Estimating the proportion of Irish ancestry in the US and Massachusetts

[See Estimating the proportion of Puritan genes in America's white population for links to census data.]

"A Survey of Irish Surnames 1992-97" (pdf) lists the following as the 10 most common surnames in Ireland in the 1990s:

1. Murphy 2. (O)Kelly 3. Walsh(e) 4. (O)Connor 5. (O)Sullivan 6. (O)Byrne 7. (O)Brien 8. Ryan 9. Smith/Smyth 10. (O)Neill

We'll exclude Smith/Smyth for obvious reasons. The remaining 9 most common names, all of Gaelic origin, cover 7.85% of the 1990s Irish population. (With the 1890 data, the number would be 7.67%; but that's leaving out some of the variants included in the 1990s survey.) Northern Ireland's inclusion in the survey might end up inflating our surname-based Irish Catholic population estimates by something like 10%, but I'm not worried about this level of error right now.

The number of US whites bearing one of the nine most common Irish surnames in 2000, from Census data: 1188571

The extrapolated equivalent total number of Irish individuals among the US white population in 2000: 15141032

Which comes out to 7.78% of the ancestry of the US non-Hispanic white population in 2000.

15 million (or maybe 13.5 million) descendants is certainly a more plausible biological outcome of 4.5 million Irish immigrants than the "40 million Irish Americans" we see from census self-identifications.

But it appears there's considerably less disconnect between levels of Irish ancestry and Irish self-identification in Massachusetts (vs. the US as a whole).

In the 1940 Census (the 2000 Census surname data is not available broken down by state), 87028 Massachusetts whites had one of the nine most common Irish names. Based on that, we can estimate the number of Irish in MA was 1108637 -- or 25.9% of the total 1940 MA white population of 4280019.

The 2005-2009 American Community Survey 5-Year Estimates put the Irish proportion of the Massachusetts population, based on self-identification, at 23.7% (vs. 11.9% for English). Or, considering only the non-Hispanic white population, something like 29% identify as Irish.

This better agreement likely reflects relatively lower levels of intermarriage in MA, as might be expected from the state's greater Irish concentration.


Average Joe said...

I think you are making the mistaken assumption that most Irish Americans have Gaelic last names. A lot of them anglicized their names, especially those who arrived in the 19th and early 20th centuries.

Average Joe said...

And don't forget that a lot of Irish Americans are only partially of Irish ancestry. Therefore, someone who is only Irish on their mother's side (i.e. Mark Wahlberg, for example) will not have a Gaelic last name.

Vanishing American said...

From this genealogy info on Wahlberg, his Irish ancestry seems to be less than his French-Canadian, Swedish, and even Puritan English.

So how does one define "Irish-American''? Is anyone with a fraction of Irish blood an ''Irish-American"?

It becomes very messy when people have multiple nationalities in their family tree. There are some people who just pick an ancestry out of the family tree, say 1/16 American Indian, and identify 100 percent with that. Same with the Irish identity; it's claimed by many Americans who have only a little provable Irish ancestry.

n/a said...

Average Joe,

We're not assuming Irish Americans all have Gaelic names. We're assuming post-famine Irish immigrants carried the nine most common distinctively Irish surnames in (roughly) the same proportion as the Irish population as a whole (this is not a perfect assumption, but it's close enough for my purposes here).

That is, if we count bearers of these names in the US, we'll know we've counted about 7.85% of bearers of Irish surnames (in terms of patrilineal descent from Irish immigrants -- we don't care about the etymology), and can estimate the full number accordingly (regardless of how many outside the top 9 have non-distinctively Irish names). Bearers of Irish surnames as a fraction of the total number of whites in the US then becomes our estimate for the fraction of Irish genes in the US white population.

(By letting each Irish-surnamed person represent the equivalent of a single Irish individual, we're overcounting mixed-Irish with Irish surnames and non-Irish wives of Irish-surnamed men, while missing a roughly complementary number of mixed-Irish without Irish surnames and women born with Irish surnames but married to non-Irish-surnamed men.)