[BlackGriffen]

Just wanted to keep everyone abreast of predictions of what's going on in the electoral college:
Electoral-Vote.com with history graph.

Pollkatz also has a projection with a history graph and map.

Note that their methodologies are very different. The first uses the most current major poll, no matter how old. The second takes the polls from the last five days and extrapolates the unpolled states using voting record correlations among the states.

Summary:

• Electoral-vote.com - Kerry ahead, 280 to 238
• Pollkatz - Kerry ahead, 315 to 203

I find Pollkatz's history graph to be especially interesting. It shows that there's little fluctuation in the hard core electoral votes for each candidate, and much fluctuation in the middle ones. Kerry's hard core: about 270 votes (270 needed to win). Bush's hard core: about 150 or so.

BlackGriffen

[SimeyTheLimey]

[Zimphire]You're projecting[/Zimphire]

[Logic]

Originally posted by SimeyTheLimey:
[Zimphire]You're projecting[/Zimphire]

[MindFad]

KERRY: LANDSLIDE.

[spacefreak]

I prefer to let the actual election take place, thanks.

[AKcrab]

Originally posted by spacefreak:
I prefer to let the actual election take place, thanks.

Why? It's the electoral that counts. All the voting is just a show.

[CRASH HARDDRIVE]

Either side getting overly-cocky about victory at this point needs to take a chill, or else they could be in for some major crow eating in November.

Several things amaze me in that electoral-vote.com chart.

In particular- California has gone from �Strong Kerry� to Barely Kerry. The number of �Strong Kerry� states has dwindled- there are currently only a small handful of �Strong Kerry� states. Unquestionably, Kerry has lost support, and Bush has gained it, if California is any indicator.

Meanwhile Bush has a slew of �strong� states in his corner, and any number of states are still very much up for grabs by either side.

A recent LA Birdcage-Liner Poll poll of registered voters shows Bush slightly leading Kerry 49% to 46%.

The poll also contradicts some assertions I�ve seen floated that Republicans en mass will vote for Kerry, and fewer Democrats will vote Bush, IE:

With independent voters splitting evenly in the survey between the two men, one key to Bush's tentative new advantage was his greater success at consolidating his base. While just 3 percent of voters who called themselves Republicans said they would vote for Kerry, Bush drew 15 percent of all Democrats, and 20 percent of Democrats who consider themselves moderate or conservative, the poll found.

All of this before the RNC.

Of course any of this will probably bounce back and forth between now and November, but the point is, it�s silly to get overly-cocky one way or another.

Whoever wins, I think it�s going to be a tight scrape again.

[soul searching]

Originally posted by CRASH HARDDRIVE:
...
Several things amaze me in that electoral-vote.com chart.

In particular- California has gone from �Strong Kerry� to Barely Kerry. ...

I found this interesting too, given that California has donated more money to the Kerry campaign than any other state and the amount of campaigning he's done in CA. Maybe the Governator has something to do with moving the state a bit to the right.

EDIT: moving not tilting since CA is still a Demi state.

[BlackGriffen]

Ouch, bad news for Kerry in the latest EC forecasts.

Electoral Vote.com: Kerry 270 Bush 259
PollKatz: Kerry 353 Bush 185

Kerry's trending down in the first projections. Pollkatz is confusing me, though. The forecast (now with state by state breakdown and victory probability in each state) has Kerry's position improved from last time I looked. The trend graph, though, paints a very different picture. In the trend graph, Kerry's nose dived to just over 270 (min needed to win), but more importantly, his hardcore states nose dived to 150 or so (down from nearly 270).

All I can say for sure is that it's far from over. We'll see what sort of Impact the RNC and attendant protests have.

BlackGriffen

[BlackGriffen]

Newest data from Pollkatz has Kerry improving ever so slightly (probably statistical noise) and electoral-vote.com has both Bush and Kerry down.

Electoral-vote.com: Kerry - 249, Bush - 232, dead even - 57
PollKatz: Kerry - 282, Bush 256

With the obligatory history graph.

