What Happened in North Carolina

Adam Martin
The Book Aisle
Published in
19 min readFeb 24, 2021


This is the eleventh (and second to last) part in my series analyzing various swing states from the 2020 presidential election. In this part we turn to North Carolina, the only other state aside from Florida that I incorrectly predicted. While there are some similarities with Florida in terms of why I got it wrong, North Carolina had other factors going on as well which will be explored in this article.

With that said, let’s dive in.

Big Picture

Joe Biden- 2,684,292 (48.59 percent, up 2.42 from Clinton in 2016)

Donald Trump- 2,758,775 (49.93 percent, up 0.10 from 2016)

FiveThirtyEight Projection- 50.5 for Biden (up 1.91), 48.8 for Trump (down 1.13)

RCP Average- 47.6 for Biden (down 0.99), 47.8 for Trump (down 2.13)

In 2008, Barack Obama won North Carolina by about 14,000 votes, making him the first Democrat to carry the state since Jimmy Carter. While he performed relatively poorly in the western part of the state as well as other historically Democratic counties, Obama forged his victory by running up huge majorities in the major cities, such as Charlotte, Raleigh, Greensboro, Winston-Salem, and the Research Triangle. At the time, this victory was seen as a turning point in the Democratic coalition. Considering the state’s fast-growing urban base and the Obama’s widespread success with metropolitan and college-educated voters, it was believed that North Carolina would be a new breeding ground for Democratic victories in the years to come.

But despite continued successes in states that have followed a similar demographic trajectory (such as Colorado and Virginia), North Carolina has become a source of continual disappointment to Democrats. Since 2008, no Democrat has captured North Carolina in a presidential or Senate race; only Roy Cooper has found success at the state level, winning the governor’s mansion in 2016 and 2020. A critical reason for this is that while Democrats have consolidated their support in the large, urban counties, these gains have been offset by consolidated Republican support in rural, more heavily white counties that Democrats were traditionally more competitive in. Indeed, while Biden made the state closer than Clinton did in 2020, President Trump’s advantage outside the major cities was a key reason why he still prevailed statewide.

As of Election Day, there have been 280,376 COVID cases in North Carolina, the seventh highest among the states. This makes sense, given that North Carolina is the ninth largest state by population; however, the incidence rate is only 2.7 percent, ranking 28th among the states. On the other hand, there have been 4,457 deaths, ranking 15th among the states, but the fatality rate of 1.6 percent ranks just 33rd among the states. And economically, the impact has been fairly consistent with the rest of the country, with the unemployment rate going from 4.3 percent in March to 12.5 percent in April (a 8.2-point increase that ranks 29th among the states). So overall, while the impact of the pandemic in North Carolina may seem large based on raw numbers, it’s actually more modest relative to its population.

As of Election Day, there have been 502 BLM protests in North Carolina, the eighth most among the states. While the state includes large swaths of rural area, the state has multiple urban centers that are ripe for protest activity, such as Charlotte, the Research Triangle, and numerous medium-sized cities. Furthermore, North Carolina has the sixth largest black population in the country by raw numbers (and the ninth highest by percentage of the total population), the tenth most college graduates (but only 26th by percentage), and the ninth largest young adult population (but 22nd by percentage).

As for the exit polls, Biden won 92 percent of black voters (up from 89 percent in 2016), as well as 57 percent of Hispanic voters (matching Clinton’s performance). On the other hand, Biden only won 33 percent of white voters (just a point better than Clinton in 2016). Meanwhile, President Trump ran away with 66 percent of white voters (up from 63 percent four years ago), which allowed him to carry a majority of the electorate.

As for educational attainment, Biden won 59 percent of college graduates and 50 percent of white college graduates, both of which are significant improvements from Clinton’s 2016 performance and a reflection of his success in the Research Triangle. But ultimately, Trump carried 55 percent of voters without a college degree as well as 78 percent of white voters without a college degree, both of which are largely consistent with his 2016 performance. Overall, this shows a considerably wider education gap in North Carolina compared to four years ago, both at the aggregate level as well as broken down by race.

