No, the battleground polls aren't close

Today, the news media claimed that polling in three battleground states showed that Donald had either closed the gap or moved ahead of Clinton. This is incorrect. Let's take a look, shall we?

Florida
Quinnipiac's poll says Donald is up, 42% - 39%. Their poll was weighted 31% Republican, 29% Democrat, and 39% Independent / other. A look at Florida's most recent registration data (May 2016) shows that registrations break 36% Republican, 38% Democrats, and 26% unaffiliated / no preference. If you apply Quinnipiac's cross tab data to Florida's most recent registration data, Hillary remains on top, 44% - 43%.

Pennsylvania
Quinnipiac's poll says Donald is up 43% - 41%. Their poll was weighted 34% Republican, 35% Democrat, and 31% Independent / other. A look at Pennsylvania's most recent (July 11, 2016) shows that registrations break 38% Republican, 49% Democrat, 13% unaffiliated. If you apply Quinnipiac's cross tab data to Pennsylvania's most recent registration data, Hillary remains on top, 47% - 41%.

Ohio
Well, Ohio doesn't have breakdowns of voter registrations by party* so I can't really share with you how Quinnipiac's polling is wrong, but suffice to say, there is a trend to add too much weight towards Republicans.

Finally, Gallup's Party Affiliation tracker, well over a decade long, currently shows Democrats + Leaners at 48% and Republicans + Leaners at 42%, and if you've paid attention to my posts in the past, this tracker tells you a lot about the direction of the country and which party will win the White House. In 2004 the lead for party affiliation flipped nine times between that January and voting day; in 2008 Democrats + Leaners led all the way from August to voting day; in 2012 Democrats + Leaners led (or tied) all the way to voting day, for 13 months. This season, Democrats + Leaners have led since last September.

* -- A note to Ohio's SoS: You have a publicly available zipped file with the names, birthdate, party registration and mailing address of every registered voter. I stumbled upon it while trying to find aggregate registration data. You need to remove it, immediately!