Why was North Carolina Bernie's best state in the Old Confederacy?

[Adam Griffin]

Closed primaries: there are lots of ancestral Democrats, Dixiecrats & otherwise federal R voters locked into voting D because of party reg, particularly in western NC. It's the same reason he did so well in WV, KY, OK, etc.

[Adam Griffin]

Closed primaries: there are lots of ancestral Democrats, Dixiecrats & otherwise federal R voters locked into voting D because of party reg, particularly in western NC. It's the same reason he did so well in WV, KY, OK, etc.

WNC ain't like WV, OK, or KY.  Those are real granola eating birkenstock wearing people until you get past Jackson Co.  Buncombe is pretty liberal on every level now and there are wealthy retirees south of there.  Plus many of he mountain counties have a R history and the Ds are really D.  The housing is pretty expensive (and not in trailer parks)  and the rate of college degree attainment is pretty high.  

I disagree, at least broadly speaking. Of course Buncombe is filled with yuppy liberals (and Watauga is a college town), but the vast remainder of this area is either standard rural swathes of Appalachia (where ancestral D voting behaviors aren't uncommon) or traditional Southern suburbs. I think you might also be confusing Eastern TN with Western NC in terms of historical patterns; the latter (at least the core area, mentioned and defined below) had a fairly strong tendency to lean Democratic going all the way back to the 19th century: not as strong as most of the Deep South, but that can be said about any part of Southern Appalachia.

With that being said, the stretch of terrain from the Eastern Band to the eastern counties adjacent to Buncombe does still have a higher level of Dem support than most of the remaining area - however, even it is not enough to explain the higher levels of Dem primary turnout. Part of this can be explained by the Eastern Band that augments the D vote share in Swain, Jackson & (somewhat) Haywood - these individuals are for the most part "real" Democrats, but we know that Native turnout in primaries is abysmal in general, meaning they're not primarily responsible for Dems outnumbering Reps in primary turnout.

In that stretch of terrain I mentioned earlier, there were several counties where Obama got 40% or more of the white vote in 2012, putting it out of step with most of the region. However, in 2016, shifts among white voters were larger here than pretty much anywhere else in NC, with Clinton falling into the low 30s in most of them. A large segment of people who voted in the Dem primary here in 2016 were never going to vote for the Dem nominee - in this case, including a fair number who had voted Democratic through 2012.

County	2012 White D%	2016 White D %
Jackson	45	39
Swain	40	31
Haywood	40	33
Transylvania	37	34
Madison	43	34
Yancey	40	31
Henderson	31	31

NC in general has a lot of Democratic primary voters who don't inherently vote Dem in general elections (you also see it in the eastern portions of the state; in line with multiple Southern counties that are adjacent to the Black Belt and may have at one point been considered part of a greater, expanded Black Belt where Democrats dominated locally):



Republicans barely outnumbered Democrats statewide in the 2016 presidential primary, and in the 2018 primary, Democrats were 56% of all ballots cast - far exceeding their share of actual support in the state. While my primary vote by county project couldn't find discernible reports for turnout by party and county for 2018, I imagine these counties still had artificially high levels of Democratic primary participation compared to their GE vote shares.

[Adam Griffin]

I would guess that even apart from the type of primary, Tennessee is more racially polarized - i.e., the Democratic party is almost totally black in Tennessee, therefore Clinton won.

Adam Griffin?

TN's Democratic electorate is actually a white-majority electorate; black voters are such a relatively small share of the population for a Southern state (unlike most Deep South states, the overwhelming share of blacks are confined to the 4 major cities, with only 1 majority-black rural county - Haywood - and 2 others - Hardeman and Lauderdale - with sizable black population) that they cannot comprise a majority unless Democrats start getting less than 30% of the vote there.

In 2012, the Obama coalition was 63% white and 35% black; in the 2016 general, it was a meaningfully closer 53/41 (because Clinton hemorrhaged white voters there). I can't say for sure what the 2016 presidential primary looked like, but it was probably somewhere in between those two sets of numbers.

It's obvious though that both white and black voters in TN voted for Clinton. Probably something like 55% and 80%, respectively.

[Adam Griffin]

The stretch of counties from Jackson to Polk have college educated rates of 30% or higher though they are quite old also.  There are lots of old people in rural WV, OK, and KY but nothing approaching that level of college education.  There is also less reliance on resource extraction in WNC than those other states and in fact more outdoor recreation opportunities in WNC.  They aren't diverse and they are old, but WNC skewed towards Bernie for different reasons  than  the aforementioned states.

