The General Election Map

Hmmm...

You seem to be bucking the 538 crew.

Interesting. Would you care to share your reasoning? Are you suggesting the Obama Landslide Projectors cool their jets?

Posted by Prematurely Grey | June 20, 2008 11:55 AM

Marc, take a look at the recent polls. Do you see why some might say that you are horribly biased against Obama?

Posted by Joe | June 20, 2008 11:55 AM

The fundamental problem McCain has is that Obama's money and registration drive and the voter targeting information he has (recent as it is from the primary process) opens a lot of doors for him.

He needs a big play to open a door for himself.

Posted by Rhoda | June 20, 2008 11:56 AM

Georgia is NOT a base state for MCain any longer.......with the HIGH AA registrations and the BARR vote, GA is definetly a battleground state....Everyone else has NV leaning Obama

Here's a good look at the map

http://www.fivethirtyeight.com/

Posted by tr | June 20, 2008 11:59 AM

Georgia is NOT a base state for MCain any longer.......with the HIGH AA registrations and the BARR vote, GA is definetly a battleground state....Everyone else has NV leaning Obama

Here's a good look at the map

http://www.fivethirtyeight.com/

Posted by tr | June 20, 2008 11:59 AM

Georgia is NOT a base state for MCain any longer.......with the HIGH AA registrations and the BARR vote, GA is definetly a battleground state....Everyone else has NV leaning Obama....I think you are way off

Here's a good look at the map

http://www.fivethirtyeight.com/

Posted by tr | June 20, 2008 11:59 AM

Sorry, Marc, I agree with the others here, esp the one referencing the 538 site. The recent polls can of course be suspect and reflect an Obama bounce. But the longer term analysis, including the money AND organizational strength of the Obama campaign is just so strongly skewed in his direction, don't you think?

May be best to leave the predictions to other sites. Notwithstanding this diversion, your other insights and scoops have been almost always outstanding ... so we will all buck you up. And don't forget to continue having those sessions with Andrew S when he returns from the Cape!

Posted by Mike | June 20, 2008 12:03 PM

I think Nate Silver has made this whole gut-based approach to election forecasting pretty obsolete.

A better use for Marc's particular skills would be to record the most recent buzz he's been hearing regarding each state from insiders, without trying to turn it all into numbers.

Report on the subjective side, that is, rather than the more objective side; at any rate, stop trying to conflate the two.

My 2 cents.

Posted by lampwick | June 20, 2008 12:04 PM

I have to go to a meeting, but I bet if you did the math, Obama's average lead in the "toss-ups" would be bigger than McCain's average lead in the "lean McCain" states. That should tell you a little bit about Marc's biases -- he's essentially saying states that are narrowly McCain are leaners, but states that are narrowly Obama are toss-ups.

(Just eyeballing it, the only "toss-up" states where McCain is ahead have him up by 0.5%. Whereas a few states where Obama is polling ahead by 8-10 points are called toss-ups.)

Posted by Joe | June 20, 2008 12:04 PM

Marc, if you include Washington and Oregon as only tilt Obama, simply because McCain wants to put them in play in absence of any polling evidence that they are, why wouldn't you do the same for Georgia?

Posted by Greg | June 20, 2008 12:09 PM

Pennsylvania, Iowa, and maybe even Ohio lean Obama.

Posted by Kathryn | June 20, 2008 12:09 PM

You're trying to make a horse race where one doesn't currently exist. There may yet be a horse race but based on current polling you're simply wrong. Go to fivethirtyeight.com for some decent analysis and then come back and report what you found.

Iowa - There has not been an Iowa poll going back to February '08 where McCain has been in the lead. The closest Iowa poll was Rasmussen where Obama had a 2 point lead. Rasmussen's most recent extended Obama's lead to 7. Tilt Obama.

NH - ARG's most recent (I know it's ARG) showed a 12 point lead. Rasmussen announced a poll today showing a 5 point lead. Again, this has to be in the lean Obama column if you're going to be objective and impartial.

PA - Q poll - Obama by 12. Last four PA polls show substantial high single digit to low double digit Obama leads. McCain has not been ahead since April and then only in a Strategic Vision (R) poll. Again this should be lean Obama.

Colorado - Polls show a small but consistent Obama lead.

Basically, if you want to project the general election at this point, you should include in the tilt Obama category these states which total 41 EVs. That brings the EV to 253 Obama. Given that Obama has legitimate possibilities in Virginia, Ohio, Florida, North Carolina, Michigan, New Mexico and Missouri the road to an Obama victory is easier than is McCain's path.

Posted by rts | June 20, 2008 12:09 PM

I think Georgia is quite safe for McCain.

The fivethirtyeight.com site is very thorough but fundamentally biased in a lot of it's assumptions (the guy who runs it is an Obama supporter).

What interests me about this campaign is the issue of timing. McCain said not long ago that he would wait until mid-August to choose his running mate. This is comparativel late (Kerry annoucing on July 6, Bush on July 25). It seems like the McCain camp understands that Obama is on a wave. He is riding very high, he is the establishment candidate, he is expected to win. For McCain to put off the critical and narrative defining announcement of his running mate until mid-August suggests that he understands that this early time does not matter too much. He is pacing himself, Obama needs to be careful.

Posted by Andrew | June 20, 2008 12:10 PM

I agree with others that this is a curious list. Take Washington/Georgia for example. You have WA as tilt Obama, because Mccain wants to contend there (I would like to contend there to, but will get zero votes). Obama wants to contend in Georgia, yet still have it safe McCain. Historically WA hasn't been out of the Dem column since the Reagan landslide in '84 (even Dukakis won there), Georgia has gone Dem as recently as '92 (albeit with a strong Perot factor thrown in). The polls have consistently showed Obama with a large WA lead, the same cannot be said for GA, where even before yesterday's poll indicating a GA tie Mccain had been ahead only 8-10 points. Why the discrepency in classification for Washington and Georgia (especially with Barr in the mix)?

Posted by jmscher | June 20, 2008 12:12 PM

How is Penn. a toss-up? There are no non-Republican pollster polls out there where McCain is ahead. Its not a safe Obama but he's clearly favored.

Posted by Greg | June 20, 2008 12:14 PM

Andrew, please state where the 538 assumptions are biased.

You are just making that up. Yes, they support Obama, but they have not allowed it to have any effect on their assumptions.

Posted by bob | June 20, 2008 12:18 PM

I love that all the commentary for the "tilt" states, on both sides, is written from the perspective of McCain's campaign. "McCain should be doing better." "McCain wants to contend here." "This one may be getting away from McCain."

Ambinder, be honest with your readers and with yourself. Stop advertising yourself as nonpartisan. Let your readers know where you're coming from so that they may place your so-called reporting in context.

Posted by xx | June 20, 2008 12:18 PM

I'd like to second (eighth?) the request to hear more about why you put Pennsylvania and Iowa as toss-ups. I'm guessing it's not from polling but from "interviews, reportage, and guesswork". And hell, Washington isn't close right now in anyone's estimate.

Posted by Xenocrypt | June 20, 2008 12:19 PM

Emailed this, didn't see comments were back up...

As of today, Georgia, per InsiderAdvantage -

McCain: 44%
Obama: 43%
Barr: 6%
Undecided: 7%

With this being Barr's home state, this will be a tough one to gauge.

Posted by Ben | June 20, 2008 12:20 PM

What makes you think PA is a toss-up? Every poll conducted in the last two months has had Obama up, the Pollster.com average has Obama up by 8 points. The real base number for Ohio, with Pennsylvania included, should be 233.

