Thousands of Trump supporters gather for a rally in Wilkes-Barre, Pennsylvania.
After the 2012 Election, the Beltway consensus was that for a Republican to take the Presidency, he or she would need to find a way to reach out to Hispanics and other minorities. As they tell it, the reason Mitt Romney lost the 2012 Election was because he got an anemic 27% of the Latino vote, much less than the 44% George W Bush received in 2004. Indeed, to the chagrin of the Republican base, in 2013 four Republican Senators – including former Presidential hopeful Marco Rubio – teamed up with the Democrats to sponsor an Immigration Reform Bill, likely a move aimed at improving their dismal 2012 numbers with Latino voters.
Does the data actually confirm the idea that it’s minority voters specifically that are dooming the Republican Party in Presidential elections? Not really. While it is incredibly difficult to win the national popular vote while losing badly among African Americans, Hispanics, and Asian Americans, as is the case currently with Trump, most of the competitive “swing” states actually have very small minority populations relative to the national average. Indeed, according to detailed demographic and census analysis at The Upshot, the 2012 electorate in the swing states of Ohio and Pennsylvania was actually 84% white. In New Hampshire and Iowa, more than 93% of the electorate was white. In fact, out of the nine main swing states during the 2012 election, only two states, Florida and Nevada, had electorates that were more than 30% non-white.
That brings us to the 2016 election. Once again, the pundits exclaim that Trump can’t possibly win the election without doing respectably with Latino voters. Joe Scarborough of MSNBC, about as Beltway insider as you can get, quipped that “There are not enough white voters in America for Donald Trump to win while getting routed among minorities”. Yet Nothing But Numbers is forecasting Trump victories in Ohio, Pennsylvania, and Florida, all states where Romney fell short. What gives?
For a better idea, let’s take a look at Pennsylvania, a state Obama won by 5.2 points that also happens to be right in the middle of the rapidly changing political battleground known as the Rust Belt. As you know from my previous post, Nothing But Numbers is using Facebook “Likes” to predict the 2016 election in all 50 states. Since I made that post, however, a demographic component has also been added to the model, and the model is now 80% based on Facebook “Likes” and 20% based on demographics (the full 50 state version of the new model will be published tomorrow). Here is Nothing But Numbers’ 2016 Pennsylvania projection, shown in comparison to Romney’s 2012 performance, on a county by county basis:
| County | Romney Vote% | Obama Vote% | Projected Trump Vote% | Projected Clinton Vote% | Trump Margin G/L% |
|---|---|---|---|---|---|
| Adams | 63.1% | 35.5% | 60.6% | 34.4% | -1.5% |
| Allegheny | 42.2% | 56.6% | 39.5% | 55.5% | -1.6% |
| Armstrong | 67.9% | 30.7% | 66.3% | 28.7% | 0.4% |
| Beaver | 52.6% | 46.0% | 58.1% | 36.9% | 14.6% |
| Bedford | 77.0% | 22.1% | 69.7% | 25.3% | -10.5% |
| Berks | 49.5% | 48.9% | 54.4% | 40.6% | 13.1% |
| Blair | 66.1% | 32.6% | 64.8% | 30.2% | 1.1% |
| Bradford | 61.5% | 36.8% | 67.4% | 27.6% | 15.2% |
| Bucks | 48.8% | 50.0% | 47.9% | 47.1% | 2.0% |
| Butler | 66.8% | 31.9% | 64.4% | 30.6% | -1.1% |
