Not everyone that reads the Right on Daily Blog is a friend of mine on Facebook. I spent a couple months writing rants about polling there. So, let’s have a look at a nice post-mortem of the polling data here on the blog now that the results are basically finalized everywhere.
The conundrum for pollsters was their inherent bias towards Hilary Clinton and the basic problem of trying to figure out who was going to vote.
The most common thing biased pollsters do is screw with the samples. In the case of many polls, they were overweight with Democrats 7-10 points. Meaning, they were polling 7 to 10% more democrats than Republicans. They were also keen on oversampling women, who were less likely to support Trump than Clinton.
Let’s have a look at some of the National Polls the week before and the week of election day:
Name of Poll Week Before Week of Election
ABC/WaPost T+1 H+4 Democrat +7 Sample
Bloomberg H+9 H+3 Did not publish details
Economist / YouGov H+3 H+4 Consistent Poll Heavy on Dems
Fox News H+2 H+4 Consistent Poll, Heavy on Dems
Gravis Marketing H+2 H+4 Used bizarre model that featured a lot of “soft” voters
NBC/Wall Street Journal H+6 H+4 One of the consistently highest numbers for Clinton (hit +11 more than once)
Reuters H+8 H+3 They changed their methodology midstream when their numbers were showing Trump leading
CBS H+3 H+4 Consistent Poll
IBD H +1 T +2 They were one of the most favorable to Trump
Rasmussen H+3 H+2 They nailed it. H+2 actual result.
I did not include Monmouth University who were so biased against trump that they blew 2/3 of their polls. So did Marist.
Look at The italicized polls – they all showed a collapse of Gary Johnson and Jill Stein and a surge for Clinton. This is what happened. Johnson had been pulling 5 points or more in every poll, he got a little more than 3 as Republicans and Democrats returned home. This is why I was not as cocky on election day.
Rasmussen used a D +4.5 model and the actual, final result was D+3.8. This is why he nailed it.
IBD missed because of the late surge of dems.
Democrats turned out in massive numbers, higher than in 2012 when Obama won by 3.8. Dem turnout drills were motivating low propensity voters with loss of welfare, deportation and war fears.
Trump was able to win despite this because evangelical voters also voted in record numbers, plus grossly inaccurate exit polls missed the obvious fact that a lot more Democrats voted for trump than anyone wants to admit.
See the consistent 2 point difference between the final numbers and the actual result? This is the silent trump vote – I had figured it to be 2-3% and it ended up on the low end of that.
Remember, polls can only track the popular vote – which is why I am going to break some state polls down in a bit.
This election has also proven a polarization of population. Dark Blue States like California, New York, Illinois, Massachusetts and Maryland spot a dem candidate an 8+ million vote lead. Trump took advantage of the system in place and beat Clinton in 30 states. Let that sink in… welfare recipients know where the easy money is. Illegal aliens know where the sanctuaries are.
The State Polls were awful. The pollsters in my opinion knew that the state polls were more important to the narrative about Hillary Clinton and what happened with them is legendary. Those are coming up next…