Also fun: a graph comparing Gore/Kerry poll history and a map of middle earth.. I mean "America"

BlackGriffen

[zigzag]

Originally posted by BlackGriffen:
Newest data from . . . electoral-vote.com has both Bush and Kerry down.

Maybe by November, support will have dried up for the both of them.

Unless something unexpected happens at the RNC, Bush will probably get a bounce, and then it will be up and down until election day.

This reminds me of technical stock analysis, in which people make pantloads of money saying essentially nothing more than "I expect this stock to go up, unless it goes down."

[itai195]

Originally posted by CRASH HARDDRIVE:
In particular- California has gone from �Strong Kerry� to Barely Kerry. The number of �Strong Kerry� states has dwindled- there are currently only a small handful of �Strong Kerry� states. Unquestionably, Kerry has lost support, and Bush has gained it, if California is any indicator.

It doesn't look much different than the 2000 results to me. There also seems to be a trend in CA politics over the last decade that polls trend more and more Democratic as the election nears. In particular there have been a couple of recent Senate and Governor's races that appeared close until the Democrat pulled away in the last few weeks. Probably the more important point is that there is still a substantial, committed conservative base in CA, which makes this state somewhat unusual and probably an inappropriate starting point for evaluating largescale trends. Republicans can win here, but they need to be moderate enough to attract a few Democrats... they tend to lose here because the CA Republican base is anything but moderate.

[BlackGriffen]

Originally posted by zigzag:
Maybe by November, support will have dried up for the both of them.

Unless something unexpected happens at the RNC, Bush will probably get a bounce, and then it will be up and down until election day.

This reminds me of technical stock analysis, in which people make pantloads of money saying essentially nothing more than "I expect this stock to go up, unless it goes down."

Heh, what you say might come true. The Columbus Dispatch (reg required, secondary source) has the race tied up at 46%, MOE 2% (nearly 3,200 people polled). That will knock Bush down by another 20 on electoral-vote.com, without giving Kerry anything.

The good news is that the trend is in Kerry's favor - previous results from the same poll had Kerry 45% Bush 47%. Well, as much as a shift that much smaller than the MOE could be considered a trend, anyway.

BlackGriffen

[itai195]

Originally posted by zigzag:
This reminds me of technical stock analysis, in which people make pantloads of money saying essentially nothing more than "I expect this stock to go up, unless it goes down."

LOL, so maybe Bush's RNC bounce has already been 'priced' into poll results by prospective voters, thus his results will drop after the convention. Good thing elections don't work like the stock market!

[chris v]

What's interesting is how the results vary by who's doing the polling. Observe the "Bush Dissaproval" timeline:

http://www.pollkatz.homestead.com/fi...2_image001.gif

Zogby is near the top of the trend always, and Fox is near the bottom. Gallup seems to run right up the middle between them, most of the time.

[OreoCookie]

Originally posted by SimeyTheLimey:
[Zimphire]You're projecting[/Zimphire]

That one was good.

[SimeyTheLimey]

Originally posted by BlackGriffen:
Ouch, bad news for Kerry in the latest EC forecasts.

Speaking of Electoral-vote.com . . .

Of course, I still think these predictions are meaningless.

[Spliffdaddy]

Dubya 280 Kerry 242

Looks like four more to me.

[Spliffdaddy]

Originally posted by MindFad:
KERRY: LANDSLIDE.

At heart, we're all dreamers.

Thanks for reminding me that people can still dream - even if that dream has a rather remote chance of surviving in the realm of reality.

[greenamp]

These electoral forcast things are all completely absurd. TN going "barely Kerry?" riiiight.

[BlackGriffen]

Originally posted by SimeyTheLimey:
Speaking of Electoral-vote.com . . .

Of course, I still think these predictions are meaningless.

D'oh! I was waiting for that one to blow over.

Pollkatz has some good news for Kerry, though: the graph shows a slight resurgence in the number of deep blue states.

