The Day After: What Happened?

ELECTION POSTMORTEM 

Well...that was a long night.

And let me say, with all sincerity, congratulations to President-Elect Donald Trump. He ran the most unconventional (successful) campaign in modern presidential history and overcome what was just yesterday seen as near overwhelming odds.

So, what happened? Every prediction site was projecting a win for Hillary Clinton, this one included. The talk amongst everyone wasn't whether she would win, but by how much.

Except in the final tally (as of this morning), Trump beats Clinton 310 to 228.

There are four theories I have that account for this victory, two related to polling, and two of a political nature:

1. Polling in rural areas was off by often large margins, as non-urban turnout far exceeded what was expected,
2. Polling methods themselves are now outdated and increasingly inaccurate (both phone and internet-based),  
3. The United States is not nearly as progressive as many (most?) in the mainstream media had assumed and/or hoped for, 
4. Rural and suburban voters were fed up with the political status quo of both parties and were ready to burn down the proverbial house and start fresh.

Donald Trump, ridiculed by much of the mainstream media, disliked by much of the Republican establishment, and vilified by liberals and progressives as a racist, misogynistic buffoon, rose above all of those groups to tap into a core of anger, resentment and disillusionment to propel himself to victory. Trump openly dismissed many of the previous standards of campaigning, both politically and organizationally. 

And won. 

We'll be talking, writing, and reading about this campaign for decades. But for all the nitty-gritty analysis that will be coming starting today, one question will loom over everything: is this a personality-driven victory or an actual political movement?

Can Trump win over skeptical members of the Republican Party and reshape the GOP in his image? Can he reach across the aisle to moderates and progressives? Will he be able to build on his campaign success to govern successfully?

These are all questions that cannot be answered now, but will be in the forefront of everyone's mind as we move into the next four years under the leadership of President Donald Trump. Whether you support him or not, no one can deny one simple thing: we have absolutely no idea what's going to happen next. May we live in exciting times, indeed.

POLLING POSTMORTEM 

The entire point of this blog was to test out my customized statistical projection algorithm. So, did my algorithm work on a day when so many projections seemingly turned out to be wrong?

Yes and no. 

Yes, in the sense that when I had solid data to work with, it seemed to do no worse than projection methods used by the other major sites. And no, in the sense that it still missed far too many states by far too large a margin. My comfort there is that I am far from being alone in that respect, as the other sites also badly missed on those states.

The chart below shows my projections, the actual results, and the difference between the two. I've placed the states into five categories: "Good Call" (up to 2% difference), "Pretty Close" (2 to 4%). "Missed A Bit" (4 to 8%), "Missed A Lot" (8 to 12%), and "Not Even Close" (over 12%).

All told, I had 23 states predicted within the standard 4% margin-of-error, which is not bad but not great. On the flip side, I missed 13 states by over 10%, which is not good at all.

So what happened? 

Let's take a look at the states I completely whiffed on: Wyoming, West Virginia, Tennessee, and Alaska. All are conservative, mostly rural, and polled very lightly due to their not being remotely considered swing states. My theory is that the groundswell of rural support and turnout for Trump was vastly under-reported (and unexpected), which lead to such high-percentage misses. I also think that polling firms are going to have to rethink and redesign their own models to adjust for a more conservative landscape than they had assumed.

The same theory holds true for the "Missed A Lot" group. Again, mostly conservative (with the exception of D.C and Hawaii), mostly rural (again, with the exception of D.C), and except for Utah and to a smaller extent South Carolina and Missouri, lightly polled due to not being swing states. And Utah was a special case with the three-candidate aspect there making polling modeling more difficult.

But what about the states I got right? There, polling worked. And the states in the first two sections are a mix of urban and rural, swing and non-swing states. What's the main commonality? Look at the party affiliation: 17 out of 23 states leaned to the left, politically. 

Could it be that polling firms aren't as good at gauging support levels for conservatives as they are for progressives? I don't have an answer to that, but the results do rather support asking the question, at the very least.

