Current 2020 map based on Morning Consult tracking

[pbrower2a]

Link to tracking website: https://morningconsult.com/tracking-trump/

The pollster Morning Consult has been tracking Trump's favorable and unfavorable numbers in every state starting in January 2017 and all the way till May 2018, a total of 17 months thus far. I have added together all the raw numbers for each state, and then ranked each state correspondingly. After 17 months, this is the 2020 map that is starting to emerge:

Results for this specific map:

Dem: 266
Rep: 235
Toss up: 37



This is the current full ranking state by state after Trump's first 17 months in office (from lowest to highest approval):

1. Hawaii: -463
2. Massachusetts: -428
3. Vermont: -422

4. Maryland: -369
5. California: -321
6. Rhode Island: -306

7. Washington: -288
8. Connecticut: -269
9. Oregon: -251
10. Illinois: -243
11. New York: -221
11. New Jersey: -221
13. Minnesota: -208

14. New Hampshire: -176
15. Colorado: -166
16. Delaware: -161
17. Wisconsin: -154
18. Michigan: -141
19. Maine: -112

20. New Mexico: -94
20. Iowa: -94
22. Pennsylvania: -41
23. Virginia: -27
24. Nevada: -14
25. Ohio: -9

26. Arizona: +43
27. North Carolina: +58
28. Georgia: +92
29. Florida: +96
30. Utah: +99

31. Missouri: +113
31. Indiana: +113
33. Nebraska: +147
34. Texas: +154
35. Kansas: +156
36. Montana: +161
37. North Dakota: +162

38. South Carolina: +205
39. South Dakota: +231
40. Alaska: +244
41. Arkansas: +263
42. Idaho: +276
43. Kentucky: +283
44. Tennessee: +287

45. Oklahoma: +320
46. Mississippi: +322
47. Louisiana: +360

48. West Virginia: +406
49. Alabama: +440

50. Wyoming: +544


Does the map above seem like a fair map come the fall of 2020?

No. Later maps should matter more due to the effects of policies and personalities. The 2016 election should have disabused us all of any idea that early polling matters much more than it does after later polls come in. Going even farther back, polls from early 1977 suggested that Jimmy Carter was a shoo-in for re-election.

If current polls show that the President is behind 43-54 in approval in Ohio, that he was up 48-46 early in 2017 becomes irrelevant.

What's with Iowa?

Iowa reflects in part that rural voters are turning against Trump. Donald Trump has done nothing for agriculture except to drive agricultural pay down. He has offered tariffs that will hurt farmers by cutting their commodity prices and raising fuel costs. Agriculture simply devours energy.

Iowa can vote for a city-slicker, but Donald Trump has shown himself the worst sort of city-slicker, the sort who sees rural people as people to fleece and humiliate.

[pbrower2a]

I see a problem with aggregated results -- that they would overestimate the chances of someone who had relatively high approval early and lost it steadily. Just think of the results for Carter in 1980 (late polling would be trivialized), suggesting at the least that he would not lose in a landslide -- or those for Dubya that would include approvals for him  immediately after 9/11 that suggest a blowout re-election instead of the close win that he got as his support faded to mediocrity.

Politicians are elected or rejected in the here-and-now of the election on conditions then in existence. Maybe we can use polls to predict competence. OK, things would get better with Reagan after he took measures that brought short-term pain and near-term improvement... but that is not how I see Trump.

Politicians win or lose elections to no small extent due to the conditions of the time.  President Trump needs major changes in the basic reality of the political realities to have a chance of winning.   

[pbrower2a]

Approval ratings in 2018 do not, by any means, indicate 2020 electoral results.

The ones that Trump has indicate a troubled Presidency that needs to either turn  much around or cheat to be re-elected. Weird events could get President Trump re-elected, but I wouldn't bet on such happening.

Sure, Reagan was this far behind in 1982, but look what happened... But Reagan  successfully put an end to stagflation, and by 1984 price stability made many more comfortable. Those who got stuck with low pay in jobs? They could take second jobs in the fast-food places and shopping malls then in bloom.
 
 

[pbrower2a]

Approval ratings in 2018 do not, by any means, indicate 2020 electoral results.

To add some fuel to this, you really only need about 42% approval as an incumbent to win an election.

