When most people think of battleground America, they think of Florida and Ohio, two of only three states (along with Nevada) that have voted for the winner of every presidential election since 1996. They tend not to think of Pennsylvania as a classic “swing state” — it has voted for the Democrat in every election since 1992, and it didn’t even crack the top 10 in 2012 campaign ad spending.
But in 2016, Pennsylvania could be the keystone of the Electoral College and the ultimate arbiter of whether Donald Trump or Hillary Clinton resides at 1600 Pennsylvania Avenue.
The term “swing state” can be a bit fuzzy. For example, we often call Michigan and North Carolina swing states. But if a Republican is winning Michigan, they’ve likely already won the election; the same is true for a Democrat winning North Carolina. In other words, after all the sturm und drang, these states tend to be icing on the winner’s cake. But a few years ago, Nate Silver articulated a much more valuable concept: the “tipping point state.” This is defined as the state that provides the presidential winner his or her 270th electoral vote when all states are rank-ordered by his or her highest to lowest margin of victory. Over time, this state has shifted: In 2000, it was Florida. In 2004, it was Ohio. In 2008, it was Colorado.
In 2012, President Obama’s three closest wins were Florida (by 0.9 percentage points), Ohio (by 3.0 points) and Virginia (by 3.9 points). But even if Mitt Romney had flipped all three states to his column, he would have still fallen four Electoral College votes short of the White House. The “tipping point state” was a virtual tie between Colorado and Pennsylvania, which both voted for Obama by about 5.4 percentage points.
In 2016, Florida and Ohio will likely remain necessary for Trump to obtain 270 electoral votes. Predictions that 2016’s Clinton vs. Trump showdown could “scramble” the traditional red/blue map are probably overblown; political scientists1 John Sides and Andrew Gelman have found that over time, year-to-year swings between the states are getting smaller. That said, the ordering of the battleground states — from most Republican leaning to most Democratic leaning — is unlikely to stay the same, especially because 2016 is an open election.
I’d argue Pennsylvania has leapfrogged Colorado and Virginia as the next most winnable state for Republicans. In fact, it may be on pace to claim sole “tipping point” status.
I arrived at this conclusion by calculating each state’s “six-cycle trend” — a measure of how much more Democratic or Republican-leaning each state has trended relative to the nation as a whole since 1992. To do this, I compared Obama’s 2012 two-party vote share in each state to the Democrat’s share in each of the five prior elections, then took the average of those differences, controlling for national swings and the number of four-year intervals.2
For example, California voted 5.1 percentage points more Democratic than the nation overall in 1992, but 9.9 points more Democratic than the nation in 2012, for a trend of 0.96 percentage points towards Democrats every four years. I recalculated this differential versus 2012 for every year between 1996 and 2008, and took the average. California’s resulting “six-cycle trend” was 1.46 percentage points towards Democrats. By contrast, West Virginia’s was 4.66 percentage points towards Republicans.
As it turns out, Colorado and Virginia are among the top 10 fastest Democratic-trending states in the nation — they are, respectively, getting about 0.9 percentage points and 1.2 points more Democratic-leaning compared with the country every four years. By contrast, Pennsylvania has gradually migrated in the opposite direction. It’s gotten about 0.4 percentage points more Republican every four years.
Projecting this trend forward another four years from 2012’s results would reorder the existing battleground states on the 2016 electoral map.
Pennsylvania, where the projected Democratic share of the two-party vote would drop to 52.3 percent, would become the next most winnable state for Republicans after Florida (50.6 percent) and Ohio (51.9 percent). In fact, after Pennsylvania, the next most winnable states for Trump would be New Hampshire (52.9 percent) and Iowa (53.0 percent), followed by Virginia (53.2 percent), Wisconsin (53.4 percent) and Colorado (53.7 percent).
Western Pennsylvania is driving the state’s rightward drift — its voting patterns now resemble greater Appalachia’s more than those of the Philadelphia suburbs. Once dominated by steel towns and union Democrats, the region has reveled in a fracking/natural gas boom that has more recently experienced a downturn and has revolted against EPA regulations. Obama’s infamous “bitter clingers” remark in 2008 didn’t do Democrats any favors either.
