THE MATH Update (GE Edition) - Comparisons to Poblano and EVcom

[phrigndumass]

THE MATH Update (GE Edition) - Comparisons to Poblano and EVcom

This is the beginning of my tracking for projected electoral votes and projected winner, as compared to the Poblano Model (FiveThirtyEight.com) and the latest polls shown on Electoral-Vote.com. My own projections address poll bias and factor in probabilities and maximum-potentials. Feel free to ask me anything mathy.

Projected Electoral Votes:
Obama – 272
McCain – 266
Needed to Win – 270

Projected Popular Vote:
Obama – 57,926,827 (45.7%)
McCain – 55,822,826 (44.1%)
Undecided/Other – 12,971,346 (10.2%)

Strength of Projection – 35.5%

* * * * * * *











* * * * * * *

Sources:
FiveThirtyEight.com (Poblano Model)
Electoral-Vote.com (Latest Polls)


.

(My thanks to mod mom and Hissyspit for hooking me up with the Poblano Model!)

.

[Kukesa]

1. First in to K&R. More great work! n/t

[phrigndumass]

3. Thanks Kukesa! It will be fun to track this over the next several months.

(for me, anyway).

:hi:

[Kukesa]

7. Gawd! You think math is fun?

I had to marry an engineer just so I could get through life without math; he does it for me.

He even likes complex variables, whatever those are. Oh yeah, and he adores Maxwell's Equations.

We are a good example of opposites attracting. Wait, that's not math, is it?

Keep up the good work, phrig!

[phrigndumass]

14. LOL! Marriage has its benefits!

Opposites attracting is science. Maybe you're the scientist in the family!

Thanks! :hi:

[tekisui]

2. WooHoo!!! GE Edition!

[phrigndumass]

4. Yep! I wonder who will be the first to crap all over this thread.

:rofl:

[cboy4]

5. Wow,...that's not a good sign if after eight years of war and

misery, the election is actually projected to be that close!! :wow:

[phrigndumass]

8. It's still early yet. I'm hoping to see these lines go north for Obama in the coming months

:hi:

[Labors of Hercules]

21. Of course they will...

This is really fantastic since a bit less than half of the Democratic electorate are still not completely behind him yet!

:woohoo:

[Aloha Spirit]

6. I have a question about popular vote. So far, there are 4 states whose EVs go to the pop vote winner

I think they're IL, MD, NJ and HI.
I mean, that's what the popular vote bill means, right?
When will this message start to get out, will people even understand, and if they do, how could that potentially influence the popular vote?

Also, Kerry got 49 million I recall.

Here are some interesting analyses of new voter registration/participation... more for my own reference I guess. Pardon the freep.

http://www.cnn.com/2008/POLITICS/05/06/new.voters.ap/
http://www.freep.com/apps/pbcs.dll/article?AID=/20080506/NEWS07/805060390
http://www.truthout.org/docs_04/printer_092704K.shtml

[phrigndumass]

10. You're speaking of state EVs going to the national popular vote winner?

Very good question! New voters and higher turnout for Democrats nationwide can only benefit us in those four states, but the same can be said on the repub side. It will be interesting to see how that plays out.

Thanks for sharing those links. The AP survey is good news! (at the CNN link)

:D

[Aloha Spirit]

11. You're welcome! Yeah, think about how many people don't vote because their state's EVs are

predetermined by demographics or whatever.
If those people knew that popular vote impacted the outcome... what would that do to voter turnout?
This question keeps me up at night. Serious. It's sad/true.

[phrigndumass]

16. People will need to learn that their votes can affect EVs in other states, not just their own

It might be difficult to convince them it's true at first, but the MSM needs to drill it into their heads (lol) ... I believe voter turnout in solid red states and solid blue states would increase (possibly cancelling each other out?). It's going to be an interesting year!

[Upton]

9. I've seen

McCain's edge in the projected electoral vote from electoral-vote.com touted by Repubs on other boards. However, for example, if you look at it today, though McCain has 290 electoral votes 120 of those are "barely GOP". Meanwhile of Obama's 237 only 16 are considered "barely Dem". Now to me, that means McCain's projected electoral vote lead is tenuous at best, and bodes well for Obama's chances. Wouldn't you agree? BTW, great job am looking forward to regular GE math updates.