BlackGriffen

[BlackGriffen]

Originally posted by greenamp:
These electoral forcast things are all completely absurd. TN going "barely Kerry?" riiiight.

That's because the latest poll he's using is Zogby (he always goes with the most recent major poll, regardless of source, averaging if there are multiple form the same day, making his data somewhat unstable). If you check out this graph (from Pollkatz), you should notice that Zogby is always leaning anti-Bush by a couple of %. TN may not be "barely Kerry" but it isn't very heavily in Bush's favor, either.

FWIW, the PollKatz forecast, if the election were held today, predicts Kerry's vote share at 49.5%, with a 33% chance for Kerry to pull out an unexpected win.

BlackGriffen

[BlackGriffen]

Originally posted by SimeyTheLimey:
Speaking of Electoral-vote.com . . .

Of course, I still think these predictions are meaningless.

Interesting - this is electoral-vote.com's actual projection of what will happen in Nov (based on linear regressions of the state by state polling data).

Kerry - 298
Bush - 220
Too close to call - 20 (MN and WI)

Considering what I said about how unstable his data is (mixing different polls and all that), I wouldn't trust that predicted much, myself. This map, too, will probably go Bush's way soon after the RNC.

BlackGriffen

[BlackGriffen]

The only thing really new for today is the update to the Pollkatz graph. If it is to be believed, the race is tightening as states exit both the "Strong Bush" and "Strong Kerry" camps.

BG

[greenamp]

Originally posted by BlackGriffen:
The only thing really new for today is the update to the Pollkatz graph. If it is to be believed, the race is tightening as states exit both the "Strong Bush" and "Strong Kerry" camps.

BG

"the race is tightening"

I have a feeling that statement will go down in history as one of the biggest media spins of the decade. Right next to the early poll callings of 2000.

[greenamp]

Actually, I am really surprized that you post anything from that website as truth. You seem to be a little more open minded than that to me.

pulled from top of page on www.pollkatz.homestead.com:

"Photo Mosaics:
Bush from dead soldiers
Ashcroft from X-rated
cheesecake pictures
Which is more obscene? "

[BlackGriffen]

Originally posted by greenamp:
Actually, I am really surprized that you post anything from that website as truth. You seem to be a little more open minded than that to me.

pulled from top of page on www.pollkatz.homestead.com:

"Photo Mosaics:
Bush from dead soldiers
Ashcroft from X-rated
cheesecake pictures
Which is more obscene? "

Either his raw data is wrong or his analysis is wrong. I know it's not easy, but the only way to discredit him is to critique his mathematics. He isn't doing the polling, so he can't be biasing those.

Even a blind partisan can do math right.

BlackGriffen

[greenamp]

Originally posted by BlackGriffen:
Either his raw data is wrong or his analysis is wrong. I know it's not easy, but the only way to discredit him is to critique his mathematics. He isn't doing the polling, so he can't be biasing those.

Even a blind partisan can do math right.

BlackGriffen

All I'm trying to say is, it's obvious who he wants to win. The fact that his bias is clearly seen on his website is enough to discredit his info.

I would say the same about partisan Bush polls too.

[BlackGriffen]

Originally posted by greenamp:
All I'm trying to say is, it's obvious who he wants to win. The fact that his bias is clearly seen on his website is enough to discredit his info.

I would say the same about partisan Bush polls too.

Thing is, he isn't conducting any polls. He's just taking poll data, from all the major polls, and analyzing it. You can show fault with his analysis. You can say he's falsifying the poll data he claims comes from major polls. You can even disagree with how he analyzes his data (ie he uses correlations between state voting records in order to extrapolate data from states being polled to states that aren't). In other words, you either have to make a technical argument, or call him an outright liar. Any other claims are moot.

BlackGriffen

[BlackGriffen]

Things still looking bad for Kerry on electoral-vote.com (history graph): Current status: Kerry 237, Bush 275, tied 26.

