I finally realized I could simply look up the answer in Nathaniel Weyl's endlessly fascinating 1989 book Geography of American Achievement, which compares ethnic surname from various reference books listing different kinds of big shots compared to the total number of those surnames on the Social Security rolls in 1984. We can't identify every surname by ethnicity but we can know that just about everybody named, say, "Caruso" is descended from Italians in at least the direct male line, while somebody named "Weber" has a German name.
In the 1985 book listing 127,000 American Men and Women of Science, people with ten specific Italian surnames were represented only 54% as often as their fraction of the total American population. However, in a more elite book listing 16,700 prominent scientists, Frontier Science and Technology, Italians were 98% of the average.
So, that may suggest that Italian-American culture isn't as science-oriented as the American average, but that those Italians who do go into science do just as well as the national average.
In case you were wondering about other ethnic groups based on characteristic surnames, here are some (but not all) of the groups as of the mid-1980s, with the national average as 100. The broader database of professional scientists listed first and the more elite listing of top scientists second:
In case you were wondering, Weyl notes:
"The higher achievement index of Swedes and Danes than of Norwegians is not a statistical aberration, but a reality. This is indicated by the magnitude of the difference and by the fact that Swedes lead Norwegians by significant numbers in the great majority of those rosters of achievement in which the comparison could be made."
Alert Garrison Keillor!
Names from the British Isles were statistically adjusted to exclude the impact of African-Americans (Weyl used the name "Washington," which is now found among blacks about 85%-90% of the time to estimate the African-American coefficients of achievement.) Names from the British Isles are likely to also include various non-British people of European descent, such as, to list a couple of non-scientists, John Kerry and Woody Allen.
As you might expect, the Sikhs in Weyl's study were all named "Singh" and the Koreans all named "Kim." The Sikh/Singh sample size is pretty small (only 22,000 Singhs in the Social Security database, so there are about 75 Singhs in the broader reference work of scientists), but most of the other sample sizes are more reliable, such as 295,000 Italians, 804,000 Jews and 1,937,000 Scots.
(The size of Weyl's SS samples depend in part on how diverse names are within an ethnic group because Weyl only used a limited number of names for each group that he could be sure belonged to members of that group. For example, there are fewer Jewish-Americans than Italian-Americans in America, but there are fewer Jewish than Italian names (for example, there are almost three times as many people in the U.S. with the last name of Cohen in the U.S. as there are with any single Italian surname).
Of course, Weyl's choice of surnames to focus upon could bias the national indices to some extent. One of Weyl's most amazing findings is that people with old English clerical names (Clark, Clarke, and Palmer) that indicate their direct male line ancestors were literate around the time surnames were adopted (about 1300) are 50% more likely to show up on lists of high achievement than people with other English names.
He also found that people with the kind of surnames common among 19th Century Chinese immigrants (e.g., Chan and Chang) tended to perform at a lower level than those with the surnames common among 20th Century Chinese immigrants (e.g., Chen and Cheng). Earlier Chinese immigrants were more often recruited to be laborers, while more recent immigrants were often students.
Immigrants are not necessarily representative of the countries from which they came. For instance, the mediocre achievement indices of Franco-Americans stems from the snootier sort of Frenchman's horror at the idea of emigrating to unrefined America (Jacques Barzun is close to being the exception that proves the rule), so Franco-Americans tend to be descended from French Canadian fur-trappers, lumberjacks, and the like.