This was originally published on 5th March, 2017, and was republished here on 4th January, 2022.
In part one of our election review, we looked at the results of the election by state and region, and what they indicated for the Democrats, Republicans and third parties. In this second part, we will be looking more closely at some of the states that mattered, to discover exactly where it was that Trump won the election.
States are in alphabetical order. Maps are courtesy of the New York Times, voting figures are from the respective state electoral commission.
California
Dem – 8,753,788 (+899,503) Rep – 4,483,810 (-356,148) Oth – 943,997 (+599,693)
California stands out at this election for how unrepresentative it is of the election as a whole. It is one of only five states (plus D.C.) in which the Democrats gained votes and the Republicans lost votes, and the only of those states in which the Democratic gain was over 100,000 votes.
Unsurprisingly, the shift in the votes predominately came from the major urban centres. Amongst the inner-city counties of Los Angeles, Sacramento, San Diego, San Francisco and Santa Clara, the blue vote increased by 485,727, more than half of Clinton’s gains in California. There was also some increase in her vote in suburban/coastal northern California, with a collective increase of 146,277 in Alameda, Contra Costa, Marin, Monterey, Napa, San Mateo, Santa Cruz, Solano, Sonoma and Yolo Counties. But it’s in southern and eastern California, traditionally stronger Republican territory, where Clinton made the most impressive gains. Removing San Diego and Los Angeles, her vote increased in SoCal by 236,937, including flipping Orange County with a 180,000 vote turnaround.
Donald Trump’s unpopularity in the OC is, I think, indicative of the enormous political shift that has occurred, as Orange County has not voted Democrat since 1936. It is reflective of both a cultural shift that has occurred within the county, and indeed across the other coastal counties, and also of a political shift that has occurred within the Republican Party, which is not appealing to these counties like, say, Mitt Romney was. On the other hand, Trump’s vote in Kern, Riverside and San Bernardino actually increased, while his overall vote in the counties he won was more or less exactly the same as it was in 2012. It was in the urban centres, where Clinton and the third parties gained votes, that his losses occurred, presumably due to the aforementioned cultural and political shifts.
Colorado
Dem – 1,338,870 (+15,768) Rep – 1,202,484 (+17,241) Oth – 238,866 (+177,690)
I’ve put Colorado in here as another example of a state acting contrary to the ‘big story’. In this case, it’s because Colorado exemplifies the overall story of the western states: nothing happened. The campaign and the election itself was not really fought on issues or themes that were relevant to these states, and as a result very little within them changed.
Four counties in Colorado flipped, three of which (Conejos, Huerfeno and Las Animas) are small, rural counties, but one of which is Pueblo, home to the city of the same name. Beyond this, Trump made gains in rural areas but lost in Boulder and Denver, Clinton made gains in Boulder and Denver and had losses everywhere else, and the third parties had a strong showing everywhere. This comes out to a higher turnout, but otherwise evens out.
Florida
Dem – 4,504,975 (+267,219) Rep – 4,617,886 (+454,439) Oth – 297,178 (+224,202)
In my prediction, I mentioned that the election would be decided on turnout. Florida is definitive proof of this being the case. Turnout increased from 71.5% to 75%, which flowed to each of the three options, but the Republican increase was 200,000 votes greater than the Democrats. This was large enough to flip the state.
As in Colorado, only four counties flipped in Florida, all to the Republicans, and none of which became heavily Republican – all four are lightly shaded red on the map, indicating a smaller margin. The question all along was whether turnout in traditional Republican areas would outstretch turnout in traditional Democratic areas, and in the early voting it didn’t seem like it would. But this measure completely underestimated the intensity of the turnout not in the panhandle – what Steve Schale identifies as the ‘state’ of North Florida – but in the Orlando and Tampa media markets.
Trump got 2,156,633 votes across the counties in these markets. Mitt Romney got 1,887,874. That’s an increase of 268,759 votes, a larger rise than Clinton got in the entire state. What would drive these two regions to come out so strongly for Trump? It seems likely to have been driven by the same thing that caused the mid-west and the rust belt to swing his way, whatever that may be, because many of those who live here hail from these regions. Trump appealed to their ‘home’. Additionally, the Orlando market has been affected by the loss of manufacturing – and we know which candidate such voters favoured.
