Courtesy of the Politico, Barack Obama’s campaign has telegraphed the electoral focus of its fall campaign. Three are its highest profile defenses (PA, WI, MI), one is a Kerry state in which McCain runs particularly well (NH), and 14 are Bush 2004 states where Obama thinks he has an opportunity to gain on the previous map (IA, NM, OH, CO, VA, FL, MO, NV, IN, ND, NC, MT, GA, AK).
Let's take a look at how FiveThirtyEight breaks down each of these states in terms of polling average, trend adjustment, 538 regression, and 538 prediction.
This map shows a baseline of Obama 200, McCain 139.
John McCain's best slicing and dicing of the public polling data within the 18 states is the polling average. Obama leads in Iowa (+6.2%), Wisconsin (+6.0%), New Hampshire (+6.0%), Pennsylvania (+5.5%), New Mexico (+4.3%), Ohio (+4.0%), Colorado (+2.5%), and Michigan (+1.9%) for a total of 93 EVs. McCain's states are North Dakota (+0.6%), Florida (+1.4%), Indiana (+1.8%), Missouri (+2.2%), Nevada (+2.8%), North Carolina (+4.7%), Alaska (+5.7%), Montana (+7.3%), and Georgia (+8.6%) for 93 EVs. Virginia is dead even, so after adding in the base EVs, the tally would be Obama 293, McCain 232, Unallocated 13.
When we adjust for trends in the polling data, we find ourselves in landslide territory. Obama leads in Iowa (+9.3%), New Hampshire (+9.0%), Wisconsin (+8.8%), Pennsylvania (+8.5%), Ohio (+7.1%), New Mexico (+6.1%), Colorado (+5.2%), Michigan (+4.6%), North Dakota (+4.4%), Virginia (+3.7%), Missouri (+3.3%), Indiana (+2.1%), and Florida (+1.0%), for a total of 158 EVs. McCain holds North Carolina (+0.5%), Nevada (+0.8%), Alaska (+2.8%), Montana (+3.1%), and Georgia (+5.4%), for 41 EVs. Under trend adjustment, the tally adding to our base would be Obama 358, McCain 180.
Under the 538 regression model, Obama leads in Iowa and New Hampshire (both +10.0%), Wisconsin (+8.7%), Pennsylvania (+8.6%), Michigan (+8.0%), Colorado (+7.1%), Ohio (+7.0%), Nevada (+5.4%), New Mexico (+5.2%), Virginia (+3.0%), Florida (+1.6%), and Missouri (+0.5%) for 149 EVs. McCain holds Montana (+0.3%), North Dakota (+2.0%), North Carolina (+2.5%), Alaska and Georgia (both +3.6%), and Indiana (+3.7%) for 50 EVs. Added to our base, the regression tally would be Obama 349, McCain 189.
Finally, the 538 projections as of June 25 show Obama winning Iowa (+9.4%), New Hampshire (+9.3%), Wisconsin (+8.8%), Pennsylvania (+8.5%), Ohio (+7.1%), New Mexico (+6.1%), Colorado (+5.8%), Michigan (+5.2%), Virginia (+3.6%), Missouri (+2.7%), Florida and Indiana (+1.1%) and Nevada (+0.7%) for 160 EVs. McCain projects to hold North Carolina (+0.8%), North Dakota (+0.9%), Montana (+1.8%), Alaska (+3.0%), and Georgia (+5.0%) for 39 EVs. With the baseline, Obama 360, McCain 178.
Under these numbers, McCain needs to gain 2.5% on the polling averages to win (adding Virginia, Michigan and Colorado). He needs to gain 5.2% on the trend analysis to win (to flip Florida, Indiana, Missouri, Virginia, North Dakota, Michigan and Colorado). Percentage-wise, McCain's toughest task is the regression model, since to win he has to gain 7.0% all the way up Obama's list to claim Ohio (Michigan, Florida, Virginia, New Mexico, Nevada and Ohio). Under our projection figures, McCain needs to reclaim Nevada, Indiana, Florida, Missouri, Virginia, Michigan and Colorado to win the presidency. That's a polling gain of up to 5.8%, again with Colorado being the final clinching state (working back up Obama's list) for the third of our four models.
The bad news for McCain in these numbers that among the "safe" Obama base states, Obama holds double-digit projection leads in every one of those states, with the lone exception of Oregon (+9.4%), discussed yesterday. Minnesota is next closest, projected at Obama +10.9%. The rest are just not remotely in play without a huge game-changing event in the race. Moreover, 538 currently projects a number of McCain's "safe" base states in single digits (alphabetically Arizona, Kansas, Louisiana, Mississippi, Nebraska, South Carolina, South Dakota, Tennessee, Texas, and West Virginia - see sidebar). While Obama isn't likely to win these states, our models show Obama coming closer to winning them as we sit here in late June than McCain is to winning safe Obama states like Delaware, Maine, Washington, etc.
Finally, there is the matter of the one EV in Nebraska's 2d congressional district, which Obama's camp explicitly added to the list of eighteen. Eventually, we will add broken-out data for this district and refer to the Obama-defined battlegrounds as "the 18.2."
If John McCain's camp identifies a similar roster of states it sees as battlegrounds, we will analyze that list as well.
Wednesday, June 25, 2008
The Obama Eighteen
-- Sean Quinn at 8:45 PM
Labels: battleground states, electoral math, methodology
55 comments
Great analysis Sean- Love the breakdown of the polling, trend, and regression- helps a lot.
