When Brian Kemp launched his 2018 campaign for governor, his first ad caught everyone’s eye. With promises to “blow up” government spending, a massive truck that he claimed would be used to “round up criminal illegals and take ’em home”, and a gun pointed at his daughter’s boyfriend that “no one [was] taking away”, Kemp’s campaign seemed at odds with many things previously stated by Republican presidents, ranging from George W. Bush’s public appeals to “compassionate conservatism” and self-proclaimed support for DREAMers to his self-proclaimed support for “common-sensical” gun laws. But as a testament to how Donald Trump had redefined Republican policies in both rhetoric and policy, Kemp’s ad catapulted him to prominence and a primary victory and set up his rise to the governor’s mansion.

Trump redefined the politics of conservatism in 2016, shaking up both political parties’ coalitions. This reality begs an interesting question: can one quantify conservatism? Split Ticket’s research on ideology would say yes.

Ideology is a function of demographics and partisanship. This baseline allows us to identify the congressional districts most vulnerable to changes in ideological representation. Districts like CO-03 and VA-05, where mainstream conservative incumbents were replaced by far-right newcomers, are perfect examples of this phenomenon in action.

The Cutting Lines

To understand Trump-era politics one must understand educational polarization. The concept explains differences in belief between those who have college degrees and those who do not. Polarization along educational lines has existed for quite some time, but it has grown starker since Trump’s 2016 victory.

Let’s look at an example. Rural Belmont County in eastern Ohio has a college attainment rate of 16.2% per the Census Bureau. Despite being a Democratic bastion for years, Trump’s appeal to working-class, non-college whites snapped the county rightward — Belmont voted for Trump by 38 points in 2016 and 44 in 2020; Mitt Romney won it by just 8 points in 2012.

The reverse is true in highly-educated areas. In southwestern Connecticut’s Fairfield County, where nearly 50% of the population holds a college degree, Obama beat Romney by 11 in 2012; Clinton won it by 20 and Biden by a whopping 27.

Although these examples underscore the electoral impacts of polarization, they fail to define sweeping changes that have affected political discourse. The education divide has reduced ancestral loyalties of the parochial and regional sort that long persisted in American politics. As local issues fade away, more divisive subjects – like culture war issues – move to the fore.

The second biggest factor in analyzing modern political alignment is the whiteness of an area. In general, white voters vote for more conservative candidates than nonwhite voters; in 2020, Trump won whites by 12 points, while Biden won non-white voters by 50.

It is important to note that this does not mean that all Democratic voters of color are liberal, but rather that minor ideological misgivings are subordinate to a desire for political cohesion. Thus, the representatives of extremely white areas tend to be more conservative than those of extremely nonwhite areas.

Exceptions do exist. Cuban Latinos in southern Florida and the Lumbee Tribe in southeastern North Carolina are well-established pockets of nonwhite conservatism. Overall, though, it can be said that the average white voter will be far more conservative than his or her non-white counterpart.

The final important factor we will consider is raw partisanship. Split Ticket’s focus is on demographically-identical areas that have different electoral preferences. A perfect example is the dichotomy between the neighboring towns of Dracut, Massachusetts and Pelham, New Hampshire.

• Dracut is 88% white and 32% of residents have a college degree
• Pelham is 92% white and 28% of residents have a college degree

As one can see, there is demographic similarity between these Merrimack River towns, but that is where the relationship ends; in 2020, Biden lost Dracut by 1, while he lost Pelham by 20.

Because race and education, the two most predictive factors, are already accounted for in this case, one must conclude that there is ideological self-selection taking place between the two towns. In other words, the liberal subset of the same demographic has aggregated in Dracut, while a more conservative subset has aggregated in Pelham.

The Variables

Our data include congressional district results for the last three Presidential elections, the white percentage of the total population according to the 2015-2019 ACS, and DailyKos’s educational attainment figures by House seat. Because we do not yet have adequate data to analyze new congressional districts by these standards, this analysis will focus on current seats only.

To begin, we outline three auxiliary variables: the nonwhite percent of total population, composite district partisanship from 2016-2020, and the 2012-2020 swing.

• The first variable is a simple remainder, calculated as 1 minus the white population.
• The second is a weighted average of the margin of victory in 2016 and 2020. 75% of the composite relies on 2020 while the remaining 25% is dependent on the 2016 numbers. This accounts for sharp swings among Latinos and potential for regional reversion.
• The last is the difference in margin of victory between the 2012 and 2020 elections. Composite partisanship defines district tilts in the post-Trump era and the swing estimates how receptive a district’s residents are to populist conservatism.

It is important to understand that Trump’s conservatism accelerated racial polarization among white voters. Because there are more non-college educated whites than college-educated whites, Trump often pushed culturally conservative non-college white Democrats into the GOP camp.

Furthermore, we have created a polarization determinant to function as a proxy for racial polarization. This is calculated by subtracting the product of the composite Republican vote share and the white share of the population from the product of the composite Democratic vote share and the nonwhite share of the population.

