Where did the national debt come from? Just ask!
If your question inspires a column, we’ll send you an official Department of Data button and ID card. This week, we should really send yet another one to Scott Clement.
Posts
No comments:
Post a Comment