great electoral vote analysis from user on Kos

[montana500]

http://www.dailykos.com/story/2004/10/30/22728/128#40

Time for my second-to-last state poll analysis and electoral college projection (if you're someone who is really tired of poll analysis--and even I am half the time--you probably don't want to read this whole post). Also bear in mind that what follows (at least the numbers part) doesn't account for Osama--who, in my opinion, is highly unlikely to move the race at all, and if he does, it'll be detrimental to Bush. Basically what I've done:

Assigned Bush and Kerry their safe states

Briefly examine five states that are pretty strongly leaning Bush but could conceivably go Kerry: Missouri, Nevada, Colorado, West Virginia & Arkansas

Take a closer look at the nine states that, in my view, are 99% certain to determine the election: Ohio, Florida, Pennsylvania, Wisconsin, Iowa, Minnesota, Michigan, New Mexico, & New Hampshire. Please note that I DO NOT consider these states toss-ups, like much of the ignorant SCLM does--but they are the states that the campaigns are playing for right now.

More below

Diaries :: mattb25's diary ::

For each of those nine states, I've taken all polls conducted in October and averaged them. I've excluded any partisan polls, Rasmussen polls, and Zogby Interactive--I've used the most recent batch of Zogby's state phone polls.

Generally speaking, whenever I had a choice between RV and LV numbers for a given poll, I took the RV numbers. Additionally, in the states in which Nader was on the ballot and I had a choice between two and three way numbers, I used the two way. Now, this doesn't mean that I don't think there could be third party effects. But, I don't like to use third party poll numbers because I'm of the belief that they very often over-represent the minor candidates by simply listing them as a choice--that is, you'll get some folks who say "Nader" just because they hear a name first or last, or something to that affect, even if they have no intention of voting for Ralph. I'm also of the belief that Bush and Kerry are going to be hurt basically equally by third parties, but not substantially in either case. So really my numbers should be thought of as shares of the two party vote.

In addition to computing the mean averages of the polls for each state, I also calculated the median. The median is an excellent measure for reducing the effects of outliers and for, generally, given a more complete summary of a distribution. In addition, I checked the numbers for outliers (in the statistical sense, margins that fluctuated by 2+ standard deviations from the mean), and re-computed new means and medians minus outlying polls. For certain states, I've allocated undecideds based on varying projections.

****************
Safe Bush: Alabama (9), Alaska (3), Arizona (10), Georgia (15), Idaho (4), Indiana (11), Kansas (6), Kentucky (8), Louisiana (9), Mississippi (6), Montana (3), Nebraska (5), North Carolina (15), North Dakota (3), Oklahoma (7), South Carolina (8), South Dakota (3), Tennessee (11), Texas (34), Utah (5), Virginia (13), Wyoming (3)

191 Electoral Votes

*Note: if Kerry gains another point or two nationally before Tuesday, or if Dem turnout is truly overwhelming, or if basically all of the undecideds break for Kerry, then Arizona, North Carolina and especially Virginia will come into play, but Kerry will have already won before they do.

Safe Kerry: California (55), Connecticut (7), Delaware (3), DC (3), Hawaii (4), Illinois (21), Maine (4), Maryland (10), Massachusetts (12), New Jersey (15), New York (31), Oregon (7) Rhode Island (4), Vermont (3), Washington (11)

*Note: I'm not particularly worried about Hawaii, but am glad that the ad time has been bought and that Gore and Alex Kerry are headed there. No need to take things for granted out there.

190 Electoral Votes

*****************

Strongly Leaning Bush

*NEVADA (5 EV--6 polls)
-Mean: Bush 49.8, Kerry 44.2
-Median: Bush 50.5, Kerry 44.5

Analysis: I've been arguing all along that NV would be one of those huge surprises, that turns blue even if the polling doesn't show it in advance. And I still won't be shocked if Kerry makes a pick up here. I expect turnout to be tremendous and the Yucca issue should be pretty motivating. But, while I generally think that Kerry is under-represented in the polls of Western states (not enough pollsters who speak Spanish, and certainly, the robo-calls don't), the Bush margin here is such that it has to stay as a lean-Bush for now.

Bush 196, Kerry 190

*MISSOURI (11 EV-6 polls)
-Mean: Bush 49.8, Kerry 45.1
-Median: Bush 49, Kerry 45

Analysis: I think MO is competetive, and that it'll be closer than the polling suggests; as with many other states, I think the Dem ground game will make major inroads into Bush's margins. But the fact that the campaigns aren't really allocating any resources here anymore does suggest that even the Kerry camp thinks it's going red.

