Fashions change. For example, the fraction of Americans who report English ancestry has dropped drastically since 1980 – so much that so that you would have to wonder about secret death camps if you took it seriously. But it’s fashion. ... This means that people in the US claiming a particular ethnicity can not only have limited ancestry from that group, but be oddly unrepresentative as well.The GSS shows 15.2% and 9.0% of respondents identifying as English or Welsh in 1980 and 2012, respectively. The US Census shows 18.5% and 8.0% in 1980 and 2012, so the GSS parallels the census results pretty well.
I figured the boiling off that occurred over the last three decades would be perceptible by looking at the political orientation profiles of the two cohorts, but (excluding moderates to make the comparisons easier to digest) those of English/Welsh ancestry were 32% liberal, 68% conservative in 1980 and, similarly, 38% liberal, 62% conservative in 2012. To the extent any disproportional shifting did occur with the shaking off of English/Welsh affiliation, it was more among political conservatives than among liberals. I would've guessed the opposite to have been the case. The only ancestry less attractive to SWPLs than English is probably German. Okay, and "American", too, of course.
Greg Cochran gives a clue in the fashion bit, though--the English/Welsh affiliates of today are a noticeably older lot than were their counterparts 30 years ago. In 1980, the mean age of English/Welsh adults was 46.7. As of 2012 it was 53.2.
GSS variables used: YEAR(1980, 2012), POLVIEWS(1-2)(3-5)(6-7), ETHNIC(8), AGE