If disapproval is also low. 100-disapproval is a reasonable ceiling to what an incumbent can get.  Disapproval is rejection, and changing the minds of people who already disapprove of the performance of an incumbent politician is difficult. Talk to them as a canvasser for your candidate and you will find that you give them even more cause to vote  against yours.

Politicians usually get re-elected, and when they don't, then something is amiss. An incumbent can usually pick up a huge chunk of the undecided with a spirited campaign, finding out what the undecided have as concerns, and offering solutions. That sounds like what Harry Truman did in 1948, and it obviously worked.

[pbrower2a]

Approval ratings in 2018 do not, by any means, indicate 2020 electoral results.

To add some fuel to this, you really only need about 42% approval as an incumbent to win an election.

If disapproval is also low. 100-disapproval is a reasonable ceiling to what an incumbent can get.  Disapproval is rejection, and changing the minds of people who already disapprove of the performance of an incumbent politician is difficult. Talk to them as a canvasser for your candidate and you will find that you give them even more cause to vote  against yours.

Politicians usually get re-elected, and when they don't, then something is amiss. An incumbent can usually pick up a huge chunk of the undecided with a spirited campaign, finding out what the undecided have as concerns, and offering solutions. That sounds like what Harry Truman did in 1948, and it obviously worked.

Except that a presidential election isn't an approval/disapproval poll...it's a binary choice between two candidates. If Trump's opponent has similar disapproval/unfavorability numbers, the metric becomes meaningless.

There are better metrics, and as the election gets closer, those metrics become available. Of course we have the previous election, which is usually highly relevant. Few states switched sides from 1992 to 1996, from 2000 to 2004, or from 2008 to 2012. So the last three Presidential elections involving an incumbent have been similar in result to the election in which the incumbent President was elected. But go back in time, and you find that the elder Bush, who won his first election lost by a large electoral margin lost  his re-election bid by a large margin. Before that, Reagan won by a landslide margin in 1980 only to win by a crushing landslide margin in 1984 -- and defeated a troubled incumbent who barely won in 1976. In fact one does not find an incumbent President winning in much the same 'electoral style' as did Clinton, the younger Bush, and Obama until the two elections involving Eisenhower in the 1950s.  

Just because a pattern has held three times does not establish a likelihood of repetition.

Elections involving an incumbent usually have the achievements and failures of the incumbent as the focus of the election. Of course, should Trump lose only two states that he won in 2016, he wins -- unless one of the states that he loses is Florida (less likely than Michigan, Pennsylvania, or Wisconsin) or Texas (36 electoral votes, and otherwise a joke as a prospect). But should he lose Michigan, Pennsylvania, and Wisconsin or one of those states and Florida and picks up nothing, then he loses despite having an electoral result similar in result.

So what are better metrics than approval and disapproval? Obviously, match-ups between nominees, but that availability will not emerge  until 2020. The Democrats have a huge 'quarterback controversy' unlikely to resolve itself. Indeed, we may have an election in which a significant Third Party or independent campaign emerges, and we cannot tell whether that would take votes from the Republican nominee or from the Democratic nominee or even whether such would have a significant effect.

Better than what I have for 43 states, Dee Cee, and the five independent-voting Congressional  districts is the answer to a polling question, essentially "Do you expect to vote to re-elect Donald Trump or do you expect to vote for someone else?" I have that for seven states (AZ, FL, MI, MN, NH, OH, and WI), which is no random scatter of states of at most marginal relevance to Election 2020. All of them are either traditional swing states or were very close in 2016. Except for one of the states (New Hampshire, with four electoral votes) the question is asked by a pollster with which I have some familiarity and respect (Marist).  

Results for six of the states are:

AZ 35-57
FL 35-54
MI 28-62
MN 30-60
NH 41-50 or 41-57
OH 34-58
WI 31-63

Except for Ohio (a usual swing state in which Trump did extremely well for a Republican nominee) all of these states were within  5% of going one way or another. We all have good reason to believe that in a close election in 2020, a basic premise that most of us had in 2016, that these states would as a group generally be close in 2020.  

The problems with this measure in New Hampshire are

(1) that the question is asked by a pollster new to us, and
(2) that this pollster subdivides answers into "strong" and "weak" likelihood of voting for or against Donald Trump. Only 37% of polled voters in New Hampshire say that they have a high likelihood of voting for Donald Trump, which is awful for a Republican. Only 4% say that they have a slight chance of voting for Donald Trump, which is itself weak... and I can add those because strong support and 'squishy' support is what one gets in an election. 50% say that they will definitely not vote for Donald Trump and 7% say that they are unlikely to vote for him.

