08 November 2016

Statistikhengst's Final 2016 Electoral Landscape (No. 3 of 3): Clinton vs. Trump

Statistikhengst's FINAL Electoral Landscape (No. 3) for 2016:

Clinton vs. Trump
(Johnson, Stein, McMullin as sidenotes)

Data through November 7, 2016, 1 day before the election


Clinton 323 / Trump 215 / Margin: Clinton +108 EV

Clinton 268 / Trump 197 / Tossups 73 / Margin: Clinton +71 EV

Monday, November 7th, 2016, was exactly 1 day before the General Election on November 8, 2016. This is the  FINAL ELECTORAL LANDSCAPE, which will be released on election day.  For comparative purposes, HERE is the Final Electoral Landscape from 2012.

Democratic Presidential Nominee Hillary Clinton is projected to win the General election and become the 45th President of the United States of America. 

She will win nationally with a margin between +4% and +5.6%. According to the polling aggregates for the end-polling, her best case scenario actually goes past the map with 323/215, because there is still the possibility that Ohio, technically a tossup state, could go her way. But the most reasonable reading of the data is the first electoral map you see. Her worst case scenario, based on these numbers, would be were Trump to win all of the tossup states, including the ones technically leaning toward Clinton, plus ME-02, and in that case, Clinton still wins, 272/268.

Complete polling results for all states for 2016 to-date are here as an EXCEL table for 11/07/2016 in GOOGLE DOCS. All polls are hyperlinked.

That EXCEL table contains 5,056 presidential matchups from a total of 2,879 polls collected and analyzed since early 2013.  190 of those polls and 195 matchups were collected and analysed on 11/07, a banner day for polling.  A small smattering of polls came in on 11/08 as this report was being written. At first glance, they make no difference in the overall statistic and show the same direction in the few states polled, also nationally. I will include them after the election.

Just for easy comparision, here is the EXCEL table from the 2012 election.

So, my methodology for doing averages in this wild-and-wooly-sometimes-3-way election was explained in the first Battleground Report, from 11 October, 2016. It is somewhat different as was with the 2012 election.

The Battleground table is pretty darned detailed. The LEGEND for said table is HERE. In fact, you need to read it in order to develop an "eye" for the Battleground tab. However, I have changed the format a little bit - I think you will see how and why.

The long and short of it is that four states are in the very narrowly defined (+1 in margin or less) tossup zone: NV, NC and FL on the Clinton side and OH on the Trump side. You will also notice that I removed states with aggregates of +10 or more from the battleground table, if you compare it to the table from November 6th:

I think it would be instructive to place this battleground table up against the battleground table from the end-polling 2012:

This makes for an outstanding and easy-to-see comparison.

Let's concentrate on the four tossups and see how what they looked like on election day 2012:


Florida 2012: Romney +0.61, actual result: Obama +0.88
Florida 2016: Clinton +0.90, actual result: ???
Florida was the only state in the Union in 2012 where the actual result did not match the end-polling aggregate. Most all pollsters got Florida wrong. Clinton's end-polling statistic in FL is 1.51 points better than Obama's from 2012.


North Carolina 2012: Romney +2.40, actual result: Romney +2.04
North Carolina 2016: Clinton +0.92, actual result: ???
North Carolina was on of the 4 states where the actual result came stunningly close to the end-polling aggregate. Clinton's end-polling statistic in NC is 3.32 points better than Obama's from 2012.


Nevada 2012: Obama +4.67, actual result: Obama +6.68%
Nevada 2016. Clinton +0.67, actual result:???

Obama did better than his end-polling average in NV, most likely because of the undercalculation of the Latino vote. But Clinton's NV end-polling average is 4 full points under Obama's, not a good place to be. However, based on the early voting statistics from this state, the wisdom in the field is that Clinton has the state locked up. After tonight, we will all know.


Ohio 2012: Obama +3.24, actual result: Obama +2.97
Ohio 2012: Trump +0.91, actual result: ???
Ohio is obviously going strongly against the grain this time and in the case of a Clinton national win, which I am predicting, but also a Trump win in the state, will lose it's coveted bellwether status. Trump is doin 4.15 points better in Ohio than Romney did.

