Political Junkie's Guide To The Elections - Week Ending September 9, 2016 (Premier 2016 Edition)

September 10, 2016 | Political Junkie Too

[Political Junkie Too]

Previous Editions:

November 6, 2012 Post Mortem

This is the premier edition of my 2016 state-by-state presidential election model.

I made several significant changes to the model from the previous election cycle:

• The model no longer uses Rasmussen as the polling source. I moved to a poll averaging model based on the polls listed at Real Clear Politics. I do not use their poll averaging, I use my own method, but I use their listed polls.
• I am not using a random Monte Carlo simulation of 32,000 iterations. I now use a 1,000 iteration methodology based on Stochastic Information Packages developed by the Probability Management Institue.
• I am now correlating the randomizations. In order to reduce the likelihood of infeasible combinations, I am correlating each state to the national polls. The correlation set includes the final election results of 1996, 2000, 2004, 2008, 2012, and the current poll average.
• I am aging the polls, giving more weight to recent polls. Older polls that are still included in the averaging have less weight than newer polls.

I don't have access to secure anonymous file hosting anymore, so I might be light on the graphics this time around. I won't be able to post the animated maps that show the poll changes over time.

The Race for the White House

I started this cycle over Labor Day and updated the model with changes throughout the week in order to show a first week delta. Below are the state rankings as of Labor Day followed by the state rankings as of Friday.

As of now, Donald Trump has an expected Electoral Vote count of 231 versus Hillary Clinton's 307. Trump's probability of winning is 29%, up from last week's 21%.

I've been posting that I want to see movement in the state polling, and this week the movements are starting.

In Colorado, a poll of 500 likely voters by Magellan was the first to be within the margin of error (Clinton 41%-36%, MOE 4.4%). This moves Colorado from Safe to Strong for Clinton.

In Missouri, a poll of 1,250 likely voters by Remington Research was the first to be outside the margin of error (Trump 47%-38%, MOE 3.0%). This moves Missouri from Lean to Strong for Trump.

In New Hampshire, a poll of 600 likely voters by Emerson was the first to be within the margin of error (Clinton 42%-37%, MOE 3.9%). This moves New Hampshire from Safe to Strong for Clinton.

In New Jersey, a poll of 800 likely voters by Emerson (Clinton 47%-43%, MOE 3.4%) is replacing all previous polls (two of registered voters), the most recent being from June. This moves New Jersey from Safe to Strong for Clinton.

In Ohio, a pair of polls from Quinnipiac of 775 likely voters shows Trump leading 46%-45% in 2-way race, and leading 41%-37% in a 4-way race (MOE 3.5%). This swings Ohio from Leans Clinton to Leans Trump.

In Rhode Island, its first poll is of 800 likely voters by Emerson (Clinton 44%-41%, MOE 3.4%). This moves Rhode Island from Safe to Strong for Clinton.

Using an average of polls from Real Clear Politics, the results of simulated elections are listed in the table below. The definitions of the columns are:

• "GOP Electoral Votes P10” – the value where the probability of the actual number being lower is 10%.
• ”GOP Electoral Vote EV” – the expected value (mean) or probability-weighted average of the simulation.
• ”GOP Electoral Vote P90” – the value where the probability of the actual number being lower is 90%. Or, it is the value where the probability of the actual number being higher is 10%.
• ”Probability of 270” – The probability associated with value 270 (the Electoral College majority goal).

Week	GOP Electoral College P10	GOP Electoral College EV	GOP Electoral College P90	Probability of 270
03-Sep-16	148	215	295	21%
09-Sep-16	152.2	231	298	29%

State Rankings Definitions

• Safe - 100% probability of winning
• Strong - 75% to 99.99% probability of winning
• Lean - 62.5% to 74.99% probability of winning
• Toss-Up - 50% to 62.49% probability of winning

9-Sep-16 State Rankings

Clinton - 279								Trump - 199
Safe		Strong		Leaning		Toss-Up		Leaning		Strong		Safe
EV	State	EV	State	EV	State	EV	State	EV	State	EV	State	EV	State
55	California	9	Colorado	6	Nevada	29	Florida	11	Arizona	3	Alaska	9	Alabama
7	Connecticut	16	Michigan	13	Virginia	16	Georgia	18	Ohio	6	Iowa	6	Arkansas
3	District of Columbia	10	Minnesota			15	North Carolina			6	Kansas	4	Idaho
3	Delaware	4	New Hampshire							10	Missouri	11	Indiana
4	Hawaii	14	New Jersey							9	South Carolina	8	Kentucky
20	Illinois	7	Oregon							6	Utah	8	Louisiana
4	Maine	20	Pennsylvania									6	Mississippi
10	Maryland	4	Rhode Island									3	Montana
11	Massachusetts	10	Wisconsin									5	Nebraska
5	New Mexico											3	North Dakota
29	New York											7	Oklahoma
3	Vermont											3	South Dakota
12	Washington											11	Tennessee
												38	Texas
												5	West Virginia
												3	Wyoming
166		94		19		60		29		40		130