Indiana
Dem – 1,033,126 (-119,761) Rep – 1,557,286 (+136,743) Oth – 144,546 (+93,442)
Would you believe, looking at the intensity of the red on this map, that Indiana was won by Barack Obama eight years ago? That happened because John McCain could only eke out slim margins across the state, while Obama carried the cities by big enough margins to overcome the regional counties. Mitt Romney returned it to Republican hands with a 12% margin, but Trump blew that away by winning the state by nearly 20%, and reducing the Democratic held counties from nine to four.
Trump’s supremacy in the state is reflective of the breadth of his support in states that voted for him. Suburban Republicans in Indiana, perhaps the most likely to turn away from him, were easily appeased by Gov. Mike Pence as his running mate (or, otherwise, by the prospect of voting for Clinton); northern conservatives were likewise appeased by both Pence and Trump’s noises about issues of social conservatism; while in the south many working class voters were swayed for the same reasons that other rust-belt states went towards him. Indiana was the perfect storm of Trump voters, and it seems amazing to think that there was talk of this being a swing state at one point.
Iowa
Dem – 653,669 (-168,875) Rep – 800,983 (+70,366) Oth – 111,379 (+82,360)
While Indiana went Democrat two elections ago, Iowa in 2016 rather resembles Indiana in 2012. Iowa went blue in 2012, but not this time. Why not? Look at the map, and see how many light red counties are there. They correlate very closely with counties that voted for Obama in 2012. If we calculate the vote tallies only in counties that flipped this time around, the result looks like this:
Dem – 160,139 (-66,014) Rep – 210,502 (+32,619) Oth – 31,425 (+21,757)
Now, that represents a bit over a third of the drop in the Democratic vote. More importantly it accounts for nearly half of the Republican rise, demonstrating that the rural Trump surge was real. There was something about the difference between Trump and Clinton that genuinely appealed to the same Iowans that were willing to vote for Barack Obama instead of Mitt Romney. Whatever Clinton represented, a great number of voters in this mid-west state felt that it did not represent them.
Maine
Dem – 357,735 (-43,571) Rep – 335,593 (+43,317) Oth – 54,599 (+35,001)
Where better to go to demonstrate the splitting of the American electorate than Maine, which doles out it electoral votes by Congressional district as well as by the overall result? Maine has not split its electoral votes since introducing the system in 1972, but the differences between the rural north and the metropolitan south reared their head in such a way as for this result to occur. The map used here records shows results by towns, rather than by counties, but much as in Iowa, those lighter red counties were all blue in 2012. Perhaps the greatest single feature of this division in Maine is that the drop in the Democratic vote is almost exactly the same as the gain in the Republican vote.
Michigan
Dem – 2,268,839 (-295,730) Rep – 2,279,543 (+164,287) Oth – 250,902 (+199,766)
Indiana was a total victory for Trump, but Michigan (despite its close margin) was perhaps an even greater example of how the rust belt states shifted. Trump flipped twelve counties, most of which are light red in south-centre of the state. He also held on to Grand Rapids, while turning the Republican leaning north into deep red. But the kicker for Clinton was not that Trump increased his margins in these areas – it was that she didn’t get an Obama-like turnout in Democratic areas.
In 2012, Obama received 1,402,063 votes from the five big south-east counties: Genesee, Macomb, Oakland, Washtenaw and Wayne. In 2016, Clinton got 1,270,065 votes from the same counties, and Macomb actually flipped. Bear in mind that these counties include Detroit and Flint, cities that are meant to be absolute strongholds of the Democrats, yet Trump gained 27,775 votes in the two counties that they belong to.
In the counties that flipped (Bay, Calhoun, Eaton, Gogebic, Isabella, Lake, Macomb, Manistee, Monroe, Saginaw, Shishawee and Van Buren), Trump gained 61,336 on Romney while Clinton lost 77,089 from Obama. That difference can almost entirely be accounted for in Macomb County, which is an almost stereotypical old industrial working class area – exactly the type of county Trump was aiming for.
Minnesota
Dem – 1,367,716 (-178,451) Rep – 1,322,951 (+12,726) Oth – 254,146 (+183,977)
Minnesota has not voted for a Republican since 1972, so in some sense it isn’t surprising that Clinton carried the state. But when you look at the map, it looks remarkably similar to Iowa. Clinton lost nineteen counties, and many counties that only voted for Mitt Romney by a slim margin voted in much greater numbers for Trump. Yet Trump could only must 12,000 more votes than Romney, and couldn’t take advantage of a nearly 180,000 vote slump for the Democrats. Why?
