How did the fake news media polls do with GA6? Not very well. The final margin was Handel by 3.8.
Here were the final predictions by the polls:
Atlanta J. Constitution: Ossoff +7
Fox: Ossoff +1
WSB Landmark: even
Trafalgar (2nd best poll in 2016, called PA & MI) is not considered a “mainstream” polling organization. Trafalgar had Handel +2. To be fair, Trafalgar’s previous poll had Ossoff by 3, so its trend was moving the right way, very fast.
If you include Trafalgar, the average was an error of 3, with only Trafalgar in Handel’s direction. If you exclude Trafalgar, the average error is 3.4, all in the Democrat’s direction.
What about us “untrained” people or those not accepted by realclearpolitics.com? Here is my band of “Deplorables” on Twitter:
Deplorable Richard Baris, People’s Pundit Daily: This was the most accurate poll of 2016. Baris admitted he did not poll right before the election): even.
Me: Around 9 a.m. on election day, before any votes were counted, I changed my “Handel +1-2” to “Handel +3-4.”
“Deplorable Greg,” the Obgyn on Twitter who was part of our election night team got it closest of all: Handel +5.
So of the “established” pollsters they were all off by an average of 5 points, and even if you throw out the older AJC poll, they were off 3.4.
Trafalgar was low by 1.8.
I was low by .8 (low estimate) to .2 (high),
Greg was high by 1.2.
Conclusion: Both myself and Trafalgar were still low, and only Greg slightly overshot, but we all averaged just over 1 point error. Generally speaking, there is still more Trump support out there that is not being counted or properly estimated. This is what I have been picking up in the voter registration changes since November, where out of 11 battleground states, all but two (OR and CO) have trended Republican, and some significantly so.