Donald Trump: polls of approval and favorability

[pbrower2a]

The Gallup tracking poll that just appeared indicates a Presidency intensely unpopular in America. It cannot say whether the unpopularity arises from policies or from assessments of a personality. Most likely, both personality and policies both interact.

We need more statewide polling to establish whether the trend is confined to a few states reliably D over several elections or states rapidly lurching D (Colorado and Virginia?) I notice that a recent poll of Alabama suggested that although support for President Trump is stronger than opposition, the net support for the President is far lower than is the norm for previous Republican Presidents (Reagan and both Bushes, and anything before that is practically ancient history). I do not predict any statewide polling. Trump would be OK in a re-election bid if his best 'losing' state (he will lose Michigan in 2020) would give him 32% of the vote and his worst 'winning' state is Pennsylvania, which lets him squeak by with 20 popular votes but is the difference between 252 electoral votes for him and 272 electoral votes. But that is a probabilistic freak, something indicating that someone rigged the election.

At this point I predict that the best projection of the 2020 vote is that the President gets 100% less the disapproval rating in a binary election. Thus if the disapproval rating for Trump is 57% in Michigan, then the most that Trump can get in Michigan is 43%. Could he win with 43% of the vote? Only if his opposition is split, as between conventional Democrats and a new and powerful Socialist or even Communist party. But such does not yet loom.

At this point I must treat any immediate response to an outrageous deed or unpopular policy of the President or a polarizing and unpopular piece of legislation as a temporary and reversible event. My main map is cautious and slow to respond, reflecting the paucity of polls so far. I expect that to change in about a month as Senate races and control of the House of Representatives will become more frequent as is the norm in all midterm elections. I do not make inferences of 'likely polling' from extrapolations of national tracking unless I so indicate.

OK. It is a reasonable assumption that the Democrats are not going to cut into the white vote for Republicans enough to put states in the Deep or Mountain South in play for any potential Presidential nominee in 2020, and that there is going to be no sudden lurch toward the 50-50 split in states that Republicans just do not win except in landslides like 1980. 1984, and 1988. I think that we can agree that at this point nobody needs concern himself with any state that Trump won or lost by 10% or more or in which he got less than 45.5% of the vote (you will see what state that means.

States that I deem potentially in play at this stage based upon 2016 results are in colors other than gray. Here we go:



Utah 2016, Trump getting much less than a majority of the vote (45.05%)
Trump won by 8.0 to 9.99%
Trump won by 4.0 to 7.99%
Trump won by 1.0 to 3.99%
(white) Trump won by less than 1%
Clinton won by less than 4%
Clinton won by 4.0 to 7.99%
Clinton won by 8.0 to 9.99%

I have Florida in a really pale shade of blue because Trump won the state by 1.22% and New Hampshire in a really-pale shade of pink because Clinton won it by less than 1%.

I don't have a split of Maine by popular vote in the two Congressional districts, but I would guess that ME-01 is very solidly Democratic, and that ME-02 went to Trump in the same range as did Georgia. That's conservative on my part. But with that I get to add NE-02 for political symmetry, and I am guessing that Trump won it by slightly less than 10% and that Clinton won ME-01 by a similar margin.

Why do I show Utah? All that it would take for Trump to lose Utah would be for a Third Party or independent candidate to get Democrats to decide to not waste their vote on the Democratic nominee for President and vote instead for some conservative who better fits LDS (Mormon) values than does Donald Trump.

4% is the usual margin of error. In that range anyone who makes any prediction of absolute certainty is a fool and the only definitive result is the election itself. This is with the dynamics of 2016 which I must consider relevant. But this is assuming that results of 2016 have relevance in 2020. In the last three presidential elections involving an incumbent President, the statewide maps of the election changed little from the election that put the incumbent in and the one in which the incumbent sought re-election. That in mind, electoral history of the states is relevant.

Going back to 1992, as there are seven Presidential elections bearing some possible relevance to this one (and anything before that is either the GOP landslides of 1980, 1984, and 1988 which will not be approached and elections from 1976 or earlier when the Democratic party was stronger in the Middle and Deep South and the Democrats usually lost states outside the South except for a few scattered from Minnesota to Massachusetts)



Democratic seven times
Democratic six times
Democratic five times
Democratic four times (white)
Democratic three times
Democratic once
Republican seven times

No state has voted for the Democratic nominee in exactly two elections on this map.

I know, of course, that this history is relatively crude. It will be relevant, demographics notwithstanding, that the Democrats have a very weak position for winning any statewide election in Arizona or Georgia (one-time Bill Clinton wins, and close to somewhat close in 2016), let alone Texas. Utah is obviously a pipe dream for any Democrat or independent under all but the most extreme circumstances. Trump getting so close in Minnesota is itself a shocker, but I see plenty of ways for Minnesota to be a very bad state for Trump in 2020 and copious evidence in polling data that he will fare badly there. That Democrats usually win Michigan, Pennsylvania, and Wisconsin indicates that Donald Trump must achieve his promises without offending too many sensibilities to win those states. Likewise, that Bill Clinton lost Colorado and Florida once in the 1990s may not be particularly relevant. In view of the 2016 election it is absurd that Colorado and Ohio should be in the same category, but they are on this map of electoral history. The same applies to Florida and Virginia.



So how can I combine relevant issues such as the way the state voted in 2016 (which must matter greatly in an objective projection of the 2020 election) and the last seven elections?

Assign one point for the number of times in which a Democratic nominee won the state and one point for belonging in a category most unfavorable to Trump (7 for Hillary Clinton winning by 8% or more, 6 for her winning by 4% to 7.9%, 5 for her winning by less than 4%, 4 for a loss by her under 1%, 5 for a loss between 1% and 1.99%, 6 for a loss of 2 to 5.99%, and 7 for a loss of 6% or more. I am arbitrarily assigning a value of '4' to Utah because Trump got a smaller share of the total vote in Utah than he got in Michigan.

This is a probabilistic construction based upon the direction and margin of the vote in 2016 and the electoral history of the states beginning in 1992:


0 points (Texas)
2 to 4
5 to 7
8
9 to 11
12 or 13 (Maine, New Mexico, ME-01)

Before someone says that Iowa is lost to the Democrats, one must remember why the state voted in five of the last seven Presidential elections for a Democratic nominee. With Ohio, one must ask why the Democrats won four of the last seven such contests. On the other side, one must ask anyone who thinks that either Arizona, Georgia, or North Carolina will go for just about any Democrat due to demographics or some other state trend why those states voted in six of the last seven times for Republican nominees for President. One can of course distinguish Florida and Virginia... and Colorado and Ohio.