What Do Dense Trump Areas Look Like? Sparse Clinton Areas?

The 2016 Presidential Election was almost certainly the most polarized election along urban/rural lines in American presidential history. While the swing between 2012 and 2016 could be best explained by education levels, no single census tract-level variable other than race better explained the precinct-level Clinton-Trump percentage margin than degree of urbanization:

Independent Variable(s) 2016 R-squared 2012 R-squared
Race and Ethnicity (White, Black, Hispanic, Asian, Native American) 0.5185 0.5954
Area Classified by Census Bureau as Urban 0.3933 0.3140
Population Density 0.2792 0.2672
Median Age 0.2591 0.2575
Households on Welfare (TANF, WIC, etc.) 0.2567 0.3158
Proportion on Medicaid 0.2531 0.3114
Proportion Employed in Construction Industry 0.2267 0.2034
Proportion Employed in Manufacturing Industry 0.2161 0.2014
Gini Coefficient (Inequality) 0.2043 0.2009
Proportion Employed in Scientific or Technological Industries 0.2000 0.1811
Median Household Income 0.1936 0.2387
Labor Force Participation Rate 0.1932 0.1835
Proportion with College Degree 0.1870 0.1869
Disabled Population 0.1857 0.1820
Proportion Employed in Finance Industry 0.1818 0.1860


By comparing the two columns, we can see that the 2016 General Election was substantially more polarized along urban/rural lines, while less polarized along racial and income lines than the 2012 General Election. This urban/rural polarization was not a symmetric process, but largely occurred due to dramatic swings to Trump in less dense areas, without a strong countervailing impact in dense areas:

Despite this trend, the plurality of voters for both candidates lived in areas with between 2,000 and 5,000 people per square mile, just as in 2012, although Trump’s peak declined and Clinton’s rose:

As indicated by the first chart above, there are non-negligible areas in the tails of each candidate’s distribution; Trump won some quite dense precincts while Clinton won some very sparse areas. What do these precincts look like? Defining a threshold of 10,000 people per square mile as “dense” and below five people per square mile, Trump won 709 dense precincts nationally and Clinton won 762 sparse precincts. Let’s look at each category:

Dense Trump precincts

The average dense Trump precinct was 71.2% white, 14.8% Hispanic, 9% Asian, and 3.3% black with 39.4% holding college degrees and a median household income of $61,728. They tend to be higher in people of Italian, Russian, Ukrainian, Polish, and Arab descent and lower in Irish, British, and Swedish descent compared to other majority-white dense precincts. They have more people employed in education and healthcare and fewer employed in accommodations and science or technology-related fields, are older, and have both a higher median household income and more non-seasonal housing vacancies and people not in the labor force than average.

The large majority of these precincts exist in New York City, with some others scattered in parts of Philadelphia, Chicago, Miami, and Orange and San Diego Counties in California. They are parts of El Cajon, Huntington Beach, Costa Mesa, Cypress, La Mirada, Corona, and Santa Clarita in southern California; Cuban areas in Hialeah, West Miami, and Sweetwater, but also coastal communities like Hallandale in Broward County; highly Polish communities in Summit and Harwood Heights, Illinois; a number of precincts in South Philadelphia near the intersection of Broad Street and I-76 plus two other large clusters on the north city of the city (one near Fox Chase and Jeanes Hospital, the other just west of Longmead Farms near Archbishop Ryan High School); Lindhurst and North Arlington in Bergen County, New Jersey.

New York City’s dense Trump precincts span numerous neighborhoods: Arden Heights, Great Kills, Westerleigh, Heartland Village on Staten Island; Borough Park, Dyker Heights, Gravesend, Sheepshead Bay, Coney Island, Brighton Beach, Gerritsen, Eastern Parkway, and Broadway Triangle in Brooklyn; Middle Village and Whitestone in Queens; Floral Park, Locust Point in the Bronx; Franklin Square, Williston Park in Nassau County, and near the Yonkers Raceway in Westchester County.

Borrowing Patrick Ruffini’s penchant for Google Street View, Trump won areas that look like:

Sparse Clinton Precincts

The average sparse Clinton precinct is 40.4% white, 27.5% Native American, 20.6% Hispanic, and 2.1% black. The vast majority of these precincts are either on reservation land or along the Mexican border in Texas. They are much poorer with a median household income of $37,117. Only 24.5% hold college degrees. These precincts have very high rates of labor force non-participation (43.3%, just slightly lower than the total employed population) and high rates of government assistance usage (37.6%). Employment that does exist is primarily in education, social services, and public administration.

Equally interesting, if not moreso, are sparse Clinton precincts in majority-white census tracts, of which there were 291. These precincts tend to have about the same income as a sparse majority-white Trump-supporting precincts ($49,059), but have 37% with college degrees compared to 29.9% in similar areas Trump won. White-majority sparse Clinton precincts exhibit a much higher rate of seasonal housing than sparse Trump precincts, 30.1% to 17.4%. These tend to be areas where lots of people go to vacation; they have much higher rates of employment in accommodation, travel, real estate, construction, arts, and recreation but much lower in agriculture and forestry, mining and oil extraction, manufacturing, and finance. Ethnically, these areas are very German with relatively low levels of English, Dutch, and Danish ancestry (all three are heavily associated with conservative and highly religious communities).

Sparse majority-white precincts can be found throughout Alpine, Inyo, and Mono Counties in California and montane Colorado; outside Jackson, Wyoming; throughout Washington’s Methow Valley and near North Cascades National Park; near Ashland, Oregon; outside Moab, Utah; around Sedona, Arizona; in rural areas in northern Nevada and outside Reno; near Sun Valley and Stanley, Idaho; in far northern California in Del Norte, Humboldt, and Trinity Counties; near Big Bend, Glacier, Grand Canyon, and Yellowstone National Parks; and throughout rural western and northern New Mexico. Outside the west, the only significant cluster of such precincts is along the Lake Superior coast in Minnesota plus a few scattered in northern New York through Maine.

Some examples of sparse Clinton precincts:

Other than perhaps the image from Dunton, Colorado, I’d be hard pressed to guess which belonged to a Trump precinct and which belonged to a Clinton precinct just by looking.