But this said, "Will you or will you not vote to re-elect Donald Trump?" is unambiguous enough for six states, and... the President is not going to pick up Minnesota (which, if he did, would suggest that he is gaining support), he is unlikely to win New Hampshire even with the most generous interpretation of the question... and he is not going to win any of the five states in Marist polling that he lost, including a state that he won by 8% in 2016.

Just look at the Ohio result. 58% of polled voters suggest that they will not vote for him. That means that Trump has a ceiling of 42% of the popular vote in Ohio in 2020, and that Democrats have invariably cracked that level except in three-way Presidential races or blow-out losses. (Bill Clinton ended up with just less than 40% of the popular vote in Ohio, but that was in a three-way contest with Perot getting over 20% of the popular vote).

I have Trump approval at 42% in Ohio in an even later poll than that in which Marist asked the "re-elect/do not re-elect" question. Disapproval is at 49%, which is awful for a state that the President won by 8%. At this point I predict that the President will lose Ohio -- and Arizona, Florida, Michigan, New Hampshire, and Wisconsin.  

How often does an incumbent lose a state that he won by 8% or more in the election that got him to the Presidency in his re-election bid?

The elder Bush lost a raft of states (FL, NV, GA, NH, TN, AR, NJ, KY, ME, OH, and PA) that he won in 1988 -- but we can attribute that to partisan fatigue, lacking an idea of what to do in a second Presidential term, and an unusually-strong opponent in the next election following an opponent who stumbled at every turn. To that we can reasonably treat Michigan (7.90%) which is very close to 8% anyway.
 
Sixteen years earlier we go back to Jimmy Carter, who ended up losing Alabama, Arkansas, Massachusetts, North Carolina, South Carolina, and Tennessee in the 1980 disaster for him. Carter did not win many states by huge margins, but certainly lost many by small margins in 1976.

Democrats do not need to nominate another Ronald Reagan or Barack Obama to defeat  Donald Trump.      

[pbrower2a]

Were it for cumulative polling, we would have Presidents Mike Dukakis, John McCain, are the ones to most likely predict the results. and Hillary Clinton.

The last polls are the ones most relevant to the electoral results... and as a rule the match-ups matter far more than approval and disapproval.

Not one vote has been cast in the 2020 Presidential election. Opinions are being set already, and at this point I see Trump more likely to lose as did the worst electoral losers among incumbent Presidents of the last century (Hoover and Carter) than that he barely win.  

He is more likely to lose Texas than to win any one of Michigan, Pennsylvania, or Wisconsin.

[pbrower2a]

Where the President's approval is at the latest is what usually matters most.

I see polls that have the President enduring approval levels under 40% nationwide, which may relate to the Border Wall that he isn't going to get. His failure to get it may allow him to recover some. Enough? Who knows? I am not in the prophecy business.

Cumulative approval from January 2017 to October 2020 will matter far less than will what people think as they make up their minds to vote.  Let us suppose that the cumulative disapproval in New Hampshire in October 2020 is -406... but we see approval for the President like this:

New Hampshire +7

for October alone. It was +4 in September, +1 in August, and -3 in July.  The cumulative number was -418 after July results. The -418 is meaningless in October as New Hampshire voters start casting absentee ballots.

So what of the other side?

Let us suppose that Florida is at +120 in July, but approval numbers for Trump go from +4 in July to even in August to -2 in September and -4 in October.  Do you think that Trump wins Florida in November? 

Consider this: Barack Obama had some impressive approval numbers early in 2009. Those obviously took a tumble, and he ended up winning re-election by a narrower margin in popular and electoral votes in 2012 than in 2008. (Most incumbents gain some, but Obama got elected while the economy was in a tailspin, and Romney was an unusually-strong challenger against an incumbent who had not fouled up as President). It is how Obama was perceived in the autumn of 2012 that decided whether he would be a on-term or two-term President. 

The average to now is already irrelevant. Most Presidents have their ups and downs, and I see no reason to expect differently with this President. What I see in January 2019 is incredibly horrid for a President seeking re-election. Maybe what we saw in August 2018 will be closer to the reality of October 2020, as it seems not to have been as 'down' as such few statewide polls as I saw in January show.