Pollkatz's forecast is a bit more forgiving to Kerry: Kerry 271, Bush 267.
Note that Pollkatz's graph, which seems to be updated before the forecast is, has things just about back to where they were: Kerry 282, Bush 256, with no EVs in the barely region, and an equal number in the weak regions of both sides.

It still looks like a horse race, to me. The only difference being that Bush is now slightly ahead in overall national polls instead of slightly behind.

Wouldn't it be ironic if Bush won the popular vote, but lost the electoral college?

BlackGriffen

[dcolton]

Originally posted by BlackGriffen:

Wouldn't it be ironic if Bush won the popular vote, but lost the electoral college?

BlackGriffen

Even I would laugh at that! But seriously BG, do ou really believe in polls or this electoral college predictor? Just curious.

[Logic]

Originally posted by BlackGriffen:
Wouldn't it be ironic if Bush won the popular vote, but lost the electoral college?

BlackGriffen

Poetic justice

oh, and I like your sig

[BlackGriffen]

The Wall Street Journal has a summary of Zogby polls of the battleground states. The verdict? Kerry 307, Bush 231. The caveat? Zogby tends to be on the extreme end of being anti incumbent of all the pollsters (Pollkatz article and graphic).

I was surprised to find out that Zogby polling wasn't liberal leaning, since he was on the extreme of anti-Clinton, too. Fox polling isn't conservative, just pro incumbent leaning.

That is, if Pollkatz's investigation of Clinton era data is to be believed.

BlackGriffen

[RAILhead]

Originally posted by BlackGriffen:
The Wall Street Journal has a summary of Zogby polls of the battleground states. The verdict? Kerry 307, Bush 231. The caveat? Zogby tends to be on the extreme end of being anti incumbent of all the pollsters (Pollkatz article and graphic).

I was surprised to find out that Zogby polling wasn't liberal leaning, since he was on the extreme of anti-Clinton, too. Fox polling isn't conservative, just pro incumbent leaning.

That is, if Pollkatz's investigation of Clinton era data is to be believed.

BlackGriffen

Here's the full text of the article as it isn't *quite* as Pro-Kerry as BG would have everyone believe in his "verdict"...

Coming out of the Republican convention, President Bush has his best showing since John Kerry accepted his party's nomination in late July.

The latest Zogby Interactive poll still shows Mr. Kerry well ahead, leading in 12 of the 16 battlegrounds in Zogby's twice-a-month polls. But Mr. Bush took the lead in two states -- Arkansas and Tennessee -- since the poll conducted a week before his convention. And there are other signs of strength for the president.

Here's how the numbers could play out on Election Day.

To analyze Zogby's results, we start by assuming that the District of Columbia and the 34 states that aren't in the battlegrounds poll will vote for the same political party that they did in the 2000 election. Thus, President Bush begins our calculations with 189 electoral votes and Mr. Kerry with 172. A candidate needs 270 electoral votes to win the White House.

Then, we add in the electoral votes from the latest poll, regardless of the margin of error or the spread between the candidates. Mr. Kerry's 12 states control a total of 135 votes, while Mr. Bush's four have 42. If you add up the numbers, you find that Mr. Kerry would win the Electoral College 307-231.

Sifting through the numbers, though, the results aren't as clear cut.

First, Mr. Kerry's leads in three states -- Florida, Missouri and Nevada -- are less than one percentage point.

Overall, looking only at the states where the results are outside of the margin of error, Mr. Kerry is ahead by just four states to three -- and his Electoral College lead shrinks to 40-36. In fact, more states' results are outside of the margin of error in this poll than in any since Zogby began conducting polls for WSJ.com in late May.

Another sign of shift: In the July 12 poll, which was the leader until now in terms of the poll with the greatest number of states outside of the margin of error, Mr. Kerry had six states (worth 80 votes) outside the margin. At that point, Mr. Bush had none.

