So, this is my last election update for this cycle. Let it be known that this is how I think election day is going to play out. Let's see how good my methods are compared to the result.
Aggregate polling - 3.0% Clinton
Aggregate polling with third party candidates - 3.3% Clinton
Electoral College
All states not listed here are considered "safe", which means they have an 8%+ margin in favor of one candidate, which should net them a 98%+ chance of winning.
| States | Winner/Margin | SD | % Clinton Win | % Trump Win | Electoral Votes if Clinton Wins | Electoral Votes if Trump Wins |
| Oregon | 8.0% Clinton | -2 | 97.78% | 2.22% | 182 | 363 |
| Wisconsin | 6.5% Clinton | -1.63 | 94.84% | 5.16% | 192 | 356 |
| Minnesota | 6.0% Clinton | -1.5 | 93.32% | 6.68% | 202 | 346 |
| Virginia | 5.0% Clinton | -1.25 | 89.44% | 10.56% | 215 | 336 |
| New Mexico | 5.0% Clinton | -1.25 | 89.44% | 10.56% | 220 | 323 |
| Maine | 4.5% Clinton | -1.13 | 87.08% | 12.92% | 223 | 318 |
| Michigan | 3.4% Clinton | -0.85 | 80.23% | 19.76% | 239 | 315 |
| Colorado | 2.9% Clinton | -0.73 | 76.73% | 23.27% | 248 | 299 |
| Pennsylvania | 1.9% Clinton | -0.48 | 68.44% | 31.56% | 268 | 290 |
| New Hampshire | 0.6% Clinton | -0.15 | 55.96% | 44.04% | 272 | 270 |
| Florida | 0.2% Trump | -0.05 | 48.01% | 51.99% | 301 | 266 |
| Maine CD2 | 0.5% Trump | 0.13 | 44.83% | 55.17% | 302 | 237 |
| Nevada | 0.8% Trump | 0.2 | 42.07% | 57.93% | 308 | 236 |
| North Carolina | 1.0% Trump | 0.25 | 40.13% | 59.87% | 323 | 230 |
| Iowa | 3.0% Trump | 0.75 | 22.66% | 77.34% | 329 | 215 |
| Ohio | 3.5% Trump | 0.88 | 18.94% | 81.06% | 347 | 209 |
| Arizona | 4.0% Trump | 1 | 15.87% | 84.13% | 358 | 191 |
| Georgia | 4.8% Trump | 1.2 | 11.51% | 88.49% | 374 | 180 |
| South Carolina | 6.3% Trump | 1.58 | 5.71% | 94.29% | 383 | 164 |
and what the electoral college will look like if a candidate wins that state and every previous state they are more likely to win.
Most likely outcome: Clinton wins 272-266
If I had to implement my best guess, no toss up map, that is what I would guess. Clinton wins, narrowly, in Florida and New Hampshire, but Trump takes Maine CD2.
A Clinton victory is not certain
As you can see by my chart above, the three states, New Hampshire, Florida, and Maine CD2 are very much toss ups. Either candidate can win them since the margin in them is by less than a point. While the terrain favors Clinton and she's been playing defense all election, if Trump launches an offense in those three states and wins, he wins the election 270-268. While my larger model doesn't take into account nuance like this, we might even see a tie in the electoral college. If Trump takes Florida and New Hampshire, but loses Maine CD2 (total opposite of what is most likely to happen in these three states), we could actually see an electoral college tie of 269-269. This would likely be a de facto victory for Trump since the election would go to the house of representatives, which is controlled by the republicans.
All things considered, if we follow my model in the chart above, I predict a 56% chance that Clinton will win, and a 44% chance Trump will win, with New Hampshire being the deciding state.
Another Methodology
Another way of going about calculating the probability of a win is by focusing on the five toss up states and treating them as separate races not related to each other, rather than viewing all the races together as trends. This means we view the probability of Trump carrying the states needed to win and then calculating the probability of winning all 5. This is relatively easy. Out of the combinations of those states and their outcomes, only one outcome means Trump will win: he needs to win all 5. Likewise, I can calculate the probability of Trump winning all of them except Maine CD2, which is needed for a tie. Any other outcome means Clinton will have enough electoral votes to win.
That said, probability of Trump winning all 5 states: 3.18%
Probability of a tie: 2.58%
Probability of a Clinton win: 94.24%
By this particular model, Trump has an extremely low chance of winning. This is because among the 5 states in play, there are 32 outcomes of how these states can play out, and only 1 of those leads to a Trump win, and one of them leads to a tie. The other 30 outcomes lead to a Clinton win because if she can hold at least one state besides Maine CD2, she wins. By this model, Trump's chances are bleak indeed.
Which methodology is more accurate?
Well, it depends on how you view the race. Are all of the races interconnected in which general trends shift the electoral college? Or are they all like coin flips, each independent of the other? Considering how the current scenarios are playing out similar to my old Trump +2 Trump +3, and Trump +4 models, there is a lot of legitimacy to the trend model on the whole, which is a huge reason I don't even consider states that aren't in the toss up region in my calculations. However, given the close margins of this race and how these 5 states are effectively toss ups, since they have leads of 1% or less, maybe the independent trial theory should be given more consideration. After all, trend or not, it really is too close to tell and the slightest variation from the polling data can shift the state to the other candidate.
All in all, I think my trend model is too forgiving for Trump, and potentially overestimates his chances, and the toss up model is too strict and underestimates them. We won't know until tomorrow.
Other ways Trump can win
I'm not going to get into the probability of these, but I will say this, Florida seems essential to a Trump win. It's 29 electoral votes, and if Trump does not win here, he cannot win the election unless he carries Maine CD2, New Hampshire, North Carolina, Nevada, Pennsylvania, and Colorado to make up for those electoral votes (scenario here). The chance of him carrying all of these of these is slim. If Trump carries Florida but loses one other state, he can still win by picking up Colorado or Pennsylvania.
Summing it up
Looking at this a few months ago, I never expected the election to get this competitive. Whereas before the chance of Trump winning involved a lot of outlandish hypotheticals, now he is within striking distance of the white house, and what used to be Trump +3-4 scenarios are now the real thing or Trump +1 at best. However, the terrain still favors, and has always favored, Hillary Clinton. I expect her to win. I don't think it will be by a huge margin, and she will get it by the skin of her teeth, but she should be able to pull it off. However, there is a very real chance Trump can be the next president and it's anyone's game.
No comments:
Post a Comment