Posted by Hatch | June 20, 2008 12:22 PM

What makes you think PA is a toss-up? Every poll conducted in the last two months has had Obama up, the Pollster.com average has Obama up by 8 points. The real base number for Obama, with Pennsylvania included, should be 233.

Posted by Hatch | June 20, 2008 12:22 PM

Horse. Race.

Posted by adam | June 20, 2008 12:23 PM

Here are my disagreements with you:
- Pennyslvania and Iowa are not tossups. They are at least lean Obama.
- Washington is a base state. If McCain wins Washington, he will have won this race in a landslide.
- Alaska should be lean McCain

I know these are all pro-Obama shifts, but it is just how I see it.

Posted by Mike | June 20, 2008 12:26 PM

You think the Obama campaign's push into Alaska is bluster and/or an attempt to help Mark Begich?

I think I dispute the notion that Washington state is tilt McCain. No poll has shown him within 10 in quite some time. He's probably the Senator most responsible for killing the Boeing tanker deal. I don't see how he wins here.

Otherwise this seems about right.

Posted by Nicholas Beaudrot | June 20, 2008 12:29 PM

Marc, the previous post you made is a shocker.

It's bad enough you have given up even acting objective, you also seem to have given up writing coherent sentences - or even spelling correctly. in this ONE post - we have at least five spelling errors. "naughy", "berudges" and "ambitoins" in one sentence alone. This is a disgrace that the Atlantic has someone who can't even write to a high school level.

Posted by Ambinder | June 20, 2008 12:30 PM

This is the most rational assessment I've seen anywhere in the media thus far. My fellow Democrats are getting way too excited about June polls. (That said, there is a definite possibility that Obama wins EVERY state you've listed as a TOSS-UP, with no such possibility for McCain.)

I would probably move Washington to Safe Obama and Georgia to tilt McCain, but that's about it.

Good job, MY.

Posted by ASDem | June 20, 2008 12:31 PM

Not a complaint about bias:

Marc, these posts would be much more helpful if you were to include maps with shading. Graphics are good! All it would take is some simple work for an Atlantic intern.

Posted by Dan Miller | June 20, 2008 12:35 PM

Polls now are basically useless. Still, I think your gaming is not correct. Washington should be base Democratic and PA should not be a toss-up but lean.

Would like to see more Georgia polling before we start bullshitting about Obama's chances there.

Would also like to see what his voter registration drives do. 600,000 unregistered black voters in Georgia. Let's get them to the polls!

Posted by Arabella | June 20, 2008 12:37 PM

E-Z Money (Obama +10.0% or more): CA (55), WA (11), IL (21), NY (31), ME (4), HI (4), MD (10), DE (3), DC (3), CT (7), MA (12), VT (3), RI (4) - 168 EVs

Nearly in the Bag (Obama +3.6% to +9.9%): OR (7), OH (20), PA (21), CO (9), NM (5), IA (7), MN (10), WI (10), NH (4), NJ (15), MO (11) - 119 EVs

Toss-up (Obama +3.5% to McCain +3.5%): AK (3), MI (17), VA (13), NC (15), NV (5) - 53 EVs

Still Possible (McCain +3.6% to +9.9%): MS (6), GA (15), MT (3), LA (9), FL (27), KS (6), KY (8), SC (8), AR (6), ND (3), TX (34), AZ (10), WV (5), IN (11) - 151 EVS

Maybe next time (McCain +10.0% or more): UT (5), ID (4), NE (5), TN (11), AL (9), WY (3), OK (7), SD (3) - 47 EVs

Posted by Ajax in NYC | June 20, 2008 12:38 PM

I really dont see how he can get to these figures given the figures that the polls are currently showing. If Marc is disregarding the current polls perhaps to discount the 'bounce', ok, but he doesnt give any clue.

But the end result is Obama is 8 electoral votes behind? Eh?

State by state it makes even less sense.

Unless with Independents you are assuming a McCain majority, I really don't see how you came to this result.

Posted by James | June 20, 2008 12:40 PM

Pennsylvania, Michigan and Iowa are not tossup states. They lean (strongly) to Obama.

Washington, Oregon, Wisconsin and Jersey are not "tilt Obama" states. They're states he will win, as clearly as McCain will win West Virginia or Georgia.

Posted by TH | June 20, 2008 12:42 PM

I'd say the toss-ups are leaning Obama. I think you need more detail in your analysis if you want to dispute the analysis at 538.

Posted by Julie | June 20, 2008 12:45 PM

Marc, I'm fine with this conservative (for an Obama supporter) estimate, I worry that people are overly-optimistic and it's breeding complacency.

However, listing Iowa as a toss-up is pretty ballsy.

Since head-to-head matchups began, NOT ONE SINGLE Iowa poll out of the 20-30 that have been run has shown McCain with a lead. Obama leads have ranged from 4-9, but I think the pattern is pretty clear.

A state that has had numerous polls, and every single one showing one candidate in the lead, it's hard to call that a toss-up.

Posted by Dan | June 20, 2008 12:47 PM

To clarify: It's fine, I guess, to think of current polls as useless, and go off the state party chair or longtime activist, or whatever, but if the opinions of such folks contradicts the polling consensus, it'd be informative to know their rationale, as you provided with New Hampshire.

Posted by Xenocrypt | June 20, 2008 12:59 PM

Marc is wrong. He is doing his stuff using the old styled math.

This election is all about how energized the base is for their candidate. It is as simple as that.

Obama's base is not only "super" energized but they also have strong infrastructure and hundreds of millions to convert it into a wave. On the other hand the conservative base is with McCain half-heartedly and they are not at all energized.

A very simple question.. who has energy, money and organization to capture more votes than in 2004? The answer is very simple: Obama. In addition to that Obama has issues on his side. Wait till he starts releasing round the clock ads about merging McCain's face into Bush's and call McCain Bush's third term.

Eventually it will be a very comfortable Obama victory.

Posted by Victor | June 20, 2008 1:00 PM

This is the kind of "reporting" that makes people turn off to politics. Exactly what is the story here? How is it any different than predicting who will win the Super Bowl by looking at draft picks?

Do some reporting or change the subtitle of your blog.

Posted by KevDog | June 20, 2008 1:04 PM

I think your McCain bias is showing. I disagree with you on almost every thing.

Posted by rosie | June 20, 2008 1:09 PM

The general trend for the map in in Obama's favor. Georgia is super-competitive and his decision to forgo public cash will mean virtually unlimited ad time and ground efforts in plenty of toss-up and red states.

http://www.political-buzz.com/

Posted by matt | June 20, 2008 1:12 PM

I get that polls aren't terribly useful this early. The thing is they aren't useful because lots of stuff we don't know about will happen. That doesn't mean you can just ignore them and make up your own silly story. There is absolutely know reason to trust what McCain wishes was true over what people say they will do when the get into the both. The fact that you don't get this suggests that you are in the wrong business.

Posted by Michael Foody | June 20, 2008 1:14 PM

lampwick - I agree with you. And this particular "gut check" gives gut checks a bad name.

WI, IA and PA are not lean-Dem, but GA is safe-Repub? Whatever.

Posted by evie | June 20, 2008 1:15 PM

Personally, I hope the McCain people take this reporting and go with it.

This is your calendar, Team McCain, and don't let anyone tell you it's not!

Posted by evie | June 20, 2008 1:18 PM

Marc is also wrong about NH. Obama has huge lead over McCain in latest Rasmussen poll. Rasmussen is usually a republican leaning org.

Rasmussen, June 20th poll:

NH-President

Obama (D) 50%, McCain (R) 39%

---------------------

NH-SENATE

Shaheen (D) 53%, Sununu (R) 39%

Posted by vic | June 20, 2008 1:22 PM

And suddenly comments are back on? Wonderful.

