Another day – another blatant example of the media manipulating polls to discourage Trump voters.
Now the Atlanta Journal Constitution does the same pretending that Hillary is going to win Georgia.
Polling firms contact registered voters. The number of voters who answer the phone and agree to the survey are put into a number called the “unweighted base.” The polling firm then applies their own “formula” based on who they think is going to vote. In the example below, 278 Republicans in Georgia answered the phone for a poll in May. The AJC’s polling firm thinks that’s too many Republicans, so they move that to 245, “adjusting” the numbers with an “average weighted response.”
2 1/2 months, later, the same polling firm substantially altered their “secret formula.”
Compare these two crosstabs. This is a link to the Aug 1-4 poll. Here is the poll by the same outfit back in May. The first number is the raw data. The second is the weighted sample. Talk about a giant leap!
There are 5 million registered voters in Georgia. That 6% of Republican voters represents almost 200,000 Republicans. Keep in mind the change in the estimated percentage of Republicans voting is a made-up number. It has nothing to do with the voter preferences. This poll has Trump at 81% of the Republicans contacted. That includes those who say they are sitting out, voting third party, or voting neither candidate. The difference in the results is in the weighted average, which is the Atlanta Journal Constitution’s polling firm putting it’s thumb on the scale. To believe this poll is remotely true, you’d have to believe that hundreds of thousands of registered Republicans in Georgia plan to sit out the election entirely.
In September 2014, the AJC ran a poll claiming that Republican Nathan Deal was at 42% and tied with Jason Carter in the governor’s race. The result was a 52% blowout for Deal.