One word: Minneapolis. In both Minnesota and Wisconsin, the talk was always that suburban Republicans who like Mitt Romney and Paul Ryan – classic liberals, essentially – did not want to vote for Trump at all. We will get to Wisconsin later, but in Minnesota that seems to be the thing that cost Trump from an extra ten electoral votes. In the two counties of Minneapolis (Hennepin and Ramsey), Mitt Romney received 326,872 votes. In those same two counties, Trump received 262,664 votes. Clinton, in comparison, got almost exactly the same amount of votes as Obama.
Nevada
Dem – 539,260 (+7,887) Rep – 512,058 (+48,491) Oth – 74,067 (+54,089)
This is another state to file away in the ‘nothing happened on the west coast’ file, which can at least show us where the election was not won. County by county, the result is almost exactly the same as 2012. The only difference worth noting is that Trump did better than Romney in Washoe County, where Reno is.
New Hampshire
Dem – 348,526 (-21,035) Rep – 345,790 (+15,872) Oth – 49,980 (+38,487)
New Hampshire is an interesting state, almost a curiosity. Much like Maine, the divisions here are by town, not by county, and one thing that stands out almost immediately is how few towns are deep red. In Maine, there’s quite a bit of deep red, even if that doesn’t represent many votes. Here, it’s a rarity, and similarly it doesn’t represent many votes, but it perhaps explains why Trump couldn’t quite overcome the necessary margin. It may have been less than 3,000 votes, but if he had won more towns by big margins, rather than simply solid ones, it would’ve turned the state.
New York
Dem – 4,547,562 (+61,821) Rep – 2,814,589 (+324,158) Oth – 296,865 (+191,608)
Could this state be more typical of the election? Red across most of the geographical area, but wherever there is a city dotted, so there is a blue county that contains it. What actually interested me here was the sharp rise in the Republican vote. Where did it come from? Some of it can be accounted for in the southern counties, bordering on Pennsylvania. Some of it also comes from NYC, where Trump managed to flip Staten Island quite comfortably. But the most notable portion of it comes from upstate rural New York, where most of the nineteen counties he flipped are.
Ohio
Dem – 2,394,164 (-433,545) Rep – 2,841,005 (+179,568) Oth – 261,318 (+169,617)
Hillary Clinton lost 430,000 votes in Ohio. 430,000 votes. If she’d had a net loss of zero votes she would’ve had almost exactly the same tally as Trump ended up with, but those 430,000 didn’t just go to Trump. About a two-fifths went red, about the same went to others, and the remainder…just disappeared. You may be thinking that these lost votes were from places like Cleveland and Columbus, as in Michigan, where people just didn’t want to turn out for her like they did for Obama. But across the six cities shown on the map, Clinton got 1,332,293 votes, 28,180 votes less than Obama. That still leaves 50,000 votes to find.
They are, of course, in the rest of the state. In 2012, only the western-most counties near the Indiana border voted strongly (ie. over 60% of the vote) for Mitt Romney; all the other Republican voting counties had relatively small margins. Not so in 2016. The counties that vote strongly for Romney voted even stronger for Trump, with the latter getting 80% in Mercer County. But the counties that voted with a weak majority for Romney also came out very strongly for Trump, turning 50-45 results in 2012 into 70-25 results. But even then, that’s only part of the story.
We can divide the Trump counties into three different regions. The first, I have already mentioned, is the western, strongly Republican counties. The second is the central-to-south eastern counties, in the Appalachian region. In both of these regions, turnout was about the same as in 2012. The third region is in the north, which accounts for almost all of the Ohio counties that are not deep red. All but three of those light red northern counties voted for Obama in 2012, and turnout plummeted at this election.
Removing the counties of Cleveland, Toledo and Akron, Clinton received 441,201 votes in these northern counties. Obama got 549,385 votes in the same counties. Trump and ‘others’ account for most of that gap, but around 20,000 votes disappeared here too. Combine that with the small drops from county to county in the red remainder, and Clinton is down by 430,000.
Pennsylvania
Dem – 2,926,441 (-63,833) Rep – 2,970,733 (+290,299) Oth – 218,228 (+135,266)
Speaking of Appalachia, here’s a state that is almost entirely within the region. Only the south-east corner – which you’ll notice is mostly blue, with a bit of light red – is not part of the region. If you look at a map of Pennsylvania in 2012, it looks a lot like eastern Ohio did in the same year. And now, much like eastern Ohio, it’s turned deep, deep red. This helps explain why Trump’s rise in the vote is greater than Clinton’s drop in Pennsylvania. Unlike Ohio, in which Appalachia is only part of the Trump majority, in Pennsylvania it is, in essence, all of it. In many of these counties, turnout increased, and it increased so that voters could vote for Trump.