As for age, there’s evidence that Biden consolidated support with younger voters while containing Trump’s advantage with older voters. Biden won 57 percent of voters in the 18–29 bracket (matching Clinton’s performance) and winning 56 percent of voters in the 30–44 bracket (up from 49 percent for Clinton). And although Trump still won 52 percent of voters in the 45–64 bracket and 59 percent of voters over 65, these margins are slightly smaller than what he received in 2016. To that end, Biden made some improvements.

Geographically, Biden won 69 percent of urban voters, a 9-point improvement from Clinton. While Biden did make slight improvements with suburban voters, President Trump still won this group with 60 percent, which matches his own 2016 performance. Trump also won 59 percent of rural voters, one percent better that he received four years ago. So overall, this election exacerbated North Carolina’s urban-rural divide. While Trump kept his margins from 2016 in the suburban and rural areas, Biden significantly consolidated support in the urban areas.

Socioeconomically, 45 percent of North Carolina voters said their family’s financial situation is better than what it was four years ago (compared to 41 percent nationwide), of which 82 percent voted for President Trump. Meanwhile, 20 percent said their family’s financial situation is worse than what it was four years ago (the same as the rest of the country), of which 91 percent voted for Joe Biden. And 34 percent said their family’s financial situation is the same as what it was four years (compared to 39 percent nationwide), of which 69 percent voted for Biden. Regarding income brackets, Biden won 53 percent of voters making less than $50,000; however, him and Trump tied voters in the $50,000-$99,999 bracket and Trump won 52 percent of voters making over $100,000.

When asked about their most important issue, 35 percent of North Carolina voters said the economy (compared to 35 percent nationwide), of which 82 percent voted for President Trump. Meanwhile, 14 percent said the pandemic was the most important issue and 21 percent said racial inequality, of which 84 percent and 94 percent voted for Biden respectively. And when asked what was more important to accomplish, 52 percent prioritized containing the virus, of which 78 percent voted for Biden. On the other hand, 43 percent prioritized reopening the economy, of which 80 percent voted for Trump.

Now that we’ve gone over the exit polls, let’s run some regressions. First, here are various regression results for Biden’s county-level vote share.

For the most part, nothing is out of the ordinary for North Carolina. While there’s some evidence that the COVID incidence rate is negatively associated with Biden’s vote share, this relationship is not consistent across all the regression models. Aside from this, none of the main variables attributed to the pandemic’s effects are statistically significant with Biden’s vote share in North Carolina. But similar to the rest of the states covered in this series, there is no effect of the number of protests at the county level on Biden’s vote share.

Aside from this, nothing too exceptional emerges with demographic factors. Most particularly, the education gap remains persistent in North Carolina, with the more highly educated counties voting more for Biden and the less highly educated counties siding more with President Trump.

Now, let’s look at the regression models for the Democratic vote share change between 2016 and 2020.

For the most part, these regressions reinforce the trends seen in Figure 1, with the education gap widening even further from its 2016 level. Aside from this, however, nothing too extraordinary emerges from these regression results.

So with all this established, let’s dive into each of the subsections for North Carolina.

Solidly Democratic Counties

Anson- Biden Hold

Biden- 5,789 (51.7 percent, down 3.85), Trump- 5,321 (47.53 percent, up 4.80)

Bertie- Biden Hold

Biden- 5,939 (60.51 percent, down 1.31), Trump- 3,817 (38.89 percent, up 1.92)

Cumberland- Biden Hold

Biden- 84,469 (57.40 percent, up 1.25), Trump- 60,032 (40.80 percent, up 0.59)

Durham- Biden Hold

Biden- 144,688 (80.42 percent, up 2.75), Trump- 32,459 (18.04 percent, down 0.11)

Edgecombe- Biden Hold

Biden- 16,089 (63.15 percent, down 2.04), Trump- 9,206 (36.13 percent, up 2.94)

Guilford- Biden Hold

Biden- 173,086 (60.84 percent, up 2.86), Trump- 107,294 (37.72 percent, down 0.38)

Halifax- Biden Hold

Biden- 15,545 (60.35 percent, down 2.23), Trump- 10,080 (39.13 percent, up 3.25)

Hertford- Biden Hold

Biden- 7,097 (66.74 percent, down 1.10), Trump- 3,479 (32.72 percent, up 2.29)

Mecklenburg- Biden Hold

Biden- 378,107 (66.68 percent, up 4.39), Trump- 179,211 (31.60 percent, down 1.29)

