What does this really mean?
So everyone is familiar with the pollster.com graph above. That is a little old, showing C +3.9. Right now, it’s C +4.0. However, their customization tools don’t seem to work for me in that I get some weird results that make no sense if I take something out. So I wanted to point out some stuff.
First — let’s look at the most recent national polls:
Here are the last national polls per pollster.com:
C+3 (ARG)
C+2 (Yougov)
C+6 (NBC/WSJ)
C+5 (iCitizen)
Tie (Ipsos/Reuters)
C+1 (UPI/CVoter)
C+5 (NBC/SurveyMonkey)
C+4 Morning Consult
C+5 St. Leo
Average: C +3.4. Median (what Sam Wang uses) — C +4. Mode — C +5.
Now — I conveniently left out 1 poll. Rasmussen (T+5). Rasmussen has a very specific agenda — they are intentionally skewing their polls to show a Trump lead in order to defeat the poll aggregators and to give the Fox News crowd something to talk about. If you want to add them back in — the average drops to C +2.6 and the median drops to C +3.5. But they aren’t just a bad pollster, they are a pollster with an agenda. Unlike, say, PPP which actually has a slight R lean even though they themselves are liberals.
Let’s look at the last 100 polls and identify every one that has either a tie or a Trump lead:
Rasmussen T44 C39 (T+5) 9/20-9/21
Ipsos/Reuters T39 C39 (Tie) 9/15-9/19
Rasmussen T42 C40 (T+2) 9/12-9/13
UPI/CVoter T49 C46 (T+3) 9/5-9/11
CNN T45 C43 (T+2) 9/1-9/4
Rasmussen T40 C39 (T+1) 8/29-8/30
UPI/CVoter T48 C48 (Tie) 8/14-8/20
Penn Schoen Berland T44 C41 (T+3) 7/22-7/24
CBS T44 C43 (T+1) 7/22-7/24
CNN T48 C45 (T+3) 7/22-7/24
Morning Consult T44 C40 (T+4) 7/22-7/24
Gravis Marketing T51 C49 (T+2) 7/21-7/22
Rasmussen T43 C42 (T+1) 7/18-7/19
Out of the last 100 National polls at Pollster, Hillary leads in 87 of them. There have been 2 ties, and 11 Trump leads. Yes — there have been more Trump leads in September (5) than August (2) — but 2 of those 5 are from Rasmussen.
6 of the 11 Trump leads came between 7/18 and 7/24. That’s the Republican Convention.
4 of the 11 Trump leads came from one pollster — Rasmussen. 1 of them came from a pollster with an equivalent goal of changing the narrative — Gravis.
Ignoring Rasmussen and Gravis — let’s look at the other pollsters that gave Trump a lead or have a tie during the last 100 polls:
Morning Consult — was Trump +4 7/22-24. Most recent — Clinton 46-42 (C+4) 9/15-16
CNN — was Trump +3 7/22-24, Trump +2 9/1-9/4. That is their most recent poll
CBS — was Trump +1 7/22-24. Most Recent — Clinton 46-44 (C+2) 9/9-13 with the Times. Most recent standalone poll Clinton 47-41 (C+6) 7/29-31.
Penn Schoen Berland — was Trump +3 7/22-24. Most Recent — Clinton 45-40 (C+5) 7/29-8/1.
UPI/CVoter — was Trump +3 9/5-9/11. Most Recent — Clinton 48-47 (C+1) 9/12-18.
So — of the 6 polls other than Rasmussen or Gravis that have had a Trump lead, 4 of them came from pollsters that have shown a Clinton lead *after* they showed a Trump lead.
Finally — if you were to assume for a moment that the entire universe of polling was only the 13 polls showing a tie or Trump Lead — basically his best 13 polls out of the last 100, Trump would be polling at a whopping 44.7%, and Hillary would be at 42.6%. So totally cherry-picking the absolute best polls for Trump — he can’t break 45% and he has a 1.9% lead against Hillary.
Now imagine if you took these 13 polls out and looked at Hillary. This is where the customization at Pollster doesn’t work for me. If I take Rasmussen out, the polls go from 45.8-41.8 (C+4) to 44.5-40.7 (C+3.8). This even though the most recent poll is a T+5 at Trump 44 Clinton 39, and the Ras polls before that all had Clinton under her current average. So I can’t just back out the “bad” pollsters using their tools.
I believe that we should look at not only the polling averages, but take out all the noise. Popular opinion is not changing week to week. What changes week to week are who is willing and interested in being polled. I do not believe that Trump got a “bump” out of his convention. Few people who didn’t already support Trump, or who weren’t hard core Republicans, decided to support Trump because of “Lock her up” chants at the convention. Yet he got a “bump.” The Democratic convention was better, but I don’t think it moved the needle much either. I don’t think pneumonia moved the needle, and I don’t think that the Trump Foundation having paid off Pam Bondi or paying Trump’s legal bills move the needle. The country is basically set. Clinton is ahead by 3 points. She’s ahead in enough states to get to 270. She’s got potential to win in enough states to get to 365-370 if everything broke her way (which it won’t). The question is — who is going to get out the vote. Clinton has been invested in the ground game for a long time. Trump is finally investing. The Koch’s are putting money into GOTV. Saw this week where big donors have stepped up with nearly $20MM to help us GOTV. Will our voters show up? That’s the only question.
If they show up everywhere — it will be over early. If they pull a 2014 and don’t show up at all — Trump will be announced as the winner when the Dakotas and Idaho report.
