GA-PPP: Hillary leads in Georgia

[Mr. Morden]

THese state polls with Clinton should translate to her having a much bigger lead than she does in the few national polls. But as we have learned, state polls are more important...Hillary could really realign the map.

I think your second point answers your first point.  It's not necessarily that the national polls and state polls are in conflict, as it is that she realigns the map so much that the state and national polls match up in different ways than we're expecting.  For example, if the Democratic states trended Republican and the Republican states trended Democratic, then you'd see close-ish races everywhere in the country, and it could help explain why we're getting these polls in Republican states that show Clinton doing only marginally worse than the national average.

What we need is more polls in Democratic leaning states, to see if they're also closer than expected.  PPP already polled MN, and found Christie losing there to Clinton by only 6 points, when he's losing nationally by 5, so that helps support this theory, but we need more such polls from other Democratic leaning states.

I speculate here about why "Clinton vs. X" and especially "Clinton vs. Christie" might be skewing the map so much:

https://uselectionatlas.org/FORUM/index.php?topic=169110.msg3620773#msg3620773

In brief, I speculate that it could just be a consequence of there being an unusually large number of crossover voters.  Part of that could just be the fact that it's early, and partisan loyalties haven't locked in yet.  But part of it could also be the huge gender gap we're seeing in these polls, which could mean that some voters who might normally lean towards one party switch their preference to match up with their sex.  Since the male:female ratio is virtually the same everywhere, that would help explain why you'd get close races everywhere.

[Mr. Morden]

Also thought I'd mention that, like some of the other 2016 polls that include Clinton, this one shows a flattening of the age gap.  In the Clinton/Rubio matchup for example:

age 18-29: Clinton +4
age 30-45: Rubio +4
age 46-65: Clinton +5
age over 65: Clinton +3

[Mr. Morden]

I find it very hard to believe that Clinton is underperforming in the traditional democratic states. But we will see when we get more data

Well, underperforming in the sense of trend, not necessarily swing.  I mean, let's looks at what we have so far.  We've seen that Christie vs. Clinton produces some weird results, though PPP isn't polling Christie anymore.  They're still polling Clinton vs. Rubio though.  Here's the margin in all of PPP's polls of Clinton vs. Rubio since Jan. 1:

national, Jan. 3-6: Clinton +14
Florida, Jan. 11-13: Clinton +4
Minnesota, Jan. 18-20: Clinton +13
Texas, Jan. 24-27: Clinton +1
national, Jan. 31-Feb. 3: Clinton +8
Alaska, Feb. 4-5: Clinton +1
Louisiana, Feb. 8-12: Clinton +3
Georgia, Feb. 15-18: Clinton +3

If you assume that the true national margin is the Clinton by 8 seen in that Jan. 31-Feb. 3 poll, then relative to the 2012 election, the trend would be:

Florida: R+1 (Rubio's home state)
Minnesota: D+1
Texas: D+12
Alaska: D+11
Louisiana: D+16
Georgia: D+7

But, OK, maybe, as you say, the true national popular vote margin is larger than that.  Perhaps something more like the 14 point margin from the early January poll.  In that case, move each of those trend numbers a few points towards the Republicans.  You still have a decent Democratic trend in at least Texas, Alaska, and Louisiana.  That *has* to be made up somewhere.  That's the nature of trend.  It all has to balance out.  Minnesota, then would have a bit of a Republican trend.  I'm just suggesting, maybe there's a bit of a Republican trend many other Democratic leaning states as well.  Not sure, just a thought.