Community Spring Cleaning week is here! Join your fellow Maveryx in digging through your old posts and marking comments on them as solved. Learn more here!

Analytics

News, events, thought leadership and more.
DrDan
Alteryx Alumni (Retired)

Part I of a three part series, continue reading part II and part III

 

In my previous blog post, I examined trends in county level political partisanship in the post World War II period. That analysis indicates that there have been dramatic shifts across regions of the country in that time period, and since the 1970s, there has been increasing partisan polarization consistent with the idea of there being a "big sort" of people into enclaves of politically like-minded individuals.1 What was only very briefly covered in that post are the factors that are driving the observed current partisanship patterns, which is the topic of this upcoming series of posts. For this analysis, the partisanship measure used is the county level Partisan Voting Index (or PVI) following the 2012 election.

 


"...there has been increasing partisan polarization consistent with the idea of there being a "big sort" of people into enclaves of politically like-minded individuals."


 

Drivers of political preferences and behavior have been heavily researched by both academics and practitioners. Below I will discuss some of the important findings in this area, but before doing that, let's look at some initial data to help frame the discussion. The data consists of two summary tables. The first contains demographic, socioeconomic, and religious participation data for the 13 most Republican leaning counties in the US, while the second provides the same information for the 12 most Democratic leaning counties. The socioeconomic and demographic data represents 2012 third quarter estimates from the Experian CAPE data.2 The religious participation data comes from the 2010 U.S. Religion Census, which is conducted by the Association of Statisticians of American Religious Bodies (ASARB).

 

Characteristics of the Most Republican Leaning U.S. Counties Based on the PVI

 

County PVI Population Density Non-Hispanic White African American Native American Hispanic Evangelical or LDS Other Religion Per Capita Income
King, TX R+49 0.2 83.6 0.4 1.3 13.7 100.0 0.0 $29,836
Franklin, ID R+45 19.5 91.7 0.2 0.4 6.6 94.5 0.0 $17,620
Madison, ID R+45 81.3 90.8 0.5 0.3 6.1 98.4 0.0 $15,133
Roberts, TX R+45 0.8 90.0 0.3 0.4 8.2 63.0 0.0 $36,172
Glasscock, TX R+44 1.4 66.1 1.5 0.3 31.8 15.8 0.0 $31,135
Ochiltree, TX R+44 11.8 49.1 0.4 1.0 49.1 62.5 0.0 $23,382
Sterling, TX R+44 1.3 63.6 1.3 2.1 32.4 49.8 0.8 $20,157
Wallace, KS R+43 1.7 90.9 0.3 0.6 7.3 27.5 0.0 $25,429
Grant, NE R+43 0.8 97.2 0.3 0.2 1.4 33.0 0.5 $19,968
Borden, TX R+43 0.7 83.9 0.0 0.3 15.0 100.0 0.0 $50,042
Hansford, TX R+43 6.0 54.7 0.6 1.0 43.6 46.1 0.0 $23,969
Motley, TX R+43 1.3 83.8 2.2 1.0 13.0 82.5 0.0 $18,828
Oldham, TX R+43 1.4 82.4 3.0 0.6 12.1 75.1 0.0 $22,519

 

 

Characteristics of the Most Democratic Leaning U.S. Counties Based on the PVI

 

County PVI Population Density Non-Hispanic White African American Native American Hispanic Evangelical or LDS Other Religion Per Capita Income
Oglala Lakota, SD D+41 7 2.8 0.0 96.0 2.2 19.0 4.7 $8,768
Washington, DC D+40 10,337 35.0 50.5 0.3 9.1 22.8 10.1 $45,290
Bronx, NY D+39 33,173 10.8 36.4 1.4 53.6 12.6 10.9 $18,171
Prince George's, MD D+38 1,816 14.7 64.4 0.5 15.2 47.3 5.6 $32,344
Petersburg, VA D+38 1,402 15.2 78.8 0.3 3.9 40.7 13.3 $18,936
Jefferson, MS D+37 14 13.9 85.4 0.2 0.4 25.9 0.0 $12,723
Baltimore City, MD D+36 7,655 27.9 63.6 0.4 4.3 30.7 10.8 $24,759
Claiborne, MS D+36 20 14.4 84.0 0.1 0.8 23.8 0.0 $12,177
Macon, AL D+35 34 15.2 82.6 0.1 1.1 30.6 4.7 $16,826
Menominee, WI D+35 12 10.7 0.4 87.3 4.4 4.7 0.0 $14,719
San Francisco, CA D+34 17,495 41.9 6.0 0.5 15.1 12.8 14.7 $48,486
Starr, TX D+34 51 4.2 0.1 0.1 95.5 9.6 0.0 $11,584

 


"…strongly Republican leaning counties having much lower population densities than Democratic leaning counties."


