Decennial Census 1980–2000 · American Community Survey 2010–2023
You may have heard the claim "The USA is more German than English". Based on historical migration patterns, this cannot possibly be true (the English had a 200 year head start on settling North America), but it is often repeated because Americans report more German ancestry than English ancestry in the Census (or the American Communities Survey, which replaced the decennial census for ancestry surveys in recent years)
My hypothesis was that this under-reporting of English ancestry was due to social desirability bias, that this was recent, and therefore a decreasing trend of reported English ancestry might be visible in ancestry data over time
Ancestry questions were first asked in the 1980 Census, and in that year reported English and German ancestry were approximately equal. There was a big decline in reported English ancestry from 1980 to 1990, except in New England and Utah. Unfortunately, because of methodological differences the 1980 results are not fully comparable to the results from 1990 and subsequent years (see below). There is no support for a decreasing trend after 1990 - the English ancestry pattern established in 1990 stayed roughly constant, until around 2020, when reported English ancestry started to increase again.
Another well-documented trend in US ancestry reporting is the tendency of white Americans in the South and Appalachia to report their ancestry as simply "American", rather than English or Scotch-Irish (note "American" was not an available option in 1980). Starting in 2020, some of these areas have seen an increase in reported English ancestry. Most of the increase in reported English ancestry nationally occurred in these former "American" majority counties
Each county is colored by its dominant European ancestry category — the category with the highest share of the total population. Opacity reflects that share. Click a legend item to isolate a particular ancestry group, click the [+] to expand the group and see component ancestries.
Only European ancestries are tracked on this map. Non-European ancestries (Hispanic, Asian, African, Middle Eastern, etc.) account for roughly 35–40% of the national population in recent years and are excluded from this plot to make the shifting patterns of European identification more legible. (I'm working on an all-ancestries presentation, but there are several tricky issues to resolve)
Respondents could report only a single ancestry. This suppresses totals relative to later years — someone with mixed German-Irish heritage could only pick one. German and English shares are therefore underestimates. No "American" ancestry category existed yet.
Respondents could report up to two ancestries, so county totals can exceed 100% of population (each person may contribute to two categories). The 2000 census used a write-in format with a comprehensive coded ancestry list.
The ACS replaced the decennial long form. It surveys ~3 million households per year and rolls up 5-year windows for county-level estimates. Respondents can report multiple ancestries. ACS years shown: 2010, 2011, … 2023.
Ancestry responses are grouped into 10 categories following the ACS B04006 table structure. "British" covers Scottish, Scotch-Irish, and Welsh (distinct from English). "E. European" covers Slavic and Baltic ancestries. "Other European" covers Dutch, Greek, Hungarian, Swiss, Austrian, Portuguese, Belgian, and similar.
1980–2000: IPUMS NHGIS decennial census county extracts.
2010–2023: US Census Bureau ACS 5-year estimates, table B04006.
There are a small number of counties with missing data in some years, due to changes in county names or coding. Data for Connecticut is absent for 2022 and 2023, as the number and shape of the counties completely changed, and I haven't sourced new county shapes to use for those years.