The first four (primary) states: What past and present tell us about future

It seems like just yesterday, we (I) were (was) waiting for Bernie to announce his second bid for president and hoping Hillary would not do her William Jennings Bryan imitation.

Nine months later, it’s a boy!

Just kidding. It still won’t be anything for three more months. But we don’t have to wait impatiently for states to vote. We can look at data. (NOT early polling. Early polling is atrocious. Early polling is so bad that it ought to be illegal. Anyone who tries to surmise anything from early polling other than “some people are doing early polling. Idiots.” should be laughed out of politics and off Twitter.)

What follows is a data-based, county-by-county look at the first four states of the political season: Iowa, New Hampshire, Nevada and South Carolina.

1. Iowa

Iowa is a bad contest. The winner usually wins the presidential nomination AND then usually loses in November, which is a good reason to stop caring about Iowa. (Carter finished second, and Clinton basically didn’t bother; only Obama won Iowa and the presidency.) But so long as the political media is obsessed with a state whose opinion even the Des Moines Register acknowledges is atrocious (“The Iowa caucuses have a poor record of picking presidents”), let’s do this thing.

Kerry did it in 2004, Obama in ’08 and Hillary in ’16, and for however much any Democrat wants to insist that whichever nominee wuz robbed, losing is losing is losing.

Harkin won Iowa in 1992 on his way to watching Bill Clinton get the nomination, which is what happens when you win your home state and two other states (one of them by less than a point).

Gephart won in 1988 (or, at least, he was credited with the win: we don’t have the full results from that contest) on his way to three victories, total. And then Mondale and Carter (1980) won in the two tries before that. So Iowa’s kind of bad at discerning ability to win in November.

The super-fun thing about Iowa is that for all the hype, practically nobody gets delegates. Three of the eight candidates in 2008 got delegates. Three of the six in 2004 got delegates. So the notion that seven of the seventeen (siiiiiiiiiiiiiiigh) candidates still in the race are going to walk away with something to pin their hopes on is a little silly.

But how do you win Iowa?

Get ready for data tables, y’all.

John Edwards ran for president in 2004 and finished second in Iowa, winning 41 counties — most of them in the southern part of the state. He repeated in 23 of the 41 in ’08, with Obama taking 10 of the rest, Hillary taking six and the final two splitting.

The demographics of how those Edwards counties voted is pretty sensible: Obama grabbed a chunk of the highest-population counties, including crown jewel Polk, and the smallest Edwards county he flipped (Greene) had 8,981 people in it, per Census data. Hillary grabbed fewer Edwards counties, but two of hers (Osceola and Worth) were smaller than Greene. The largest counties Edwards kept from his ’04 run had about 51,000 (Warren), 37,000 (Jasper) and 35,000 (Wapello) people. Here’s how it all looks:

The problem for Edwards wasn’t that he didn’t hold his original turf. It was that he couldn’t “two Americas” his way into many Kerry counties, possibly because they’d heard his pitch four years before and hadn’t been interested then either. Of the 49 counties Kerry won in ’04, Edwards won just six in ’08, and of the nine counties that split in ’04, he won just one in ’08. Meanwhile, eventual caucus winner Obama soaked up 27 Kerry counties, Hillary grabbed 15 of the rest, and one county split.

Eight years later, Hillary likely knew she’d have to grab new ground in 2016, and with Obama not allowed to run again, she did better than she’d done with the Kerry counties in ‘08:

But her real strength was in taking the counties Edwards had won in ‘08:

Population isn’t a great predictor of whether a county flipped to Bernie or Hillary. But this map, in which round shapes are in Edwards ’08 counties that Hillary won in ’16 and square shapes are in Edwards ’08 counties that Bernie won in ’16, reveals a pretty clear pattern:

Hillary won where Edwards had won in southern Iowa and the middle-northish part.

But Hillary also had to play defense, and whereas Edwards had won 23 counties twice, Hillary repeated in just 12 of her 25 from ‘08. Population isn’t a great indicator of why Bernie took about half of her ’08 counties:

But here again, a map shows what happened, with round shapes in Hillary ’08 counties that Hillary won in ’16 and square shapes are in Hillary ’08 counties that Bernie won in ‘16):

He did pretty well in the north and the southwest, but she protected those two southeast counties.