 

Looking across the tables, there are several striking differences. The first is the population density3 difference between the two groups, with strongly Republican leaning counties having much lower population densities than Democratic leaning counties. The second factor relates to the very strong differences in the racial and ethnic makeup of the two groups. Strongly Republican leaning counties generally have a much higher percentage of the population that are non-Hispanic whites compared to Democratic counties. Interestingly, the Democratic leaning counties that have comparatively low population densities (under 100 people per square mile) also tend to have a high concentration of another racial or ethnic group (either African American, Native American, or, in the case of Starr County, Texas, Hispanic).

 

There are also important differences between the two groups of counties in their residents' religious beliefs, typically with a much higher share of individuals who belong to a religious congregation that is identified as being either Evangelical or the Church of Jesus Christ of Latter-Day Saints (hereafter designated as LDS) residing in strongly Republican areas.4 In addition, in strongly Democratic counties there is likely to be a high share of individuals in religious congregations that fall into the "other" religion group (largely non-Christian faiths such as Bahá'í, Buddhism, Hinduism, Islam, and Judaism).5

 


"None of the most Republican leaning counties are located in either a West or East Coast state, while only one of the most strongly Democratic leaning counties is in the Great Plains region of the country"


 

One other important factor is in the areas of the country where the different counties are located. None of the most Republican leaning counties are located in either a West or East Coast state, while only one of the most strongly Democratic leaning counties are in the Great Plains region of the country. This suggests that the region of the country itself may play an important role in driving political party preferences.

 

The one potential factor that exhibits the least systematic difference across the two groups of counties is per capita income. What stands out is that per capita income is fairly scattered within in each group. Notably, the three lowest per capita incomes are represented by counties on the Democratic table, but the variability within each group swamps the differences across groups.

 

Based on this analysis, the "typical" highly Republican county is a very rural one located in the center of the country with a largely non-Hispanic white, Evangelical or LDS population. In contrast, the "typical" highly Democratic county is an urban county located in an East or West coast state that has a religiously diverse, and less non-Hispanic white population. While these are the "typical" Republican and Democratic counties, there are certainly plenty of exceptions.

 


"In part 2 of this blog, we'll dig deeper into this by examining the conceptual drivers of political behavior..."


 

In part 2 of this blog, we'll dig deeper into this by examining the conceptual drivers of political behavior and what I am calling the "Three Rs of American Partisanship," namely Race, Religion, and Region.

 

1David Wasserman has recently written an interesting post on the problems associated with geographic sorting for FiveThirtyEight.

2Which, in turn, is based on Census of Population and Housing and the American Community Survey data from the U.S. Census Bureau.

3Measured as individuals per square mile.

4The grouping of Evangelicals with LDS members is not one contained in the 2010 U.S. Religion Census, but can readily be constructed from the available data.

5In the 2010 U.S. Religion Census, LDS members are included in the "other" group, but have been removed from this group for this analysis. There are some remaining Christian oriented faiths in the redefined category, such as Jehovah's Witnesses.

Dan Putler
Chief Scientist

Dr. Dan Putler is the Chief Scientist at Alteryx, where he is responsible for developing and implementing the product road map for predictive analytics. He has over 30 years of experience in developing predictive analytics models for companies and organizations that cover a large number of industry verticals, ranging from the performing arts to B2B financial services. He is co-author of the book, “Customer and Business Analytics: Applied Data Mining for Business Decision Making Using R”, which is published by Chapman and Hall/CRC Press. Prior to joining Alteryx, Dan was a professor of marketing and marketing research at the University of British Columbia's Sauder School of Business and Purdue University’s Krannert School of Management.

Dr. Dan Putler is the Chief Scientist at Alteryx, where he is responsible for developing and implementing the product road map for predictive analytics. He has over 30 years of experience in developing predictive analytics models for companies and organizations that cover a large number of industry verticals, ranging from the performing arts to B2B financial services. He is co-author of the book, “Customer and Business Analytics: Applied Data Mining for Business Decision Making Using R”, which is published by Chapman and Hall/CRC Press. Prior to joining Alteryx, Dan was a professor of marketing and marketing research at the University of British Columbia's Sauder School of Business and Purdue University’s Krannert School of Management.