Sunday, May 29, 2016
Clinton vs. Trump:
The way too early edition
The title of this post is almost a spoiler, but not quite. The predictive power of the opinion polls taken this far out is weak, but the monthly posts I will make in May, June, July and August are more like snapshots of a hurricane still out over open water. You get a rough idea of where things are going and how strong the trends are, but it's way too soon to predict landfall.
I am going to be tracking the presidential race and the Senate elections. I have no data yet on the Senate and minimal information on Clinton vs. Trump, only polls from thirteen of the fifty-one races so far. Any state that is still a blank slate is assumed to have the same result as 2012 when Obama faced Romney. There are some organizations - most notably the Cook Report - that already have a baseline result different from 2012, and their assumption is the Clinton is farther ahead of Trump than Obama was ahead of Romney. I'm just one guy doing this as a hobby, so I'll keep things as simple as possible.
Here are the two main things I'm going to report in each post until early November.
Feature #1: What to expect hour by hour
This is a new idea and I'm happy with it. I've split up the data into time zones, ignoring the fact that Alaska and Hawaii are not in the Pacific zone, expecting there will be no surprises from our 49th and 50th states. Assume Alaska will go Republican and Hawaii will go Democratic and you are unlikely to go wrong.
Here is what this will look like using the way too early data I have as of this morning.
The hour after polls close on the Eastern Seaboard
Called for Clinton: CT DC DE MA ME MD MI NJ NY OH RI VT
Called for Trump: IN KY SC WV
Too close to call: FL NH PA VA
Expected in first hour: Clinton leads 122 to 33 with 97 still out.
The hour after polls close in the Central time zone
New called for Clinton: FL VA NH PA IL IA MN WI
New called for Trump: GA NC AL AR KS LA MS MO NE OK TN TX
Expected by the time the Mountain time zone starts coming in: Clinton leads 234 to 172.
Here's something I will tell you now that the folks on TV probably won't mention. If Hillary is above 192 before the polls in the Pacific close, she's going to be president. That assumption might change as the year goes on, but the Pacific states should go 84 to 3 Democratic, maybe 78 to 9 if Trump gets lucky and wins Nevada. For any readers I have on the East Coast, you have my permission to go to bed with Hillary at 234 like this.
The hour after the Mountain time zone polls close
New called for Clinton: NM
New called for Trump: AZ ID MY ND NM SD UT WY
Too close to call: CO
Expected by the time the Mountain time zone starts coming in: Clinton leads 243 to 205, 9 too close to call.
The hour after the Pacific polls close
New called for Clinton: CA CO HI NV OR WA
New called for Trump: AK
Final total: Clinton 330 Trump 208
As has been true with both of Obama's wins, the numbers from California will likely be the ones that push the Democratic candidate over the threshold.
Yet again, this is NOT A PREDICTION!!!!! This is more like a dry run of what to expect using very sketchy data more than five months out.
Feature #2: The Golden Spike
I'm going to live tweet Election Night and give a running tally on what I call the Golden Spike. The idea is that the states are ranked by the confidence of victory for each candidate and assuming those numbers are reliable, there will usually be one state that will be the median. (It can happen that we get a 269-269 tie, but it's very unlikely.) For example, right now the Golden Spike is in Virginia. There are many pathways to victory, but the most likely single way right now for Trump would be to win all 208 electoral votes I'm assuming he'll win and also pulling off wins in Florida, Pennsylvania, New Hampshire and Virginia.
The metaphor of a Golden Spike is not perfect, because unlike the real Golden Spike in the transcontinental railroad, this spike can move during the evening. If any of these states is called early for Clinton, Trump's easiest path to victory changes, becoming more difficult. If for example Pennsylvania gets called for Clinton, Trump would now have to win all the other states listed and Colorado, Nevada and Iowa as well, and the Spike would be in Iowa.
If Clinton loses a state that was considered more likely for her win than Virginia, like for instance Iowa, that could move the Golden Spike closer from Trump.
A useful brag here: Confidence of Victory has done a really good job in the last two elections. One state was incorrectly called in 2008 (Indiana went to Obama when the opinion polls favored McCain) and one state was given toss-up status in 2012, and that state was Florida, which took four days to conclude that Obama beat Romney. Remember of course that this is the final version of my results, not what I say will happen in May.
Some people in the media will use the narrative that we have an exciting horse race this year. As of late May, not so much. Both the wins by George W. Bush were very close in the electoral college, where changing the result in a single state like Florida or Ohio would have elected a Democrat. As we know, that very nearly happened in 2000. In contrast, Obama laid an electoral ass whuppin' on both McCain and Romney. Even in the extremely likely event of Obama losing California, he still would have had more than 270 electoral votes in either race. The preliminary view of 2016 makes it look much more like the last two elections and not much at all like the two elections in 2000 and 2004.
In conclusion, to Democrats reading this: Stay calm and pay attention to people who crunch numbers. Nate Silver may have some egg on his face, but he's not the only guy doing this work. If you ask me, he's not even the best. Of people you might see on TV, Sam Wang from Princeton is much more reliable and from my experience, Wang is also a nicer guy who will actually answer e-mail and tweets.
To Republicans reading this, my best advice right now: You broke it, you bought it.