Looking forward to 2020, a big area of concern for Bernie should be south-central Iowa counties, and he should focus hard on those: As we saw above, he can’t expect to repeat in two dozen of the 36 counties he won in 2016. Edwards won just 23 of his 41 counties twice, and Hillary repeated in only 12 of her original 25.

He should also target the second-largest block of counties won by one (not-him) person, which is the block won by Kerry, Obama and Hillary. Have another map:

Some of those — Webster, Hardin, Bremer — dip slightly into that area of Edwards counties where he struggled against Hillary, but others — Dubuque, Jackson, Plymouth — are more doable. And Bernie has shown that he can win counties won by Kerry and Obama.

The 20 counties Edwards won twice and Hillary won once constitute the largest block. The trouble for Bernie is that those folks simply might not be into him: Edwards repeated in 23 counties, so for Hillary to win 20 of them speaks to a communal lack of feeling the Bern. Still, though, challenging in the ones where he was close might be wise. The counties that are bolded rejected him by just single-digit margins:

These bolded Kerry-Obama-Hillary counties offer promise as well:

And he has options in Hillary country too:

2. New Hampshire

New Hampshire isn’t as bad as Iowa, but it still shouldn’t be listened to. Before 2016, New Hampshire hadn’t picked the eventual Democratic winner or rejected the eventual loser since 1984. That’s a long losing streak.

But since everyone gets to vote — no matter how wrong they are — let’s look at New Hampshire’s primary voting history since 1992 (said no one ever, until now):

As you can see, four bolded counties have picked the primary winner every year since 1992. Here’s how that looks geographically:

That those four counties keep getting the primary winner they want isn’t a coincidence: Hillsborough is the largest county in the state, and its closest rival, Rockingham, is 33 percent smaller. Stafford is the fourth-largest, so in that map, you have three of the four biggest counties (plus Belknap) all picking the primary winner four times in a row.

Merrimack, the third-largest county by population, also has roughly half the population density of the other three; whether that’s what led its voters to pick Obama over Hillary in ’08 is possible, since the other Obama counties also have relatively low population densities:

Now, question: Did proximity to Massachusetts helped former Massachusetts Sen. Paul Tsongas beat Clinton in 1992? Clinton won Coos, Sullivan and Cheshire counties, and Cheshire is on the New Hampshire-Massachusetts border. Sullivan is directly north of Cheshire, while Coos is New Hampshire’s hat. So while proximity helped some, Clinton winning a border county shows that geographical proximity isn’t everything.

A quick look at the data shows that the bigger factor is population density. Clinton won three of the five counties with the lowest population densities — the most rural counties.

Where geography helped Tsongas quite a bit, it made the 2004 primary a straightforward affair (Dean counties in blue):

Dean won three of the four counties on the Vermont-New Hampshire border. But he won nowhere else in the state, thanks to John Kerry’s 18 years as border state Massachusetts’ junior senator.

Moving ahead now to 2016, I think it’s pretty clear that Bernie’s margins came due to geography, as happened with Dean, versus population density. Here’s a chart comparing density to Hillary’s share of the vote:

If Hillary were more popular in urban areas, we wouldn’t expect to see her share of the vote fall in Strafford versus Merrimack counties, and it definitely wouldn’t fall in Sullivan and Cheshire counties versus Grafton, let alone Coos. More likely is that geography played a role:

Vermont is to New Hampshire’s left. The farther you get from Vermont, the better Hillary does (except for that blasted Strafford County).

So Vermont and Massachusetts politicians are both helped by their proximity to New Hampshire. Unfortunately, I don’t see enough in the data to offer a strong guess on how Bernie will do against Elizabeth Warren in those populous New Hampshire-Massachusetts border counties. Given her re-election map here, in which the New Hampshire-Massachusetts border isn’t nearly as blue as the western and central parts of the state are, I mightn’t pencil her in as the winner of Hillsborough or Rockingham counties yet, though.

Having said that, Bernie’s Vermont border advantage will be of limited value: Even ludicrously high margins in the border counties of Coos, Grafton, Sullivan and Cheshire — total population slightly less than 240,000 — won’t win the day unless he can at least do well in the more populous counties. Rockingham alone has more than 240,000 people, and Hillsborough is bigger still.

3. Nevada

Oh, look. Another bad contest.

“Oh, shut up. You think every contest is bad.”

Wyoming, Alaska, Vermont, D.C. Listen to the folks there and ignooooore Nevada (and especially Iowa).