Northampton- Biden Hold

Biden- 6,069 (60.03 percent, down 2.36), Trump- 3,989 (39.46 percent, up 3.08)

Orange- Biden Hold

Biden- 63,594 (74.82 percent, up 2.04), Trump- 20,176 (23.74 percent, up 1.20)

Vance- Biden Hold

Biden- 12,431 (59.20 percent, down 2.02), Trump- 8,391 (39.96 percent, up 3.26)

Wake- Biden Hold

Biden- 393,336 (62.25 percent, up 4.88), Trump- 226,197 (35.80 percent, down 1.36)

Warren- Biden Hold

Biden- 6,400 (62.18 percent, down 2.99), Trump- 3,752 (36.45 percent, up 3.81)

Washington- Biden Hold

Biden- 3,396 (54.73 percent, down 2.20), Trump- 2,781 (44.82 percent, up 3.22)

First, there are the fifteen solidly Democratic counties in North Carolina. In 2016, Clinton won this subsection with 61.6 percent of the vote, which was an improvement from the 61.8 percent President Obama won in 2012 and a net gain of 62,678 votes. While this wasn’t enough for Clinton to win statewide, it was a clear affirmation that the state’s Democratic backbone held strong. This time around, Biden continued to expand his support in this subsection, winning all fifteen counties with almost 65 percent of the vote and a net gain of 238,142 votes over Clinton’s total.

Most of these gains were made in Charlotte and the Research Triangle, the subsection’s main population centers. Biden received more than 60 percent of the vote in both Wake and Guilford counties while making gains in Mecklenburg as well. Even in Durham, where Clinton got more than 75 percent of the vote, Biden nudged this county above 80 percent. These population centers are filled with voters who are college educated, non-white, and metropolitan, all of which were likely to vote for Biden over Trump. And to this end, Biden was incredibly successful in consolidating support in these areas.

On the other hand, Biden actually underperformed Clinton in nine of the counties, most of which are more rural. Granted, these counties are fairly small, with the largest (Edgecombe) having a population just over 50,000. Interestingly, these counties are more predominantly black than the larger, urban counties; however, they are considerably poorer and have fewer college graduates. And while the losses were not as substantial as those sustained in other states, two counties (Anson and Washington) actually slid out of the solidly Democratic column and into the swing column. This minor development demonstrates how much of Biden’s gains in this election were centered in the suburbs and major population centers of the country.

As of Election Day, there have been 97,901 COVID cases in this subsection, setting the incidence rate at 2.6 percent (about the same as the state average). About a third of these cases came from Mecklenburg while a decent chunk also came from Wake. On the other hand, there have been 1,389 deaths in this subsection, setting the fatality rate at 1.4 percent (slightly below the state average). Regarding timing, there was a massive spike in cases from the early months to the summer months. And while the cases in the early fall months were not as high, they were still considerably higher than those from the spring. And for unemployment, the impact was similar to that of the rest of the state, going from 4.1 percent in March to 12.1 percent in April. Some counties experienced fairly mild increases, particularly the rural counties in the northern part of the state. Other counties took more substantial hits, particularly the larger, more urban counties where Biden picked up ground. While this subsection has largely recovered in the months since then, the unemployment rate was still at 7.3 percent as of September.

Finally, as of Election Day, there have been 219 BLM protests in this subsection, with 65 occurring in Wake, 61 in Mecklenburg, 33 in Durham, and 29 in Guilford. In fact, all of the protests that happened in this subsection were in counties where Biden improved upon Clinton’s 2016 performance, which were disproportionately in the Charlotte area or the Research Triangle. Interestingly, these counties were not as predominantly black as the rural counties in the subsection, where Biden underperformed Clinton. Still, these more urban counties have considerable pockets of black residents and other racial and ethnic minorities. On top of that, these counties where protests were prevalent also have large young adult populations and large college graduate populations, which make sense given the immense number of universities between Charlotte and the Research Triangle. Overall, the favorable demographics in this subsection (specifically the counties where Biden made improvements) explains the immense protest activity.

