Cook PVI ratings:
Color and intensity will indicate the variance from a tie (ties will be in white) with in a 50-50 election with blue for an R lean and red for a D lean. Numbers will be shown except in individual districts
int var
2 1-4%
3 5-8%
5 9-12%
7 13-19%
9 20% or more
Cook PVI assumes a 50-50 Presidential election, reasonable since 2000 because except for the 2008 Presidential election all such exception, those five all were basically even for almost the entire electoral season. One can use polling to predict whether the next Presidential election will be a 50-50 proposition, and if not, how far the likely reality diverges from that assumption.
Based on 2012 and 2016 Cook PVI shows that the average Republican nominee will carry Alabama 59-41 and Florida 52-48; that the nominees will tie in New Hampshire and Pennsylvania; and that the Democratic nominee will win Michigan by 51-49 and New York 62-38. Of course, even in a 50-50 election, cultural affinities and the emphases will matter greatly. So could demographic trends. We will talk about that for Arizona and Texas.
For DC (not measured) and Congressional districts that vote independently of states, I have common sense for Dee Cee and the congressional votes for those districts.
DC -- way out of reach for any Republican.
ME-01 D+8
ME-02 R+2
NE-01 R+11
NE-02 R+4
NE-03 R+27
(data from Wikipedia, map mine)
............
Here is the Gallup data for 2017. Figuring that this is an average from early February to late December, I will have to assume an average date of July 15 or so for the polling data. This is now rather old data, and in some cases obsolete. For example, I see Trump support cratering in the Mountain and Deep South (seer below). the Mountain South and Deep South are going back to a populist phase (the South has typically oscillated between the two) or whether The Donald is beginning to appear as a bad match for either part of the South. This data (or later polling) is not intended to show anything other than how support appears at some time or at an average of times. As a general rule, new polling supplants even better old polling.
So here is the Gallup polling with a number of 100-DIS reflecting what I consider the
ceiling for Donald Trump. This is lenient to the extent that I assume that he can pick up most undecided voters but recognizes that undoing disapproval at any stage requires miracles. By definition, miracles are unpredictable.
Gallup data from all polls in 2017 (average assumed in mid-July):
*Approval lower than disapproval in this state
for barely-legible numbers for DC and some states -- CT 37 DC 11 DE 42 HI 40 MD 35 RI 38
Lightest shades are for a raw total of votes that allows for a win with a margin of 5% or less; middle shades are for totals with allow for wins with 6 to 10% margins; deepest shades are for vote percentages that allow wins of 10% or more. Numbers are for the projected vote for Trump.
I assume that President Trump will reach (1) just short of winning if he is behind, or (2) the ceiling that 100-DIS suggests if he is ahead. Basically I expect him to pick up the undecided vote to a large extent if he is behind because as badly as he is behind in some states, the undecided are clearly right-of-center. That is not charity; it is caution. People who have given up on him already are unlikely to give him a second chance. The mirror image would apply if the Democratic incumbent were having trouble with outrageous behavior.
.................
This is polling from
October or later.
I use 100-DIS as a reasonable ceiling for the Trump vote in 2020. Thus:
Lightest shades are for a raw total of votes that allows for a win with a margin of 5% or less; middle shades are for totals with allow for wins with 6 to 10% margins; deepest shades are for vote percentages that allow wins of 10% or more. Numbers are for the projected vote for Trump.
I use a 'favorability' rating for Illinois because such is all that is available and at this stage, favorability and approval are close when there is no active campaign.
Note -- if Trump is underwater in the polling, then the results come out in pink.
..........
Now, for the variance between 100-DIS from recent polling and Cook PVI.
Variation from PVI (polls from October 2017 and later):
Orange implies that President Trump projects to do better than Cook PVI based on 100-DIS. In Minnesota I have a 49-47 poll with which to work, and people in that thread tell me that the pollster who got those results is suspect. So the President is doing 2% better in California in accordance with 100-DIS than Cook PVI suggests. Not significant, obviously, because that is the difference between losing the Golden State 60-40 instead of 62-38.
This is likely the last that you will see of my analysis of polling based on deviation from Cook PVI. I think we can be assured that President Trump is doing worse, in general in polling, than something consistent with a 50-50 split of the popular vote.
This data is from mid-February.