Still Unmelted after All These Years

With acknowledgments to Michael Novak, The Rise of the Unmeltable Ethnics (1972) and Paul Simon’s 1975 album and song.

Are Polish Americans or Italian Americans or African Americans uniformly distributed through the United States? No; in fact, America is stunningly "unmelted". Even a cursory exploration of the Census data reveals this. For example, check out the American Factfinder at the US Census cite.

A bowl of raw meat and uncooked potatoes, celery, carrots and onions is not per se appetizing. But even in a well simmered and tasty soup or stew, you can tell by looking that there are carrots, potatoes, celery, onions and meat. So let’s not despair. Let’s investigate.

The Ancestry Question on the US Census has produced a stunning array of information about how Americans self-describe themselves. The self-describing aspect of the US Census, especially The Ancestry Question is an highly important feature of data collection in a pluralistic democracy. Unlike the Race Question on the US Census which was constitutionally and historically imposed and rooted in pseudo-scientific and political assumptions of exclusion, the Ancestry Question emerged from a more current understanding of ethnicity[FN1] and its organic character and growth through the self-determined iterations rooted in the person, family, household and neighborhoods that constitute the American experience of immigration, urbanization and the attendant cultural pluralism of democratization and freedom fostered by a wide range of forces that accompanied American political development especially for the past seven decades. These social economic, political, and personal dynamics make the demography of ethnicity in America seem messy. Indeed, the ostensible messiness of immigration, the articulation of ancestry and identity rooted in ethnicity may well explain the slow evolutionary process and the significant impediments to collection of demographic information. Uniform data would be achieved by replacing the variety of Race and Ethnic Origin Questions associated with Hispanic, Asian, Indigenous Peoples with a single Ancestry question and the tabulation of the multiple responses that are clearly evident in America. Nonetheless, now that Ancestry data has been collected for the last three Censuses and the computer driven computational revolution is firmly in place, demographic analysis can employ standard protocols and verifiable methods that enable a fresh look at the data and thus establish connections, patterns and places and further discussion, interpretations and a scientific understanding of American pluralism.


FN1. Although, in some ways the Ancestry Question is arguably back to the future. See Saint Francis College v. al-Khazraji, 481 U.S. 604 (1987). A unanimous Court held that that persons of Arabian ancestry were protected from racial discrimination under Section 1981. The history of the definitions of “race”, presented by the Court, is well worth reading because it shows how prior to the 20th century “race” and “ancestry” were synonymous concepts. After outlining the history and usage of the term "race", Justice White and the Court rejected the claim that “a distinctive physiognomy” is essential to qualify for 1981 protection and concluded: "We have little trouble in concluding that Congress intended to protect from discrimination identifiable classes of persons who are subjected to intentional discrimination solely because of their ancestry or ethnic characteristics." William J. Brennan, Jr., in a separate concurrence, added that "Pernicious distinctions among individuals based solely on their ancestry are antithetical to the doctrine of equality upon which this nation is founded." (Emphasis supplied).

This article investigates one such method: State Similarity Scores. A Similarity Score investigates the“distance” between States. Consider three cities: Baltimore, MD; Washington, DC and Chicago, IL. Baltimore is about a 40 mile drive from Washington. The driving distance between Washington and Chicago is roughly 710 miles. Chicago is 720 miles from Baltimore on the interstates. Knowing these distances, we can conceive of the triangle that these cities form and how they are geographically related.

In the two dimensional space of a map, a computer can now easily crunch out distances from a simple formula derived from Phythagoras.

Distance^2 = a^2 + b^2, or
Distance = SQRT(a^2 + b^2 )

For example, using latitude and longitude to get the distance between Chicago and Baltimore, we find the difference between Chicago’s latitude and Baltimore’s latitude and the difference between Chicago’s longitude and Baltimore’s longitude.

a = LatChicago - LatBaltimore and
b = LongChicago - LongBaltimore

So,
Distance = SQRT[(LatChicago - LatBaltimore)^2 + (LongChicago - LongBaltimore)^2]

In three dimensions, we’d add c^2, to handle perhaps altitude for Google Earth. The theorem isn’t limited to our spatial definition of distance. It can apply to any orthogonal dimensions: space, time, movie tastes, colors, temperatures, and even ancestry responses. There is no limit to the number of variables. The focus, however, of this research is race, ethnicity and ancestry data form the US Census 2000. Appropriately, this type of investigation is also known as Nearest Neighbor Analysis. To find out how closely related any two states in terms of ethnicity, our equation would look like this:

Distance = SQRT[ (Ancestry1State 1 - Ancestry1State 2)^2 + (Ancestry 2State1 - Ancestry2State2)^2 + ... + (Ancestry NState1 - Ancestry NState2)^2]

For this paper I used 56 of the largest ethnicities [FN2]as orthogonal dimensions: Asian Indian, Asian Multiple Response, American Indian, “American”, Arab, Austrian, Black or African American, Belgian, British, Canadian, Chinese, Cuban, Czech, Czechoslovakian, Danish, Dutch, English, Finnish, French excluding Basque, Filipino, French Canadian, German, Greek, Guamanian and/or Chamorrian, Jamaican, Japanese, Korean, Hawaiian, Hispanic or Latino Other, Hungarian, Irish, Lithuanian, Mexican, Native Not Specified, Norwegian, “Others”, Other Asian, Other Pacific Islander, Puerto Rican, Polish, Portuguese, Russian, Samoan, Scandinavian, “Scotch Irish”, Scottish, Slovak, Slovene, “Some Other Race”, Sub Saharan African, Swedish, Ukrainian, Vietnamese, Welsh, and West Indian.[FN3]

FN2 Some of these categories also come from the Race and Hispanic origin questions of the Census. Even though the Ancestry Question captures ethnic responses like Japanese, Korean, Cuban, Mexican and Black or African American, the Census Bureau sanitizes its Ancestry data, so that these responses are only readily available from the Race and Hispanic origin questions.

FN3 Older analysis of ethnic disimilarity differs from this method because it grouped ancestry responses into larger but far fewer categories such as “Old Stock”, “Eastern and Southern European”, “Asian”, etc. Calculating similarity in 56 dimensional space was simply not possibly with hand calculations employed by previous researchers

For any two states, we can calculate a measure of similarity. A measure of 0, would mean that the two states are identical, i.e. they have exactly the same percentage of Polish American, Italian Americans, Irish Americans, African Americans, etc. The largest “distance” between two states was between DC and North Dakota at 91.429. The closest “distance” between two states was between Tennessee and Arkansas 3.720. At the end, Table 1 shows each state’s “nearest cultural neighbors” and the “distance” metric.

If we look at only the closest connection for each of state, some distinct networks or groupings emerge. The largest of these clusters happens to correspond roughly to "The South".

Finally, we can also measure the distance of each State to the United States as a whole. Illinois and Florida are very similar to the entire US, while North Dakota, DC and Hawaii are furthest in “distance” from the US in our 56 dimensional ethnic space. See Fig. 1. Table 1. Nearest Neighbors along 56 dimensions of Ethnicity/Ancestry. Table 1, three nearest ethnic neighbors not posted here.