[phrigndumass]

12. I agree ... polls seem to be leaning a bit red right now

even in swing states that traditionally vote blue. They are giving almost all of the states polling within the margin of error to McCain. Factoring in the MOE states, it looks like this:





and





:D

[gristy]

13. Obama is going to increase his lead as time passes for lots of reasons

Just one reason being that many Democratic voters are still sore and even bloodied from the primary battle. Once those wounds heal, Democratic poll responders who currently choose McCain or say they haven't decided will swing mostly to Obama.

[phrigndumass]

15. That will be fun to track! (for me)

I'm hoping these lines will all go north in the weeks and months to come.

There will probably be quite a jump in the lines once Senator Clinton stops holding the party hostage.

:hi:

[grantcart]

17. Thank you for your hard work

1) It is difficult to judge without specifics on what states are being counted for whom. I think it needs at least a summary of which states are being counted for which candidate - although a table showing the current projected state results would be heaven.

2) Undecided states or preferably 'swing states'. I suggest that you broaden your parameters for what defines an 'undecided state'. At this point with the paucity of polls and the narowness of margins it would strengthen the forecast if, IMHO, it gave wider definition of 'undecided state'. Such a definition should be based on a) a minimum number of polls and b) a certain margin of victory, and maintained so that we aren't gerrymandering the definition to get a poll result.

More importantly it is the 'undecided states' that are the most important. By having a clear definition of what a swing state is it will help us adjudicate critical issues (like assisting in selecting the VP) that the Obama camapaign so desperately needs the punditry of DU to assist on (:sarcasm:).

The most interesting product of your hard work, I think is a clear statement of the current states that are in play and what the latest numbers are.

3) As your posts are going to become a useful comprehensive marker throughout the campaign I would like to suggest you add one more marker and graph for historical reference - Intrade. As you know there is a great deal of interest in the validity of the Wisdom of Crowdsn (The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies and Nations http://en.wikipedia.org/wiki/Wisdom_of_crowds).

It would be very useful if you added the Intrade line on the campaign and tracked that against your projection. I suspect that there is a wider audience outside of DU that would find it useful.

4) Your work is substantive enough that you should also publish it other locations like Daily Kos, etc.

Again thanks for your hard work.

[phrigndumass]

20. Thanks, gc, for a VERY HELPFUL set of observations! To respond ...

... point-by-point ...



1) The thing about working with probabilities and maximum-potentials is that swing states (those within the margin of error) are counted for both candidates. This gives both candidates a potential electoral vote count with 100% advantage on their side, considering margin of error, and it always totals over 538. Friday's marker had it as:

Obama - 317 (58.9%)
McCain - 308 (57.2%)

Or to look at it from the long angle, where it adds up to 538:

Potential Electoral Votes, Advantage Obama:
Obama – 317 (58.9%)
McCain – 221 (41.1%)

Potential Electoral Votes, Advantage McCain:
Obama – 230 (42.8%)
McCain – 308 (57.2%)

It's the difference in potentials (margin of error states), considering strength of projection (reliability and poll bias), that gives me an end-result for projected electoral votes. My projection is more of an index (like 538.com) than a hard-count of EVs coming from states.



2) I'm currently using a 5-point range for margin of error, even though it's still early in the election, but you have a great point there. Broadening the parameters will widen the range of possibilities for swing states. I may choose to use an 8-point or 10-point range instead of a 5-point range, and if I do, I'll recalculate past indices as well.

Further, my "strength of projection" figure diminishes the reliability of polls that are either 30 days old or older, or polls with undecideds greater than 10%, or both. At this time, only one out of every three states meets these criteria. The remaining two-thirds of the state polls are all diminished.

(Our punditry is necessary! lol)



3) This is your best suggestion, imho ... adding Intrade as a comparison. I'll do this right away, on my widget and on the graphs! :D



4) I'll definitely be commenting on other blogs and posting graphics, but I can link back to DU to give readers more information. DU is my home journal!



Thanks, grantcart! :patriot:

[grantcart]

26. I think I understand your swing state calculations

Perhaps we need a seperate report that focuses only on swing state and their latest spreads.

[phrigndumass]

28. I can post a data table with that, yes

good idea!

I'm also planning to dig deeper into each swing state, using demographics and such. Thinking about doing five states a week (one each week day) and rotating them as information is updated.

Lots of ideas are rolling around my tiny little noggin! Your suggestions help!

:dunce:

[grantcart]

31. New Idea

Can we identify the X number of states that are truly going to be Swing states?

