Expected State Polls Based On National Polls

by: Chris Bowers

Fri Aug 15, 2008 at 11:23


Looking at the twenty states most commonly described as swing states in this election, the following table compares current, actual state polling to "expected" state polling. The "expected" numbers take the 2004 state-level result, and shift it 5.96% in favor of Democrats, given that Kerry lost the national popular vote by 2.46% but Obama leads the national popular vote by 3.5%. The current or "actual" state polling is taken from Pollster.com.

Expected Vs. Actual State Polls
State Expected Actual Difference
Alaska McCain +19.6 Obama +2.7 Obama +22.3
Colorado Obama +1.3 Obama +0.9 McCain +0.4
Florida Obama +1.0 McCain +1.4 McCain +2.4
Georgia McCain +10.6 McCain +5.5 Obama +5.1
Indiana McCain +14.7 McCain +0.6 Obama +14.1
Iowa Obama +5.3 Obama +6.0 Obama +0.7
Michigan Obama +9.4 Obama +5.4 McCain +4.0
Minnesota Obama +9.4 Obama +9.4 As expected
Missouri McCain +1.2 McCain +2.8 McCain +1.6
Montana McCain +14.5 Obama +2.9 Obama +17.4
Nevada Obama +3.4 McCain +2.7 McCain+6.1
New Hampshire Obama +7.3 Obama +6.1 McCain +1.2
New Mexico Obama +5.2 Obama +8.2 Obama +3.0
North Carolina McCain +6.5 McCain +4.4 Obama +2.1
North Dakota McCain +21.4 McCain +2.4 Obama +19.0
Ohio Obama +3.9 Obama +3.6 McCain +0.3
Oregon Obama +10.1 Obama +6.8 McCain +3.3
Pennsylvania Obama +8.5 Obama +8.5 As expected
Virginia McCain +2.2 Obama +0.9 Obama +3.1
Wisconsin Obama +6.3 Obama +10.0 Obama +3.7

There is a lot to say about this table, but I am struck by Alaska, Florida, Montana, Nevada and Virginia. The reason these five stick out to me is that their changes actually result in the state slipping to the unexpected party. Overall, this a 32 electoral vote gain for Republicans, and a 19 electoral vote gain for Democrats, making for a net gain of 13 "expected" EVs for Republicans.

McCain is over-performing the most in Michigan and Nevada, while Obama is heavily over-performing in Alaska, Indiana, Montana and North Dakota. Otherwise, the states are petty much all where one would expect them to be. What do you see in these numbers?

Chris Bowers :: Expected State Polls Based On National Polls

Tags: , , , , (All Tags)
Print Friendly View Send As Email
Expected State Polls Based On National Polls | 26 comments
I don't understand this .. (0.00 / 0)
is Obama losing any state that Kerry won in 2004?  So how is it a 32 EV gain for the Republicans?
versus (0.00 / 0)
where the national polls would expect Obama to be in EVs.

Really it's all about Florida still being favored for McCain.

[ Parent ]
Confused (0.00 / 0)

You have Kerry winning Florida, Colorado, Iowa, or am I interpreting "expected" incorrectly?

Oops (0.00 / 0)

I should read more carefully!  I get it now.  
[ Parent ]
It means (0.00 / 0)
Add Obama's national margin over Kerry of 5.9 (Kerry lost by 2.4, Obama is winning by 3.5) to Kerry's share of the vote in each of those states with an even distribution.  So, Kerry lost Ohio by 2%, add 5.9 percent to that, and Obama "wins" by 3.9.  He is currently ahead on Pollster by 3.6, so he's "underperforming" in Ohio as compared to if we just adding his margin over Kerry to Ohio specifically.  

[ Parent ]
What they say to me is (4.00 / 1)
When Obama campaigns hard, and with little pushback, he does well.  Montana, Indiana, North Dakota, Alaska.  Michigan he does worse in.  I suppose 3 cheers for the primary.  Regardless, I think starting on his return from vacation, Obama is going to be a workhorse, the money is going to go in full force, and the ground operations go in full force from now until the election (75 days or so, good god).  

This is why I'm confident.  We can fret about television ads, the Obama response, etc. all we want.  However, when I see numbers like these, it tells me that locally Obama is a good campaigner.  The campaign may know how to work local media really well, which I think is one of the more underreported facets of the election season.  There is a reason, I think, that we keep hearing about all of these local negative ads in Ohio et. al.  Local stations may play them and talk about them, but Obama doesn't get hit for going negative on a national level.

I don't think it's as much about campaigning as (4.00 / 1)
it is about geography. This has more to do with the idiosyncratic natures of regions and indididual states as anything else.
[ Parent ]
Both (0.00 / 0)
Geography may set some limits. But those limits are a bit squishy and leave a lot of room to expand.  The campaigning takes advantage of the opportunities left.

But you're 100% correct to caution folks that campaigning alone is not the explanation, and cannot vanquish all other factors.