This "analysis" is total garbage. Take a look at any of the numbers - as reported by pollster.com, fivethirtyeight.com, electoral-vote.com, etc. - and you will find a very different story. Heck, even granting the fact that it is early and some of these polls may overstate the Obama bounce, I find it incredibly unlikely that so many states swing back to the Toss-up or Lean-McCain categories. Where is the logic? Where are the facts?

The answer - you sir are a hack.

I've done my own "analysis" and here's where I see things as of today:

McCain Base
*Marc Ambinder

Posted by Edward | June 20, 2008 1:24 PM

If McCain's campaign is counting on Iowa and Pennsylvania as "toss-ups" he might as well start working on his concession speech now.

Posted by Leo | June 20, 2008 1:24 PM

There was a theory floated a while ago (by Ygelias?) that everyone thinks Ambinder is biased because, O paradox, he's such a straight shooter.

After reading this post, I've moved from the "lean toward disbelief" column to the "definite bullshit" one.

Posted by Estrien | June 20, 2008 1:25 PM

To those giving the love to the 538.com site. It has its good and its bad. Before the primary ended, the analyses on that site were excellent.

But now they have changed their methodology for the general election by incorporation national election polls and superimposing those findings onto state polls as well. This gives a huge electoral vote advantage to candidates ahead in national polls in states where the polling is mixed or split. That's why Obama is up so big in their current analyses. It's bad analysis, and it is clear the site is very biased toward Obama given the way they write.

They appear to be more objective than they really are. Their changed methodology that led to a dramatic increase in Obama's likelihood of winning is a real problem.

Posted by Statistician | June 20, 2008 1:29 PM

The fivethirtyeight.com site is very thorough but fundamentally biased in a lot of it's assumptions (the guy who runs it is an Obama supporter).

Prove it or STFU.

Posted by SavageView | June 20, 2008 1:30 PM

I think the polling is undereporting older voters. I seen polls out of Penn that had the 18-29 vote equaling the Senior Vote. That poll is a joke as Seniors out number the Youngies by 10% in the Dem primary and when you add Republicans to the mix that tend to be older it will likely be a 13% lead for the grayhairs on election day. My theory is caller ID is throwing off the polls as older people are less likely to pick up and answer a number they do not know, meanwhile younger people I know have no problem answering unknown numbers.

Posted by Paul8148 | June 20, 2008 1:36 PM

Statistician: But the nice thing about 538 is he (I think) shows all his work. If you don't like the "trend-adjusted" measures, just use the non-trend-adjusted ones.

Posted by Xenocrypt | June 20, 2008 1:38 PM

Statistician? Calling yourself this does not mean you know what you're talking about.

Nate Silver (538.com) knows what he's doing and the logic behind his new methodology is reasonably sound.

However, even if you dismiss his numbers, how can also dismiss pollster.com (which operates from the same office building as Marc Ambinder) or electoral-vote.com.

The facts aren't on your side.

Posted by Edward | June 20, 2008 1:45 PM

"This gives a huge electoral vote advantage to candidates ahead in national polls in states where the polling is mixed or split."

This is false. It is only a time trend adjustment, which means it only affects the result when the polls are OLD. It has nothing to do with mixed or split polling.

If you want to take comfort in the fact that some old polls from before Clinton dropped out showed McCain doing better, go right ahead. People who want to actually understand where the race stands right now, though, are unlikely to find your approach very compelling.

Posted by Leo | June 20, 2008 1:46 PM

Virginia, Missouri, and Ohio are true tossups. All of Marc's other tossups are 'lean Obama' states, some quite precipitously so. (Where can I get a bet down on Obama winning Iowa at 50-50 odds?)

Washington's solidly in the Dem base now. NJ, all but. (Call it a 'lean Obama as much as one can lean without falling over.')

And Alaska is definitely 'lean McCain' territory. Obama ought to visit the far north this summer; he could steal that state.

Marc also has an addition error. The states he regards as Obama's base have 157, not 159, EVs.

Posted by low-tech cyclist | June 20, 2008 1:46 PM

By the way, I had time to do the math. Using Pollster and averaging all April-or-later polls for Indiana and Montana (where there aren't enough polls for Pollster to come up with a composite), the average margin in the "Tilt McCain" states is McCain by 3.6. The average margin in the toss-ups is Obama by 3.5. The average margin in the "Tilt Obama" is Obama by 12.7. Basically, Marc's methodology appears to be:

McCain by more than 5 = "McCain's base"
McCain by less than 5 = "Tilt McCain"
Obama by less than 5 (except for NH and PA) = "Toss-up"
Obama by 5-15 = "Tilt Obama"
Obama by more than 15 = "Obama's base"

Marc, do you see why people might accuse you of bias?

Posted by Joe | June 20, 2008 1:47 PM

And the states Ambinder lists as McCain's base have 160, not 159, EVs.

Posted by low-tech cyclist | June 20, 2008 1:48 PM

It's absurd to look at polls released in the past few days and conclude oh the election is over rating systems should take that into account and close shop.
Polls have to be read, interpeted to account for various factors.

These ratings are solid. The only thing i disagree with is Washington -- it looks like as solid an Obama state as any at this point, and it was already clearly going blue in 04 -- and Obama's willigness to play in Alaska has to put that in play.

Posted by Dan | June 20, 2008 1:55 PM

Marc, the fact that you listed solidly-Democratic Washington and Oregon as "tilt Obama" merely because "McCain wants to play here" yet kept competitive-polling Alaska and Georgia in McCain's base despite the fact that "Obama wants to play there," makes your entire analysis suspect and fundamentally useless.

Posted by Travis Mason-Bushman | June 20, 2008 2:14 PM

"These ratings are solid. The only thing i disagree with is Washington"

Really? You agree that PA and NH are both "toss-ups"? Even though polls have Obama with a 5-10 point lead in both, and the Democrats carried them in 2004 (in fact, in every election since 1988 except NH in 2000)?

Look, I have no problem with saying that states that would have to flip from 2004 are toss-ups, even if a candidate has a small-but-significant lead according to the polls (IA, CO, OH). But if you are going to play that game, there's no way that PA and NH are "toss-ups."

Posted by Joe | June 20, 2008 2:15 PM

Isn't it at least possible that this reflects the attitudes of the people involved in the campaign? Marc cites talking to New Hampshire Democrats as saying the state is a toss-up as opposed to leaning Obama.

I think it is very possible that this summary represents the way the campaigns are feeling right now. It seems like both sides would have an incentive to think it is going to be a close election - for those in the lead to not take anything for granted, and for those behind to feel like they have a realistic shot.

This - as Joe outlines above - represents a view of the electorate as it might have existed if Obama got no bump from winning the primaries. It is systematically moved 4-6 points to McCain from the current polling. One thing that Nate has talked about quite a bit at 538 is that states tend to drift together and that sometimes the campaigns have information that differs from the publicly available polling. So maybe this is information that is ahead of information soon to be available.

I guess my point is that I agree that this appears tilted towards McCain compared to other sites relying on the current polling. But there are other possible explanations than Marc being in the tank.

Posted by Curtis | June 20, 2008 2:15 PM

What "The Love Guru" is to cinema Marc Ambinder is to Political analysis.

Posted by Michael Foody | June 20, 2008 2:15 PM

I don't think Marc is in the tank (that is the most fun explanation though). I do think that if he is to put states in categories different then the polling indicates he could at least attempt to justify (and the aspirations of the campaign really is a weak reason to do so).

