The tendency and tumble of split ticket voting in NC

I recently joined WUNC's Due South to talk about ticket-splitting voters, especially in North Carolina. The dynamic of ticket-splitting is often associated with how presidential and congressional candidates do within a district: we may hear of a Biden-Republican congressional representative district, or a Trump-Democratic district as a sign of the voters willing to divide their votes for different parties on the same ballot. We may also hear about a 'split' in how a state votes for president (for one party) and a U.S. Senate seat (for the other party).

But with North Carolina being one of eleven states that holds a gubernatorial election in a presidential election year, those of us who study NC politics have a natural experiment that lends itself to studying the impact of split-ticket, or the opposite dynamic of straight-ticket, voting on a state-wide scale.

As an example of this bi-polar partisan behavior, Greene County demonstrates what North Carolina experienced at the start of the 21st Century and the changes leading up to this year. In 2000's election, Greene County gave Republican presidential candidate George W. Bush 57.5 percent of its vote, but immediately below on the ballot, Republican gubernatorial candidate Richard Vinroot got only 40.4 percent of the vote--a difference of 17 percentage points, making it one of seven counties with a 17 point or greater difference between the GOP presidential and gubernatorial two-party vote percentages.

The largest difference, at 19 percentage points, were Tyrrell and Columbus counties, followed by Johnston and Nash at an 18 point difference and Bladen and Camden, along with Greene, at 17. Overall, the mean difference between Bush and Vinroot across the state was 11 percentage points. 

This shows, in pretty clear form, the difference in voting at the presidential and gubernatorial level, with North Carolina voters in 2000 willing to split their tickets between a Republican presidential candidate and a Democratic gubernatorial candidate. In that election, Bush won at 56 percent state-wide, with Democrat Mike Easley winning 53 percent.

Four years later, incumbents Bush and Easley easily won re-election in North Carolina, with both claiming a 12 point victory in their respective contests in 2004. A large swath of Tar Heel voters were very comfortable voting overwhelming for a Republican for president and turning around and voting Democratic for governor--thus, a huge bloc of ticket-splitters. 

However, in plotting how a county voted at the two-party presidential level versus the gubernatorial two-party vote, a strong relationship is demonstrated in 2000 and 2004. To demonstrate, all of NC's counties are identified in Scatter Plot 1, which gives Bush's two-party county vote percentage (on the horizontal, or x, axis) against Vinroot's two-party vote (on the vertical, or y axis) in 2000.

Scatter Plot 1: Bush 2000 Presidential Two-Party Vote to Vinroot 2000 Gubernatorial Two-Party Vote by NC Counties

For a clearer sense of the dynamics, here's the same scatter plot with only the top seven counties with the largest percentage difference indicated (note Greene County circled in both scatter plots).

Scatter Plot 2: Top Seven NC Counties with largest GOP pres-gov difference between Bush & Vinroot 2000 two-party vote percentages

While the relationship is at 0.909 (meaning, Bush's two-party vote in a county explains nearly 91 percent of Vinroot's two-party vote, which is a pretty good measure), there is some healthy 'distance' between counties and the solid line (or line of fit). 

Scatter Plot 3 demonstrates 2004's election dynamics, again with Bush's two-party county vote on the x axis to Republican gubernatorial candidate Patrick Ballentine's two-party county vote on the y axis. 

Scatter Plot 3: Bush 2004 Presidential Two-Party Vote to Ballentine 2004 Gubernatorial Two-Party Vote by NC Counties

In 2004, Greene County gave Bush 59 percent of their two-party vote, while Ballentine got 43 percent, thus a 16 percentage point difference between the Republican presidential and gubernatorial candidates.

In 2008, however, something dramatic happened that brought what I contend is a new political dynamic to the state's politics: Barack Obama turned a 12-point Republican presidential state into a half-a-percentage point Democratic presidential state, the first time since 1976 that North Carolina voted for a Democratic president. 

More notably, the relationship between the Republican presidential and gubernatorial two-party vote percentage dropped from its previous 90 percent 'explanation' down to 73 percent between John McCain's presidential performance and Pat McCrory's gubernatorial performance, as demonstrated in Scatter Plot 4. 

Scatter Plot 4: McCain 2008 Presidential Two-Party Vote to McCrory 2008 Gubernatorial Two-Party Vote by NC Counties

For ease of viewing the overall trend, here's a scatter plot without the counties identified, except for 2000's top seven (and Greene County noted):

Scatter Plot 5: Top Seven NC Counties with largest GOP pres-gov difference in 2000 and McCain & McCrory 2008 two-party vote percentages

In scatter plot 5, the widening of the counties out from the line is quite evident, and thus the lower level of 'explanation' (as reflected in 0.73). Greene County gave 53 percent of its vote to McCain, a drop from 2004, but only 37 percent to McCrory, another sixteen percentage point difference between the two GOP candidates. 

However, 2008's election wasn't just notable for Obama's win of the state, but the effect it, along with 2010's Tea Party Republicanism and the mid-term election, would have on the state. The dynamics of nationalization of American politics has a real impact on North Carolina's voting patterns, and is evident in the 2012 and subsequent elections: meaning, the 'difference' between a county's vote for GOP presidential and gubernatorial candidates decreased--thus voters were picking one party for both slots, and not splitting their tickets.