Labor Day State Rankings

Clinton - 297								Trump - 181
Safe		Strong		Leaning		Toss-Up		Leaning		Strong		Safe
EV	State	EV	State	EV	State	EV	State	EV	State	EV	State	EV	State
55	California	16	Michigan	6	Nevada	29	Florida	11	Arizona	3	Alaska	9	Alabama
9	Colorado	10	Minnesota	18	Ohio	16	Georgia	10	Missouri	6	Iowa	6	Arkansas
7	Connecticut	7	Oregon	13	Virginia	15	North Carolina			6	Kansas	4	Idaho
3	District of Columbia	20	Pennsylvania							9	South Carolina	11	Indiana
3	Delaware	10	Wisconsin							6	Utah	8	Kentucky
4	Hawaii											8	Louisiana
20	Illinois											6	Mississippi
4	Maine											3	Montana
10	Maryland											5	Nebraska
11	Massachusetts											3	North Dakota
4	New Hampshire											7	Oklahoma
14	New Jersey											3	South Dakota
5	New Mexico											11	Tennessee
29	New York											38	Texas
4	Rhode Island											5	West Virginia
3	Vermont											3	Wyoming
12	Washington
197		63		37		60		21		30		130

[Political Junkie Too]

This is the first edition of 2016.

I intend to produce this weekly, just as I did in 2012. I may add the Senate races later.

-PJ

[Political Junkie Too]

Reviving my 2012 ping list.

Let me know if you want on or off the list.

Comments are welcome and much appreciated.

-PJ

[be-baw]

I would like to be on your ping list. Clearly you take this seriously, and I’m curious to see how it all pans out. Thanks.

[CatOwner]

Well, that was depressing. :(

[MEG33]

On, please

[whiterhino]

I would love to see your summaries as you have them plus any adjustments for polls that over/under sample based on gender, race and political affiliation. That would be very interesting so we can see the level of adjustments these pollsters make to generate their desired outcome.

[matt1234]

Interesting. All the polls you specifically named are moving in Trump’s direction.

[bobfeland]

Please add me to your ping list!

[Political Junkie Too]

I've played with adding a bias adjustment in the past, but have turned it off for now. I was simply adding a +1 or +2 to the totals. I once even made it a random uncertainty, since it wasn't backed up with any evidence.

I don't have the resources to look at in that detail, and I can't really compete with sources like fivethirtyeight.com. Nate Silver looks at polling "house effects," as well as historical quality, turnout predictions, and an extra special sauce.

I am taking the polls as they are, and assuming that the polling companies have already taken into account various turnout assumptions. I hope that by averaging different polls together, that the bias can be smoothed out somewhat.

The best I can do is show the changes over time to the polls as they are released, and leave it the reader to assume the rest. If state polls lag too much behind national polls, or they don't align with people's on-the-ground experiences, then let that show up as comments on websites and calls to radio shows.

-PJ

[fieldmarshaldj]

FL, GA & NC should all be under Trump.

[Political Junkie Too]

Well, that was depressing. :(

See it as an 8% growth from last week.

-PJ

[Political Junkie Too]

Interesting. All the polls you specifically named are moving in Trump’s direction.

That's what I'm hoping to see more of.

The national polls showing Trump closing or leading are of no consequence, since it's the state's Electoral College votes that matter. If the national vote moves, I expect the states to move, also.

-PJ

[Political Junkie Too]

FL, GA & NC should all be under Trump.

Show me the state polls, and I'll add them to the model.

-PJ

[mrsmith]

Oh you’ll see plenty more movement as the pollsters update their “likely voter” profile.
Heck, many are still using the 2012 turnout.

[fieldmarshaldj]

Which ones ? They’re all over the map. You can safely put GA and NC based solely on the 2012 results alone. Trump will carry every Willard state as a bedrock.

[InterceptPoint]

Bless you. We need this and I am so glad you’re up to the task.

A couple things you might consider that might make your job easier. (But they might make it harder as well.)

1. You could consider putting your tables into Excel and using Snipping Tool to save as a jpg file. Post the image at TinyPic.com and then stick it in your post.

2. If you do the above you could color code the changes Week over Week. Pink for a favorable movement in our direction. Light Blue for the Dems. Red and Dark Blue for a 2 Category Jump.

Using the Excel/jpg scheme would also facilitate posting your very valuable Weekly Chart to Twitter and Facebook and the like and including them in email to the WP, LAT, Fox News and your real friends.

Just a thought. In any case I am glad to be on your Ping List and very happy to see you continuing your project. And your methodology looks first rate.

[Nifster]

Ignore all media

Get Out The Vote

[Political Junkie Too]

They’re all over the map.

Which is why the model shows it as a toss-up right now.

You can safely put GA and NC based solely on the 2012 results alone.

I will, as soon as the polls show it. This isn't a model of what I want, it's a model of what the data says. If the data is corrupt, then people will call the pollsters out on it sooner or later.

-PJ

[AU72]

I don’t see 2016 turnout in any way mirroring 2012. There will be an inherent Hillary bias of 3-5% until late October.

[EDINVA]

Please add me to your list, thanks!