Skip to main content
ABC News
It’s Not All About Clinton — The Midwest Was Getting Redder Before 2016

Here’s one of the biggest questions Democrats face in the wake of Hillary Clinton’s loss: Did Clinton have an Electoral College problem or does the Democratic Party have an Electoral College problem?

After all the votes are counted, Clinton is going to win the national popular vote by more than 2 percentage points, but her votes were incredibly inefficiently distributed in the Electoral College — she ran up huge margins in coastal states with major cities while narrowly losing in several Midwestern swing states. Clinton might have won if she’d hammered on a more consistent economic message aimed at working class white voters in states like Michigan, Wisconsin and Pennsylvania. She might have won if she spent a little more time campaigning in those states. But there’s also reason to think that Clinton’s weak position in the Electoral College was not solely the result of her own campaign and that, instead, the Democratic Party has a problem.

Before we begin, let’s be clear about what we’re trying to measure. We’re interested in how Democratic- or Republican-leaning each state is relative to the country as a whole, and we want some measure independent from the 2016 election. If Democrats generally are losing strength in the Midwest, that shift should show up in data other than the 2016 results. It should show up, for example, in President Obama’s job approval ratings.

And it does. Gallup gathers approval ratings on the president in all 50 states every year.1 (They haven’t collected data for 2016 yet.) Obama’s relative strength in each state (compared to his job approval rating nationally) in 2015 versus 2012 largely tracks with where Clinton outperformed and underperformed Obama’s margin in 2012.2

Let’s take Maine as an example. Obama won it by 15 percentage points in 2012. Clinton squeaked out a win there by only 3 points. So Clinton did much worse in Maine even though both she and Obama won the national popular vote by similar margins. That is, Maine shifted right over those four years. You can see the same shift in Obama’s approval rating in Maine. In 2012, Obama had a 48 percent approval rating nationally and a 50 percent rating in Maine, according to Gallup. In 2015, he had a 46 percent approval rating among all Americans and a 41 percent rating among Mainers. The state had swung away from him compared to the country overall.

The strong relationship between Obama’s relative shift in approval ratings and Clinton’s performance held in most of the swing states too.3

STATE 2012 2016 CHANGE 2012 2015 CHANGE
Maine +15.3 +2.7 -12.6 +2 -5 -7
Iowa +5.8 -9.4 -15.2 +2 -5 -7
New Hampshire +5.6 +0.4 -5.2 -2 -6 -4
Wisconsin +6.9 -0.7 -7.6 +2 -1 -3
Ohio +3.0 -8.1 -11.1 -1 -4 -3
Michigan +9.5 -0.2 -9.7 +3 +1 -2
Pennsylvania +5.4 -0.7 -6.1 -1 -3 -2
North Carolina -2.0 -3.7 -1.7 -1 -2 -1
Minnesota +7.7 +1.5 -6.2 +2 +1 -1
Florida +0.9 -1.2 -2.1 0 0 0
Georgia -7.8 -5.2 +2.6 -2 -2 0
Nevada +6.7 +2.4 -4.3 -2 -1 +1
Arizona -9.1 -3.5 +5.6 -4 -2 +2
Colorado +5.4 +4.9 -0.5 -2 0 +2
Virginia +3.9 +5.3 +1.4 -2 0 +2
Clinton underperformed more where Obama’s approval rating dropped most

Sources: Gallup, Dave Leip’s Atlas of U.S. Presidential Elections, David Wasserman

Clinton did better in 20164 than Obama did in 2012 in three swing states5: Arizona, Georgia and Virginia. Those three states, along with Colorado and Nevada, also happen to be where Obama saw the biggest improvement in his relative approval rating in 2015 compared to 2012. That suggests that at least some of the movement toward Democrats in these states wasn’t about Clinton and Trump but was reflective of a longer term trend of these states becoming more Democratic-leaning.

At the other end of the spectrum, Obama’s approval rating relative to the nation dropped in all the Midwestern states that were key to the 2016 election, including Michigan, Minnesota, Ohio and Wisconsin. It also fell off in Pennsylvania. The movements in some of these states were not that large, but they didn’t need to be: A number of the states that made up the mythic “blue wall” were never that much bluer than the country as a whole. Pennsylvania, for instance, was less than 2 percentage points more Democratic than the nation in 2012.

The shift in Obama’s relative approval rating isn’t a perfect predictor.6 Obama’s rating was better in Nevada compared to the nation in 2015 than it was in 2012. Yet, Clinton did worse there. Obama’s rating was actually far worse in New Hampshire in 2015 relative to the nation than it was in 2012, and yet Clinton did only somewhat worse than Obama in the Granite State. (She still narrowly won the state.)

Overall, though, the changes in Obama’s approval ratings and Clinton’s performance mostly went hand-in-hand. Obama’s relative approval rating dropped a point or more in 25 states, while it rose a point or more in only 12 states (including California and Texas). That is, Democratic support was becoming deeper but more narrow before the 2016 general election really got underway. That’s a sign that Clinton’s problems reflected the electoral drawbacks of the evolving Democratic coalition at least as much as her own inabilities as a candidate.7

So if a weakening in Democratic support in a few key states cost the party its edge in the Electoral College, what should Democrats do? Perhaps the best answer is “wait and see.” It took a few years after Obama’s re-election in 2012 for Trump’s “Midwestern” path to the White House to emerge as viable. Trump won running primarily on a hardline stance toward immigration, which is exactly the opposite of what Republicans argued for in their assessment of what went wrong for them in 2012. The point is that things can change quickly in politics. States like Arizona and Georgia could quickly turn blue. Some states, like Pennsylvania and Wisconsin, might flip back into the Democrats’ column. We just don’t know yet. The only thing we do know is that, historically speaking, any one party’s advantage in the Electoral College doesn’t last. Four or eight years from now, something will probably emerge that gives Democrats the opportunity to win the Electoral College while losing the national popular vote.


  1. This poll’s consistency — same question wording, same methodology — allows for better comparisons of year-to-year data.

  2. The correlation is +0.88.

  3. These are the 15 states where the 2016 margin rounds to 5 points or less (that is, the margin was less than 5.5 points) or that were won by Obama in 2012 and by Donald Trump in 2016.

  4. Not all states have certified their results yet. This data is as of Dec. 7.

  5. David Wasserman’s results-tracking spreadsheet uses a 5.0 margin as a hard cutoff for determining what’s a swing state. For this analysis, I’m using any margin that rounds to 5 or less (that is, less than 5.5 points), which brought Georgia and Virginia into the realm of swing states, although they’re not marked that way in Wasserman’s spreadsheet.

  6. Maybe the 2016 approval ratings will prove to be more predictive.

  7. If you ran a simple linear regression in which you predicted Clinton’s margin knowing only Obama’s 2012 margin in a state and how Obama’s approval rating shifted relative to the nation as a whole, the only two states that would be predicted incorrectly are Michigan (barely) and New Hampshire. These were the two closest states in the election. Pennsylvania and Wisconsin would have been projected within a percentage point. If you didn’t take into account the shift in Obama’s approval rating and re-ran the regression knowing only Obama’s 2012 margin, you’d have missed five states, including Iowa, Michigan, Pennsylvania, Virginia and Wisconsin.

Harry Enten was a senior political writer and analyst for FiveThirtyEight.