Bush 207, Kerry 190

*COLORADO (9 EV--9 polls)
-Mean: Bush 49.1, Kerry 44.0
-Median: Bush 49.0, Kerry 44.0

Analysis: Polling all over the map, particularly Zogby. I think that it's close here, and that the Kerry campaign would like to really play for it, but in my view, it's a sound judgment to focus the resources in the upper mid-west. Again, I expect a fantastic ground game, and I think Hispanic turnout increases for Salazar; Nader won't poll close to his 2000 totals. Moreover, CO strikes me as full of paleo-cons as opposed to wingnuts, so it wouldn't surprise me to see a strong Badnarik showing here. Will it be enough? It could be, but I don't think it's quite a toss-up, so I'm giving it to Bush for now.

Bush 216, Kerry 190

*ARKANSAS (6 EV-6 polls)
-Mean: Bush 49.2, Kerry 44.3
-Median: Bush 49.5, Kerry 44.5

Analysis: I don't think it's over here, not by a long shot, and especially with Clinton visiting. I really thought AR was a great pick-up opportunity that wasn't really pursued early this cycle. I do think that's understandable from the campaign's point of view, though, and it'll be close here--but I can't see the election turning on it. I don't think Kerry wins here unless he's already won. I'm giving it to Bush for now, but very reluctantly.

Bush 222, Kerry 190

*West Virginia (5 EV--1 poll)

Analysis: Why this state hasn't been polled more, I have no idea, but obviously one poll isn't nearly enough for any kind of in-depth analysis. The one poll is the right-leaning Mason-Dixon that has it Bush 49, Kerry 44. Obviously, Dems enjoy a tremendous registration advantage here--it's great to see Byrd getting active, and from everything I hear anecdotally, the ground game on our side is far superior to the Thugs'. The DNC has shown some recent interest in getting back in the game here, which leads me to believe that it's tight in the internal polling. I think it's close to a toss-up. But since I'm not leaving any toss-ups on the board--if you held a gun to my head, I'd say Bush wins a squeaker.

Bush 227, Kerry 190

*****************

The Nine States That Will Pick the President

*FLORIDA (27 EV--19 polls)
-Mean: Bush 46.5, Kerry 45.8
-Median: Bush 46, Kerry 46

Poll Note: In one of only three technical outliers of my analysis, Gallup's FL poll gets tossed; and the LA Times comes awfully, awfully close. Tossing Gallup:

-Mean: Bush 46.2, Kerry 46.0
-Median: Bush 46, Kerry 46

Analysis: of the 18 polls remaining, Bush makes it over 50 exactly once, and that poll, from SUSA, is the oldest of the bunch and has since been replaced by newer SUSA data. Bush's mode percentage share (most frequently occurring) is a paltry 46. Based on the numbers alone, if even 55% of the undecideds break against Bush, Kerry wins the state. Projections of the two-party share with undecideds projected:

55% for Kerry--Kerry 50.3, Bush 49.7
65% for Kerry--Kerry 51.1, Bush 48.9
75% for Kerry--Kerry 51.9, Bush 48.1

The anecdotal evidence of turnout is excellent; and the tiny piece of scientific evidence as to who has already voted (Kerry 56, Bush 39 amongst the 16% who have already voted, according to Quinnipac) is fantastic news, if it's true (which it may or may not be). Dems have vastly outpaced repubs in new voter registrations and ACT is really humming down in FL. Additionally, looking at various demographic groups, Bush will almost certainly lose some support amongst the Cuban population; and increased Repub voters from the exurban communities ought to be matched, if not outpaced, by increased Dem-leaning non-Cuban Hispanics. Bush has really hurt himself with seniors--not only is the prescription drugs plan a fiasco, but Jeb has banned re-importation of Canadian drugs, an unpopular move. Social Security and the flue vaccine shortage are huge issues here. Throw in the fact that the hurricanes largely hit GOP strongholds. I can't see where Bush does better here than in 2000--maybe slightly better amongst the Orthodox Jewish community, but the actual evidence of that is slim. Now, obviously, the potential for fraud is very real and scary. I say that Kerry wins by enough here that's it a pretty evident win--about 2% if I had to guess