This quantity is highest in districts like WV-03 and TN-01, both extremely white and Republican. It is lowest in VRA-mandated minority seats such as NY-15 and MD-07, where the reverse is true. In other words, the score rises and falls in tandem with white voting power.

The Methodology

With these variables, our final score is calculated with the following steps:

1. The Trump vote share is shifted by subtracting the product of the district’s 2012-2020 swing and the non-college rate. We assume that lower-education areas are more favorable to Trump and will thus have the damping coefficient remain closer to 1.
2. The value from step 1 is subtracted from 0.5 and subjected to the Cube Root. This means that a Trump vote share lower than 50% will yield a negative result. Rapidly right-trending seats held by Democrats show up as positive, indicating that moderate Democrats could be replaced by conservative Republicans.
3. The values from step 1 and step 2 are added together, to get our final value, which we will term the Q-score, which ultimately helps us quantify modern conservatism.

Examining the Results

The full map of Q scores is printed below:

For qualitative reasons, we have categorized the districts into 6 distinct categories with the score thresholds listed below:

• No Q Risk (negative Q-score)

• Minimal Q (Q-score in between 0 and 0.5)

• Low Q (Q-score in between 0.5 and 1)

• Medium Q (Q-score in between 1 and 1.25)

• High Q (Q-score in between 1.25 and 1.5)

• Extreme Q (Q-score over 1.5)

There are 14 current districts in the Extreme Q category, and they are displayed below:

This map is unsurprising because it features some of the reddest congressional turf in the country. The 14 districts above had some of the largest rightward swings between 2012 and 2020 (i.e. Ohio-06). With new coalitions in mind, Split Ticket would expect the replacement-level congressional candidate in such a district to be the most conservative candidate.

Many of these seats, including Missouri’s 8th and Oklahoma’s 2nd, used to be bellwether districts for Democrats, but have since become utterly inhospitable.

Political junkies will certainly note that most of the districts in the extreme Q category have ancestrally-Democratic electorates. These voters are primarily concentrated in the rural Midwest and South, but can be found elsewhere. The one thing that unifies these geographically-disparate voters is their recent propensity to side with Republicans.

Why are ancestral Democrats fleeing the party? Logic would point to the national party’s increasingly-liberal cultural values. Historically, economic fault lines allowed ideologically-diverse sets of Democrats to work together. Changing political tides encapsulated by Trump’s first campaign have since given conservative Democrats a populist new home within the GOP.

This is not to say, however, that every seat influenced by Trump is ancestrally Democratic. Many historically-Republican districts also clock in at the top of our list. Already extremely Republican in 2012, GOP tallies in districts like these actually increased in the ensuing two presidential races. Districts such as Ohio’s 4th, Kentucky’s 5th, and Pennsylvania’s 15th all fit into this category.

With the standards out of the way, we can now examine primary implications: which districts are most vulnerable to rapid ideological turnover?

Conclusion

Split Ticket’s numbers show where potential members of Congress could be seen as “insufficiently conservative” and vulnerable to right-wing primary challengers (with the appropriate redistricting caveats).

Alabama’s 4th and Kentucky’s 5th are the first seats in this category that come to mind. Both seats gave Trump more than 80% of the vote and are represented by veteran incumbents. Robert Aderholt (R-Haleyville) first won his seat in 1996, and Hal Rogers has held his ancestrally Republican district since the 1980s.

The 4th and 5th are both ultra-white and ultra-low-education districts historically producing quiet, workhorse politicians. Aderholt and Rogers fit this low-profile Republican brand perfectly. Both members are extremely conservative, but focus on pork-barrel spending and constituent services more than bombastically attracting media attention.

But in this age of supercharged conservatism, a successor to Aderholt or Rogers may espouse more sanguine rhetoric. The same goes for Bill Johnson of Ohio’s 6th and Adrian Smith of Nebraska’s 3rd. Both of those members are quiet, conservative politicians focused more on work than bomb-throwing. As for KY-05 and AL-04, though, Split Ticket’s Q scores suggest that future representatives of OH-06 and NE-03 could be even more conservative.

Changes in representation would be triggered either by an incumbent retirement or an outright primary challenge. It is normally difficult to dislodge an untarnished sitting congressman, but it becomes easier when an incumbent is “invisible”.

The best example of this was the 2020 Tipton vs. Boebert primary in CO-03; Tipton was an influential chair of Trump’s Colorado campaign and boasted a solidly conservative voting record, yet he was beaten because Boebert’s style was more in tune with modern conservatism within the GOP.

The absorption of millions of populist-conservative ancestral Democrats into the GOP coalition should continue move GOP rhetoric rightward. To use our earlier example, imagine a party more in tune with Belmont County than Fairfield. For better or for worse, American conservatism is fundamentally moving in a new direction.

A full list of Q scores for districts is available here.