"You know what they say -- those of us who fail history... doomed to repeat it in summer school." -- Buffy The Vampire Slayer, Season 6, Episode 3

[ Parent ]
I am struck by the fact (0.00 / 0)
that there are four states -- Alaska, Montana, North Dakota, Indiana -- where Obama is showing a flabbergasting 10-20 point gain -- and none of the other states show anything close to this. What's so special about these four states, as opposed to all the rest? Montana and North Dakota are right next to each other, so maybe one could buy that there's some underlying demographics or whatever which causes Obama to have much greater appeal in this area; and they say Montana and Alaska both have a libertarian bent (and failing that, Alaska is so far from anything else that it really could be in a class of its own); but Indiana? Why? Why such a huge swing in Indiana but not Ohio, Michigan, or Kentucky? (Kentucky isn't shown here, but I'm assuming it's not very favorable for Obama either). Is it because Indiana borders Illinois? If so, why aren't we seeing the same thing in Missouri, Iowa, and Wisconsin? (Iowa and Wisconsin are favorable, but +0.7 and +3.7 are not +14.1).

I think it's very important to note, however, that Obama isn't the only possible source for these discrepancies; there's three others: John McCain, John Kerry, George W. Bush. Any of these four candidates could have some special appeal or (what's the opposite of appeal?) to certain states/demographics, and any combination of the four could be leading to the numbers we see above. Hell, it's possible to have the quadruple-whammy of a state where George W. Bush has more appeal than a generic Republican while John McCain has less, and John Kerry has less appeal than a generic Democrat while Barack Obama has more, in which case all four factors would be causing the state to swing in Obama's direction. (The reverse, of course, is also possible.)

I also suspect that, logically, there should be a number of other states among the remaining 30 (South Dakota, for example?) where Obama should be showing the same 10-20 point improvement over his 'expected' performance; or is it possible that there are states where it exists but we haven't 'discovered' it, that some states are actually in play without anyone knowing it? (Seems doubtful...)

Obama campaigned hard in Indiana (4.00 / 1)
For the primary and has a very good field operation there.  Plus it's midwest, where he does well.  Iowa he has a higher lead but it is more like expectations.  Maybe that is just getting cxlose to his ceiling there.  Afterr all, he campaigned there a lot in 2007 and again right after he won the nomination.  In Ohio Hillary fought harder.  He may have some residual problems in Southern Ohio. I'd expect he will improve there and in Michigan, though.  He'll take the midwest except for the Plains and maybe MO.

John McCain--He's not who you think he is.
[ Parent ]
None of those states are very "white-evangelical" (0.00 / 0)
I ask the same question below... but a thought occurs.

Obama seems to do worse in states with large numbers of white, conservative christians, and the White 'Murican identified voters of appalachia. This carries over to states with historical emigration from those regions.

So, parts of the Central South and Western South were settled by Southerners and Appalachians (OK, MO, TX, CO, and even CA... remember the Okies?). I'm guessing that ND and MT were settled more like MN, with scandinavian and "regular" protestants.

I hesitate to think it is some kind of racial issue. But, perhaps states with very low numbers of African-Americans are less motivated by race. Or, maybe it is more about the brand of conservative Christianity, or the influence of certain right-wing preachers.

AK is a bit iconoclastic; lots of rugged individualists and libertarians.

[ Parent ]
What I see in these number... (0.00 / 0)
is that if you mix national and state polling, mix numbers from 4 years apart for completely different candidates in a completely different environment, and use an aggregation of polls with different margins of error versus actual election results, you can come up with "expectations" that are mostly incorrect.
Also, you can see up your own asshole if you really try.
Using Kerry's national poll instead of the election result (0.00 / 0)
doesn't make much of a difference. He consistently polled two percent behind Bush. Looky here:
http://edition.cnn.com/ELECTIO...

The question is, does this method make any sense? Chris should check if the same method, used on Kerry's polls and the 2000 election results, shows any resemblance to the actual results of 2004.

[ Parent ]
Problem with methodology (0.00 / 0)
"The "expected" numbers take the 2004 state-level result, and shift it 5.96% in favor of Democrats, given that Kerry lost the national popular vote by 2.46% but Obama leads the national popular vote by 3.5%."

Wouldn't it be more accurate to take Kerry's polls in each state at the same time in the campaign, calculate the difference to Obama's numbers now, and adjust the 2004 results with this factor? Using the same national correction factor of +5.96% for all states seems to be overly simplistic to me...

Btw, how does this correlate with Paul's electoral maps? (0.00 / 0)
Just askin', but this seems to paint a totally different picture. Looks like Obama will still end behind McCain, and desperately needs a VP who can deliver Florida. Hmm, who could that be?
Apples and oranges (4.00 / 1)
This doesn't predict a McCain win; Obama doesn't need Florida.
[ Parent ]
Misunderstanding on my side? (0.00 / 0)
"Overall, this a 32 electoral vote gain for Republicans, and a 19 electoral vote gain for Democrats, making for a net gain of 13 "expected" EVs for Republicans."