Bush 227, Kerry 217

*IOWA (7 EV--9 polls)
-Mean: Bush 47.6, Kerry 46.0
-Median: Bush 47, Kerry 46

Analysis: Two of these polls, one from SUSA and one from Mason-Dixon, had Bush up 6; those are awfully close to being statistical outliers, but they barely make the cut. Of the other seven polls, Bush leads in four by no more than 2 points, and Kerry leads in three, all by 1 point. Bush makes it above 50 only once, and that's in the aforementioned SUSA poll; and he makes it to 49 in only two other cases. This, for me, may be the toughest state to predict. The biggest piece of evidence in my mind comes from Vilsack, who indicates that the Dems have a pretty massive early voting lead. That's what happened in 2000, as well--Repubs actually won on election day, but not enough to make the difference. I expect tremendously high turnout here, which could cut both ways, given that R's do slightly outnumer D's. But the state does have a history of going blue in presidential elections. If the current polling is to believed, Kerry would need to take 65% of the undecideds for a victory; that's certainly a reasonable number, though I wouldn't want to bet my life on it. But I don't fully believe those numbers--bear in mind that they're based on the two Bush +6 polls. All of the most recent data points show total dead heats. I say Kerry wins here behind the surge of Dem early voting, and the fact that his people really, really know the state (think Michael Whouley).

Bush 227, Kerry 224

*OHIO (20 EV--13 polls)
-Mean: Kerry 48.1, Bush 46.2
-Median: Kerry 49, Bush 46

Poll Note: the second outlier comes in the form of a Fox News Poll that had OH as 49-44 for Bush. Eliminating it produces these totals:

-Mean: Kerry 48.4, Bush 46.0
-Median: Kerry 49, Bush 46

Analysis: Basically, I've felt the whole cycle that the polls have been missing swaths of newly registered folks who really are going to vote. I still believe that, and yet Kerry already has a substantial lead here. Of all the polls, Bush never gets to 50, and the only time he hits 49 is in the outlying Fox Poll. Kerry, on the other hand, hits 50 three times, and 49 five times. I think there are certainly some fraud issues here, but not nearly to the same extent as in FL. It's just not the same when your brother's not the governor. I think that Blackwell is a hack, but I also think he doesn't want to see his career end as the man who stole Ohio; his opposition to the challengers in the polling places, along with the court order, ought to at least partially minimize the potential voter suppression. I think Buscho knows they've lost here and are desperately trying to find a way to get to 270 by other means. Kerry wins here, solidly.

Kerry 244, Bush 227

*MICHIGAN (17 EV--9 polls)
-Mean: Kerry 47.9, Bush 44.2
-Median: Kerry 48, Bush 44

Poll Note: The third and final technical outlier is the Mitchell Research poll that had Bush up 46-42. Tossing it out produces these numbers:

-Mean: Kerry 48.6, Bush 44.0
-Median: Kerry 48.5, Bush 43.5

Analysis: Obviously, Zogby has scared some folks here by having Bush ahead in his tracking, and Bush is clearly attempting to make a desperate play for the state--but it's just not going to happen. This state isn't shifting 5% from 2000, and Kerry has led here the entire cycle, really even during the best days of Bush's bounce. You've got polls out just today that have Kerry +6 and Kerry +4. In my view, the Bushies are taking a shot here only because it's 17 electoral votes--somehow swinging it would largely compensate for losing OH. Maybe also trying to divert Kerry's focus there. But I don't think the Bush camp, or anyone else, for that matter, thinks there's much chance that MI goes red. Kerry by about the same margin as 2000, if not more.

Kerry 261, Bush 227

*MINNESOTA (10 EV--6 polls)
-Mean: Kerry 47.2, Bush 44.3
-Median: Kerry 47.5, Bush 43

Analysis: MN is probably closer than alot of us would like it to be; but Bush's numbers don't suggest any ability at all to win the state. Of the six polls taken in Ohio, Bush's highest number is 47. Meanwhile, in half of the polls, Kerry leads by 5+. I think it'll be close here, but that it goes Kerry pretty much without a doubt.

Kerry 271, Bush 227

*NEW HAMPSHIRE (4 EV--11 polls)
-Mean: Kerry 48.2, Bush 44.5
-Median: Kerry 49, Bush 45

Analysis: NH is pretty clearly going Kerry. The Kerry folks know the state exceedingly well, the Repubs there are much more the reasonable type, and the incumbent R governor has been a complete disaster for his party. Nice pick-up for Kerry, and one that could put him over the top (not by itself of course).