Now, those are the four tossups, but both NC and FL are barely inside the tossup zone and that mainly because of some partisan Republican polling thrown into the mix at the last minute. Both teams are admitting that the Clinton ground game and organization in all of the battleground states is superior to Trump's ground game, and that is the factor that influences GOTV. This is why I am, in light of electoral history, very confident that Clinton will retain Florida for the Democrats and also flip North Carolina, but lose Ohio and most definitely Iowa. I am also confident that Ohio and Iowa will flip because the highest of quality pollsters, the so-called "gold standard" pollsters, are showing margins too larger for Clinton to overcome. Likewise, in NC and in FL, states with heavy early voting, polls from quality pollsters are showing Clinton ahead and her team is certainly not looking worried about those states like the Gore camp was in 2000, in the case of Florida.

Now, you can look at the other states on the list and you will see that in most cases, Clinton is doing comparable or better with her end-polling statistic than Obama did, for instance, in Pennsylvania:

Pennsylvania 2012: Obama +3.88, actual result: Obama +5.34
Pennsylvania 2016: Clinton +3.68, practically the same.

Clinton is doing considerably better in Virginia (Clinton +5.85, Obama +0.77), Colorado (Clinton +4.30, Obama +1.87) and Wisconsin (Clinton +6.13, Obama +4.52) than the end-polling aggregates showed for Obama.

The other state where I see the press playing the "maybe, just maybe game" is Michigan, but let's compare:

Michigan 2012: Obama +4.79, actual result: Obama +9.48
Michigan 2012: Clinton +5.86...

So, in light of seeing how the Democrats have consistently outperformed their end-polling aggregates in 2008 and 2012, I see these current numbers in that light.

For instance, it looks tight in New Hampshire, but the last two polls both showed Clinton back at Clinton +11 over Trump. So, no, it does not look like New Hampshire is going to flip.

Now, as to the red states that Clinton would like to flip (Georgia, Arizona), they are close, but probably not close enough. However, in the case of Georgia, the likelihood of an automatic runoff election in December is extremely high, because most likely neither candidate will get over 50% - and Georgia is the only state in the Union with the "50% +1 vote" rule. Wait and see. Also, in the case of Utah, in spite of what is going to be an underwhelming margin for Utah standards, it looks like enough Republicans have returned to the fold in order to make sure that the Beehive state goes to Trump.

The other reason why these numbers are more solid than they look is that the national polls are almost all universal about the number "4", meaning that they are pointing to a +4 race for Clinton. And a +4 race is very likely to mean easily coming over 300 EV.

Here is the PENUMBRA chart for both candidates:

What do we know for sure, based on this data? We know that Clinton has 14 states in her column posting a landslide aggregate of +10 or more. 19 such states also post such an aggregate for Donald Trump.  Clinton's 14 states, from DC down to Oregon, are worth 174 guaranteed EV.  Donald Trump's 19 landslide aggregate states, from Wyoming down to Texas, are worth 145 guaranteed EV. The rest goes into single digit margins, most of them very safe for both sides.  So, the stability of the inner-columns for both parties has remained stable.

On the Democratic side, 4 states are posting aggregates that underperform vis-a-vis Obama 2012: HI, RI and to a lesser degree, CT and DE. Some of that is understandable, since HI is Obama's birth-state and DE is Biden's birth state. Plus, not all Vermonters are happy that Sanders did not get the Democratic nomination.  On the other hand, the polling aggregate in California bodes a huge motherlode in raw vote margin.

On the Republican side, 4 states are posting aggregates that underperform vis-a-vis Romney 2012: both UT and ID (Mormon populations), but also TN and to a lesser degree, KS. None of these details will affect the outcome of the race, but they can have an effect on the overall margins.

I thought a long time about writing in great detail about each of the battleground states, but I have a better idea. I encourage you, the reader, to go to the excel document linked above (for 2016) and simply tab through the data and see the sea of either RED or BLUE by certain states. Fix your eyes on the margin numbers and see how long the lists sometime are.

What remains to be see is whether the so-called "Obama coalition", which is simply the new Democratic coalition, will hold, or not, or whether perhaps a number of new coalitions are forming. Already there is heated talk about our society splitting into four factions: Traditional Republicans, Trumpublicans, Moderate Democrats and the Sanders coalition.  Maybe this election is a warning sign that our society is about to encounter a techtonic shift once again.