Posted by Greg | June 20, 2008 2:20 PM

Obama explained his decision to rely on private donations in a video posted to his website. 23/6 has added hilarious closed-captioning for any disillusioned supporters: http://www.236.com/news/2008/06/20/conservatives_confused_mccain_7260.php

Posted by eliana | June 20, 2008 2:24 PM

As a strong Obama supporter, I fail to see how estimates that may not show an Obama landslide amount to pro-McCain bias. I think it is healthier from the perspective of an Obama supporter to suck up and address as much potential weakness and bad news as possible, not just dismiss polls not showing Obama ahead as "pro-McCain propaganda".

Let's be cautious in this election. Hubris and overconfidence will lower the energy of the campaign. "It's an Obama wave. I don't really need to work to get him to him" That's one way Democrats will lose this year. Remember that McCain has a lot of maverick appeal still, however unjustified, and even independents who hate Bush will take a serious and long look at McCain.

Posted by Chuck | June 20, 2008 2:27 PM

As the obvious underdog, I think it was wise for Obama to opt out of public financing.

Posted by Gwendy | June 20, 2008 2:28 PM

Marc, after reading your post, but before reading the comments, I kind of scratched my head and thought that you seemed to be awefully generous to McCain and a little hard on Obama. As some have already commented, is the "McCain campaign wants to contend here" relevant or any basis on which to measure electoral strength of probability? You're a smart guy. I can't believe you could defend this.

I'm glad you turned "comments" back on. I'm not going to call you names if I perceive that you have a bias. I don't think that is constructive. But I do want to challenge you to be more upfront with your biases and, with regard to the Obama candidacy, abandon claims of reasoned impartiality. I started reading this blog about 4 or 5 months ago. It was fairly obvious to me, the casual observer, that you favored Clinton over Obama. I've kept up with your blog posting for the last several weeks as well. This post on the electoral map is a good example of favoring McCain over Obama. I'm beginning to think that you don't have a pro-Clinton or pro-McCain bias, but instead have an anti-Obama bias. What I'd really like to know is, why? I wouldn't begin to suggest that you aren't right or entitled to espouse whatever view you want, or support whoever you want, or disfavor a particular candidate; but can you explain your thinking on why Obama gets the short end of the stick with you? I'd be interested to know.

Now, in order for you to answer this question, you, of course, have to give up the ghost and implicitly acknowledge that, in your blog postings, you disfavor Obama. If you're not willing to do that, if you're not willing to give up the claim to objectivity or impartiality when it comes to Obama, well . . . I guess you won't be answering my question. On your little "Table" discussions with Andrew, you come off as much more neutral and impartial. I wish you'd be that way in your "reported blog."

Posted by Andy | June 20, 2008 2:28 PM

Edward said, in response to my previous post criticizing the Obama-backing 538.com: "However, even if you dismiss his numbers, how can also dismiss pollster.com (which operates from the same office building as Marc Ambinder) or electoral-vote.com.

The facts aren't on your side."

I didn't say McCain is ahead. But a fair look at state polls combined with knowledge from campaign internal polling suggests the race is much closer than 538.com suggests. It's methodology slants pro-Obama as long as Obama is slightly ahead nationally. It is a fair critique, and the guy running the site is reporting figures on the probability of Obama winning based on a different methodology than used during the primary (when the same methodology would have hurt Obama dramatically).

Obama is slightly ahead in many polls, but many of the state's that 538.com has for Obama are actually too close to call or slightly lean McCain once you exclude the national poll methodology he imputes to state results.

The 538.com site isn't as great as it looks. People who know stats can be fair, or they can use them for their candidate. It's a mix of both right now.

Posted by Statistician | June 20, 2008 2:29 PM

Taken alone I might agree that this doesn't indicate that Mark is "in the tank" ... but given the pattern of his other posts ... a clear bias is emerging whether Mark intends it or not.

Posted by Edward | June 20, 2008 2:30 PM

This counts as analysis!?!?!? Way off. Just to show the heavy, heavy bias:

GA (supposedly a McCain base state, according to this clown):

InsiderAdvantage -

McCain: 44%
Obama: 43%
Barr: 6%
Undecided: 7%

NH (supposedly a toss up, not even a leaner):

Rasmussen, June 20th poll:

NH-President

Obama (D) 50%, McCain (R) 39%

What credibility do you have with this kind of hog-wash?

Posted by Bob from Michigan | June 20, 2008 2:35 PM

"Isn't it at least possible that this reflects the attitudes of the people involved in the campaign? Marc cites talking to New Hampshire Democrats as saying the state is a toss-up as opposed to leaning Obama."

In other words, rather than analyzing the facts, he's passing on some amalgamation of competing spin, regardless of the factual basis underlying each side's claims. That's a plausible story, as that seems to be largely how "objective journalists" view their role. But it's not much of a defense. There are facts available, and its not too much to ask to have journalists report them.

Posted by Leo | June 20, 2008 2:35 PM

What, now comments are closed again?

fuck you - either all comments or none.

Posted by Ambinder | June 20, 2008 2:37 PM

Hey Ambinder, you'll note that most of the commenters in this thread have been respectful and apparently hesitant to assume the worst about your motives.

Should you ever re-enable comments on the rest of your posts, you shitbag, beware that some of us will not be so kind.

Posted by xx | June 20, 2008 2:43 PM

Marc, your analysis runs somewhat contrary to a slew of recent polls which show Pennsylvania, Iowa, and Michigan with leads for Obama. While I don't believe either of these are "safe" for Obama, they aren't quite toss-up states.

Posted by Natascha | June 20, 2008 2:49 PM

I ask the same thing as I did earlier: Would you please share your reasoning? If you won't, I may begin to wonder if the commenters who question your neutrality are on to something.

Posted by Prematurely Grey | June 20, 2008 2:51 PM

Michael Foody wins

Posted by Post of the Day | June 20, 2008 2:56 PM

Agree with the sentiments of most commentors. How you can identify IA, PA, and NM as toss-ups is beyond me. They lean strongly Obama. Also, Nevada is a toss-up, not a lean McCain.

Posted by reispace | June 20, 2008 2:57 PM

Marc seriously ... Joe's eexample aside, mentioning Pawlenty without noting that THREE of the folks on Obama's shortlist hail from Virginia is pretty darn telling.

Posted by Dirk | June 20, 2008 2:58 PM

Marc seriously ... Joe's example aside, mentioning Pawlenty without noting that THREE of the folks on Obama's shortlist hail from Virginia is pretty darn telling.

Posted by Dirk | June 20, 2008 2:58 PM

And comments are back off again. Wonderful.

Inconsistent much?

Posted by Daniel | June 20, 2008 2:59 PM

You people don't understand. Polls don't mean anything. All they can tell us is how a random sampling of voters will vote if the election were held today. What's really important is what the campaign aides and insider-journalists feel in their gut.

For example: Two polls in the past week show Obama up by double digits in NH? Who cares? That tells us nothing. Now, a couple of party hacks talkingn about "waves cresting"-- now there's data you can set your watch to.

Posted by Oracle | June 20, 2008 3:03 PM

Ambinder's analysis is good. As he noted, it's based on past polls, current polls, and talking to campaigns on the ground. If you don't like it, go read others. There might be errors, but there might be errors in the polls too. It's one of many configurations of state swing possibiliites. I'm glad Charlie Cook doesn't allow comments, or you all would be trashing him for his objective analysis too.

Posted by Relax people | June 20, 2008 3:18 PM

this "analysis" is a steaming pile.

it's quite obvious that you are in the tank for Old Man McWar.

Posted by jimmy | June 20, 2008 3:28 PM

How is Marc's analysis "good"? He talks to "people" and then overrules the polls based on that. HUH?? Why not throw a dart at the map to determine toss-up states?