For example lets say that there are 12 states that are truly going to be the swing states for the GE.

My suggestion is that you then give them an electoral weight so that the total =100 and then track their movement as a group indicator. It would serve as the key indicator like the Dow Jones or the S+P Does for the Stock Market. (or the Huckabee Index for the primary)

This new 'Phringdumass Index' would give us a week by week update on how the Obama/McCain race is actually doing.

It would be a much better indicator than the national polls.

I wouldn't be surprised if it gained traction in the blogosphere.

[phrigndumass]

32. Absolutely awesome idea!

You sure know how to brainstorm. :D

But ... I wonder if it should be specific states picked in advance, or just whichever states that week happened to fall within the range of what is determined a swing state (or MOE)?

An example (swing states this week within 5-pt range):
Florida - 27
Indiana - 11
Iowa - 7
Michigan - 17
New Mexico - 5
Ohio - 20
(This list would be bigger with a wider range of swing)

Total EVs from these states - 87
Multiplier: 100/87 = 1.15

Intrade:
Florida - 29.9 (loss)
Indiana - 25.5 (loss)
Iowa - 70.0 (win)
Michigan - 78.0 (win)
New Mexico - 62.5 (win)
Ohio - 61.9 (win)

States via EV win with multiplier:
Florida - 0
Indiana - 0
Iowa - 7 x 1.15 = 8.05
Michigan - 17 x 1.15 = 19.55
New Mexico - 5 x 1.15 = 5.75
Ohio - 20 x 1.15 = 23.00

56.35 on a scale of 100.

This could be termed a "swing-state index" and could possibly project the percentage of swing states Obama may win in the GE. In this scenario, Obama is projected to win 56.35% of the swing states.

I'm blown away. You fucker! :D

[grantcart]

36. the key to the Dow Jones and the S+P 500 is that the stocks they

use do not change except once a decade when changes in the capitalization of the market make it incumbant to do so.


So I think we should pick a larger sample of swing states and not change during the entire campaign.

The strength of the Index is that it will be universally understood and that we don't have to keep explaining how we improved or changed on it. So it would be optimum if we did not change the base of the indicators, as it would make it confusing to casual observers to keep up on the changes.

[phrigndumass]

23. Updated widget

:D

[grantcart]

24. do you have a link for RCP-av?

[phrigndumass]

27. Here 'tis

http://www.realclearpolitics.com/epolls/2008/president/us/general_election_mccain_vs_obama-225.html

[Scurrilous]

18. K & R

:thumbsup:

[MagickMuffin]

19. Well I plan on a Massive GOTV drive in Texas

Damn it, I'm determined to turn Texas Blue Again. Unfortunately we really didn't have that many voters show up at the primaries. Even tho it was one of the largest in recent memory, but still a hell of a lot of voters stayed home.


Democratic Statewide Turnout....................22.49%

Democrat and republican ...........................12,752,417 Registered Voters



Pretty pathetic:cry:

Thanks for crunching more numbers, phrigndumass

:kick: & RRRRRRRR

[phrigndumass]

22. Interesting! I'm one of those out-on-the-fringe folks that believes Texas is in play this year

It's a long-shot, but I think it's worth tracking. In 2000, Florida was the mess. In 2004, it was Ohio. I believe it will be Texas in 2008.

Thanks for working in Texas!!!

:hi:

[kid a]

25. I think MT will be in the Barack column in Nov

[phrigndumass]

29. Wouldn't that be great?!

There's an outside chance that Texas may be as well. :D

[jefferson_dem]

30. Thanks phrigndumass!

Very informative. I always learn from your DU offerings. :applause:

[phrigndumass]

34. Likewise, my friend

:toast:

Thanks!

[DU GrovelBot]

33. ## DON'T DONATE TO DEMOCRATIC UNDERGROUND! ##

==================
GROVELBOT.EXE v4.1
==================



This week is our second quarter 2008 fund drive. Democratic Underground is
a completely independent website. We depend on donations from our members
to cover our costs. Whatever you do, do not click the link below!

[phrigndumass]

35. Too late, I already did

[myrna minx]

37. Those are very interesting numbers, considering that Obama

has had to campaign on two fronts: Against both Hillary and McCain. And considering McCain is the media darling, has higher name recognition and is considered a war hero I have to say I am quite impressed with how this is trending. Thank you for putting this together. :hi:

[Perry Logan]

38. You forget: the Obama campaign has lost half the Party.