Saturday, October 15, 2016

Election Update: 10/15/16

 Okay, let's get to it!

Aggregate polling - 6.7% Clinton

Wow, Clinton is way ahead now.

Aggregate polling with third party candidates - 5.3% Clinton

We see a slight "spoiler effect" here, but it's not a huge deal.

Electoral College

Okay, a few changes. Since the polling is much better for 4 way races now, I'm going to be using those numbers when available instead of the 2 way numbers for the sake of accuracy. The reason I used 2 way polls up to this point is third party polling was spotty with fewer data points to go by (which makes it harder to detect trends), and because most of these polls were within 1-2% of the 2 way polls anyway. Still, I've noticed a very significant uptick in third party polling lately, so I'm switching over for accuracy's sake. This *IS* a four+ way race after all in practice, even if only two candidates have a chance of winning. I'm also not going to count states that go more than 8 points in one direction any more because those are outside of the margin of error assuming a very generous margin of error of 4 points.

Arizona - 1.0% Trump

Colorado - 7.3% Clinton

Florida - 2.9% Clinton

Georgia - 5.5% Trump

Indiana - 4.5% Trump

Iowa - 3.7% Trump

Maine - 4.7% Clinton

Maine CD2 - 5.4% Trump

Minnesota - 4.3% Clinton

Missouri - 8.0% Trump

Nevada - 1.5% Clinton

New Hampshire - 3.6% Clinton

North Carolina - 2.9% Clinton

Ohio - 1.6% Clinton

South Carolina -  7.6% Trump

Utah -  7.5% Trump

Virginia - 6.7% Clinton

Wisconsin - 6.7% Clinton


Election Scenarios

I'm going by a 4 point margin of error instead of 3 now because most polls are around 4 points margin of error, not three. This means I will be doing 1%, 4%, and 8% handicaps instead of 1%, 3% and 6% in these scenarios.

Most likely scenario (best guess) - 340-198 Clinton 

Last week's most likely scenario - 322-216 Clinton

Most likely scenario with no swing states (<=1% lead) - 340-187 Clinton

Clinton + 1 - 351-187 Clinton

Trump + 1 - 340-216 Clinton

Most likely scenario with no swing states (<=4% lead) - 268-181 (no winner)

Clinton + 4 - 357-181 Clinton

Trump + 4 - 268-270 Trump

Most likely scenario with no swing states <=8% lead) - 223-128 (no winner)

Clinton + 8 - 410-128 Clinton

Trump + 8 - 223-315 Trump

Change in method

After some consideration, I decided to change the margin of error from 3% to 4%. This is because the hypothetical 3% MOE assumes perfect sampling conditions, and most polls don't have them. If you look at the MOE on the actual polls used in this analysis, the average is actually closer to, say, 4%.

I will warn you, this will make the percentage a candidate will win a bit more conservative. Since a standard deviation is now 4% instead of 3%, that means the number of standard deviations required to flip the election will be lower, and this will lower the percentage. We might also see some changes to the percentage because there is a small "spoiler effect" that benefits Trump. That being said, any change to Clinton's chances should not be taken too seriously this particular analysis because these changes will appear to benefit Trump's chances. I will be changing the previous analysis' margin of error to 4% and adjust the percentages accordingly though. That way the only real change will be the switch from two party polling to multi-party polling.

Current statistics 

Current tipping point: Trump + 3.6% - 268-270 Trump

The state that tips it: New Hampshire (+3.6% Clinton)

Standard deviations: 0.9

Current chance of Clinton winning: 81.5%

Current chance of Trump winning: 18.5%

Previous statistics

Previous tipping point: Trump + 4% - 262-276 Trump

The state that tips it: Minnesota (+4.0% Clinton)

Standard deviations: 1.0 (according to new methodology)

Previous chance of Clinton winning (revised): 84.1%

Previous chance of Trump winning (revised): 15.9%

Discussion

I would not put much stock into Clinton apparently declining slightly. Once again, I changed some of my methods and while I compensated for that somewhat, I cannot compensate for the change in switching from two party to multiparty polling data. As I said, there is a SLIGHT spoiler effect that works against Clinton and for Trump. If anything, looking at the two party polls while switching over, Clinton's chances are improving. So it's possible if I used third party data for my last analysis that her chances would have been lower than they are now, not higher. I mean she did gain 18 electoral votes in the newest scenario, after all. Regardless, 81.5% is a good number for Clinton and a bad number for Trump. It seems very likely she will win the election.

No comments:

Post a Comment