Trump approval ratings thread 1.5

[pbrower2a]

I want to point out, as I occasionally do, that outliers are important and still tell you something. If you have 5 polls that are within 3 points of each other and then you have another one that's 10 points higher, yes the latter is probably overestimating the support significantly. But it also means the other 5 are probably collectively underestimating it.

Sudden and severe change in the reality of approval ratings will first appear as seeming outliers. Events can shape the image of the President (or a Representative, Senator, or Governor). So imagine that Governor "Ima Grafter" has been limping along with a 42-47 set of approval and disapproval ratings and the largest big-city newspaper breaks a story about the Governor taking kickbacks for highway contracts. That makes the news at 6PM on TV one day in the preceding week and does not go away. That weekend some pollster polls the state for the Presidential election as usual and the 42-47 approval pair becomes 39-54. The new poll is out of line for other polls, but which is the reality? The older polls may have suddenly become obsolete even if they were valid at one time.

So let us suppose that a state that gets polled a lot, in which the real objective is the Presidential race of 2020 and a question about approval for the Governor is an afterthought gets polled this weekend after the President has some really-bad stories breaking earlier in the week. Let us suppose that the state is Arizona, and the state that has typically shown itself as a battleground  (polls are typically close) suddenly has an approval poll that shows the President going from 47-51 (still consistent with a hot contest) to 41-57.

I am cautious enough to suggest that a change outside of the margin of error could be an outlier -- or that it could reflect a known event. Maybe the Senate race of 2020 is still close, and the approval rating for the (R) Governor slips very little.

OK -- so other pollsters start asking Arizona voters some questions... and we see much the same. The first poll is no outlier; it reflects real change before any other poll suggests such.  

Scandals emerge and political campaigns collapse. People start to give up on politicians and causes. Polling is like a map; it shows how things are at the moment. A map of Europe from 1910 is at best a period piece or a reflection of historical realities that may no longer exist. For example, the Romanov dynasty no longer rules over Russia (not since 1917, anyway), there are seven independent republics dominating territory that used to be called the Austro-Hungarian Empire (the rest of which has become parts of Italy, Serbia, Poland, Romania, and Ukraine), and Germany no longer extends north and east along the Baltic to a city called Memel (it is now Klaipeda, Lithuania). But you would expect the map of Europe to change greatly over 110 years, would you not, in view of some of the destructive wars and the vengeance that victors have exacted upon the vanquished, would you not?  

Do not expect to take a cross-country trip on US 80 from Savannah, Georgia to San Diego, California.  The route now ends in Dallas, Texas.  

[GeorgiaModerate]

I want to point out, as I occasionally do, that outliers are important and still tell you something. If you have 5 polls that are within 3 points of each other and then you have another one that's 10 points higher, yes the latter is probably overestimating the support significantly. But it also means the other 5 are probably collectively underestimating it.

I agree with this to an extent, but I'd phrase it as the other 5 "may be" collectively underestimating it.  "Probably" is a little too strong unless you start to get more than one outlier in the same range.

[emailking]

I want to point out, as I occasionally do, that outliers are important and still tell you something. If you have 5 polls that are within 3 points of each other and then you have another one that's 10 points higher, yes the latter is probably overestimating the support significantly. But it also means the other 5 are probably collectively underestimating it.

I agree with this to an extent, but I'd phrase it as the other 5 "may be" collectively underestimating it.  "Probably" is a little too strong unless you start to get more than one outlier in the same range.

I'll rephrase it like this. It's more likely than not that true support is larger than the average of the other five.

[Skye]

CNN:

Nevada Trump approval (RV):
Approve 44
Disapprove 49

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_sc.pdf

South Carolina Trump approval (RV):
Approve 51
Disapprove 46

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_nv.pdf

[JG]

CNN:

Nevada Trump approval (RV):
Approve 44
Disapprove 49

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_sc.pdf

South Carolina Trump approval (RV):
Approve 51
Disapprove 46

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_nv.pdf

Net approval seems high for Nevada and low for South Carolina.

[pbrower2a]

CNN:

Nevada Trump approval (RV):
Approve 44
Disapprove 49

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_sc.pdf

South Carolina Trump approval (RV):
Approve 51
Disapprove 46

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_nv.pdf

Net approval seems high for Nevada and low for South Carolina.

CNN has reputable polling. It is Nevada that is really tough to poll, likely the result of many people being unreachable by the polls. The story is that Greater Las Vegas is the literal "city that never sleeps", and that people are much more likely to be at work when poll calls are made. Hispanics are the ones most likely to have shift jobs and weekend jobs that keep them from answering calls for political polls, and that would of course distort some polls. Nevada was close in 2016, and it closes its polls very late in the evening... people might forget to vote should the Democratic nominee is winning Virginia early or Colorado (a surprisingly-good analogue for Nevada in polling) is a quick call after Michigan, Pennsylvania, and Wisconsin are quick calls for the Democratic nominee.

