The electoral vote trend continues to improve: Kerry still ahead

[swag]

From my favorite electoral vote trend site (but most others are saying the same thing):

http://synapse.princeton.edu/~sam/pollcalc.html

Results current as of Saturday, September 11, 2004 (click here for trends over time)

A few more polls from Survey USA. We now have a nearly full measure of the post-convention bounce - and perhaps even a small rebound. The history is updated and shows that things stand nearly the same as before the Republican convention. Note the mismatch between this electoral calculation and Bush's lead in national polls. It is indeed possible for battleground states to diverge from overall national opinion. In terms of electoral votes, the Democratic convention bounce was three times the size of the Republican convention bounce - and longer-lasting.

Today's median (expected) outcome: Kerry 279 EV, Bush 259 EV (map)

Kerry 95% confidence band: 248-322 EV

Kerry Electoral College win likelihood: 76%

[TruthIsAll]

1. Professor Wang's model and mine( 295 EV, 85% Prob) are close.

Our state models confirm each other.

I use Monte Carlo simulation and electoral-vote.com for latest state poll data.

Professor Wang uses combinatorial probabilities and a variety of sites.

One difference should be noted: As a base case, I assume 60% of the undecided will vote for Kerry. I also show the outcomes for the cases 50, 55, 67, 75%.

Professor Wang makes no such assumption, so that would account for why my numbers are slightly better for Kerry. So to compare apples and apples, compare his results to the 50% undecided allocation case.
The results for this pessimistic case: Kerry 280 EV, 67% probability.

http://www.geocities.com/electionmodel/

[swag]

2. I like your site, TIA and visit it every day

http://www.geocities.com/electionmodel/

for those who don't remember.

I get a bit puzzled that people keep posting these idiot botched national beauty contest numbers that sink so many into despair.

Repeat after me: the president is not elected by popular national vote.

[swag]

3. One more site on electoral vote count I like is www.mydd.com

www.mydd.com has some great analysis of the national beauty contest polls that people are bemoaning relentlessly in other dreary threads.

It also has this electoral vote outlook

http://www.mydd.com/outlook/president

Electoral Projection
Kerry Bush
Solid 179 196
Lean 112 51
Total 291 247
States changing hands from 2000: FL and NH to Kerry

Sorry to kick my own thread, but I can't stand to see electoral vote projections threads go below those fucking panty-twisting Time and Newsweek poll threads.

[TruthIsAll]

5. I kick my threads ALL the time. I am shameless. Like Bush.

tia

[BillZBubb]

19. Ha! You may be shameless...

but thank goodness you are NOTHING like Bush*!!!

[Mr_Spock]

4. So the popula vote doesn't decide it? Hmmm, then are those Freepers quotin

poll numbers for the popular vote here? Thanks for the realistic and upbeat post - we need more of these here!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[KerryWilWin]

6. I think the media

want to keep this as close as they can until the last minute

[swag]

7. Yeah, they have shaving supplies and deodorant to sell, after all

[DemMother]

8. Another kick

Because I can't stand to see them drop below the doom and gloomers either. All my posts on Donkey Rising die a quick death.

And, by the way, thanks for the link to this site. I'm amazed at the skills of these number-crunchers.

[Awsi Dooger]

9. Disagree, there is too much reliance here on state polls

State polls lag and are much less reliable than national polls. Zogby nailed the 2000 presidential numbers nationwide but botched numerous state projections, including Gore barely leading in California and Hillary nip and tuck with Lazio. In 2002, he screwed up many state polls especially Colorado, insisting Strickland was far ahead of Allard. That was directly a fault of a survey of independent voters who supposedly favored Strickland by a 3/1 ratio. Zogby assigned a rash proportion of undecided independents to Strickland and missed the call by more than 10%.

Just look at the Wisconsin primary race last February between John Kerry and John Edwards. Right before the primary, an American Research Group poll showed Kerry with a 53 to 16 percent lead over Edwards, and an MSNBC/Reuters/Zogby poll had Kerry with 47 percent to Edwards' 20 percent. Edwards got within six points of Kerry in the actual primary election.

State polls funded by cash strapped local radio and TV stations are notoriously poorly executed and not worth reporting. Generally, those are not given much play on DU.

I would argue polling in general is a mess right now, with too many of them and every news organization clamoring for a new big trump card story every day. There is also such a split in methodology, from 3-day tracking polls which can swing crazily from day to day because different types of people are home on a given weekday or weekend night, to RDD (Random Digit Dialing) surveys and IVT (Interactive Voice Technology) methods, the latter using a computer. Response rates vary dramatically and are far less than decades ago.

