As analysts and campaign staff try to make sense of the polls and the state of the 2016 campaign, one question looms over everything: Is there a hidden Trump army of working class whites laying in wait in battleground states?
The hidden Trump army theory goes something like this. Polls, particularly those showing Hillary Clinton with a lead, are missing a big group of disaffected voters who have long been quiet, but who are going to engage this fall and push Donald Trump into the White House.
The idea isn't out of left field. Throughout his campaign, Donald Trump has made a point of reaching out to alienated voters who feel left behind in the new global economy. A great many of those voters fall into a specific demographic category, whites without a college degree, and those voters make up big parts of crucial states on the electoral map, such as Florida and Pennsylvania.
The argument isn't completely new. Long before the Trump campaign, Republicans have been arguing about the existence and size of a "hidden white vote" that could be activated with the right candidate to win the presidency for the GOP.
Digging into the data, however, raises some questions about how big or real that army is.
To get a sense of the potential hidden-Trump impact we looked at key counties in Florida and Pennsylvania. We singled out the five counties in each where whites without a college degree made up the largest percentage of the 18 and over population - those are places that are demographically predisposed to lean toward Trump. We then singled out five counties with the largest percentage of minorities in the 18 and older population - those places are least likely to lean toward Trump
We looked at how voter registrations in those counties had changed since last fall since last fall. The numbers are not especially promising for the Trump campaign.
In Florida the five most Trump-friendly counties had seen a 4.5% uptick in voter registrations. But that number lagged behind the 5.7% bump in registrations in the least Trump-friendly counties.
In Florida, the percentage increase in the five most Trump-friendly counties was also below the increase in voter registrations statewide, 4.7%.
In Pennsylvania the numbers are slightly less encouraging for Team Trump. Again there is a small increase in the counties that should be best for Trump (2.79%), but there is a much bigger increase in the racially and ethnically diverse counties that will likely be tougher terrain for him (5.62%).
And, like Florida, in Pennsylvania the most demographically friendly counties for Trump are far below the 4.8% increase in voter registrations overall.
None of this is conclusive, of course. We can't say for certain that bigger voter registration numbers here are necessarily bad news for Trump. Maybe it's the Trump voters that are registering, even in the counties that are least demographically suited to him.
But the bigger minority populations in those places should be of concern to the GOP nominee. Mr. Trump's demographic strengths (among working-class whites) and weaknesses (among minorities) have been pretty consistent in the polls during this campaign. And often when a candidate works hard to pull one demographic slice closer, he says and does things that push another slice away.
Furthermore, beyond demographics, the counties above follow a very consistent pattern with their 2012 vote.
Four years ago, Republican Mitt Romney won all 10 of the counties (five in Florida and five in Pennsylvania) that have favorable Trump racial and ethnic demographics. President Barack Obama won 9 of the 10 counties that have unfavorable Trump demographics - all but Florida's Hendry County, where turnout was far below the state average.
So, yes, there may indeed be a hidden Trump army lurking in the 2016 electorate. But going by the numbers, if there is one it is hidden extremely well right now.