It could be that the South Carolina poll is more accurate. I saw a report that suggested that new voters in the Palmetto State heavily lean D. Note that polls for Trump approval have been tepid at times in Kentucky and Missouri, and that neighboring Georgia and North Carolina are really close to even or lean a little against Trump.

...Not that I predict polling results, I am tempted to believe that statewide polling this weekend could look absurd (outliers?) unless one follows the news.

Trump approval:

40% or less or disapproval over 52%
41-44% or disapproval over 50%
45-49% and negative


Note: these are colored green to the same shade if I believe that an independent or third-party nominee has a chance to win the state, but a Democrat has no meaningful chance. That is Utah for now.

tie (white)
45-49% and positive
50-54%
55% or higher

[Nhoj]

CNN:

Nevada Trump approval (RV):
Approve 44
Disapprove 49

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_sc.pdf

South Carolina Trump approval (RV):
Approve 51
Disapprove 46

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_nv.pdf

Interestingly both of those results line up with the Gov races in 2018.

[Dr. Arch]

CNN:

Nevada Trump approval (RV):
Approve 44
Disapprove 49

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_sc.pdf

South Carolina Trump approval (RV):
Approve 51
Disapprove 46

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_nv.pdf

Interestingly both of those results line up with the Gov races in 2018.

Never trust Nevada or Michigan polling.

[Dr Oz Lost Party!]

CNN:

Nevada Trump approval (RV):
Approve 44
Disapprove 49

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_sc.pdf

South Carolina Trump approval (RV):
Approve 51
Disapprove 46

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_nv.pdf

>Nevada polling

Add 2 points to disapprove and deduct 1 from approve and I’ll believe it.

[Person Man]

CNN:

Nevada Trump approval (RV):
Approve 44
Disapprove 49

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_sc.pdf

South Carolina Trump approval (RV):
Approve 51
Disapprove 46

http://cdn.cnn.com/cnn/2019/images/09/28/rel1_nv.pdf

>Nevada polling

Add 2 points to disapprove and deduct 1 from approve and I’ll believe it.

This could mean that if Trump has another "good enough" night, and seeps through the "blue wall", that could mean that Nevada still will not be called by daybreak and that though he might of had one of the best nights in decades for a Republican (somewhere between Bush's and Obama's reelection margins in both way), he still will have done perhaps only average in South Carolina.

What this probably means is that South Carolina isn't overwhelmingly Republican anymore, probably never really was, and now certainly isn't. It will still not vote for a Democrat for years if not decades.

Nevada continues to be a hard to poll state that is becoming increasingly inelastic. Much like Colorado is. The only way these two states have a reasonable chance of going Republican is if both of these states have  a combination of two trends converging. Those two trends are pollers figure out who actually votes in those states and that those they expect to vote actually vote or don't vote.

[GeorgiaModerate]

AP/NORC, Sep. 20-23, 1286 adults (1-month change)

Approve 40 (+4)
Disapprove 59 (-3)

Strongly approve 24 (+3)
Stronlgy disapprove 42 (-4)

This poll was completed before the Ukraine transcript was released.

[pbrower2a]

Nevada/South Carolina:

I said enough about Nevada. I have never been in South Carolina, so I cannot speak from experience. On the other hand, a President who won a state 55-41 looks likely to win it if his approval is at 51% and that does not collapse due to events (I am not predicting such in polling -- yet... we shall find out soon enough)... but that decline of approval bodes ill for the president nationwide, especially when the two neighboring states (GA and NC) seem to have turned against the President. 

[pbrower2a]

High Point University, North Carolina. Adults, September 13-19.

Approval: Donald Trump, 42-50

Governor Roy Cooper (D) 43-30

Senator Tom Tillis (R) 27-38

Trump cannot afford to lose North Carolina unless he wins Michigan or Pennsylvania. He would lose with these numbers.   


http://www.highpoint.edu/src/files/2019/09/66memoC.pdf

Trump approval:

40% or less or disapproval over 52%
41-44% or disapproval over 50%
45-49% and negative


Note: these are colored green to the same shade if I believe that an independent or third-party nominee has a chance to win the state, but a Democrat has no meaningful chance. That is Utah for now.

tie (white)
45-49% and positive
50-54%
55% or higher

[GeorgiaModerate]

File this one under weird results:

Quinnipiac, Sep. 27-29, 1115 RV (change from last week)

Approve 41 (+1)
Disapprove 53 (-2)

Strongly approve 35 (+6)
Strongly disapprove 48 (nc)

But...

Do you think that President Trump should be impeached and removed from office, or
don't you think so?