Regardless, state polls will automatically align with the national mood, as always. A legit 3 point swing toward either candidate tilts the states in the same direction and by comparable percentage, almost without exception. Kerry needs to win the popular vote. Anyone who wouldn't give 1/5 odds that the same candidate who wins the popular vote also wins the electoral vote is a prototype idiot.

[swag]

11. Bush lost the popular vote, and (let's assume FL was legit) won

the electoral vote.

National polls are as much lagging indicators as are state polls, and as you say, are just as much of a mess.

If I had to pick a horserace poll, though, I'd pick the Economist's:

http://www.economist.com/media/pdf/YouGovJ.pdf

A new one will be out this week.

The Economist: like Time Magazine, only for adults.

[Awsi Dooger]

12. My point is we cannot rely on plucking individual states

While losing nationwide. That CAN work, like a tennis player who loses the majority of the points and two decisive sets but otherwise pulls out close tiebreakers. If you depend on that, with near zero margin for error, I prefer the other guy's position every time.

2000 was a forfeited/stolen fluke, certainly much less than 1 in 5 odds the split scenario will repeat.

I continue to have more faith in national samples than state polls, based on studying both types since '96. Margin for error in national polls invariably holds, while MOE statewide is ludicrously understated. Also I'm sure Kerry is not doing as poorly nationwide as some polls continue to suggest. But on November 2 he needs to carry the popular vote, not contend.

[featherman]

13. I count state polls as part of the equation

but history and regional political identification trends count for much more. This is why I was singularly unexcited, for example, by a state poll showing Kerry leading in TN and mostly unconcerned with any Bush "lead" in PA. These states will vote their regional identification in the end. State polling is a helpful tool in confirming that which is likely.

Former DEM strongholds like WV, AR, LA and former battlegrounds like MO seem to be trending "southern GOP". Whether a strong Kerry campaign nationally can reverse this and squeak out a victory in one or more is a big question to be settled on Nov 2.

Pro GOP (2000) states like NV, AZ, CO, VA, NC seem to be trending DEM (whether it will be "enough" is another question for Nov 2).

Unlike many, I see FL as trending DEM and trouble there for Bush
Ohio, however, is an enigma and hard to call. Is it culturally more PA or more IN and KY? What is the trend there? I guess we will find out.

[Awsi Dooger]

15. Terrific post, featherman, I completely agree with virtually everything

"history and regional political identification trends count for much more"

Exactly, in fact I'm continually frustrated by the posts and polls that imply a radical shift in allegiance. It's always baloney unless there has been an underlying regional trend, like your examples. My only qualm is with Colorado, which has seen an influx of transplanted upper income Californians. I do not believe it is shifting our way. Nevada has also felt some of that, particularly northern suburbs of Las Vegas.

Here is an old thread of mine you may be interested in, listing state voting trends in relation to the national average. I may post at least some of that again closer to November 2, as a reference: http://www.democraticunderground.com/cgi-bin/duforum/duboard.cgi?az=show_thread&om=5453&forum=DCForumID22

[TruthIsAll]

14. That's why I have a National (average) AND a State EV model.

I weight the state polls by voter population to derive a national voter split. The average of the 17 national polls and the equivalent aggregated state polls should, in theory, agree with each other.

And they are very close right now (within 1%).

[Awsi Dooger]

16. I'm very impressed by your new election model

In fact, I've tried to borrow some of the concepts to update and improve my own Excel stuff, which is admittedly somewhat crude. My MIT-educated brother-in-law is working on it now, since my expertise and background is primarily sports related.

Utilizing so many national polls is an excellent idea. Most averages I see on the internet only incorporate 5 polls, or less.

[featherman]

10. I am comfortable with a 3 point lead for Bush in the horse race

because the DEMS still have the electoral college advantage. It is likely that if the 3 point spread does not improve by Nov 1 and is something like 48-45-7 then(ala Gore 2000) the actual vote will likely again come in at (say) 49-49-2. This is due to:
-DEMS in general and challengers in particular closing stronger historically
-Some systemic undercounting of DEMS in national polling
-Skewed popularity margin for Bush in his base states - i.e It doesn't matter if OK votes Bush 62-37 but OR only votes Kerry 50-47. Each candidate gets the same 7 total EV's in the end

I still see Kerry pretty solid at 264 (Gore states +NH) needing to pick up 6 more EV's from NV, FL, OH, WV, AZ, CO, MO, AR, VA,NC, LA

[yardwork]

17. This is a fan letter from a loyal reader of these threads!

With the news media hopelessly controlled by corporate advertising, White House bullying, and the hubris of ego-driven talking heads, the internet has become the nation's new town square.