Swing Counties

Buncombe- Biden Hold

Biden- 96,515 (59.74 percent, up 5.43), Trump- 62,412 (38.63 percent, down 1.47)

Chatham- Biden Hold

Biden- 26,787 (55.12 percent, up 2.26), Trump- 21,186 (43.59 percent, up 0.67)

Forsyth- Biden Hold

Biden- 113,033 (56.16 percent, up 3.18), Trump- 85,064 (42.26 percent, down 3.44)

Granville- Trump Hold

Biden- 14,565 (46.09 percent, down 1.10), Trump- 16,647 (52.68 percent, up 3.00)

Hoke- Biden Hold

Biden- 11,804 (54.55 percent, up 1.07), Trump- 9,453 (43.69 percent, up 1.02)

Lenoir- Trump Hold

Biden- 13,605 (47.89 percent, up 1.44), Trump- 14,590 (51.36 percent, down 0.05)

Martin- Trump Hold

Biden- 5,911 (47.14 percent, down 1.73), Trump- 6,532 (52.09 percent, up 2.80)

Nash- Biden Flip

Biden- 25,947 (49.64 percent, up 0.89), Trump- 25,827 (49.41 percent, up 0.49)

New Hanover- Biden Flip

Biden- 66,138 (50.17 percent, up 4.61), Trump- 63,331 (48.04 percent, down 1.42)

Pasquotank- Biden Hold

Biden- 9,832 (49.41 percent, down 0.12), Trump- 9,770 (49.10 percent, up 2.07)

Pitt- Biden Hold

Biden- 47,252 (53.96 percent, up 2.02), Trump- 38,982 (44.51 percent, up 0.19)

Robeson- Trump Hold

Biden- 19,020 (40.31 percent, down 6.23), Trump- 27,806 (58.93 percent, up 8.11)

Scotland- Trump Flip

Biden- 7,186 (48.64 percent, down 3.91), Trump- 7,473 (50.58 percent, up 5.66)

Watauga- Biden Hold

Biden- 17,122 (53.14 percent, up 5.99), Trump- 14,451 (44.85 percent, down 0.83)

Wilson- Biden Hold

Biden- 20,754 (50.95 percent, down 0.61), Trump- 19,581 (48.07 percent, up 2.10)

Next, there are fifteen swing counties in North Carolina. For the most part, this subsection is a mixture of rural area and medium-sized cities. In 2016, Clinton won nine of these counties with 50.7 percent of the vote, a step down from President Obama, who won twelve of these counties in 2012 with 52.6 percent of the vote, and a net loss of 1,673 votes. While this isn’t bad, it wasn’t particularly great either.

Now in 2020, Biden won ten of these counties with 53.2 percent of the vote, a slight improvement from Clinton and a net gain of 79,444 votes. While Biden did allow Trump to flip Scotland (a fairly rural county), he made up for it by flipping both Nash and New Hanover (both of which contain medium-sized cities). Nash was already tightly contested in 2016, meaning that Biden only needed a slight improvement in order to flip it; however, capturing New Hanover was more significant. Biden became the first Democrat since Jimmy Carter to carry New Hanover which, despite containing the medium-sized city of Wilmington, is one of the less racially diverse counties in the subsection. Furthermore, Biden also picked up ground in nine counties, many of which are either in the medium-sized cities. He also converted three counties (Buncombe, Chatham, and Forsyth) into solidly Democratic counties. To that end, Biden’s performance was an effort to consolidate support in areas where he was likely to do well.

On the other hand, President Trump picked up ground in other key areas in this subsection. As mentioned above, he flipped Scotland, making the first Republican to carry the county since Ronald Reagan. Trump also expanded his margin of victory in Robeson, to the point that he converted it into a solidly Republican county. Given the large black and Native American populations in both of these counties, Trump’s success in this region is pretty remarkable. Specifically speaking, the region is home to the Lumbee, the largest Native American tribe east of the Mississippi River, which has rallied around President Trump. While Robeson has traditionally voted Democrat, many of its voters are socially conservative and working class. The region had been hit economically in recent years with shuttered factories and loss of manufacturing jobs, which many residents blamed on NAFTA and other international trade agreements. These factors presented Trump opportunities to capitalize, which he has now done for two elections. To this end, it demonstrates Trump’s continued success in rural, working class counties, including those with significant nonwhite populations.