It's worth noting that the exit polls for the NC Democratic Primary actually suggest the opposite should be true based on what you're saying: Bernie did worse with white college voters than he did with white non-college voters, and of course did worse with older voters as well. In fact, the only reason Bernie did as well/better among college grads overall was because black college voters were 20+ points more in favor of him than black non-college voters. Given the latter group isn't really a presence at all in western NC, it's hard to say they're responsible for improved margins relative to the rest of the state.

There could of course be some regional variation in how groups voted, but this would have to be above and beyond what would seem possible - and given the solid share each group comprised of the electorate, I think dismissal of this based on large margins of error and the like is not realistic.

[Adam Griffin]

I disagree, at least broadly speaking. Of course Buncombe is filled with yuppy liberals (and Watauga is a college town), but the vast remainder of this area is either standard rural swathes of Appalachia (where ancestral D voting behaviors aren't uncommon) or traditional Southern suburbs. I think you might also be confusing Eastern TN with Western NC in terms of historical patterns; the latter (at least the core area, mentioned and defined below) had a fairly strong tendency to lean Democratic going all the way back to the 19th century: not as strong as most of the Deep South, but that can be said about any part of Southern Appalachia.

With that being said, the stretch of terrain from the Eastern Band to the eastern counties adjacent to Buncombe does still have a higher level of Dem support than most of the remaining area - however, even it is not enough to explain the higher levels of Dem primary turnout. Part of this can be explained by the Eastern Band that augments the D vote share in Swain, Jackson & (somewhat) Haywood - these individuals are for the most part "real" Democrats, but we know that Native turnout in primaries is abysmal in general, meaning they're not primarily responsible for Dems outnumbering Reps in primary turnout.

In that stretch of terrain I mentioned earlier, there were several counties where Obama got 40% or more of the white vote in 2012, putting it out of step with most of the region. However, in 2016, shifts among white voters were larger here than pretty much anywhere else in NC, with Clinton falling into the low 30s in most of them. A large segment of people who voted in the Dem primary here in 2016 were never going to vote for the Dem nominee - in this case, including a fair number who had voted Democratic through 2012.

Maybe another factor is that while these places aren't dominated by the same kind of people who set the trends in places like Buncombe County, they still contain a large enough share of those types to be decisive in lower-turnout Democratic primaries. Any of the communities along the Blue Ridge Parkway will have more urban retirees, tourism, and recreation than other rural parts of the state, right? Along with the usual assortment of weirdos who are drawn to places that are isolated and marginal enough to have cheap land?

A place like Floyd County, Virginia, one of a handful of Sanders counties in that state, would be a prime example. It has a low population, and it's rural and pretty isolated by East Coast standards, but it's also known as a minor center for folk music. There aren't enough yuppies and other back-to-the lander types around to dominate elections, to say nothing of whatever image outsiders hold of the place, but those people are numerous enough to make a difference in a Democratic primary as long as the locals aren't too lopsided in their support.

Perhaps - though I would point out that this combination of factors isn't inherently unique to this part of Appalachia or the region. Just a stone's throw over the border in NE Georgia, you'll find the same broader dynamics (rural and mountainous area with relatively cheap land; heavily reliant on tourism; most who move there are older and come from outside the region; relatively well-educated compared to most other rural areas nearby)...yet I hypothesize that those who moved into that part of my state actually made the area more Republican (we're talking 80% R as opposed to perhaps 70-75% R, but alas). It's also important to note that North GA in general was Sanders' best segment of that state, but he lost by 5-15 points there as opposed to winning by 5-15 points across the border; I think it's safe to say that the North GA performance was a good baseline for Western NC in terms of what racial dynamics alone would produce sans any protest voters.

I'd also be cautious in assuming that low turnout could produce these effects: regardless of national climate, Democrats have tended to outvote Republicans in state and presidential primaries in this area for a long, long time. We're talking about some counties where Clinton lost by 20-25 points but where Democrats were 55-60% of the primary electorate in 2016, with similar primary compositions existing both before and after the 2016 presidential (potentially indicating 15-20% of voters were not "true" Democrats in a federal sense). If low turnout in the Democratic primary causing the results we saw were a specific concern, then I imagine the likelihood of there being even more registered D voters who vote R would only grow (and given the dynamics of 2016, those who wouldn't show up to vote in such a contest would probably be just as pro-Sanders - whether that be through genuine support or protest-voting). Perhaps this would be cancelled out to some degree by Nat-Am voters' low turnout in the cluster around the Eastern Band, but that's only one segment of the broader area.