Many states show improvement for the president. For instance, Mr. Bush increased his lead in Ohio, a key contest, to 10.9 percentage points, his biggest lead there yet and up sharply from the 5.6-point edge he had in mid-August.
At the same time, Mr. Kerry's lead in Pennsylvania slipped to its weakest since the Zogby polls began in May. Mr. Kerry was ahead there by 2.8 percentage points, inside the margin of error. In four of the five prior polls, Mr. Kerry's lead was outside of the margin; his lead hadn't fallen below 6.5 points in any of the prior Zogby polls.

All told, Mr. Bush's numbers improved in 12 of the 16 states, most notably Tennessee. That state, which has been volatile in prior polls, gave the president a 9.6-point lead. In the mid-August poll, Mr. Kerry was up 1.9 points.

Still, Mr. Kerry picked up ground in Minnesota, Washington, Michigan and New Mexico. His leads in the latter two states moved outside of the margin of error. In Washington, his lead has been outside of the margin in every poll since early June.

And, of course, even with his convention-time gains, the president is still short of the strongest results he got in earlier Zogby polls. At one point in June, Mr. Kerry was ahead just nine states to seven -- and Mr. Bush led the Electoral College analysis, 285-253.

Also, Electoral Vote.com has the spread as Bush 275, Kerry 237, so the bottom line is that no one knows anything any more than anyone else, I guess.

[BlackGriffen]

Originally posted by dcolton:
Even I would laugh at that! But seriously BG, do ou really believe in polls or this electoral college predictor? Just curious.

Prediction of Nov? Nah, not really. Politics is just as bad as the weather in terms of predictability. It is an interesting look at the present status, though, and you can bet that the candidates are looking at this kind of stuff to decide where to spend their resources and time.

As the election draws closer, the pollsters' predictions become better. From what I understand, Survey USA (SUSA) has had a good track record over the last couple of years, though their use of automated pollsters is controversial.

BlackGriffen

[E's Lil Theorem]

Originally posted by RAILhead:
....

Also, Electoral Vote.com has the spread as Bush 275, Kerry 273, so the bottom line is that no one knows anything any more than anyone else, I guess.

They have Kerry at 237 not 273.

[BlackGriffen]

Originally posted by RAILhead:
Here's the full text of the article as it isn't *quite* as Pro-Kerry as BG would have everyone believe in his "verdict"...



Also, Electoral Vote.com has the spread as Bush 275, Kerry 273, so the bottom line is that no one knows anything any more than anyone else, I guess.

I didn't say anything about the article, I'm only interested in the electoral college counts. So, the count is the verdict. I also added the proviso that Zogby tends to lean anti-incumbent by a few points.

Please note, that the numbers you provided don't even add up to the total number of electoral votes. The number needed to win is 270. Electoral vote.com has the count at: Kerry 237, Bush 275, tied 26.

Also note that there is a difference in the methodology of the two aggregators (e-vote and pollkatz): e-vote just takes the newest major poll from the state, averaging in the case of same day polls. That means that his data tends to be rather volatile and some of it is hopelessly out of date. Here is Pollkatz describing his method:

August 2004
Here's how I calculate my estimates of the state-by-state Kerry vote if the election were held today.

My method takes advantage of the state-by-state correlation of the popular vote over the past seven elections (1980-2000). _ The correlation means that a poll in State X implies something about what the poll results would have been in State Y had the poll been conducted there, even if the states are very different -- say, Alaska and Virginia._ For each pair of states, the question reduces to "What is the predicted Kerry percentage in State Y, given that the State X poll percentage is p%?"_ Simple linear regression gives the best easy answer to this question._ I use the "alpha" and "beta" from the regression of each state's vote on each other state's vote to generate 2500 estimates of the form State Y vote | State X vote._ Then I take the median of each estimated State Y vote, update the map graphic, and also update the table of estimated electoral votes._
The actual alphas and betas are calculated directly from the means and covariances of the actual votes in the seven elections.
One problem is the differences among the polls -- whether Nader is offered as an option, the poll's age, and so on._ I have experimented with controlling for these discrepancies but have discovered that I get pretty much the same estimates anyway._
To construct 95% confidence intervals, use +/- 5%._ This is approximately the median estimate of 2*stdev.

