A data-driven blog that uses chart, maps, cartograms, and other visualizations to understand important societal issues.

Diversity, Elections

The Changing Face of America: Where did Trump & Clinton Perform Better & Worse than 2012?

Description: County-level illustration of where Trump and Clinton did better (dark Red and dark Blue) or worse (light Red and light Blue) that Romney and Obama, including winning counties their predecessors did not (Orange and Green). Comparison is based on county vote percentages (not counts); percentages are adjusted to include only the two major candidates.
Analysis: Broadly, the 2016 election saw urban/Blue counties become more Blue and rural/Red counties become more Red. Key points include:

  • Third party candidates reduced the winning margin of the two major candidates in Utah (Trump: Evan McMullin), New Mexico (Clinton: Gary Johnson), and Vermont and parts of surrounding states (Clinton: Bernie Sanders). Also, compare the 2012-2016 change counties with the specific  impact of third party candidates; see http://chartedterritory.us/2018/04/29/2016-election-in-maps-and-cartograms-impact-of-third-party-votes/.
  • High-minority population counties, including along the Texas-Mexico border and the US Black Belt1, saw Clinton performed worse than Obama. The US Black Belt is known for ~100 million-old geology impacting modern-day voting patterns2.
  • Moderate-to-large urban counties were most likely to flip from Red (2012) to Blue (2016). These included Orange (CA; Anaheim), Salt Lake (UT; Salt Lake City), Lancaster and Douglas (NE; Lincoln and Omaha), Fort Bend (TX; Sugar Land), Anne Arundel (MD; Annapolis), Cobb (GA; Marietta), Montgomery (VA; Blacksburg), and Gallatin (MT, Bozeman) counties.
  • Swing states saw the most concentrated county flipping, particularly Wisconsin and Michigan, but not Pennsylvania. Maine, arguably, has the largest statewide impact.

Methodology: 2016 election data were collected manually from the Secretary of State website for each state. 2012 election data were sourced from GitHub3. Spatial data was visualized using ESRI ArcMap4. Data was processed using custom VBA scripts in Microsoft Excel5. The comparison only looks at the vote percentage of the two major counties. In each case, the vote actual vote percentage of the two major candidates in 2012 and 2016 was summed (almost always less than 100%) and then reassigned between the two candidates (always adds to 100%). This removes the direct impact of third party candidates, population growth, and voter turnout.


  1. https://en.wikipedia.org/wiki/Black_Belt_(U.S._region)
  2. http://www.deepseanews.com/2012/06/how-presidential-elections-are-impacted-by-a-100-million-year-old-coastline/
  3. https://github.com/tonmcg/County_Level_Election_Results_12-16
  4. http://www.esri.com/data/esri_data
  5. https://products.office.com/en-us/excel

 

Leave a Reply