Kerry 275, Bush 226

*NEW MEXICO (5 EV--5 Polls)
-Mean: Bush 47.2, Kerry 45.2
-Median: Bush 47, Kerry 46

Poll Note: taken by itself, today's Zogby would be an outlier--I've left it in since Zogby has shown Bush consistently ahead, and also just due to a small number of data points.

Analysis: I don't really know what's going on here. I am certain that Bush doesn't lead by 9. Of the five polls this month, Kerry leads in two of them, and one is a tie; while Zogby and Mason-Dixon have Bush up significantly. My general reaction to Zogby is that he does his best work in western states, and his numbers are all over the place (it was only Bush +3 two days ago); the state's got a popular Dem governor, and all evidence suggests that Bush will not increase his share of the national Hispanic vote. If one were to accept the polling average as is, Kerry would need to take 70% of the undecideds for a victory here--again, perfectly, reasonable, but I wouldn't bet my life on it. Overall, though everything in my gut tells me that NM is staying blue, and Bush wouldn't be spending Monday night there if he thought it was safe. The polls are almost certainly under-estimating Spanish-speaking voters. Kerry squeaks by in what may tbe closest state in terms of number of votes, as it was in 2000.

Kerry 280, Bush 227

*PENNSYLVANIA (21 EV--16 Polls)
-Mean: Kerry 48.4, Bush 44.7
-Median: Kerry 48.5, Bush 45.0

Analysis: Bush's numbers are horrible here--of all 16 polls this month, he's never been higher than 47 amongst RV's, and the Repubs have largely pulled resources from the state. Rendell has a great machine in the urban areas. Staying blue, easily.

Kerry 301, Bush 227

*WISCONSIN (10 EV--8 polls)
-Mean: Kerry 47.3, Bush 46.6
-Median: Kerry 47.5, Bush 46.0

Analysis: I long thought of WI as Bush's best shot to pick up a Gore state, but I think that's Iowa now. WI has been slowly but surely trending blue the whole month, and the Springstein rally was just the perfect way to culminate it--turnout will be massive here, lots of young voters; and the third parties will have a role. I think that Bush's best shot here was if his guys could win the ground game, and based on everything I've seen and read, I don't think they can. Kerry wins a tight race here--and it's critically important that he do so.

Total Projection: Kerry 311, Bush 227

******************

Basically, I see a very slight Kerry victory in the popular vote (between 1-2 points) coupled with a solid EC victory. I think that the numbers are favorable for Kerry in the nine critical states, and that the Repubs will simply not have an answer for the newly registered Dem voters and the ACT/Moveon/Kerry ground game. The importance of WI and NH, incidentally, cannot be overstated. I'm fairly certain that OH is going Kerry--and I think that all the facts (the fraud aside) suggest a Kerry win in FL. But I wouldn't want to bet everything on FL, nor on IA or NM. If all those go to Bush, while Kerry takes OH, then it's all about WI and NH--winning them both would earn him a 272-266 victory.

All in all, I feel very good about where things are, and Bin Laden tonight didn't change that. I'll probably offer one more (shortler) final predicition on Monday. Work your asses off the next few days, y'all, we're gonna win this thing.

[lablue]

1. cbs (or abc) analysis this morning was great

kerry looks like will keep pennsylania and win from bush (in 2000) nh and ohio... only needs to hold most of the gore states in midwest - particularly wisconsin (iowa seems to be leaning bush, minnesota kerry) and it's kerry's. Florida a "true tossup" - Kerry wins florida it's waaay over. Analyst said Cheney going to Hawaii basically because things look bleak and need every vote they can get.

[BeatleBoot]

2. I agree with your Michigan analysis

Looks positive for Kerry.

Even in the northern part of the state, which is strongly Republican, there are plenty of Kerry signs around. The economy is hurting there as it is in the auto industry around Detroit.

B/C spent alot of time on the west side of the state (Grand Rapids, Kalamazoo) I think because our Democrat Governor Granholm did well with Republicans in that area. They were trying to shore up their base.

But that area has been hit with outsourcing of jobs overseas (Greenville, MI - Electrolux Plant is a prime example).

Conservatively, I would hope for Kerry by 3%. Probably will be more like 5%.

Just my non-scientific $ 0.02 worth....

[montana500]

3. this post should be required reading before folks post about HArdball ;)

nm