All I can say is that the numbers, both at the state level and national level, point to a clear and decisive Hillary Clinton victory, due in large part to her nailing down both Virginia and Colorado, erstwhile GOP bastions.

If Florida is called for Clinton earlier than later, then that is a bad sign for Trump. Also, if Virginia is called relatively quickly. I don't expect Florida or Ohio to be called for quite a while tonight...

On the national level:

We can draw some logical conclusions about this data that both Liberals and Conservative can agree to: here are the results of 20+ pollsters. 2 of those results go against the rest, and one goes very strongly against the rest, namely, the LAT poll, which shows Trump +4.8, while the Rand poll shows Clinton +9.1. That is a disparity of almost 14 points, which cannot really exist in the same universe. One of those two polls is going to end up being very, very wrong in retrospect come November 9th.

Even with the contrarian results from both the LAT and the IBD/Tipp poll, Clinton's aggregate has risen to Clinton +3.53

Now, here is the kicker: if you remove both possible outliers (the LAT poll and the Rand poll) on both sides, the aggregate is Clinton +3.63, only a 0.10% difference. So, I split the difference between the two and see the final aggregate at: Clinton +3.58.

THIS is the reason why I love to work with the averages.

The other kicker is that the LAT poll and the RAND poll use the same methodology, totally apart from the other pollsters. And yet, they show totally disparate results.

Here is an important detail: on this day in 2012, Obama's national average stood at: +1.77.
Actual result: +3.86%. This means that Obama outperformed his national end-polling aggregate by +2.09 points. Therefore, it is entirely conceivable that Hillary's final margin could be +5 or above, when all is said and done. I just wanted to make that point.

Many weeks after the election is over with, I will run a complete comparison of the actual election results to the different pollsters to see how they fared.  I did this in 2013 for 2012 and will be doing it again this time around. A word of caution: it is wise to wait until after the final canvasses are in before engaging in such work.

The tracking list:

12 October 2016, Clinton +5.52, -0.21
13 October 2016, Clinton +6.61, +1.08*
14 October 2016, Clinton +6.75, +0.14
15 October 2016 - no battleground report posted
16 October 2016, Clinton +6,44, -0.31
17 October, 2016Clinton +7.35, +0.92*
18 October, 2016, Clinton +6.91, -0.45
19 October 2016, Clinton +6.64, -0.27
20 October 2016 - no battleground report posted
21 October 2016, Clinton +6.74, +0.10
23 October 2016, Clinton +6.02, -0.72
24 October 2016, Clinton +6.25, +0.23
25 October 2016, Clinton +6.37, +0.23
26 October 2016, Clinton +6.73, +0.36
27 October 2016, Clinton +6.55, -0.18
28 October 2016 - no battleground report posted
29 October 2016, Clinton +5.91, -0.64
30 October 2016, Clinton +5.81, -0.10

31 October 2016, Clinton +5.93, +0,12
01 November 2016 - no battleground report posted
02 November 2016, Clinton +5.02, -0.91
03 November 2016,  Clinton +3.86, -1.16
04 November 2016, Clinton +2.52, -1.34
05 November 2016 - no battleground report posted
06 November 2016, Clinton +3.05, +0.53
08 November 2016, Clinton +3.58, +0.53*Due to rounding, the difference is off by 0.01

On this corresponding day in 2012the national aggregate was Obama +1.77, so Clinton is currently +1.81 points ahead of where Obama was four years ago. Her national aggregate jumped from 11/06 to 11/08, by +0.53.

If you look at the margins, you see a gentle rising and falling - often called the sinus-curve effect. I fully expected to see this.

And some end-statistics: I first started collecting 2016 presidential polls at the beginning of October 2013 and since then I have logged 5,056 matchups from 2,879 individual polls. You can find the breakdown to this stuff at the EXCEL table, in the tab that says "poll totals".

For past reference, I have already linked to the final electoral landscape from 2012. Here is the corresponding report from 2008, known as the FINAL POLL CONVERGENCE and the DISTILLATION thereof.


No comments:

Post a Comment

Constructive comments and critique are always welcome. Please keep it polite and respectful.