The 2016 cycle has had a deluge of polling to try to get a picture of the current status of the election. Much of that gets disputed by all sides, depending on methodology, sampling, and — let’s face it — outcomes. That isn’t the only data in the field, though, and L2’s approach might offer a clearer understanding of how the electorate has expanded, and perhaps shifted. In two key swing states, L2 notes that Democrats appear to have picked up significant voter-registration advantages over the past year, modeled from their HaystaqDNA database (provided to us by L2).

Pennsylvania has added 275,901 voters to the rolls from November 4, 2015 to present. Of those, 46.8% have registered as Democrats, while only 36% have registered as Republicans. Bear in mind that Democrats usually dominate in Pennsylvania voter registration, but some tend to vote Republican in elections. Still, in 2012, the split was 45/35, slightly narrower than what L2 sees in new registrations.  The opportunity for this cycle was to improve the GOP’s numbers, which does not appear to have happened. Also, only 66.7% were white voters, below 2012’s turnout model of 78% white voters. Another 13.2% were of undetermined ethnicity, but Hispanics were confirmed at a higher rate (7.1%) than in the 2012 model (6%).

One potential piece of good news for Republicans — 75% of all registrations came from outside of the urban cores of the state. L2 mapped out the data, with blue representing Democrats and red Republicans:

In Florida, the news also doesn’t look terribly promising. L2 estimates that the state added 692,321 registered voters over the past year, with a Dem/Rep/Ind split of 33.4/28/35. That’s off from 2012’s 35/33 split in the last presidential election. Interestingly, while 18-29YOs make up a plurality of new registrants (as one might expect), that accounts for just 35.3% of all registrations, according to L2. The second-fastest-growing demo is the 50-69YO range, which accounts for 25.2% of all new registrations, about the same as the two demos between 30 and 49 years of age. That might help Trump, but another demo looks like particularly bad news: Hispanics make up 26.6% of all new registrations, far above their 17% share of the 2012 election turnout. Whites comprised 47.7% of new registrations, almost 20 points below their share of 2012’s turnout.

Unlike Pennsylvania, almost half (46%) of all new registrations came from an urban area:

What do these numbers tell us, and what are their limitations? It’s worth noting that L2 is among the elite firms that collect and maintain extensive databases on American consumers for marketing purposes, including elections. The Trump campaign used L2 at least for a time for its own strategic planning. (Full disclosure: L2 gave me limited access to their data for free while writing Going Red.) As with most data except for explicit and comprehensive censuses, these are estimates — but from files that include data on over 265 million adult consumers in the US, according to L2. Those who are inclined to dismiss media polling might find this more convincing, as L2 does not have a partisan or ideological axe to grind, and have to earn their living based on the accuracy of their research.

The data here does not state who will vote for whom in two weeks. It does suggest that Republicans have not moved the needle much in terms of party affiliation or attraction over the past year in two key states — one (Florida) where they have to win no matter who the nominee is, and the other (Pennsylvania) where their nominee was supposed to exert a specific appeal. If that translates into voting patterns — a big if, of course — then the GOP should prepare themselves for disappointment.