FiveThirtyEight Demographic Calculator Revisited: Can you make Trump win?

[Nyvin]

I just have a really hard time believing Trump will win 52% of college-educated whites, but I guess older ones might vote differently than the ones around my age.

I could see it happening. A lot of suburbanite Republicans are going to hold their nose and vote for Trump. At least half of my Republican friends between 25-35 are planning to vote for Trump, even if they supported the likes of Kasich, Rubio and Bush in the primaries. More will come around as reality sinks in. There are also a decent number of college-educated folks who have blue-collar backgrounds, e.g. those who went to school on the GI Bill, and most of them are going to be voting for Trump.

It's more than this. Think super-low turnout like 1996. College-educated white voters drop 6% to 70% turnout. However, young college-educated voters make up the majority of the drop and they are the more Dem fraction, so his percentage only drops 2% overall with college-educated whites.

I expect black turnout to drop 10% to around 56% putting them in line with non-college educated whites who drop 3% to 54% turnout. I also think that with such low turnout the Pub% could creep up 5% to 12% of the black vote.

The other minority groups become 3% more Dem, but drop 8% in turnout following the national trend of being between the white 6% and black 10% drop off in turnout.

Then all it takes is for Trump to increase his fraction of the non-college whites up to 70%. This can happen by dropping off some of the youth vote which is heavy Dem. The older non-college whites have been heavily trending Pub over the last few cycles, and its the group Trump has most concentrated on. At that point the 538 calculator gives Trump the win.

Try it.

That is completely flying in the opposite direction of every shred of evidence we have so far about this election.   

The most viewed primary debates ever?

The highest primary turnout ever?

Soaring voter registration (especially for latinos)?

Probably the two most divisive candidates ever?

Why would you possibly think this will be a low turnout election?

[Nyvin]

I don't think Hispanic turnout is going to drop 8%.

But I appreciate muon2 putting together a potential Trump win.

Yeah, I don't see Hispanic turnout dropping at all from 2012, let alone 8%. That is just totally at odds with all of the available evidence.

These are the percentages of the voter eligible population that voted since 1980 according to Dr. Michael McDonalds US Elections Project:

1980   54.2      1982   42
1984   55.2      1986   38.1
1988   52.8      1990   38.4
1992   58.1      1994   41.1
1996   51.7      1998   38.1
2000   54.2      2002   39.5
2004   60.1      2006   40.4
2008   61.6      2010   41
2012   58.2      2014   35.9

The last three presidential elections have been historically high in the era since suffrage was lowered to 18. Swings of 5-6% up or down are not unusual. 1988 had uninspiring candidates after two terms of a highly motivating president. 1996 had a Dem president fighting off scandal charges vs an uninspiring Pub nominee. Neither of those cycles had any major economic or other crisis (eg 9/11 for 2004) to move voters to the polls, which is also true this cycle.

I used an 8% Hispanic drop as something between a 5% white drop and 10% black drop. If I drop the Hispanics only 4%, less than the population as a whole, Trump still wins on the calculator assuming he takes 70% of non-college educated whites.

You assume this is a temporary blip. But to me it looks more like increased polarization or whatever has created a permanently higher level of turnout. 2012 for example wasn't super inspiring and still saw massive turnout.

It wasn't polarization but record turnout by the black population. I assume that without Obama the black turnout falls dramatically. I put it 2% above non-college educated whites instead of 7% above as it was in 2012.

The other driver of turnout recently has been the youth vote 18-30. Millennials were a big part in Obama's coalition. Neither Clinton nor Trump have done well at appealing to this demographic, so I think they also drop dramatically in turnout.

Black turnout has been rising consistently since the 2000 election,  the drop in 2012 was really the only exception (although some sources have it going up from 2008 to 2012).   I believe proportionally the turnout stayed relatively the same in 2010 and 2014, which doesn't bode well for the "Obama-Centric" AA turnout theory. 



The youth vote didn't really "spike" much in 2008, and certainly not in 2012...that's largely a myth.  It's been fairly consistent with national turnout overall.   2004 had larger youth turnout than 2012.

[Nyvin]

If the exercise is to make Trump win and be somewhat realistic, then I think it is something like this

My Demographics
College-Educated Whites: 54% Trump, 74% Turnout (Turnout and R share is down a bit)
Non-College-Educated Whites: 69% Trump, 61% Turnout (big surge for Trump with his base, he loves the uneducated)
Blacks: 90% Clinton, 61% Turnout (This go back to pre-Obama levels)
Hispanic: 79% Clinton, 53% Turnout (a bit worse than Romney but not too bad)
Asians/Other: 75% Clinton, 51% Turnout (a bit worse than Romney but not too bad)

Businessman Donald Trump (R-NY) - 49.1%, 273 EV's
S.O.S. Hillary Clinton (D-NY) - 49.1%, 265 EV's

This is by far the most realistic map in this thread.   Although when I put the numbers listed into the calculator Florida still goes Democratic (although barely),   which means this election outcome would almost be an exact repeat of 2000 with a few states swapped around (ie...it all comes down to Florida and the result isn't known for days, maybe weeks).

Also Trump really does poll terribly in Wisconsin, and I'm not that confident of his chances in NH either.   Neither state is a good match for him.