For now, I'll say I've enjoyed the past six months of running this blog. I appreciate the support of the members of Fark who encouraged me to do this, and I appreciate the (rather surprising) number of people who made this site a part of their daily routine. I'll have a final wrap-up tomorrow and then...well, who knows? 

Good Call	Projected		Actual		Change
Connecticut	DEM	13	DEM	12.4	- 0.6
New York	DEM	21.9	DEM	21.2	- 0.7
Oregon	DEM	11.2	DEM	10.4	- 0.8
Indiana	GOP	15.1	GOP	16	+ 0.9
New Jersey	DEM	13.7	DEM	12.8	- 0.9
Nevada	DEM	1.3	DEM	2.4	+ 1.1
California	DEM	25.3	DEM	26.6	+ 1.3
Alabama	GOP	18	GOP	16.6	- 1.4
Illinois	DEM	14.6	DEM	16	+ 1.4
Delaware	DEM	13	DEM	11.5	- 1.5
Rhode Island	DEM	12.8	DEM	14.6	+ 1.8
Florida	DEM	0.5	GOP	1.4	+ 1.9
Texas	GOP	7.2	GOP	9.2	+ 2
Pretty Close
Georgia	GOP	3.2	GOP	5.9	+ 2.7
Maryland	DEM	29.5	DEM	32.3	+ 2.8
Arizona	GOP	1.5	GOP	4.4	+ 2.9
Michigan	DEM	2.5	GOP	0.5	+ 3
Colorado	DEM	3.5	DEM	2.6	- 0.9
Virginia	DEM	6.4	DEM	4.7	- 1.7
Vermont	DEM	30.5	DEM	28.5	- 2
Washington	DEM	16	DEM	18.5	+ 2.5
Montana	GOP	19.4	GOP	23.1	+ 3.7
Pennsylvania	DEM	2.7	GOP	1.1	- 3.8
Missed A Bit
Wisconsin	DEM	3.2	GOP	1	+ 4.2
Massachusetts	DEM	23.3	DEM	27.6	+ 4.3
New Mexico	DEM	3.7	DEM	8.3	+ 4.6
Iowa	GOP	4.7	GOP	9.4	+ 4.7
Louisiana	GOP	14.2	GOP	19.7	- 5.5
Arkansas	GOP	19.7	GOP	26.6	+ 6.9
Ohio	GOP	2.3	GOP	8.6	+ 6.3
Minnesota	DEM	7.9	DEM	1	- 6.9
North Carolina	DEM	3.1	GOP	3.8	+ 6.9
Kansas	GOP	14.6	GOP	21.9	+ 7.3
Mississippi	GOP	10.9	GOP	18.2	+ 7.3
New Hampshire	DEM	7.7	GOP	0.1	- 7.6
Missed A Lot
South Dakota	GOP	21.4	GOP	29.7	+ 8.3
D.C.	DEM	80.2	DEM	88.7	+ 8.5
Nebraska	GOP	17.7	GOP	26.6	+ 8.9
Kentucky	GOP	19.7	GOP	29.8	+ 10.1
Oklahoma	GOP	26.1	GOP	36.4	+ 10.3
Idaho	GOP	19.1	GOP	29.5	+ 10.4
Utah	GOP	6.3	GOP	16.9	+ 10.6
Hawaii	DEM	22.5	DEM	33.2	+ 10.7
South Carolina	GOP	4.9	GOP	15.6	+ 10.7
Missouri	GOP	8	GOP	19.1	+ 11.1
North Dakota	GOP	25	GOP	36.3	+ 11.3
Nebraska	GOP	16.1	GOP	27.8	+ 11.7
Not Even Close
Wyoming	GOP	35.1	GOP	47.6	+ 12.5
West Virginia	GOP	27.6	GOP	42.2	+ 14.6
Tennessee	GOP	10.1	GOP	26.2	+ 16.1
Alaska	GOP	10.1	GOP	28.3	+ 18.2

* Maine not included because, frustratingly, I can't get the numerical breakdowns for the separate congressional districts at this time. Also note that not all states have complete returns in yet, so this chart will likely change a bit between today and tomorrow.