Nevada offers more geography fun, with California Gov. Jerry Brown winning the state in 1992 on the strength of California being its bosom buddy. Of Nevada’s nine counties that border or nearly border California, Brown won five, including Clark. He won no other counties in the state, and he needed to win no other counties in the state: He carried it with almost 35 percent of the vote.

This year, a different California politician is running for president. Since early polling is trash and you should ignore it like you ignore other irrelevance, I’ll say only that Sen. Kamala Harris could well win Clark County on the way to a popular vote win in the state.

That wouldn’t necessarily deliver her a delegate win, though: Hillary beat Obama by almost six points in the popular vote, but he got more delegates than she did because he campaigned early and with purpose in the rural counties.

Bernie can speak with some purpose and plans to rural counties.

In terms of winning the state, the best plan might be to hit Clark County hard and then play defense: Of the six counties Hillary won in ’08, she repeated in four of them in ’16, losing only two to Bernie. Meanwhile, Bernie soaked up Obama territory big time, winning eight of eleven counties that caucused for hope and change in ’08. And whereas in ’08 Hillary won the popular vote but ended up with three fewer delegates than Obama, she won the popular vote and the larger delegate haul — by five — in 2016. So any campaign that thinks you can just win Clark County and that’s the game is going to lose that game.

4. South Carolina

To write this piece, I built spreadsheets listing every county in each state. For Iowa, this was a beastly task, but I did it, and now I know things most other people don’t know.

New Hampshire was easy, and Nevada was pretty gentle as well.

But y’all.

Iowa’s counties are at least sorta easy because they go across rather than … being like South Carolina’s. But research isn’t always easy, so I figured out a rough pattern for organizing South Carolina’s counties (go down and right and you find pattern-ish stuff) and got ready to suffer a little.

Then I looked at the 1992 primary map.

A speck of hope formed in my heart.

Looked at 2004.

The speck grew into a dot.

2008.

The dot grew into a ball.

2016.

The ball was now as big as the Ritz.

See, when South Carolina Democrats vote for a presidential nominee, they apparently all get together the day before, decide on one candidate to back, and then everyone votes for that person. Here’s what I mean:

The bolded counties — all 38 of them — have voted for the primary winner four times in a row. Only eight counties have strayed in 24 years, and none of them more than once.

So it’s true that to win any of more than three dozen South Carolina counties is to win the state.

But why?

Why did six counties go for Kerry?

Why did one Edwards county hold out for him again in 2008?

Why did Horry County say #ImWithHer?

Population density doesn’t appear to have been a factor in 2004: While two of the Kerry counties were pretty sparsely populated, the other four were in the top 20 for population density. What may be more likely is that they were continuations of Georgia’s opposition to Edwards/support of Kerry:

In the red circle is the central part of Georgia, which was one of three patches of the state that didn’t break for Edwards. Kerry won the state by five points. Of the 11 counties South Carolina counties that border Georgia, three voted for Kerry, and the other three were either pretty close (Bamberg) or closeish and a fair distance from North Carolina (Berkeley, Sumter).

Getting away from talk of holdouts now, the question is: Will South Carolina break en masse for one Democrat in 2020?

Nobody has a geographical claim to this state: Vermont, Massachusetts, Delaware, Indiana, California, New Jersey, Minnesota — these states are all a hike and a half away from South Carolina.

Momentum — if it exists — also might not matter here. Bernie could win Iowa, New Hampshire and Nevada, and South Carolina could simply shrug.

Why?

Because Iowa, New Hampshire and Nevada are distinctly liberal, and they’ve been getting more liberal. South Carolina is still fairly conservative:

So all anyone would have to do to counter some “Bernie’s on a roll” narrative is say, “Those are liberal states. South Carolina is conservative.”

If this state is going to vote en masse for one person, I suspect that person is Joe Biden. But I doubt that’ll happen. Instead, I suspect the state will be split between young people voting for Bernie and old people voting for Biden. This might result in pockets of support for Bernie at HBCUs and stronger support for Biden elsewhere.

So to conclude:

Win Iowa by winning where Hillary won, not so much by defending old turf.

Win New Hampshire by challenging Elizabeth Warren in the counties that border Massachusetts.

Win Nevada by hitting Clark County hard but also being a regular in the rural counties.

Win South Carolina by winning any of those 38 counties.