| Cambria | 58.0% | 40.2% | 63.8% | 31.2% | 14.8% |
| Cameron | 64.3% | 34.4% | 58.5% | 36.5% | -7.8% |
| Carbon | 52.5% | 45.6% | 63.5% | 31.5% | 25.1% |
| Centre | 49.0% | 48.9% | 42.9% | 52.1% | -9.4% |
| Chester | 49.7% | 49.2% | 38.7% | 56.3% | -18.1% |
| Clarion | 67.2% | 31.1% | 64.7% | 30.3% | -1.7% |
| Clearfield | 63.7% | 34.8% | 66.6% | 28.4% | 9.4% |
| Clinton | 54.8% | 43.5% | 62.6% | 32.4% | 19.0% |
| Columbia | 55.3% | 42.9% | 63.6% | 31.4% | 19.8% |
| Crawford | 58.9% | 39.4% | 60.8% | 34.2% | 7.2% |
| Cumberland | 58.5% | 40.0% | 52.5% | 42.5% | -8.5% |
| Dauphin | 46.4% | 52.4% | 45.4% | 49.6% | 1.9% |
| Delaware | 38.6% | 60.4% | 38.2% | 56.8% | 3.2% |
| Elk | 57.1% | 41.4% | 62.1% | 32.9% | 13.4% |
| Erie | 40.9% | 57.8% | 49.7% | 45.3% | 21.3% |
| Fayette | 53.6% | 45.3% | 62.1% | 32.9% | 20.9% |
| Forest | 59.8% | 38.7% | 66.4% | 28.6% | 16.8% |
| Franklin | 68.6% | 30.1% | 65.2% | 29.8% | -3.0% |
| Fulton | 77.7% | 21.1% | 69.2% | 25.8% | -13.2% |
| Greene | 58.2% | 40.5% | 66.6% | 28.4% | 20.6% |
| Huntingdon | 68.0% | 30.6% | 66.3% | 28.7% | 0.3% |
| Indiana | 58.6% | 39.8% | 62.3% | 32.7% | 10.8% |
| Jefferson | 72.0% | 26.6% | 69.0% | 26.0% | -2.3% |
| Juniata | 72.0% | 26.7% | 67.5% | 27.5% | -5.2% |
| Lackawanna | 35.8% | 63.1% | 49.6% | 45.4% | 31.5% |
| Lancaster | 58.9% | 39.7% | 52.6% | 42.4% | -9.1% |
| Lawrence | 53.9% | 44.9% | 62.6% | 32.4% | 21.3% |
| Lebanon | 63.2% | 35.3% | 59.2% | 35.8% | -4.5% |
| Lehigh | 45.6% | 53.2% | 40.3% | 54.7% | -6.8% |
| Luzerne | 46.8% | 51.7% | 57.2% | 37.8% | 24.3% |
| Lycoming | 66.0% | 32.7% | 63.9% | 31.1% | -0.5% |
| McKean | 63.2% | 35.0% | 63.9% | 31.1% | 4.7% |
| Mercer | 50.6% | 48.0% | 60.4% | 34.6% | 23.2% |
| Mifflin | 72.9% | 26.1% | 66.6% | 28.4% | -8.6% |
| Monroe | 42.3% | 56.4% | 56.5% | 38.5% | 32.0% |
| Montgomery | 42.3% | 56.6% | 34.9% | 60.1% | -11.0% |
| Montour | 59.5% | 38.9% | 60.0% | 35.0% | 4.3% |
| Northampton | 47.1% | 51.6% | 52.4% | 42.6% | 14.2% |
| Northumberland | 58.8% | 39.3% | 66.0% | 29.0% | 17.5% |
| Perry | 68.6% | 29.8% | 67.3% | 27.7% | 0.8% |
| Philadelphia | 14.1% | 85.2% | 18.8% | 76.2% | 13.7% |
| Pike | 54.9% | 43.9% | 59.8% | 35.2% | 13.6% |
| Potter | 72.2% | 26.3% | 71.1% | 23.9% | 1.3% |
| Schuylkill | 55.9% | 42.5% | 65.9% | 29.1% | 23.4% |
| Snyder | 67.2% | 31.2% | 62.3% | 32.7% | -6.5% |
| Somerset | 70.7% | 27.8% | 66.0% | 29.0% | -5.9% |
| Sullivan | 63.4% | 35.0% | 57.4% | 37.6% | -8.6% |
| Susquehanna | 59.9% | 38.4% | 64.8% | 30.2% | 13.1% |
| Tioga | 66.5% | 31.7% | 66.6% | 28.4% | 3.4% |
| Union | 60.9% | 37.5% | 57.9% | 37.1% | -2.6% |
| Venango | 62.2% | 36.0% | 63.1% | 31.9% | 5.1% |
| Warren | 57.2% | 41.1% | 63.3% | 31.7% | 15.5% |
| Washington | 56.0% | 42.7% | 59.4% | 35.6% | 10.5% |
| Wayne | 59.8% | 38.8% | 64.5% | 30.5% | 13.0% |
| Westmoreland | 61.3% | 37.6% | 61.0% | 34.0% | 3.4% |
| Wyoming | 55.2% | 42.9% | 64.5% | 30.5% | 21.7% |
| York | 59.9% | 38.7% | 61.5% | 33.5% | 6.7% |
| Pennsylvania Total | 46.8% | 52.0% | 49.9% | 45.1% | 9.9% |
| State | Projected Trump Vote% | Projected Clinton Vote% | Trump Win State% | Clinton Win State% |
|---|---|---|---|---|
| Pennsylvania | 49.9% | 45.1% | 76.9% | 23.1% |
Counties in red are projected to shift 10 points or more towards Trump since 2012, while counties in blue are projected to shift 10 points or more towards Clinton.