This was a joke Marc. You should take it down before more people flame you.....

Posted by Bill Clintoni | June 20, 2008 3:29 PM

Not to pile on, but Marc, how can you expect anyone to take you seriously after offering this horribly skewed analysis? It'd be one thing to be wrong about a number of states, but in every single instance of inaccuracy you are favoring McCain.

This is a terrible post. Plain and simple.

Posted by What the heck? | June 20, 2008 3:32 PM

Marc, your analysis is nuts.

Basing these changes mostly on figures at Pollster.com:

In the 'McCain Base States', you should move WV, AK, and ND into the 'Tilt McCain' category.

In the 'Tilt McCain' category, Indiana should be moved to 'Toss-Up'. Pollster.com has Obama winning 2 of the last 4 polls there, going back to mid-April. Also, Nevada should probably be in the 'Toss-Up' category too, though I'm a little less concerned about that one, given that McCain has won 3 of the last 4 NV polls.

In the 'Toss-Up' category, PA, IA, and CO should be moved into the 'Tilt Obama' category. All three are consistently polling in favor of Obama. That leaves as the 'Toss-Up' states: NM, MO, MI, IN, OH, VA, and NH.

In the 'Tilt Obama' states, WA and NJ need to be moved into 'Obama's Base'.

Were these changes to be made, the list would look like this:

Total McCain Base + Lean = 210
Total Obama Base + Lean = 247

McCain's Base -- 149 EV

TX (34), SC (8), KY (8), TN (11), AL (9), MS (6), LA (9), AR (6), WY (3), ID (4), UT (5), AZ (10), OK (7), KS (6), NE (5), SD (3), GA (15)

Tilt McCain -- 61 EV

AK (3), MT (3), ND (3), NV (5), WV (5), NC (15), FL (27),

Toss-Up -- 81 EV

NM (5), MO (11), IN (11), MI (17), OH (20), VA (13), NH (4)

Tilt Obama -- 64 EV

OR (7), CO (9), IA (7), MN (10), WI (10), PA (21)

Obama's Base -- 183 EV

HI (4), WA (11), CA (55), IL (21), MA (4), VT (3), MA (12), RI (4), CT (7), NY (31), NJ (15), MD (10), DE (3), D.C. (3)

.

Posted by JGabriel | June 20, 2008 3:42 PM

Correction: Btwn IL (21) & VT (3), MA (4) should be ME (4).

Posted by JGabriel | June 20, 2008 3:46 PM

The scenario where McCain wins Pa is the one where he wins 48 states. I'll put my money where my mouth is. Keystone State is Obama base.

Posted by moondancer | June 20, 2008 3:52 PM

Poll: Obama Has Big Lead In New Hampshire
By Eric Kleefeld - June 20, 2008, 2:09PM
A new Rasmussen poll of New Hampshire gives Barack Obama a big lead in this swing state, which went narrowly to George W. Bush in 2000 and then switched to John Kerry in 2004.

The numbers: Obama 50%, McCain 39%. A month ago, Obama had only a 48%-43% edge.

Meanwhile, Rasmussen gives McCain a small lead in Nevada, which voted for Bush twice: McCain 45%, Obama 42%, within the ± 4.5% margin of error.

Posted by mike | June 20, 2008 4:01 PM

MoonDancer: "The scenario where McCain wins PA is the one where he wins 48 states. I'll put my money where my mouth is. Keystone State is Obama base."

I pretty much agree.

That said, PA shows a lot of early polls going for McCain.

True, it's been close to solid Obama since late April. But the fact that it flipped from McCain means we can't entirely discount the idea that, theoretically, it could flip back to him.

I consider it highly unlikely, but I'm trying to be somewhat objective here.

.

Posted by JGabriel | June 20, 2008 4:05 PM

I think part of the problem in this thread is that you guys are citing actual analysis and facts gleaned from research and accredited polling outfits. Ambinder is citing largely ancetodal evidence (McCain "wants to contend here"), conventional polling wisdom, and other foolishness.

Understand now?

Posted by JamboWanna | June 20, 2008 4:11 PM

In terms of the basic problems with Marc's analysis, I think I agree with most of the people here - particularly JGabriel's analysis.

Beyond that, I'm not going to accuse Marc of any particular bias here. I really don't think he's biased towards anything other than the conventional wisdom.

It seems clear that Marc thinks that the current polls are showing Obama stronger than he really is, and is thus being generous to McCain. Why this is, I don't know, and that's the biggest problem with the post.

The categories Marc is presenting are actually pretty controversial - they show a much closer race than the current polling suggests. But he doesn't defend this position forthrightly, and many of his choices are hilariously underexplained. (e.g. Washington being only tilt Obama because McCain wants to contest it, or the fact that in many states he has no explanation at all).

For instance, for Pennsylvania, if he wants to call it a toss-up, he should say "while Obama's been ahead by a fair margin in the most recent polls, most of the people I'm talking to are expecting McCain to rally, and this one to end up being really close - remember that Kerry only won it by 2.5 points, and Obama has had trouble connecting with voters in the Appalachian area that covers so much of Pennsylvania," or something like that. Instead, he justs puts it in "toss-up" as though it's self-evident.

I'd also add that the main bias here, I think, is towards the horse race. That is to say, the bias is towards the idea that the race is going to be very close, like the last two. Right now, this isn't really what the polls are showing, so the assumption favors McCain. But this would all be a lot better if Marc would lay out his assumptions and explain why he thinks the polls are wrong in some cases, rather than just presenting his findings from on high.

Posted by John | June 20, 2008 4:18 PM

I think the math is off here.

220+212+107 = 539, not 538

I think the totals at the top should read:

McCain's base states + his leaners: 221
Obama's base states + his leaners: 210
Tossups: 107

221+210+107 = 538

But that aside, I think Marc is dead on right. We are exactly that close right now.

Posted by Jim | June 20, 2008 4:22 PM

But that aside, I think Marc is dead on right. We are exactly that close right now.

Sure. That's why Intrade has Obama at 63% right now. Because it's just that close.

If Ambers ain't getting paid by the McCain campaign, he's ripping himself off.

Posted by goethean | June 20, 2008 5:29 PM

Go look at the pollster.com charts and then tell me that Nevada and Ohio tilt McCain. Somebody's in the tank and it ain't pollster.com

Posted by goethean | June 20, 2008 5:36 PM

And Florida! Go look at the pollster.com chart and tell me that Florida is not a toss-up.

Posted by goethean | June 20, 2008 5:39 PM

We are exactly that close right now.

Wow, that's a stretch. I don't think even Marc would say it is that close right now; rather, I think he would say it's going to be that close in November (which is still highly dubious).

Anybody who has inkling about politics in this country knows that, if the election were held today, it wouldn't even be close.

Posted by Oracle | June 20, 2008 5:56 PM

Hey, everybody, it's not only 538.com -- it's also Pollster.com, and electoral-vote.com, and usaelectionpolls.com -- they all have Obama leading by *over* 100 EV's.

Yes, June polls aren't worth a whole lot. But, still. The analysis at http://www.centerforpolitics.org/crystalball/article.php?id=AIA2008061901 describes polls as almost worthless this far out, but nevertheless projects a big Obama win based on other fundamentals (state of the economy, approval rating of incumbent, etc.)

Marc, I like your stuff, but you have to admit: your prediction here is a real outlier.

Posted by A DC Wonk | June 20, 2008 6:17 PM

I know the current polls don't necessarily hold any water for the one poll that matters in November, but if you're going to buck the current polling, give us better rationalizations. McCain wanting to compete in Washington and Oregon doesn't make it so... I'm sure he'd like to compete in Hawaii and DC, too, but it's not going to happen.

