GA - Atlanta Journal Constitution: Clinton +4 / Clinton +3 (w/ Third Parties)

[Adam Griffin]

Uh, I hate to actually give any credence to SERIOUSLY? or whatever, but there is a grain of precedent for this: 2014. Carter (and to a lesser extent, Nunn) both had surges  in July/August, with Carter's surge in late July being quite massive in polling for some time - multiple polls showed him up by 7-8 points (this was largely due to Deal's scandals, if I recall correctly). In the end, it faded and Carter was only up in a small fraction in all of the polls after that.

If this poll is correct, then there is some optimism: Clinton doing well with the groups that tend to break 2:1 or 3:1 Republican at the end. However, there's 9 percentage points more potential Trump support at the end of the day in the four-way than Clinton support (and 16% undecided in the head-to-head). A lot can change.

[Adam Griffin]

^^^ But in the meantime...

 

[Adam Griffin]

I caution those using 2014 (and seemingly -only- 2014) to debunk the polls this year--2014 was a midterm, midterms have lower turnout, and that always favors the Republicans. I don't at the moment think Georgia will flip (it would need to be consistent over a long period of time for me to believe it) but I'm going to be dismissive based on a single year's polls being wrong due in part to low turnout either.

Georgia has a lot of junk pollsters. Kind of like Michigan. I've seen this pattern more than just 2014. It happens over and over again. Republicans get underpolled and rally late.

Georgia doesn't have many pollsters. There are only three prominent and consistently active ones (IA, RS & Landmark). Landmark is mediocre; if anything, Rountree may very well consistently underestimate GOP support in order to scare GAGOP into getting off of their asses. As best I can tell, IA is the only truly garbage-tier pollster. I couldn't believe that they had the gall to publish that one poll in 2012 that showed Romney winning by 20+ points.

As far as the trend goes, 2014 is the only cycle where polling showed a fundamentally different race at one point, but this was a broader phenomenon that we saw throughout the country. The polls in 2008, 2010 and 2012 did not show Democrats ahead at any point (other than the odd, occasional poll - usually early in the cycle). The consistent trend in GA polling is that you basically take 70-75% of the undecideds in polls one to two weeks out and add them to the GOP column. This usually means that a Democrat would need to be up by 3 in late polling in order to have a shot in a presidential election and a bit more in any other election (to avoid run-off territory).

[Adam Griffin]

^^^ Yeah, it really depends on which counties you're talking about. If we use this definition, then they are largely blue-collar for the most part (with notable exceptions in Paulding, Coweta & Carroll, perhaps; Newton is definitely still in transition).

Honestly, I think many definitions of exurban/suburban for the broader ATL metro are overextended; many exurban counties are in fact still quite rural in nature, and many urban counties (if you dare to use the classification "suburban" alongside the other three) are really suburban in nature.

[Adam Griffin]

So does anyone know what the AJC defines as "Metro Atlanta" versus "Atlanta Exurbs"?

You have so many counties that are part of the SMSA, but in order to get these numbers my thought would be that Hillary is well exceeding Obama '08 numbers in Fulton, De Kalb, and Clayton.

Narrowly ahead in Gwinnett, close to even in Cobb/Henry and running significantly better than Obama in "Exurban Atlanta".

One of the other interesting things in the crosstabs indicate a (44-34 Clinton) lead in SE Georgia.... Maybe my math is all wrong but I'm trying to figure out that is possible, unless Hillary has dramatically improved numbers of rural Whites in SE GA.

Some time ago, I tried to reconstruct ABT SRBI's regional boundaries based on their cross tabs and population distributions throughout the state (first map). The map below was the closest I could come up with, with the original not splitting the "Exurbs" and "Metro". As far as population and geographic boundaries balanced were concerned, these boundaries made the most sense.

However, I went ahead and tried to break what I previously had as "Metro" (half of the state) into the two equally-sized groups of "Exurb" and "Metro" (second map). They must split some counties, or else the crosstab proportions combined with geographic common-sense just wouldn't make any sense.

The biggest difference I'm noticing between 2008 & this poll is in the Southeast: those numbers might be a bit unbelievable in terms of the difference. The rest more or less makes sense given the nature of the race.

In almost every area in the second map, Clinton is currently polling roughly where Obama finished in 2008, with Trump being 10 points below McCain's final total (with the exception of SE).

[Adam Griffin]

^^^ OK, I adjusted my original estimates and redid what is apparently their exact boundaries. A couple of counties (like Baldwin) were not in those crosstabs and with them being on the boundary or 2 or more regions, I just put them in the region that made the most sense. Some of those boundaries they have are stupid, in my opinion. Nevertheless, the 2008 numbers in these regions compared to what I tried to guestimate were hardly different (save for the exurbs).
 
At any rate, here are the boundaries and the results:



The Southeast (particularly along the coast) is the most "moderate" portion of Georgia in terms of the aggregate of political attitudes; everywhere else that seems moderate based on election results is merely the result of a roughly split, very polarized electorate. Because of this, it's not a huge shock to see Clinton performing better here in the poll relative to 2008 (based on margin) than in any other area.

Overall, though, Clinton's margins relative to Obama's in 2008 are more improved in southern Georgia than in northern Georgia, and this isn't necessarily surprising. When people talk about white voters "coming home" and the old Dixiecrats that might still be out there, they are disproportionately going to be in South Georgia (and perhaps even more so in that SE region, although they'll be further inland/in the Onion Belt rather than along the coast; think Barrow's CD).

You also see the highest number of undecideds in these two regions (20% in "Southwest" & "Southeast" combined, compared to 13% in the other three regions combined), and that very well may be due to the older white voters who are contemplating returning to the fold and going back to their old ways instead of instinctively opposing Democrats like they have under Obama. At the end of the day, though, they may very well vote how they did in 2008/2012; it's definitely normal in Georgia for the bulk of undecideds to pour into the GOP at the last minute, but seeing a higher-than-average share of them clustered in this part of the state gives me a bit of hope.

At the end of the day, though, this could just be funny margins of error and the result of small sample sizes in this part of the state.

[Adam Griffin]

Wouldn't the geographic categories be based on telephone area codes?

They don't appear to be, since every region overlaps with at least three area codes (except for North, which is wholly contained within 2; its 2 are also represented in SE & SW). They're likely just using a voter file and randomly sampling it down to (at least somehow accurately) represent each county/area.

[Adam Griffin]

If Clinton wins Georgia, it won't help Jim Barksdale, who is a underdog Democrat running against Johnny Isakson. I could see Clinton/Isakson split ticket voters. Can anyone else here see that, or can underdog Barksdale cause trouble for Isakson?

If Kasim Reed or Jason Carter had ran this year for Senate, Isakson would be in a lot of trouble.

He'll underperform her by 2-3 points, I'm thinking...which is about as much of a difference that can exist at this point between any two (white) statewide Democratic candidates who are running in the same election in polarized, inelastic GA. If it were Chambliss, it might be a different story - he was never really all that liked - but Isakson in particular is a different story.