As of Election Day, there have been 50,680 COVID cases in this subsection, setting the incidence rate at 2.9 percent (above the state average). While there is some variation between the counties, there aren’t any significant outliers in this subsection. On the other hand, there have been 824 deaths in this subsection as of Election Day, setting the fatality rate at 1.6 percent (slightly above the state average). Regarding timing, there was also a sharp rise in cases from the spring months to the summer months. And while the cases in the early fall were slightly smaller than during the summer months, the load was still well above that of the spring. And for unemployment, the impact of the pandemic was similar to the rest of the state, going from 4.3 percent in March to 12.7 percent in April. To this end, there was more variation, with some counties only experiencing minor changes (such as Chatham, Lenoir, and Martin), while other counties experienced more seismic increases (such as Buncombe and New Hanover). And while this subsection has mostly recovered in the months since April, the unemployment rate was still at 7.2 percent as of September.

Finally, as of Election Day, there have been 126 BLM protests in this subsection, with 40 in Forsyth, 32 in Buncombe, and 21 in New Hanover. This makes sense, given the subsection’s mixture of population and demographics. On one hand, there are considerable pockets of black and nonwhite residents; however, similar to Robeson, there are plenty of racially diverse counties that are less amenable to protest activity. To that end, the subsection provides its infrastructure for protest activity through its relatively larger cities, which contain its fair share of college graduates and young adults.

Flipped Counties

Bladen- Trump Hold

Biden- 7,326 (42.78 percent, down 1.60), Trump- 9,676 (56.50 percent, up 2.70)

Gates- Trump Hold

Biden- 2,546 (42.64 percent, down 1.59), Trump- 3,367 (56.39 percent, up 3.09)

Granville- Trump Hold

Biden- 14,565 (46.09 percent, down 1.10), Trump- 16,647 (52.68 percent, up 3.00)

Martin- Trump Hold

Biden- 5,911 (47.14 percent, down 1.73), Trump- 6,532 (52.09 percent, up 2.80)

Nash- Biden Flip

Biden- 25,947 (49.64 percent, up 0.89), Trump- 25,827 (49.41 percent, up 0.49)

Richmond- Trump Hold

Biden- 8,754 (42.16 percent, down 1.82), Trump- 11,830 (56.98 percent, up 3.26)

Robeson- Trump Hold

Biden- 19,020 (40.31 percent, down 6.23), Trump- 27,806 (58.93 percent, up 8.11)

Watauga- Biden Hold

Biden- 17,122 (53.14 percent, up 5.99), Trump- 14,451 (44.85 percent, down 0.83)

Next, there are eight counties that flipped from 2012 to 2016, seven of which are Obama-Trump counties and one (Watauga) being a Romney-Clinton county. In 2012, President Obama won the Obama-Trump counties with 52.7 percent of the vote while also getting about 47 percent in Watauga. Four years later, Clinton lost all seven Obama-Trump counties with just 46.9 percent of the vote and a net loss of 11,321 votes. On the other hand, she edged out Trump in Watauga with 47.2 percent of the vote and a net gain of 1,136 votes over President Obama.

This time around marked a widening of the divergence with these two types of counties. While Biden posted a net gain of 5,156 votes in the Obama-Trump counties and flipped one of them back (Nash), he actually lost ground from Clinton, only receiving 44.8 percent of the vote. This means that collectively, the Obama-Trump counties have slid into the solidly Republican column. On average, Biden lost about 1.9 percentage points in each of these counties. While only four of these counties are classified as solidly Republican following the 2020 election, three of them were already in that column following the 2016 election. Robeson, which overlapped with the swing counties, was a major contributor to Biden’s overall underperformance in this subsection, as he received over 6 percentage points less than Clinton did four years ago.

On the other hand, he did win back Nash after Clinton lost it in 2016 (albeit narrowly), plus Biden gained ground in Watauga, receiving a 6-point boost from Clinton. Considering that both these counties contain medium-sized cities, this isn’t too surprising. But on the whole, the story of this subsection is one of continued concentrated support for Trump in rural areas.