Posted by Adam Villani | June 20, 2008 6:29 PM

I'm not even going to bother arguing against you, Mr. Armbinder. You're crazy.

Posted by Are You Kidding | June 20, 2008 6:47 PM

Iowa as a toss-up screams "I'm not following the polls." McCain hasn't polled ahead of Obama there ever. PA is knife's edge, but I understand why you would keep it as toss-up for now with only two post-primary polls there. OR and WA should be Obama base. Just because the McCain camp wishes really hard, crosses their eyes and holds their breath, doesn't mean they are competitive in the Pacific Northwest. Obama's showing a fairly consistent double-digit lead in WA, and McCain hasn't come closer than 6 points to him in OR since March.

You're right about FL--smart money still says it's leaning McCain, but a couple more polls showing it tight or him behind should move it into toss-up territory.

Posted by Anon21 | June 20, 2008 7:03 PM

Is it possible that Ambinder is a member of PUMA?!

Posted by Natascha | June 20, 2008 7:10 PM

All your "Toss ups" are polling for Obama. That's not a trend, it's reality.

Posted by Colin | June 20, 2008 7:11 PM

All your "Tilt McCain" states are "Toss ups" according to all recent polling.

Posted by Colin | June 20, 2008 7:13 PM

From a purely demographic standpoint McCain will pick up PA and MI, possibly MN, and probably NH. He should win FL easily and narrowly win OH. OR and WA will be more competitive than the Obama Kool Aid drinkers want to believe, especially WA. Actually, when people really get to know Obama, a lot of formerly blue states will become toss ups.

Posted by EyeDoc | June 20, 2008 7:27 PM

EyeDoc:

Actually, if McCain runs ads attacking Obama in states like Oregon/Washington, then he'll see his numbers crater. He has no play in the anti-war northwest....

McCain's only hope is that cultural issues stick in OH/PA/MI and he holds off Obama in CO/NV/NM somehow...

This race is a joke. Everyone in both parties in DC knows that. Ambinder has his head in the clouds.

Posted by Paul | June 20, 2008 7:57 PM

You're hypothesis is so off Marc.

Penn, Ohio, Iowa, Colorado and New Hampshire should all be lean Obama.

North Carolina, Nevada and (maybe) Florida should be listed as toss ups.

You're analysis is not based off of any logical sequence of polls, previous voting or current climate, orginazational structure.

Posted by MSL | June 20, 2008 8:48 PM

Looks like you're really struggling to say that its an even field at the moment.

The latest AK poll has Mccain up by 4 and that is part of his base and yet the last poll in Washington has Obama up by 17 and that is only tilt Obama?

The only state where you don't seem to unrealistically negative about Obama would be Florida which I'd guess to be more in Mccain's pocket than current polls reflect.

Posted by Josh J | June 20, 2008 9:36 PM

I'm not going to rehash all the points that everyone else has made, most of which I agree with. I especially thought the point about Obama's lead in your "tossup" states is on par or larger than McCain's lead in his "tilt" states was a particularly good one, but I'm sure you were aware of that when you wrote this up.

My only question is about the statement about Nevada being an "unusual state where the regular rules don't apply."

What does this statement mean? The state itself has been polling very tightly for McCain and has been well within the margin of error.

Are you trying to insinuate something? Or have you heard something that would make you place Nevada in McCain's column this far out?

Posted by dannity | June 20, 2008 9:53 PM

with knowledge from campaign internal polling suggests

Division by zero posting from "Statistician."

Posted by SavageView | June 20, 2008 10:09 PM

Marc, please have a look at today's Rasmussen polling in NH before putting it in "tossup":

Obama 50, McCain 39
Shaheen 53, Sununu 39

http://bluehampshire.com/showDiary.do?diaryId=4424

Posted by Dean Barker | June 20, 2008 10:21 PM

Marc, please have a look at today's Rasmussen polling in NH before putting it in "tossup":

Obama 50, McCain 39
Shaheen 53, Sununu 39

http://bluehampshire.com/showDiary.do?diaryId=4424

Posted by Dean Barker | June 20, 2008 10:21 PM

I have to disagree with the people who criticize FiveThirtyEight.com's recent changes to methodology.

Prior to the recent changes, his methodology was profoundly flawed, and I criticized it in a few places. The fundamental problem was that, in order to form a prediction on each state, he incorporated only polls taken within that state. The (rather obvious to my mind) problem with this is that public opinion can shift dramatically, and his method failed to reflect the effects of that shift in cases where a state happened not to be polled for a few weeks, as was the case recently in OH, PA, and others. A good methodology would infer from other state and national polls that unpolled states shifted.

The thing he did to address this was somewhat convoluted, but it seems to work. His results now match up closely with those on my website, which uses a tested methodology that correctly predicted every state but WI in 2004.

I realize that Republicans tend to lack intellectual vigor, but I'd like to see any avowed McCain supporter or other conservative try to think hard about methodology, instead of spouting crap about how 538 must be biased just because a good methodological change was positive for Obama. Try thinking through the deep statistical questions: How are states correlated? How can you measure that? How volatile is public opinion? How can you measure that? How should you deal with missing data? How do you measure the reliability of pollsters? How do you deal differently with a pollster that's merely unreliable, compared to one that seems to be unreliable in one direction? How do you handle uncertainty about how undecideds will break? How do you deal with polls that include third-party candidates who aren't on the ballot, or that exclude candidates who are? If you think that the US shifts 5 points away from McCain, would that indicate that a state like OH would also shift 5 points? What about an extreme state like UT or DC? How do you define the term "swing state"? Once you've found a definition, how do you measure which states fit it?

(My call right now: Obama by 4.7%, with about 328 electoral votes. And NV is swing state, in the sense that it will be close if the nation is close; IA is not a swing state. The reason Marc thinks of IA as a swing state is probably that it was close in 2000 and 2004. Overextrapolating historical results is a common error.)

Posted by Benjamin Schak | June 20, 2008 10:54 PM

I must confess to some curiosity about your methodology. You have to wonder when every single one of your tilt McCain states (except Montana, for which RCP agrees with you) is listed as a tossup by both RealClearPolitics and FiveThirtyEight. Looking at your tossup states, Iowa and Pennsylvania are solidly in the Obama column according to both these sites, while 538 every single one tilting Obama. Meanwhile, none of your lean Obama states are listed as anything but safely Obama by these sites.

These are not minor oversights. You are giving Florida to McCain when Obama has won two of the last three polls. You have Pennsylvania as a tossup when as far as I can tell McCain hasn't won a poll against Obama since April.

It's hard to take this seriously when in every instance where you disagree with nonpartisan sites you do it in favor of McCain. Let's compare their numbers to yours:

Source: Obama - Tossup - McCain
RCP: 238 - 137 - 163
538: 317 - 69 - 152
Rasmussen: 284 - 14 - 240
You: 212 - 107 - 220

Posted by Lucas | June 20, 2008 11:11 PM

Another biased anti Obama post. Marc you are very off on your assumptions and very off from reputable bean counters like Chuck Todd and the Cook report. And living in Georgia, I can assure you that Georgia is competitive- voter registration is going thru the roof and the Bob Barr is doing a Ralph Nader on McCain. Conservative Republicans truly hate McCain more than they fear Obama and the religious right don't know what to do, many are thinking of sitting it out this time. Didn't Huckabee creamed McCain in Georgia. Watch for a big upset- and there are alot of Ron Paul lovers here too-mostly young republicans and conspiratist loving people.

Posted by sonya | June 21, 2008 12:00 AM

Marc, you may want to double-check your math. Your base states for Obama actually add up to 157 electoral votes and your base states for McCain actually add up to 160 electoral votes.