Of the 67 Counties, only 22 are projected to shift towards Clinton, compared to 45 projected to shift towards Trump. In addition, while only four counties are projected to shift at least 10 points towards Clinton, a staggering 28 counties are projected to shift at least 10 points towards Trump since 2012, including 11 counties shifting 20 points or more.
Trump Makes Headway In Places Hit Hard By Globalization
While Clinton does make gains in the much-talked about suburban Philadelphia counties of Montgomery and Chester, Trump more than makes up for this by making huge gains in Philadelphia County, northeastern Pennsylvania, and western Pennsylvania. This shift towards Trump is especially pronounced in the so-called “Rust Belt”, places defined by a loss of manufacturing jobs and per capita income relative to the rest of the United States since the post-war era. Just look how similar the map of per capita income decline relative to the national average is in comparison to the above map showing areas where Trump has gained:
Highlighted counties have experienced a significant net loss in manufacturing jobs from 1954-2002. Counties in green have experienced a positive change in per capita income relative to the national average, from 1980-2002. Counties in red have experienced a negative change in per capita income relative to the national average, from 1980-2002.
Compared to the first map showing where Nothing But Numbers projects Trump to gain and lose voters relative to 2012, the above map is shockingly similar. Indeed, of the 12 Pennsylvania counties shaded either bright red or dark red in the above map, indicating a significant loss of per capita income from 1980-2002 relative to the national average, 11 of the 12 have shifted towards Trump relative to Romney in 2012. 10 of the 12 counties have shifted more than 10 points towards Trump.
While the Philadelphia and Pittsburgh metropolitan areas have done well in a more globalized world – corresponding with good numbers for Clinton – northeastern and western Pennsylvania have been decimated by the loss of manufacturing jobs, as seen by the deep red shadings in those areas on the above map. Demographically, it’s also worth mentioning that many of the areas where Trump is gaining are almost exclusively white, with very few black and hispanic voters. This is perhaps the biggest difference between western Pennsylvania – where loss of manufacturing jobs and per capita income decline are leading to increased Trump support – and urban, diverse Detroit – where the same economic trends are not leading to a similar groundswell of support for Trump.
Voter Registration Trends Confirm Trump’s Surge in Pennsylvania
Facebook “Likes” aren’t the only thing pointing to troubles for Clinton in Pennsylvania. Since 2015, when Trump entered the Presidential race, Republican voter registration has surged. In 2015, that corresponded with 30,482 Democrats changing their registration to Republican, compared to 17,829 who did the opposite. In 2016, an astonishing 109,778 Democrats switched their party registration to Republican, while only 45,458 Republicans became Democrats, a net shift of more than 60,000 voters. In fact, if you take the sum of people changing their party registration from 2013-2016, you get a net shift of 85,000 voters shifting their party registration from Democrat to Republican.
In addition, the counties where Nothing But Numbers’ model shows Trump making gains correspond almost perfectly with the voter registration data. Below is a map showing where each party has made gains in terms of voters switching from one party or no affiliation to the other party:
Counties in red have experienced more Independents and Democrats joining the Republican Party in 2016 than vice versa. Counties in blue have had more Independents and Republicans join the Democratic Party in 2016 than the opposite.
As you can see, outside of the Philadelphia metropolitan area and Centre County (home of Penn State University), Pennsylvania experienced a pretty drastic rightward shift in 2016 in terms of party registration. And it probably goes without saying that these Democrats and Independents are mostly joining the Republican Party because of Donald Trump, who dominated the Pennsylvania Republican Primary with 57% of the vote in a three-way race.
In conclusion, Donald Trump’s anemic numbers with hispanic voters have not completely doomed him thanks to a resurgence in the Rust Belt states of Pennsylvania and Ohio. Fortunately for Clinton, she still has plenty of paths to victory that don’t require winning Pennsylvania or Ohio. Still, having Pennsylvania firmly in his column is a major boost to Trump’s chances this Tuesday.
Herbert Rosenberg 
Anonymous 
Lefty 
bnoah 
Mike Meagh 
Alvin Baron 
Steve Malloy