Trump Trades Traditional Republicans for Swing-State Voters
Is Donald Trump shifting the Electoral College map?
Before the general election started and polls really began to roll in, it was possible to argue the point either way. It would have been reasonable to think that voters mostly stay loyal to their parties and would thus produce a Hillary Clinton vs. Trump map that mirrored a Mitt Romney vs. Barack Obama or George W. Bush vs. John Kerry map. On the other hand, one could argue that Trump’s appeal to downscale whites would shake up the old coalitions and cause significant changes.
Now that we’re about a month from Election Day, it’s possible to assess which of these arguments is actually correct. And it appears that both have merits -- that is, the broad features of 2016 landscape bear a strong resemblance to those of the 2012 map, but there are significant shifts in a few swing states.
The Map Is Mostly Stable
Current polling shows that the familiar Obama-era and Bush-era political maps still mostly hold. Republicans have a strong base in the South and the Mountain West, while Democrats mostly draw their electoral votes from the West Coast and the Northeast. Some of the quadrennial swing states (e.g. Ohio, Florida, Pennsylvania) remain close.
This makes sense. For the most part, partisan Democrats back Clinton and Republicans back Trump. Both have some issues within their coalitions (e.g. Clinton’s unpopularity with millennials and Trump’s trouble with college-educated white voters) but much of the map and the party coalitions remain stable.
But Trump Has Been Trading Margins in Red States for a Swing-State Advantage
That being said, there have been some significant changes in how some swing states are leaning this cycle.
This table focuses on partisan lean and how it’s changed from 2012 to 2016 (if you’re already familiar with the Cook PVI and don’t care about the math that produced these numbers, skip this paragraph and the next two). The second column shows the Cook Partisan Voting Index (PVI) of each state. Essentially, the Cook PVI uses national and state results from the last two presidential elections to calculate how each state leans relative to the country as a whole. For example, if a Democratic presidential candidate won the popular vote nationally by five points, he or she might win a D+2 state (which leans two points toward Democrats) by seven points. And if a Republican won the popular vote by three points, he or she might win a D+2 state by one point.
The 2016 lean is calculated somewhat differently. For each state I collected two-way general election polling from this cycle, subtracted it from the two-way RealClearPolitics national average on the start date of each poll, and then averaged the results. In other words, the third column shows how much better or worse Trump and Clinton are performing in each state compared to how they poll nationally. All polls from the beginning of this election cycle to the first debate are included. Polls after the first debate were excluded because post-debate national polls were released later than state polls, which could produce a bias.
There are obvious caveats here (e.g. polls have a margin of error; there’s no adjustment for pollsters that tend to lean toward one candidate or another; outliers can skew averages, which appears to have happened in North Carolina and arguably Maine; some states, such as Minnesota, didn’t have enough polls for me to confidently calculate the lean; etc.) but these measure are rough guidelines and some amount of error is expected.
The table focuses on how the states lean relative to the nation as a whole, and how the partisan lean of each state has changed from 2012 to 2016. Trump has managed to push a number of swing states -- Iowa, Nevada, Maine and most notably Ohio -- significantly in his direction. Clinton has gained ground in Virginia.
Most of the other swing states haven’t shifted much (because of the error in these measures, I think of movements greater than two points as significant). In other words, Trump has shifted the map toward him where it counts most -- in the swing states that decide the election.
The education gap explains a sizable amount of this movement. States with a large concentration of whites without college degrees (e.g. Iowa, Maine, Nevada, Ohio) have shifted toward Trump while Virginia (which has a high percentage of whites with a college degree) has moved toward Clinton.
The relationship here isn’t perfect and it doesn’t explain everything. Colorado has moved little despite having a large number of college-educated white residents. And in some of the more racially diverse states, there are other groups of people who contribute to how the state does or doesn't move.
Trump seems to be benefiting from this tradeoff (shedding college-educated whites while appealing to non-college educated whites) in swing states, but his position has weakened in some redder states. For example, Trump is leading by 7.6 points in Texas, according to the RCP average, but Romney and McCain won the state by 16 points and 12 points, respectively. Trump also seems to have traded Mormons for blue-collar swing-state voters. He leads in Utah by 13 points, while Romney and McCain won the state by 48 points and 28 points, respectively.
This trade-off could be a net positive for Trump. In a close election, some extra support in swing states would be worth giving up some votes in safe states. But right now the election isn’t close enough for this advantage to make a big difference, and Trump would have to make up ground very quickly for this to push him over the top on Election Day.