Posted by Paul | June 21, 2008 12:52 AM

what a hack.

didn't you know that mccain intends on competing in EVERY state.

obama: 0
tossup: 0
lean/solid mccain: 538

now that's some real, solid math for you, mr. ambinder.

Posted by the bipolar bear | June 21, 2008 3:16 AM

Michigan is not a tossup state. We've got two Democratic Senators, a Democratic Governor, and have been voting for Democratic Presidential candidates for years. It is at least a "leans Obama" state. What basis do you have to claim otherwise?

Posted by Steve Collins | June 21, 2008 6:00 AM

A little more explanation of your methodology is appropriate. I can only assume you think Obama's lead is similar to Kerry's, Dukakis, or Jimmy Carter's at this point in the campaign, otherwise the commenters above do make valid points.

Posted by petty boozwha | June 21, 2008 9:13 AM

Mac Pick Palin Veep BEFORE Hil Campaigns w/Obama!

In addition to her overwhelming attraction on the oil/energy issue and the female/disaffected Hillary voters, her human interest story will generate millions and millions of dollars worth of publicity and media coverage — essentially free to the McCain campaign — more than offsetting Obama’s reported money advantage.

Posted by Ted | June 21, 2008 10:22 AM

I haven't read all the comments yet, but I'll respond to this, in Mark's defense:
"...but I bet if you did the math, Obama's average lead in the "toss-ups" would be bigger than McCain's average lead in the "lean McCain" states. That should tell you a little bit about Marc's biases -- he's essentially saying states that are narrowly McCain are leaners, but states that are narrowly Obama are toss-ups."

Barack Obama must depend on larger turn-out to change (no pun intended) the dynamics of the last decade of elections. I believe we all agree to that.

In other words, Gore lost the EC, and Kerry lost the EC. For BHO to win the EC, he has to expand the map, which he says he plans to do by expanding the voter pool.

However, just because BHO says he will be able to do that doesn't mean he can.

That's why so many of us are watching both the independents and the undecideds on these polls, far more than we're watching the raw horse race numbers.

When the poll, for example, is 45%-40%, the other 15% are VERY intriguing to many of us.
In fact, any number under 50% for either candidate should probably be disregarded with these two particular candidates running against each other.

As much as BHO is going to try to make JMc into McBush, that doesn't really matter as much as whether Republicans come home to McCain or stay away from the voting booths.

In addition, BHO was unable to get last-minute deciders against HRC in the D primaries. That should be a real concern.
On the JMc side, he MUST keep the majority of independents. They decide elections.

One caveat: I'm an independent from NH who was a Clinton supporter. I think the idea of "divided government" as far as independents go will come into play more and more, especially if the choice comes down to a veto-proof Congress and either "a moderate" and "a liberal" choice for President.

I don't think divided government, as an asset to JMc, has been discussed much.

Anyway, like others, it may be that Mark is basically betting that GOP McCain's voters will choose in the end to vote rather than to stay home.

(As an aside, I can't speak for Mark on this but many people who have studied polls for years are also doubting that Obama's actual November youth vote -- as opposed to his black vote -- will generally promise much more than they are delivering to him right now.
In certain quarters, Obama's youth following is actually a detriment to him.

So anyway, I think these catagories are right on, Mark.

For the fun of it, I have further given IA to BHO so they start at 220 v. 219. "Even Steven."

Then I gave McCain VA, CO, and NH.
I gave BHO PA and NM to reach 246 v. 245.
Even Steven.

I simply cannot even try to guess the rest.
So, in my opinion, it comes down to:
OH (20
MI (17)
MO (11)

The winner must take two out of three.
Here we go again!

Posted by Jan | June 21, 2008 11:33 AM

re: 538...
I encourage everyone to read Statistician at 1:29.

If you don't actually understand the raw numbers that are being used to give you data that you are depending on, you stand to believe anything you are told.
That's called propaganda.

I do NOT say 538 is biased, but their numbers are now being massaged a bit. So, if you are really looking for firm data on which to base a guess, you may want to start being cautious about their information.

Whether you "like" it or not, if you don't understand what Statistician is trying to teach you, you don't understand how you can be used and abused (and therefore disappointed in November).

Posted by Jan | June 21, 2008 11:50 AM

Leo, I hate to burst your bubble, but POLLS are NOT FACTS.

If you actually think POLLS are FACTS, I strongly suggest you take time to study the Mike Dukakis campaign sometime before November.

"In other words, rather than analyzing the facts, he's passing on some amalgamation of competing spin, regardless of the factual basis underlying each side's claims."
Posted by Leo | June 20, 2008 2:35 PM

Posted by Jan | June 21, 2008 12:03 PM

Does anyone else note the basic symmetry of Marc's analysis? The bias here is somewhat evident but misplaced. The bias is in favor of a horserace, not in favor of one candidate or another.

June polls aren't worthless, but they are not determinative either, so the Obama camp should not get too excited yet. That said, if I hear warnings about Dukasis' 20 point lead right after the 1988 convention as a comparison point for the 2008 campaign today (before both conventions), I'll scream.

Posted by danimal | June 21, 2008 12:15 PM

This one is too good to pass up: EyeDoc, you're blind.

The posts by "Statistician" show he either doesn't understand what 538 is doing or is misrepresenting it. As noted by Leo, the adjustments are an attempt to deal with the problem of having no recent polling in a given state when national opinion has shifted. Yes, making a time based adjustment not based on polling with a state has measurement issues, but assuming a given state is isolated from national trends (which you would have to do to take as current old polling data) has even more issues, and these will have a clear bias tending to ignore the impact of changes already present.

Posted by Judas_priest | June 21, 2008 12:16 PM

There is a lot of leverage in 538's methodology, and in the methodology used by other prognosticators such as the Republican "electionprojection.com" site.

It'd be pretty interesting to see both of those sites get the numbers from a bunch of past elections as they existed in mid-June and see what the predictions would've been.

My own belief is this: Look at the unemployment numbers for June vs. March. If unemployment goes up in the second quarter, Obama will win. If it goes down, Obama will lose. If it stays the same, it will be a very close election and fraud will determine the result, as in 1960 and 2000.

This prognostication method has worked in every election since 1948, with the exception of 1956. By the way, unemployment in March was 5.1%. In May it was 5.5%. The "internals," so to speak, suggest a very grim June, as do recent trends in gas prices, airline cutbacks, layoff announcements, and surveys of corporate expectations.

If unemployment goes up again in June, it'll be a big landslide for Obama. If it stays the same, Obama will have a comfortable victory. If it goes down to 5.1% or lower, then given what the Bush administration has been doing in other areas, I'd take a close look at the BLS methodology!

Posted by Master Debater | June 21, 2008 12:51 PM

Obama in a landslide!

Posted by Freakshow | June 21, 2008 1:33 PM

Jan @12:03 is correct. 538.com isn't outright biased, but they present their results in a very clear manner, but don't heavily note that they've changed their methodology in a way that (at present) favors Obama.

They impute national polls on states (in addition to state level poll data and demographics of the states). This imputation of national polls onto individual states is a really bad assumption, as it leads to huge shifts if one candidate goes up big in national polls (and states differ; Alabama is not the same as Nevada which is not the same as Massachusetts). For instance, 3 of the last 4 polls in Nevada have McCain with a slight lead. But because the national polls favor Obama right now, the 538.com web site has Obama winning Nevada.

They had excellent methodology before the primary ended, but they changed it. It will favor heavily any candidate ahead in the national polls without looking at states. There's a bad assumption in their current methodology, but it makes it look small because they do a nice job of reporting probabilities of winning.