A lot of good points have been brought up in this thread about what may lead to a better performance for candidates like Bernie in NC, including...

• Race
• Age
• Education
• Granola
• Economic anchors
• Migration patterns

...and some of these are definitely responsible for a better performance - but as already mentioned, you'll find multiple combinations of these factors in areas where Sanders did not win or even necessarily overperform (both inside and outside of Appalachia).

If the OP's question were "why did Sanders do so well in Appalachia?", I could see these factors being meaningful answers (though in reality, you could simplify all of this by just saying "there's a lot of white people there"). However, asking why Bernie did so well specifically in one state I think changes what we're looking at here (to be fair, you could even simplify the OP's question with the same answer - there were simply way more white people voting in the primary (62%) than comprised the Obama or Clinton GE coalitions (45-50%)). Also, let's not forget that Western NC wasn't the only segment of the state where a) Democrats did much better in terms of share of primary electorate than they do in the general, and b) Sanders did better than expected (i.e. "old/auxiliary/expanded Black Belt counties", where non-D federal voters are also registered as Dems in large numbers).

The strongest correlation in my view when looking at his performances in other areas throughout the country and what situations also existed in those areas is that NC has a closed primary system and many areas where ancestral Democrats and/or state & local D-national R voters still exist. In knowing that Bernie won those effectively described as "protest-voters" by 3:1 nationally and given how much of a difference there tends to be between federal D voting and D share of primaries (again, in some areas, equivalent to 15-20% of all primary voters), it's likely these types made a meaningful difference in both the statewide margin and the winner of the contest in multiple counties.

[Adam Griffin]

For those who don't want to read my TL;DR and want the simplest answer to the question (that only indirectly assumes anything about closed primaries and ancestral Democrats), I think this pretty much is the best summary - especially given racial polarization in the South:

If the OP's question were "why did Sanders do so well in Appalachia?", I could see these factors being meaningful answers (though in reality, you could simplify all of this by just saying "there's a lot of white people there"). However, asking why Bernie did so well specifically in one state I think changes what we're looking at here (to be fair, you could even simplify the OP's question with the same answer - there were simply way more white people voting in the primary (62%) than comprised the Obama or Clinton GE coalitions (45-50%)).

Also, let's not forget that Western NC wasn't the only segment of the state where a) Democrats did much better in terms of share of primary electorate than they do in the general, and b) Sanders did better than expected (i.e. "old/auxiliary/expanded Black Belt counties", where non-D federal voters are also registered as Dems in large numbers).

The 2016 Democratic primary electorate in NC was substantially whiter than its modern-day Democratic presidential electorate:

Contest	White D Electorate %	Black D Electorate %
2012-GE	46%	49%
2016-GE	50%	39%
2016-PR	62%	32%

Of course, for those who may disagree with my broader belief on why Sanders did so well, one has to explain why the white share of the Democratic primary electorate was so much higher in the NC Democratic primary than it tends to be in Democratic GE electorates, and why the increase was so substantially above and beyond any broadly similar differences we saw in other Southern states. In my view, the simplest explanation is that there are tons of conservative-to-moderate whites registered as Democrats, they are locked into Democratic primaries, and - who like these types of voters everywhere else - voted for Sanders en masse.

[Adam Griffin]

Out of curiosity, I decided to map out the difference in margin between the 2016 GE (Pres) and the 2018 primary (finally bothered to map that out in my 2018 primary by county project). Keep in mind that in both comparisons, it uses a two-way model (excludes any third-party votes).

The darkest shade of red is where the D-R margin in the 2018 PRIM exceeded the D-R margin in the 2016 GE in favor of Democrats by more than 45 points. There are some counties here with 70+ point differences in the two electorates (like Columbus; Trump +22 and Dems +49). In most of western NC, the differences were smaller but still more impressive in my view (in Swain, Trump +20 & Dems +24; in Haywood, Trump +29 & Dems +11; in Jackson, Trump +12 & Dems +15).

County data is available here