Thomas Dorvank is right.
Off-topic, does McCain pay for each ad clickthrough, or just one per IP address per day or some such? :)
None of those results show anything approaching an Obama landslide. Even if Obama were to win every last one of his eighteen targeted states, the resulting 399 electoral votes would still be slightly below the average electoral vote win over the past 100 years. An average win is 75% of the electoral vote. The median is slightly higher still. A 350 or 360 electoral vote win is actually quite modest in historical terms.
Oh, if I were Obama, I'd be satisfied with 270 -- and count my blessings that I can still win even if my total is "far below average."
Sure, Juris, a win is a win, and that's what matters most. My point is simply that a modest win is not a landslide, and we shouldn't be calling it one.
It may be historically common to have landslides as Mark illustrated- but looking at the last two elections and the position America is in so far in the new millennium a large win would mean a lot more than many of the landslides in the last 100 years were (considering many of those elections weren't ever close- and many landslides were incumbents winning big or times when either party had for a long time been the dominant party in America.
Although I do agree that there is a lot of talk about the historic nature of the campaign- and that fact does illustrate that sometimes many of us- largely Obama supporters- do get ahead of ourselves- especially with recent likely too high double digit polls.
As for the post- great job Sean- you have rebounded well from partisan postings to very informative and analytical posts that have been top notch- and are a great reminded for why the site is really the best online.
If Sean, known for his partisan and ideological factless-based rants, is now using data, it makes me nervous.
More suggestions the new method of predicting the outcome used on this site are biased. Sean doesn't use objective, analytical information.
This isn't about the post- but I feel it's important to say.
There is a lot of posting about this site being too partisan for Obama. I am a lifelong republican and will vote for McCain- but I understand that this is not a good election for republicans. Look at the polls- Obama is up by 5 or so- not tie like Gallup or 15 like newsweek- but pretty clearly around 5. That much of a lead means Obama win. The tide is turning- we Republicans have to deal with it.
Don't blame Sean and Nate for projecting Obama wins as insane liberal ridiculousness. Are some posts partisan? Yes. But, this site is clearly the best online- and republicans that say otherwise because the site doesn't project an Obama win are being more partisan and idiotic than they claim the site is being- because any site projecting McCain will undoubtedly be using the worst analysis and data avaialable.
Message to fellow republicans- Obama is going to win unless there is a very dramatic change (not Obama's "change" btw) in polling. Don't blame pollers or analysts of political data for the shift in American politics- it just makes you look dumb- and is embarrassing republicans who are too tired of defending your brashness.
GO McCAIN.... even if it is unlikely
I sleep better at night after reading posts like these.
One interesting thing (well, interesting to me) is to see how Nate's projections line up with what Obama views as the key states. Are there any states that Nate's analysis suggests that Obama may be missing here?
By my eyeballing, the longest shots overall according to Nate of Obama's 18 are ND (41%), MT (40%), AK (36%), and GA (29%). The best shots of the states that Obama's sort of conceding to McCain at this point are WV (42%), SD (36%), and SC (28%).
So, overall, Obama's views of where he's got a shot seem to line up quite well with what Nate's seeing. The biggest exceptions are probably that Nate sees West Virginia in play and doesn't really see Georgia in play. The other interesting thing is that Nate sees the two Dakotas as being very similar, yet Obama is only focusing on one (ND, which Nate does have as his better shot - 41% - 36%). Nate also sees Georgia and South Carolina as very similar (29%/28%), while Obama apparently sees something he likes in Georgia.
Poblano, any thinking of putting up a graph of the states in order of percent chance of obama wining them, with thickness by their EV count? Might help visualize how close states are to swinging, and be easier to read than the colors (saturation?) of the map.
K J-M I wholly agree. This is a great site, even for us Republicans.
Although- I am one of the famed Republicans dissatisfied with McCain and am pondering Obama- especially after all the gimmicky energy proposals by McCain. I like the anti-pork, and his service- but I feel like he sometimes plays too many political games with faulty gimmicks that seem like the pork he wants to stop. All of this makes me think about Obama despite his being far away from my positions.
Like K J-M I am tired of stupid Republican comments on this site. ITs not a good year- and the liberal bashing just makes it wore by polarizing the election- and when more people are D's than R's this year that spells Republican trouble.
Nice job laying it all out there, Sean.
Anyone else find it odd that they'd include MT and ND but not SD when the demographics are so similar?
And if you insist on bashing this site as partisan, please point out the flaws in Nate's statistical methodology that you think causes the bias. The fact that the model's projections don't jive with what you want doesn't necessarily mean it's flawed.
Interesting to see the Obama campaign adopt this tact. I remember they formerly put up Oregon and Washington as swing states. It seems this one is more accurate of how they should be apportioning their efforts at this point: aggressively. This map is realistic in the sense of including the four blue states and all the standard red-to-swing states, but ambitious for also including Georgia, Montana, North Dakota, Alaska, etc.
First, the McCain campaign did create a list of battleground states in their strategy briefing- if you think that such a thing reflects their actual thinking, seeing as some of these are not battleground states by any stretch of the imagination:
MI, FL, WI, NV, MO, WA, OR, CO, NM, MN, IA, PA, ME, CT, & NH
Second, I do some Internet marketing work and I know that Google CLAIMS that they are able to detect fraudulent clicks and not charge fo