They should just use state level poll and demographic data. If no recent polls exist, use the 2004 election results plus a baseline. Their site is biased towards any candidate with a 5% or so lead in national polls. If they had used these assumptions for projections in 2000 (since Gore was ahead of Bush nationally), Gore would have won more electoral votes, which of course, he did not.

Posted by A Statistician | June 21, 2008 2:07 PM

"It will favor heavily any candidate ahead in the national polls without looking at states."

See, this is just false--as you would know if you had bothered to try to understand the methodology. Their method favors candidates that have improved in national and other states' polling since the most recent poll in a particular state.

Can you give me a good reason I shouldn't take into account dramatic changes in the race when reading month-old polling? It seems to me that ignoring that kind of information would be (and was until recently) a pretty significant flaw.

Posted by Leo | June 21, 2008 3:06 PM

At the risk of repeating myself -- because I've seen no one address my question -- why is everyone picking on 538 as biased when pretty much all the other summaries of polls say the same thing? That would either mean that *all* of them are biased, or, perhaps, 538 is not biased.

As for Master Debater (at June 21, 2008 12:51) and his comments about tying the election to unemployment rates, I suggest an article at http://www.centerforpolitics.org/crystalball/article.php?id=AIA2008061901, which supports (but expands upon and refines) your theory.

Posted by A DC Wonk | June 21, 2008 9:59 PM

Master Debater: >"If it stays the same, it will be a very close election and fraud will determine the result, as in 1960..."

Really sick of this incorrect talking point.

It's based on the accusation that Daley stole Illinois for Kennedy in 1960

The problem is that JFK won 303-219 EV.

That's a difference of 84 EV. IL was worth 27 EV in 1960.

It wasn't even close.

I know R's just hate to hear the facts on this, because it robs them of their lousy 'but the Dems do it too' excuse. However:

Kennedy did not win in '60 on the basis of electoral fraud. The election wasn't even close. End of discussion.

.

Posted by JGabriel | June 22, 2008 12:54 AM

I think these numbers are good in the sense that they do not factor in the psychology of the campaigns. What I mean is that Senator McCain has been receiving average press coverage at the moment, Senator Obama is still moving on ahead due to his "star" status. Now with the extra money difference, I believe it will not be a landslide victory but at decisive victory for Senator Obama.

Posted by Crian Padayachee | June 22, 2008 9:29 AM

The problem w/ 538.com and the reason it is getting 'picked on,' though I am not sure about that, is because that web site posts summary probabilities of a given candidate winning 270 electoral votes. Last time I checked, the site had Obama with a 78% chance of winning the election.

While I would concede that the polls show Obama is favored, does he truly have a 4 out of 5 chance of winning, given that many swing state polls show both candidates within the margin of error or Obama only ahead by about 3-4%?

The 538 site is a problem because they include national and state polls together to predict state outcomes. So Nevada is just slightly favoring McCain in 3 out of 4 recent polls, but that site gives Obama over a 60% chance of winning Nevada, which is what feeds into the near 80% prediction of an overall Obama electoral vote victory.

That page changed its methodology since the primary ended, and it greatly enhanced Obama's chances of winnings with new measures. It gives Obama supporters false hopes for what is likely to occur. Obama has a better chance of winning at this moment, but he does not have a 80% chance of winning. That site is nearing Obama advocacy, while other sites just based on state polls don't advocate.

Posted by To a DC wonk | June 22, 2008 3:14 PM

It wouldn't surprise me if the 538.com web site guru is actually being paid by the Obama campaign. That would explain their changed methodology which enhances Obama's probability of winning to outrageously high %s.

Posted by FVC | June 22, 2008 3:17 PM

Yes, that 538 web site just posted a link to a state survey in Oregon showing Obama 48%, McCain 45%, so Obama is leading only within the margin of error.

Yet, the probability of Obama winning Oregon is 86% based on the model (and including the poll with Obama only +3 in the MofE). How in the world does a candidate have a nearly 90% chance of winning a state when they are tied in the most recent state poll? That site is biased.

Posted by To FVC | June 22, 2008 6:35 PM

It doesn't make sense that a respected statistician would bias his numbers just as he emerges as a force in politics. He just inked a deal with Rasmussen, and has a very good track record.
While this doesn't preclude an error, it makes the odds of him skewing the data remote at best.

Posted by moondancer | June 23, 2008 12:37 PM

To FVC: You do understand that margin of error is a 95% confidence interval, right? And you understand what that means? A 90% win percentage means that the polling is within the margin of error. A lead outside the margin of error results in a greater than 95% win percentage.

(Of course, the margin of error used on 538 is not exactly the same as the margin stated by the pollster, for many reasons.)

Posted by Leo | June 23, 2008 1:02 PM

The 538 site is a problem because they include national and state polls together to predict state outcomes. So Nevada is just slightly favoring McCain in 3 out of 4 recent polls, but that site gives Obama over a 60% chance of winning Nevada, which is what feeds into the near 80% prediction of an overall Obama electoral vote victory.

Given the fundamentals of this election season, suggesting that Obama has a 60% chance of winning Nevada despite the polls showing McCain in the lead seems imminently reasonable of 538.com, rather than absurd, as you seem to assume.

Posted by goethean | June 23, 2008 1:33 PM

A whole post about seals? are you fucking kidding me?

Joke.

Posted by Ambinder | June 23, 2008 1:41 PM

Who's afraid of the big bad comment thread?

Marc, or his editors?

Posted by MR | June 23, 2008 1:56 PM

The real issue is not how well Obama or McCain might do in the closely divided battleground states, but that we shouldn't have battleground states and spectator states in the first place. Every vote in every state should be politically relevant in a presidential election. And, every vote should be equal. We should have a national popular vote for President in which the White House goes to the candidate who gets the most popular votes in all 50 states.

The National Popular Vote bill would guarantee the Presidency to the candidate who receives the most popular votes in all 50 states (and DC). The bill would take effect only when enacted, in identical form, by states possessing a majority of the electoral vote -- that is, enough electoral votes to elect a President (270 of 538). When the bill comes into effect, all the electoral votes from those states would be awarded to the presidential candidate who receives the most popular votes in all 50 states (and DC).

The major shortcoming of the current system of electing the President is that presidential candidates have no reason to poll, visit, advertise, organize, campaign, or worry about the voter concerns in states where they are safely ahead or hopelessly behind. The reason for this is the winner-take-all rule which awards all of a state's electoral votes to the candidate who gets the most votes in each separate state. Because of this rule, candidates concentrate their attention on a handful of closely divided "battleground" states. Two-thirds of the visits and money are focused in just six states; 88% on 9 states, and 99% of the money goes to just 16 states. Two-thirds of the states and people are merely spectators to the presidential election.

Another shortcoming of the current system is that a candidate can win the Presidency without winning the most popular votes nationwide.

The National Popular Vote bill has been approved by 19 legislative chambers (one house in Colorado, Arkansas, Maine, North Carolina, and Washington, and two houses in Maryland, Illinois, Hawaii, California, Rhode Island, and Vermont). It has been enacted into law in Hawaii, Illinois, New Jersey, and Maryland. These states have 50 (19%) of the 270 electoral votes needed to bring this legislation into effect.

See http://www.NationalPopularVote.com

Posted by susan | June 23, 2008 2:37 PM

Marc,

When we had 'bitter', how many posts did you do?

A main adviser to McCain says it would be good for McCain if there was another attack on America, and you dismiss it in a paragraph, praise him for his 'honesty', and then try to make it a plus for McCain.

You are a disgrace.

Posted by James | June 24, 2008 7:31 AM

The General Election Map

Comments (139)

Recent Posts

Monthly Archives