The Hottest Models, Only on VodkaPundit
August 24th, 2012 - 8:45 am
Alert Reader bdog57 has his own model for wargaming the electoral college, and it goes a little like this:
Click for the full-size version — you’ll need it, if you value your eyeballs.
He has a fascinating methodology, although you might need to invest in a bigger monitor to appreciate it. Maybe I should have asked for the raw spreadsheet file…
Here’s the link to the comment where bdog explains his methodologies. You might want to put it in the post or even excerpt some of the explanation so that people understand the terms.
Something I do not get is why Romney has a limit based on ’08? Obama, sure. He is not going to outperform ’08 anywhere. If someone did not vote for him then, they sure as heck are not gonna vote for him now. But for Romney, the sky’s the limit, right?
MN doesn’t look right. Obama’s limit looks wrong? There’s this orange bar, too. What is that? NV is out of alphabetical sequence, btw.
Finally, as I commented in the other article, I do not see how Romney effectively gets to that number. This has him carrying every battleground State, including CO. This has him carrying PA, but not MI. I think it odd, because he seems to have a better chance in MI than in PA. I guess one could make the argument, that if he is carrying PA, he is carrying all the battleground States, too. But then, he’d likely carry some States like WA, OR, CT, and MN, if he is that strong.
Saw a poll on NM, today. Romney down by 14 there. He has an Hispanic problem. That is a bad sign for the rest of the nation. Lots of Hispanics all over the country. It’s a critical weakness. There’s no way Romney wins this big with such a weakness. Bush won, because he carried 40% of the Hispanic vote, rather than the usual 30.
I think in practice that this projected result seems over-optimistic. I’m thinking more like 282-256 for a final result.
Marc,
Follow NukemHill’s link. All methodology explained there.
Minnesota is colored orange to indicate that no matter what the model indicates, I just can’t get myself to put it in the Romney column (per point 7 at the link). Minnesota voted for Nixon in ’72 but NOT for Reagan in either ’80 or ’84. Can’t see it flipping now.
The states are listed in order of name, not abbreviation; hence, IA (Iowa) comes after IN (Indiana).
“2008 Limited” refers to two things (covered in points 4 and 5 at the link):
a) Obama is NOT getting a single vote more than he got in 2008. That column is his vote total in the given state in 2008.
b) Swings of more than 6% towards Romney are averaged; e.g., Obama won Iowa in 2008 with 54.0% percent of the vote. His “Demo + PVI” percentage for 2012 is 43.2%. This swing of 10.4% seems unacceptable. To account for this, any percent difference over 6% requires that two percentages be averaged. This leads to a percentage of 48.4%. (Which seems closer to reality, given RCP’s current number of Obama +1).
My methodology only employs Gallup’s weekly approval numbers applied to the demographics of the state; all of which is tempered by the Cook Partisan Voter Index (PVI). So, Michigan has a PVI of D+4 while Pennsylvania has a PVI of D+2. Further, Michigan has a larger percentage of non-whites than Pennsylvania. Given that non-whites have a much higher approval of Obama and the greater Democrat tilt, we can reasonably expect that Pennsylvania will swing before Michigan does.
As for the Hispanic numbers, you should know that they have a 56% approval of Obama – far less than 72%. Keep in mind that we’re the party of Susanna Martinez, Marco Rubio, and Ted Cruz as well (not that I like to talk up identity politics).
You can see the composition of the white/non-white demographics per state as the three-digit decimal values under “Demo Approval”. So, in Florida, 57.9% of the population is white and 42.1% of the population is non-white.
There are a lot of X factors out there that I don’t account for (other than Minnesota). I began this model because I recall a post by Jay Cost at RCP where he covered the fact that re-election numbers tend to closely resemble approval numbers. I don’t quite like divvying approval ratings between white and non-white, but it turns out to be a fairly useful metric (I copped this from IowaHawk’s dissection of the Texas vs. Wisconsin education numbers).
If you (or Stephen) would like a copy just give me a few days. I’m trying to automate it some more to make it even more user friendly -as well as include some documentation.