And his data has been remarkably stable, compared to e-vote.com at least. That stability interests me, but isn't a predictor of his method's accuracy.

So, like I said, we'll see.

BlackGriffen

[RAILhead]

Originally posted by E's Lil Theorem:
They have Kerry at 237 not 273.

Fixed.�

[BlackGriffen]

Today is a perfect example of what I was explaining yesterday. electoral-vote.com's method is volatile - today's verdict: Kerry 264, Bush 222, tie 52. This appears to be primarily because of the Zogby data I pointed out yesterday.

The Pollkatz graph, OTOH, merely adjusts the size of the regions inside the predicted electoral vote count. IOW, still Kerry 282, Bush 256.

BlackGriffen

[BlackGriffen]

Another wild swing in the electoral-vote.com numbers: Kerry 243, Bush 254, tied 41.

The "Strong Kerry" section of the Pollkatz Graph is at an all time low, with the weak Bush category also expanded, though the overall breakdown is still the same: Kerry 282, Bush 256.

BlackGriffen

[BlackGriffen]

Electoral-vote.com has the race narrowing with a bit of a Kerry gain. Standings: Kerry 252, Bush 254, tied 32.
Pollkatz's graph also shows a slight uptic for Kerry: Kerry 291, Bush 247, with the overall strength of Kerry's support swinging up (again, there must be a couple of states that are borderline in his "strong Kerry" definition) and the overall strength of Bush's support taking a downswing (unusual - strong Bush has been very steady).

BlackGriffen

[Zimphire]

Just wait till the report comes out that the attack against Bush was a fraud.

[BlackGriffen]

Interesting movement in both of the aggregators. Electoral-vote.com gives Kerry the lead at: Kerry 273, Bush 233, tied 32. Contrastingly, Pollkatz's graph gives Bush the lead, for the first time ever, at Kerry 266, Bush 272.

Looks like it's still a tight and interesting race.

BG

[Zimphire]

AP Poll: Bush Noses Ahead as Kerry Slips

Saturday, September 11, 2004

WASHINGTON _�_President Bush (search)_has a slight lead over Democrat_John Kerry (search)_in an Associated Press poll, but the president has a big advantage on protecting the country � the issue voters say they care about most.

"If we don't take care of the terrorists, we certainly won't have to worry about the economy," said Janet Cross, 57, of Portsmouth, Ohio, who switched from Democrat to Republican for the last election.

[CreepingDeth]

Originally posted by Zimphire:
Just wait till the report comes out that the attack against Bush was a fraud.

Wait a week for it to be cooked, wait 3 days to cool, and wait for suprise.
They just stepped in some dogshit.

Trend is what matters for now.
Here's a 2000 election reference.

[constrictor]

Originally posted by CreepingDeath:
Wait a week for it to be cooked, wait 3 days to cool, and wait for suprise.
They just stepped in some dogshit.

Trend is what matters for now.
Here's a 2000 election reference.

That's a mandate, alright. Look at all those Red counties, where 3 people live, that went for Bush!

[macvillage.net]

Holy crap, look at the population density overlay he accidentally left out (since it's always shown together with the above images):



Interesting.

[CreepingDeth]

God you guys are thick. I was just saying that this is what the county will probably vote like this year. Some states are going to stay the same, and some won't, but it give you an idea.
And you guys are fools to think the only counties that give Bush the go are ones where you have to walk a quarter mile to talk to someone. In northern Virginia, we had a decent population, and we went for Bush because of all of the people working for the military. Central PA isn't bad either. This is as close to blue collar as you can get in the NE. Look at the major red in Ohio. Eastern CA?
I'd take central PA over some city any day.
Libs live in cluster studio apartments from the look of the population map.

[Invictus]

Originally posted by CreepingDeath:

Libs live in cluster studio apartments from the look of the population map.

And are gay. Don't forget that part Creepy.

[chris v]

Sigh. Let's just give the election to the guy with the most square miles. You could win Texas, with four counties and 6000 total resdidents.