As of Election Day, there have been 15,444 COVID cases in the Obama-Trump counties and 1,447 cases in Watauga, setting their incidence rates at 3.9 percent and 2.6 percent respectively. For the most part, the Obama-Trump counties remain fairly close to the aggregate figure, with no major outliers. On the other hand, there have been 304 deaths in the Obama-Trump counties and 6 deaths in Watauga, setting their fatality rates at about 2 percent and 0.4 percent respectively. Regarding timing, the Obama-Trump counties saw a massive spike between the spring months and the summer months, and leveling off in the early fall months. As for Watagua, there was a more steady increase from the spring to the summer to the early fall. And for unemployment, the Obama-Trump counties went from 5.2 percent in March to 10.4 percent in April while Watauga went from 3.8 percent in March to 11.2 percent in April. Both of these increases are milder than the rest of the state. And while they have both largely recovered in the months since then, their unemployment rates remained at 7.9 percent in the Obama-Trump counties and 4.6 percent in Watauga as of September.

Finally, as of Election Day, there have only been 9 BLM protests in the Obama-Trump counties and 4 in Watauga. On one hand, this may appear surprising, given how racially diverse the Obama-Trump counties are, while Watauga is the least diverse in this grouping of counties. But on the other hand, Watauga is the most highly educated county in this grouping, while the Obama-Trump counties have far fewer college graduates, which can explain the lack of protest activity within the subsection.


As with the other states, there are a few reclassifications for counties based on how their voting behavior changed from 2016 to 2020. These changes are reflected below.

Alamance (Solidly Republican to Swing)

Anson (Solidly Democratic to Swing)

Buncombe (Swing to Solidly Democratic)

Chatham (Swing to Solidly Democratic)

Forsyth (Swing to Solidly Democratic)

Nash (Trump to Biden)

New Hanover (Trump to Biden)

Robeson (Swing to Solidly Republican)

Scotland (Clinton to Trump)

Washington (Solidly Democratic to Swing)

In total, ten counties were reclassified following this election. Not only are they are spread out fairly evenly throughout the state, but they’re also spread out in terms of direction. Six counties shifted to the left of their 2016 classification while four counties shifted to the right. For the most part, the counties that shifted to the right are in the southern part of the state (with the exception of Washington), are rural, and tend to have noticeably large black populations; however, they also tend to be poorer than the rest of the state. It’s possible that these poorer counties expressed stronger resistance to public health restrictions given their dire economic repercussions. But as we saw in Robeson, there was also a sense among many residents in these counties that the Democratic coalition took them for granted and advocated policies that conflicted with their views on social and economic issues. This presented an opportunity for President Trump to infiltrate these areas in a similar fashion to the predominantly Hispanic counties in Texas and Florida.

Aside from this, however, there was also evidence of Democrats consolidating support in more urban counties. Three of the counties that shifted to the left of their 2016 classification are in or near the Research Triangle, which is a significant source of jobs for the region and has been the source of population growth in recent years. Other counties, such as Buncombe, Nash, and New Hanover, contain medium-sized cities that share similar characteristics to the larger cities that Democrats have won over in recent years. To that end, these areas have also shifted in favor of Biden.


Overall, North Carolina has become increasingly defined by its urban-rural divide. Despite receiving almost half of the state’s votes, Joe Biden only won 25 of its 100 counties. Many of those 25 are among the state’s most populous and offer large batches of votes, but this increased success in the cities and suburban communities has come at the expense of the vast swaths of rural counties, including those that are racially and ethnically diverse. While one can argue this has been a key component in the election overall, it proved especially costly in a state like North Carolina, where the rural counties still act as an effective counterbalance against the major urban centers.

None of this is to say that Biden performed especially poorly in North Carolina. His improvements in the state’s Democratic backbone (Charlotte and the Research Triangle) allowed him to narrow up the statewide margin to under 75,000 votes (in a state where more than 5 million were counted). The issue was that while he made these improvements among urban and college-educated voters, this approach came at the expense of retaining levels of support in more working class and rural counties enjoyed by previous Democratic candidates (including Clinton’s level in some counties). None of this is to say that it had to be a tradeoff exactly; however, many of these rural counties were overlooked and led many of their voters (including those who are nonwhite) to be more receptive to targeted messaging from the Trump campaign. To that end, the story of North Carolina can be compared in certain respects to that of Texas and Florida, even though the voters in question aren’t necessarily Hispanic.

So that’s it for this part of this series. Stay tuned next week for the final part in this swing state series. If you enjoyed this, please like and follow the Book Aisle. Also share this article on Facebook, LinkedIn, and other social media platforms.