It’s true that Philadelphia and its well-educated, white-collar suburbs are trending blue. In 1992, Bill Clinton took 59 percent of the vote in the eight counties that make up the Philadelphia media market; in 2012, Obama took 63 percent. But the rest of the state accounted for 58 percent of all votes in 2012 and is trending red even faster. In 1992, Clinton took 53 percent outside of the Philadelphia media market. In 2012, Obama took just 45 percent.
This growing cultural distance between these two Pennsylvanias — some would say Philadelphia vs. “Pennsyltucky” — points to long-term problems for Democrats:
|SHARE OF TWO-PARTY VOTE|
|YEAR||DEMOCRATIC CANDIDATE||PHILADELPHIA MEDIA MARKET||REST OF PENNSYLVANIA||DIFFERENCE|
The challenge for Trump here is still considerable. As Cook Political Report National Editor Amy Walter pointed out, “For every blue collar and disaffected voter Trump may pick up in Western Pennsylvania, he could lose two or more women in the Philadelphia suburbs.” Furthermore, in addition to Philadelphia’s large African-American population, there has been modest Latino growth, particularly in Lancaster, Reading and the Lehigh Valley.
But there are a few key drivers behind why Trump is likely to perform better in Pennsylvania than in Colorado or Virginia, regardless of the final national outcome:
1. The Economy — Voters are more likely to turn on the party in the White House when they perceive the economy to be doing poorly. At the moment, the economy is doing a lot better in Colorado and Virginia than it is in Pennsylvania. Gallup found that in 2015, Colorado and Virginia residents had the seventh- and eighth-highest economic confidence in the nation. Pennsylvania residents’ economic confidence was well below average.There’s also a much larger blue-collar manufacturing sector in Pennsylvania, which plays into Trump’s protectionist, “Make America Great Again” mantra. The Bureau of Labor Statistics reported that as of March 2016, 10.1 percent of Pennsylvania‘s nonfarm labor force was employed in the manufacturing or mining/logging sectors, compared with 6.6 percent in Colorado and just 6.1 percent in Virginia.
2. Demographics — Older, white voters without college degrees are the bedrock of Trump’s coalition, and Pennsylvania is the sixth-oldest state in the nation. As of 2014, its median age was 40.7 years, three years older than in Virginia and four years older than in Colorado, according to the census. Pennsylvania also has a much whiter electorate. According to the Census, as of 2014, 83 percent of its eligible voters were non-Hispanic whites compared with 78 percent in Colorado and 70 percent in Virginia.But here’s the kicker: According to the Census, just 29 percent of non-Hispanic whites age 25 and older in Pennsylvania held at least a bachelor’s degree, compared with 43 percent in Colorado and 39 percent in Virginia. That’s a massive disparity, and whites without a college degree have been among the fastest GOP-trending groups nationally. All of these arrows point to Pennsylvania as a much more favorable electorate for Trump.
3. Voting Laws — Since the 2012 election, Colorado and Virginia have taken steps to expand participation. In 2013, Colorado Gov. John Hickenlooper signed legislation that allows same-day registration, which could reduce barriers to young and Latino first-time voters. This April, Virginia Gov. Terry McAuliffe signed an executive order restoring voting rights to 200,000 felons who have served their sentences, a move that The Upshot’s Nate Cohn estimated could boost Democrats by half a percentage point.Meanwhile, Pennsylvania’s voting laws haven’t meaningfully changed. Felons who have completed their sentences have long been able to vote in the Keystone State, and in 2014, then-Gov. Tom Corbett announced he wouldn’t appeal a state court ruling striking down the GOP legislature’s strict voter ID law, which was supposed to take effect after 2012.
Don’t get me wrong: Florida and Ohio are still likely to be crucial in any scenario in which Trump wins a majority of Electoral College votes. But if Clinton wants to keep him out of the White House, she may need to build a Pennsylvania firewall. Perhaps that’s a good reason Democrats are holding their convention in Philadelphia in 10 weeks.
Listen to the latest episode of the FiveThirtyEight politics podcast.
CORRECTION (May 17, 9:52 a.m.): A previous version of this article incorrectly described the states that have voted for the presidential winner in every election since 1996. Florida, Ohio and Nevada all have, not just Florida and Ohio.