Yes 47 (+10)
No 47 (-10)

Do you approve or disapprove of the impeachment inquiry? (no prior)

Approve 52
Disapprove 45

Do you think that President Trump has committed any crimes while he has been
president, or don't you think so? (change from June)

Yes 49 (+4)
No 42 (-3)

[Person Man]

The weird thing in the 538 is that Trump has lowest net approval rating of all presidents since tracking became available for this stage in his presidency.

It's still really odd that Trump is firming up in his worst polls and holding steady in the daily tracker.

[GeorgiaModerate]

Ipsos/Reuters, Sep. 26-30, 2234 adults including 1917 RV (prior poll Sep. 23-24)

Adults:

Approve 39 (-3)
Disapprove 56 (+2)

Strongly approve 21 (nc)
Strongly disapprove 45 (+5)

Trump should be impeached:

Yes 45 (+8)
No 41 (-4)

An elected official who uses the power of their office to attack political rivals should be removed:

Agree 63 (strongly 38)
Disagree 22 (strongly 7)

An elected official who uses a foreign government to attack a political rival should be removed:

Agree 66 (strongly 44)
Disagree 18 (strongly 6)


RV:

Approve 40 (-5)
Disapprove 56 (+2)

Strongly approve 23 (nc)
Strongly disapprove 45 (+3)

Trump should be impeached:

Yes 45 (+6)
No 43 (-4)

An elected official who uses the power of their office to attack political rivals should be removed:

Agree 65 (strongly 40)
Disagree 23 (strongly 8 )

An elected official who uses a foreign government to attack a political rival should be removed:

Agree 69 (strongly 46)
Disagree 19 (strongly 6)

[HisGrace]

File this one under weird results:

Quinnipiac, Sep. 27-29, 1115 RV (change from last week)

Approve 41 (+1)
Disapprove 53 (-2)

Strongly approve 35 (+6)
Strongly disapprove 48 (nc)

But...

Do you think that President Trump should be impeached and removed from office, or
don't you think so?

Yes 47 (+10)
No 47 (-10)

Do you approve or disapprove of the impeachment inquiry? (no prior)

Approve 52
Disapprove 45

Do you think that President Trump has committed any crimes while he has been
president, or don't you think so? (change from June)

Yes 49 (+4)
No 42 (-3)

I don't know that this is all that weird. I'm not expecting all that much change in Trump's approvals, just that a greater chunk of people who already oppose him (maybe nearly all of them) will support impeachment. You also might see his strongly approve numbers go up as people who already support him will circle the wagons.

[pbrower2a]

I think people well understand the concept of "legally innocent until adjuticated guilty"... it will be up to the House and Senate to determine the facts.

An elected official who uses the power of their office to attack political rivals should be removed:

Agree 63 (strongly 38)
Disagree 22 (strongly 7)

An elected official who uses a foreign government to attack a political rival should be removed:

Agree 66 (strongly 44)
Disagree 18 (strongly 6)

People recognize what is alleged as a serious and impeachable offense.

[GeorgiaModerate]

CNBC All-America Economic Survey (Hart Research/Public Opinion Strategies), Sep. 24-28, 800 adults (change from May)

Approve 37 (-3)
Disapprove 53 (+3)

This is a record low for Trump in this poll.

Impeachment hearings:

Support 44
Oppose 47

[pbrower2a]

We are going to find out what the President's floor of approval ratings are very soon.

[Person Man]

We are going to find out what the President's floor of approval ratings are very soon.

And simultaneously his ceiling.

[pbrower2a]

We are going to find out what the President's floor of approval ratings are very soon.

And simultaneously his ceiling.

I think we will soon have a different metaphoric meaning for 'ceiling'.

[pbrower2a]

Monmouth University, Sep 23-29:

approve 41 (+1)
disapprove 53 (nc)

Telephone (no mention of cell phones)

https://www.monmouth.edu/polling-institute/reports/monmouthpoll_us_100119/

It is toward the 'poor' side of polling for Trump results by Monmouth over time.

[pbrower2a]

https://twitter.com/MIRSnews/status/1178868336211431424

Crosstabs will be interesting, to put it mildly.

[GeorgiaModerate]

Morning Consult/Politico, Sep. 27-30, 2488 RV (prior poll Sep. 24-26)

Approve 41 (nc)
Disapprove 56 (nc)

Strongly approve 24 (nc)
Strongly disapprove 44 (nc)

If the 2020 presidential election were held today, would you definitely vote to re-elect Donald Trump, probably vote to re-elect Donald Trump, probably vote for someone else or definitely vote for someone else?

Definitely Trump 28
Probably Trump 8
Probably someone else 7
Definitely someone else 49

GCB: D 46, R 38