Like patriots publishing broadsheets, the analyses being done on blogs and independent sites are the only dependable sources of information. Your honesty and transparency in discussing methodologys and modeling helps the rest of us make our own conclusions. Thank you!

My opinion is that many pollsters are relying too heavily on the so-called "swing" or "undecided" voters. I am not convinced that those who are undecided will ever "break" one way or the other. I think they represent disaffected and uninformed people who may be registered to vote but probably won't vote on November 2.

I am concerned that the early absentee voting allowed in many states, including North Carolina, may be locking in votes for G.W. Bush right now. This is why Kerry's campaign is wise to start focusing their message and hitting Bush harder.

This election will depend on turnout, I think. Those who are motivated to vote early and those who are motivated to go to the polls on November 2 will determine the outcome. Kerry's campaign needs to focus on them, not the shadowy wishy-washy undecideds.

Red meat for the Democrats is what is needed now, imo.

[featherman]

18. Agree that turnout is the key

and more important than by what percentage the undecideds break. With a huge "usually not voting" population of 40-50%, this is a tremendous resource never measured by polls. I have a hope that a highly motivated anti-Bush & GOP record turnout is the DEMS secret weapon this year. All signs (broad anger, record registration, high primary turnout) seem to point to it. GOP and media are desperately working to suppress it. But for now I think we have to work with the 39-35-26 (DEM-GOP-IND) proportions that voted in 2000.

[papau]

20. Wash Post disagrees

Are only 10 states still "competitive states" (as suggested by Wash Post)


Does DU agree with Dan Balz of the Washington Post that only 10 states are still "competitive states"

Florida, Pennsylvania, Ohio, Minnesota, Wisconsin, Iowa, Nevada, New Mexico, West Virginia and New Hampshire, with West Virginia and Nevada major, major challenges for Kerry - more so than the other 8.

Are the other 11 "battleground states" really leaning too far to be called "competitive, because - per Baltz- we have -7 Leaning toward Bush which are: Arizona, Arkansas, Colorado, Louisiana, Missouri, North Carolina and Virginia - with 4 leaning toward Kerry which are: Maine, Michigan, Oregon and Washington.

And of the 10 that are still "competitive", the Bush team will be putting major money to win into Minnesota, Iowa and Wisconsin, while they view Pennsylvania and Wisconsin as states they can make Kerry work very hard to defend, and see West Virginia and Nevada as quite likely to go their way.

Is it really down to a huge effort to fight the above battle, while the winner of 2 out 3 of Florida, Pennsylvania and Ohio wins the election?

[swag]

21. I haven't read Balz' assertions

Maybe you could help us by telling us what polls or sources he's relying on.

Thanks in advance.

[papau]

22. He relies on "discussions with Bush and Kerry camps" - the

state by state polls are not really laid out in todays article.

[featherman]

23. This is kind of a silly list from this WP writer

Very much tilted to some kind of "popular President Bush" bias.

It is misinformation to list OR, WA, ME, MI, NH as competitive. More competitive than, say, Kansas or New York, I suppose, but they will all be DEM in the end. Bank on it. I'd probably add PA and NM to this list but will cover them in next group for the sake of argument.

Of his 10 so-called battlegrounds, five are historically DEM and likely to remain so, closely perhaps. I have seen no real signs, however, to indicate these will actually flip to GOP in 2004:
PA, MN, WI, IA, NM

The "anybody could win" battlegrounds at this point are probably:
NV, FL, OH, WV (all Bush 2000 states)

Kerry's possible flips include all seven of the above listed as leaning GOP. I don't disagree that historical trends indicate they are leaning GOP at this time but consider some of them more vulnerable to going DEM than his list of "leaning to Kerry" are to going GOP:
AZ, CO, MO, AR, LA, NC, VA

The GOP strategy of going after MN, IA, WI is a good one, though. My guess is that the target there is white Catholics. All are polling close (about +2 for Kerry) on the one hand but all three have voted DEM for 4 straight elections on the other. A wildly popular GOP candidate might pull it off.

Of the "2 out of 3" question so prevalent in 2000, it is a different 3 this year and the question is different: PA is likely DEM so Bush actually needs to sweep MO, OH, FL again to have a chance. Kerry taking one of these effectively locks up the electoral college majority.
Caveat: How does Bush win without taking all three? Only by winning two of MN, IA, WI.