Donald Trump: polls of approval and favorability

[pbrower2a]

Three years away from the start of the real campaigning of the 2020 Presidential election I can make the earliest possible prediction of its results. I'm seeing an early pattern, with the shift of support from Obama to Trump looking transient in the extreme. I have eighteen states with approval or favorability ratings (and favorabiliy ratings look the same as approval ratings when both are shown ; one poll of Michigan asked about both and the results  were off each other by 1$ although straddling a category).

Note that I make some assumptions.

First, that the 2020 election will be free and fair.  

Anything else (1) isn't interesting, (2) is a violation of over 200 years of precedent, and (3) indicates that the comparative few who own the assets and grab the income either have gained total power or have lost everything in a revolution or apocalyptic war.

Second, that American political culture does not change profoundly in the meantime.  

Ethnic divides and religious patterns remain much the same and have much the same general orientation. There is no trend toward fundamentalist religion or toward irreligion that would change voting patters. We haven't seen that since the late 1970s  and I don't expect to see that now.

Third, that the states change in their voting behavior only due to demographic change

The Hispanic and Asian populations are growing rapidly and making a bigger part of the electorate while black and white populations become lesser shares.

Fourth, approval-disapproval differentials remain much the same as they are now.

That assumes that President Trump does not endure even further losses of approval or make a miraculous recovery.  Could things go worse for him? Sure. Mass unrest. Economic meltdown. Military or diplomatic debacles. Scandals involving sex or financial turpitude.  I'm not saying that any one of those will happen, but I can't rule them out. If any of these happen, then Donald Trump might not even run for re-election, in which case all bets are off.  

Fifth, that President Trump will run for election.

He will not die, become incapacitated, resign, or be impeached. To be sure, if he dislikes the Presidency he might choose not to run while expressing some noble cause for not seeking a second term, as did LBJ.

Sixth, that third parties will not greatly shape the election.

If the liberal side splits significantly, then President Trump wins. If some conservative-leaning nominee gets 10% or more of the vote, then Trump loses 'bigly'.

>>>>I have enough approval and favorability ratings of states to create a skeleton of a likely 2020 Presidential election. There are states (Colorado, Georgia, Ohio, and Wisconsin) for which I have nothing so far, for which I would like some data.

So add 6% to the most recent number for approval or favorability (where I had both favorability and approval, they were basically the same -- I prefer approval) to get the likely share of the vote in any state in the upcoming election. Nate Silver has a model for elected (not appointed) Governors and Senators that suggests that they normally lose support once they start legislating or governing because they can't please everyone who voted for them, but that they typically gain about 6% of the vote from an approval rating at the beginning of a campaign season by campaigning. That's the 'average' Governor or Senator running against the 'average' challenger. It worked well with Obama, whose approval ratings were in the mid 40s early in 2016, and he barely got re-elected by the popular vote. If it applies to Senators and Governors, then why not to the President?

So here are the data. I normed some unflattering polls for Trump in Florida and North Carolina to the national average , but other than that I simply took the latest numbers. Here is the raw data:

NY - 31 MA - 25  NJ - 36 AZ - 39 FL - 34 (raise to 40) NC- 36 (raise to 40) MI-40 WV - 58 CA -34 NH - 43 VA - 38 IA - 42 AR - 60 TX - 46 VA -32 TN - 51 MD -29 PA - 32 SC - 44

Add 6 to the approval rating, and you get the following map:

     


white -- 49-51% for Trump (a virtual tie)

Trump wins:

60% or more
55-59.9%
51.0-54.9%

Trump loses, getting :

40% or less
40-44.9%
45-48.9%