And yet, that was very nearly not enough to shift Pennsylvania to the Republican column, all because of Philadelphia. The Democrats run up some extraordinary numbers here at every election, upwards of 80%, the kind of number you expect in a tiny rural county, rather than a big city. In 2012, Obama got 85% of the vote in Philadelphia County. Clinton ‘only’ got 82%, but actually got more votes than Obama – it’s just that Trump also got more vote there than Romney. In fact, she got more votes in the Philadelphia suburbs as well: 575,166 to Obama’s 514,598. She also did better in Pittsburgh and Allentown. But this was not enough to overcome the regional vote shift, which swung so heavily to Trump.
Texas
Dem – 3,877,868 (+569,744) Rep – 4,685,047 (+115,204) Oth – 406,311 (+290,427)
Demographics: the state. Only two counties flipped from 2012 – one each way – but the surge in votes and the massive margins in most counties reflects the kind of election that we saw in 2016. Cities and Mexican border counties voted heavily for Clinton, everywhere else voted for Trump. If this kind of pattern continued in future election, it is pretty easy to see how Texas could be seen as ‘turning blue’ sooner rather than later. But how many more votes can the Democrats find? The half-a-million extra votes this time around came almost exclusively from the big cities, and how likely is it that they will continue to expand at the rate they have been?
Side note: there were some rural Texan counties that were not deep red in 2012. Not this time – one county was nearly 90% for Trump.
Virginia
Dem – 1,981,473 (+9,653) Rep – 1,769,443 (-53,079) Oth – 231,836 (+171,689)
Virginia’s unusual county divisions, which allow for the creation of independent cities as separate administrative areas to the county the belong to, actually come in rather handy in maps like this. Need proof that urban voters and rural and regional voters are basically in different worlds?
Look no further than western Virginia, with all its blue cities surrounded by red counties. Western Virginia, like eastern Ohio, Pennsylvania and (duh) West Virginia, is in the Appalachian region, and, like those other areas, turned out stronger for Trump than they did for Romney. But Trump lost votes from Romney overall in the state. Why? The Washington suburbs. D.C. is on the northern border, where all that deep blue is. Clinton absolutely blitzed this area, especially Fairfax County where she got nearly 100,000 more votes than Obama did in 2012.
Given her results there, it’s mildly amazing that her result across the state is a less than 10,000 gain on 2012, especially given she also did better than Obama in Richmond, Roanoke, Norfolk and Virginia Beach. Furthermore, she didn’t really lose many votes in regional eastern Virginia. So where did she lose those votes? Well, it should be obvious by now. The Republicans didn’t just gain votes in western Virginia, the Democrat vote also tanked – a lot. This is the third Appalachian state we’ve looked at where this has happened.
Wisconsin
Dem – 1,382,536 (-238,449) Rep – 1,405,284 (-2,682) Oth – 188,330 (+148,847)
As I mentioned with Minnesota, suburban Milwaukee Republicans were tipped to be the big impediment for Donald Trump in Wisconsin. As it turned out, they did indeed turnout in smaller numbers than they did for Mitt Romney, with Trump losing 51,385 votes in Milwaukee and Waukesha Counties. But this turned out not to be the barrier it was thought to be, for two reasons. Firstly, the industrial north which turned out weakly for Romney came out much stronger for Trump, as in neighbouring Michigan. Secondly, the rural south of the state flipped from Democrat to Republican, as in neighbouring Iowa.
Clinton, meanwhile, lost between 1,000 and 5,000 votes in a large number of counties, which in a state of 72 counties, adds up pretty quickly. Glaringly, she also lost 40,000 votes from Obama in Milwaukee, some of which were scooped up by third parties, but many of which simply disappeared.
Conclusions
Across these maps, we have seen three types of county: rural and regional counties which turned from solid red to deep red; rural and regional counties which turned from blue to red; and metropolitan counties which remained blue.
Notice which counties voted which way? The maps tell the story over and over again. The more regional, the more red. There was no state which truly bucked this trend. Small town America voted for Trump, big city America voted for Clinton.
I have identified some sub-sections that have emerged in these states as well: Appalachia was softly Republican in 2012, but went incredibly hard for Trump this time. The rural mid-west was happy to vote for Obama to some degree in 2012, but was less keen on Clinton. Republican suburbs were not especially keen on either candidate. Wealthy metropolitan areas loved Clinton. The ease with which these things can be said is actually disturbing, but worth exploring.
We shall do that in part three.