Trump approval and disapproval numbers were comparatively consistent in 2018 -- bad.     

[pbrower2a]

Pbrower2a, you have completely misunderstood everything about this thread and about my project. This project has NEVER been about figuring out whether Trump or Democrats are the favourites to win, as you seem so obsessed about as an extreme Democratic hack. Nothing wrong with that at all, but you pretend like you're some kind of a neutral observer (LMFAO). This project has always been about implementing the latest science in the way of statistics - something I've always cared deeply about - in order to figure out which states are more likely to end up as the crucial battleground states come 2020. That's my whole project. You've criticized me from day one for doing this, like there's something unethical about it. The truth is it's never been done or approached before, so who are you to know whether it will work or not? I am obviously as clueless as you and everyone else about the final results, just like Colombus was when he sailed towards Hispaniola and expected to find India. The one thing I know about statistics is that accumulated statistics ALWAYS - and EVERY SINGLE TIME - trumps one time statistics. According to you, only the latest numbers are important. Well, my friend, you are sadly wrong, and it just goes to show how incredibly ignorant you are when it comes to the world of statistics.

It is the difference between on the one side a summation or integral to the results at a time. So someone has taken a road trip from Boston to San Francisco in six days and covered about 3108 miles. (The midpoint of the journey is near York, Nebraska in case you are curious). How long will it take to drive to Manila?

You are not driving to Manila. You can't! Your Amphicar isn't seaworthy, and it would have probably broken down along the Ohio Turnpike anyway!  As 2016 demonstrates, it is where the positions of the candidates are at election time that matters. Clinton collapsed and Trump surged. Such also explains why the elder Bush beat Dukakis. Your method would have Mike Dukakis as the 41st President.

With an incumbent President, the approval rating is everything unless something really strange is going on (like a rigged election).

Do we know all that will happen between now and November 2020? Of course not. I would not guess what the weather will be like in Colorado, Maine, Minnesota, or New Hampshire (the only near-swing states in which an autumn snowstorm could affect turnout at the polls), the Dow-Jones average, or what team wins the World Series (allegedly when the American League team wins the World Series, the Republican nominee usually wins the popular vote and vice-versa). It is imaginable that the Dodgers would defeat the Yankees, Klobuchar would win the popular vote, and Trump wins the electoral vote much as Hillary Clinton won the popular vote and lost the electoral vote after the Cubs won the World Series).

Am I predicting the result of the 2020 World Series?  No. This time it could be the Cleveland Indians against the Milwaukee Brewers, which would be the worst for the TV ratings, but is possible. I certainly do not expect a rematch between the Detroit Tigers and the San Diego Padres!

I try to be objective, and I use mathematical models with the caveat that those are not predictions of the result. I gave an almost equal chance to a Trump surge in New Hampshire and a Trump collapse in Florida.

At this stage we must ask such questions as "How can Trump win" -- and at this stage a positive answer requires some convoluted reasoning even if I can't believe it. Then again, I never believed that the American electorate could ever vote for so crass a demagogue as Donald Trump who had no political experience.   

[pbrower2a]

WI is D+2 state and OH is R+3 state, this map shows that Trump doesnt have a chance

This map is actually extremely generous towards Trump as it also includes his honey moon period of the first 4 to 6 months in office when he had a positive job approval in almost every state. If you read the beginning of this thread, you will see that several posters heavily criticised me for including his honey moon period as they considered it totally irrelevant and thus conflating Trump's numbers by a lot. I personally think I've struck the right balance. I want to be objective by including all numbers, not just a selective number of them.

It would be possible to

(1) dump the earliest months as they are no longer relevant, or
(2) use only he latest six months.

Obviously the most relevant polls before the election itself will be those that appear closest to the election itself.  Barring a 9/11-style event that he handles at least somewhat properly, he will never get anywhere near his 'honeymoon' numbers.

[pbrower2a]

Trump just having a balanced approval rating in Nebraska is pretty interesting, if true ...

Trump would lose NE-02 (greater Omaha) and put NE-01 (eastern Nebraska, especially Lincoln, outside of Greater Omaha).

Morning Consult: net approval, July 2019

Net approval for Trump

+10 or higher
+5 to +9
+2 to +4
+1 to -1 (white)
-2 to -4
-5 to -9
-10 or higher