Starting with 2012's election, we see a growing alignment between the GOP presidential two-party vote and the two-party gubernatorial vote, as evident in Scatter Plot 6.

Scatter Plot 6: Romney 2012 Presidential Two-Party Vote to McCrory 2012 Gubernatorial Two-Party Vote by NC Counties

Notice how the counties are now 'closer' to the line, and the explanation jumped from 0.73 to 0.94--greater than in the 2000 or 2004 elections. And notably, the seven counties that had the greatest difference in 2000 now moved markedly closer to the line, meaning their presidential and gubernatorial GOP votes are coming into alignment. 

Scatter Plot 7: Top Seven NC Counties with largest GOP pres-gov difference in 2000 and Romney & McCrory 2012 two-party vote percentages

This is an indication of the voters aligning their preferences between the two offices, and not being party defectors, or ticket splitters, at a lower rate. In fact, in this instance, the mean difference was McCrory +2, meaning the Republican gubernatorial candidate performed better than the presidential candidate in the counties.

With 2016's election, we see the alignment grow even greater in strength and visualization, as evident in Scatter Plot 8:

Scatter Plot 8: Trump 2016 Presidential Two-Party Vote to McCrory 2016 Gubernatorial Two-Party Vote by NC Counties

With a level of 0.972 'explanation' between Trump's two-party vote performance and McCrory's, and a mean difference of a little under 3 percentage points (this time to Trump's favor), North Carolina voters were clearly signaling their party loyalty, though there were enough Tar Heel voters who were willing to split their tickets to result in a Republican Trump-Democratic Cooper victory (to the tune of a little over 10,000 votes).

Without the county identifiers (save for the original seven with the greatest difference in 2000), the relationship is very clear between how North Carolina voters in counties were picking their GOP presidential and gubernatorial candidates. In fact, Greene County found itself almost on the line of fit.

Scatter Plot 9: Top Seven NC Counties with largest GOP pres-gov difference in 2000 and Trump & McCrory 2016 two-party vote percentages

Finally, with the most recent presidential & gubernatorial election, the straight-ticket voting was even greater than in 2016, as evident in Scatter Plot 9:

Scatter Plot 10: Trump 2020 Presidential Two-Party Vote to Forest 2020 Gubernatorial Two-Party Vote by NC Counties

I would encourage the reader to take a moment and scroll back to Scatter Plot 1, the 2000 election and compare it to Scatter Plot 10 of 2020's election. 

Note the level of explanation goes from nearly 91 percent to 99.3 percent--a nearly perfect relationship. And the mean difference between Trump and Republican gubernatorial candidate Dan Forest was 3 percentage points, enough to give another Trump-Cooper result.

Scatter Plot 11 gives the clearest sense of how the counties have, for the most part, aligned themselves between their Republican presidential and gubernatorial candidates:

Scatter Plot 9: Top Seven NC Counties with largest GOP pres-gov difference in 2000 and Trump & Forest 2020 two-party vote percentages

In fact, the seven counties with the largest differences in 2000 have now all 'come home' to rest on, or very near, the line of fit. Meaning, how they voted at the presidential level is a pretty good indicator of how they voted at the gubernatorial level: for example, Greene County cast 56 percent of its vote for Trump in 2020, and 54 percent for Republican gubernatorial candidate Dan Forest. North Carolina's overall county difference is down to 3 points for 2020.

To see the trend over time, an animated gif of the 2000 to 2020 scatter plots shows the post-2008 alignment between the two office contests over time.

GIF of NC's 100 Counties and GOP Two-Party Presidential-Gubernatorial Vote

Over the past two decades, North Carolina has undergone a sea-change in its politics: one from where it was commonplace for voters to electorally swing from supporting Republican presidential candidates to Democratic gubernatorial candidates. And yes, that habit of split ticket voting is still evident in the Old North State, but the 'difference' in the swing has been substantially reduced. 

So what might we expect this November as it relates to the dynamic of split-ticket voters versus straight-party voters? 

My suspicion is North Carolina will again see robust alignment in its voting patterns, from the top down. The real question is, will the Old North State continue to see a sliver of the electorate--three to four percent--willing to vote Trump-Stein, or (really unlikely in my estimation) Biden-Robinson?

Or with such clear partisan and policy differences between the candidates for the top two offices (Trump and Robinson versus Biden and Stein), could North Carolina simply see the ultimate demise of split-ticket voting? That North Carolina voters--all of them, and not just 96-97 percent of them--may finally 'pick a side' of the political aisle and we have a near perfect 'straight ticket' voting pattern? And as an ultimate result, the ballot ramifications could indicate a clearly sorted political Old North State.

Time, and the voters, will tell us in November.

Dr. Michael Bitzer holds the Leonard Chair of Political Science at Catawba College, where he is a professor of politics and history and author of Redistricting and Gerrymandering in North Carolina: Battle Lines in the Tar Heel State.