I understood this as saying that the Republicans will gain 13 more electoral votes than 2004. On second thought, this is probably nonsense. But why does Chris write "gain" when actually the GOP will lose votes???  

[ Parent ]
Gain over what was expected? (0.00 / 0)
This was a bit confusing, though.

John McCain--He's not who you think he is.
[ Parent ]
What's missing here (0.00 / 0)
is table that shows the "expected" results translated into electoral votes for these swing states. How much will Obama be ahead in the electoral college by Chris' count?
[ Parent ]
Chris Means A "Gain" Over What An Equally-Distributed Margin Would Produce (0.00 / 0)
The baseline Chris is using for comparison is one in which every state votes 5.96% more Democratic. What he shows is that state-level polling indicates the Democrats won't do that well.  They will get 13 electoral votes less than expected.

This doesn't mean they'll lose.  Just that they won't win as big as the national margin would predict. (If the election were held today, yada-yada-yada.)

Obama does pick up Ohio and hold New Hampshire, and that alone is enough to win.  On top of that is all sorts of gravy--Alaska, Colorado, Iowa, Montana, Virginia, etc.

So not bad news. Just cautionary.

"You know what they say -- those of us who fail history... doomed to repeat it in summer school." -- Buffy The Vampire Slayer, Season 6, Episode 3

[ Parent ]
What do you think about the methodology, Paul? (0.00 / 0)
Chris seems to have looked for a method determining where Obama performs differently than Kerry. Well, his calculation sure shows some surprises. But is the way of adjusting the 2004 results by a national correction factor sound? There are only three states in the list that are within one percent of the "expected" result. This doesn't really convince me that this way of "correcting"polls produces probable results...
[ Parent ]
It Seems Sound And Sensible To Me (0.00 / 0)
Just because it dosn't show uniform change doesn't mean the methodology is bad.  The non-uniform changes are probably quite real, and need to be better understood.

"You know what they say -- those of us who fail history... doomed to repeat it in summer school." -- Buffy The Vampire Slayer, Season 6, Episode 3
[ Parent ]
Hard to Say (0.00 / 0)
On the one hand, I might be a little squeemish about whether or not these big gains in places like Alaska, Indiana, Montana, and North Dakota are really durable. I doubt all this Obama support in those states is as hard as we would like, and without some additional move (like picking Schweitzer as VP in MT) I wonder if Obama's gains would survive if McCain ever breaks down and invests time and media in those states.

On the other hand, it might not matter if those gains are durable, as long as McCain has to expand the scope of his campaigning and advertising in the Fall which will pull some resources away from the really key states which appear at this juncture to be Colorado, Michigan, Ohio, and Virginia, and McCain needs to win three out of four of these states just to have a chance.

What this says to (0.00 / 0)
me is that there are a hell of a lot more states in play for the Democrats than is usually the case and that there's a lot of time left before the election and that the Democrats are going to have a lot more money than they usually have and that the enthusiasm and ground game this year is in the Demos favor and that at this point things are looking good for Obama. Obama won't lose for lack of money or organization and his expand the map strategy, while perhaps more dangerous for him, will help a lot of people down-ticket. The mere fact that places like Florida and Missouri and Virginia are in play will make them very expensive states for McCain.  At the present time, Obama's main problem seems to be that there isn't a candidate he can choose for VP without irritating a lot of people.  Can he run alone?

BTW, I love the fact that Openleft is getting money from advertisers of an anti-Obama book.  I'm sure a lot of the readers and contributers to the site will rush right out and buy a copy.  I hope you guys make a lot of money off the ad.  To paraphrase the late Jess Unruh: "If you can't drink their whiskey, take their money, and sleep with their women and then vote against them you shouldn't be in politics."

Why huge shifts in certain stages? (0.00 / 0)
What strikes me are the huge shifts to Obama in certain states: MT, IN, AK, and ND. What makes them so different from the states without big shifts? These states all have fairly high percentage of Republicans.

In contrast (as one example), Colorado marginally shifted toward McCain in your chart. CO has long been considered a red moving to blue state, even though Republican Party registration is still larger than Democratic registration. I would also say that the CO Republicans are fairly conservative while Democrats are quite liberal, and the only reason Democrats are doing so well is that a lot of Colorado Independents are really Dem sympathizers.

Rural Colorado might be thought to have something in common with rural states like MT or ND.

For whatever reason, Colorado seems to be more rigid or the battlelines are more solid or something.

Or McCain and the RNC are making more of an effort in Colorado (0.00 / 0)

[ Parent ]
Expected State Polls Based On National Polls | 26 comments





Donate to Open Left




blog advertising is good for you
blog advertising is good for you
LIBERAL INSTITUTIONS
STATE BLOGS
USER MENU

QUICK HITS
SEARCH

   

Advanced Search

Home | RSS/XML Feed | Register | Contact

  © OpenLeft 2007-2010, Powered by SoapBlox