2012 Election Result Analysis
 

I'll be a bit indulgent, because this has become a hobby of mine over the past few elections, and I don't have people over like I have in the past - my girlfriend is actually out working in a precinct near her home, and the people I hang out with lately are either not into politics at all or too partisan to make this fun.

My goal here is to identify a few key elections and watch the results very closely as they come in, and post an analysis that is purely based on what's being reported. No exit polls, no surveys. Just a look at the votes, the counties they come in from, and an attempt to identify patterns in turnout.

Please leave any commentary on anything other than the results for the other threads. In the interest of full disclosure, I wrote in a candidate for the presidency myself, and my ballot had about as many Republicans as Democrats, with a few third-party candidates thrown in. I watch this stuff because I love simulation development and a national election is sort-of a Super Bowl.

I decided to focus on eight races tonight. All eight are expected to be close if it's a tension-filled night. If there's a landslide either way, this won't be a long thread.

The races are as follows:

Presidency:

Colorado (polls indicate Obama leads by less than half a point)
Iowa (Obama, +3)
New Hampshire (Obama, +1)
Ohio (Obama, +2)
Virginia(Obama, less than 1/2 point)
Wisconsin (Obama, +4)

Senate:

Virginia (Kaine, D, +2)
Wisconsin (Baldwin, D, +1)

A brief note on the polling. These numbers are my own averages of the poll data reported in the media. It's not meant to be a controversial average. It's not based on past performance, because a sentient pollster, in my opinion, will learn from past mistakes - the worse they do, the more changes they should make, in theory. All polls are weighted equally, but polling companies are adjusted based on how far they are, on average, from the mean for a race. In no case is this adjustment more than one point for an entire race, and in only one case it is more than half a point.

Based on these numbers, the six states are the ones that will matter if it's a very close election. If Florida and North Carolina are neck-and-neck, it's probably a great night for Obama and these states won't matter as much. If Pennsylvania or Michigan is the tipping point, it's probably a great night for Romney. That's not to say that Romney can't possibly win Pennsylvania or North Carolina can't possibly go for Obama. It's just to say it shouldn't be a very tight race if either happens.

The two Senate races I've chosen are the ones that interest me the most. Not because the Senate is going to tip Republican. It's just that I have those two states modeled and these are close races.

It's like our own 538, but different.

In "calling" the other 45 states/district, the current total is Obama 243, Romney 235. There are 60 electoral votes at stake. I'm making no attempt to actually call those other 45 - just that if this is a very close race, results will probably be based on this subtotal.

According to the polls (and once results start coming in, the polls are not part of this analysis at all), Obama is ahead in all six states, so this would lead to a 303-243 victory. Romney has to outperform the polling released to the public if he is to have any chance of winning.

I saw where Iowa expects a lot of conservatives to be voting because they are trying to vote out the 4th and last Judge that ojkayed same sex marriage.
That could be a game changer for Romney.

CNN exit polls has Obama and Romney dead heat at 49%. Might be awhile before we know much there.

A quick note on Indiana, which I'm not following closely....

Mourdock is +2 over Donnelly in early results, Romney is +22. So that's a 10-point difference. The RCP average is +9.5 for Romney.

This could well be a very long night for Mourdock/Donnelly if Mourdock is 10 points behind Romney.

Not really. All that means is the number of voters that get out in California, Texas, New York and Illinois. You can have 50.1/49.9 and still have an electoral landslide, theoretically.

Sorry, was just talking about Virginia. I didn't put the state there.

I'd imagine the race could be called hours, if not days, before we know who wins the overall popular vote. Especially with the situation in NY/NJ.

Just a tiny trickle coming from Virginia so far - with 4% of heavily-Republican Chesterfield county precincts reporting, there's no real meaning in the numbers even though Obama has a lead there. County numbers aren't significant at first.

However, what's interesting is that Allen (R) is outperforming Romney by quite a bit in those precincts.

MSNBC just had an interesting analysis. The guy showed Cleveland and the county it is in. In 2008 Obama won Ohio by 260k votes. That, more or less, was how many votes he won the county Cleveland is in, by.

As I'm sure the television analysts are pointing out, the early returns from Virginia are largely coming from areas McCain won easily in 2008. My sense is that Romney is outperforming McCain, but McCain lost by 6.4 points.

Getting back to Chesterfield County in Virginia, which went 7.5 points for McCain in 2008. Right now, with 77% of the precincts in, Romney is up 8.5 points and Allen 5.6 points.

Not knowing what 77% is in, I don't know that it's significant. But the numbers closely match the polling in terms of the difference between Romney and Allen. He needs a higher margin from Chesterfield if he's to win the state.

Enjoying this thread -- thanks for putting it together!

Question: Did you give any serious consideration to tracking the MO Senate race as one of the keys? I'm guessing the poll indications shoved it off your radar, but I'll ask for your take about whether it's fair to infer much from it if it's notable closer than expected; i.e. Is there any real party/issue relevance in the outcome or is it entirely personality/image driven at this point?

That's an interesting question. No, I didn't track either MO or IN because it takes a while to set up a state model, and these are states Romney will win by a lot if he's to have a chance.

But both senate candidates made names for themselves in ways that could motivate further-right voters to come out, inadvertently boosting Romney. The question for the candidates is whether that turnout is worth it in terms of the middle-roaders who were turned off by the issue.

We saw Missouri, once considered a possible Obama pickup, move much more than most states in October. Was that measuring, in terms of likely voters, the extra right-side presence Akin may have generated? I'd look at turnout in the southern part of the state to check that. Turnout is a key part of the way I'm modeling this.

One county (a tiny one in the west) in Virginia indicates it's 100% in, and with 99.4% of 2008 turnout. Bland went +40 points for McCain in 2008, +49 points for Romney.

More a mild check than anything of serious interest so far. 3,000 out of 3.7 million voters is going to produce skew of some sort - can't measure it with any precision. But it's about what Romney needs to put Virginia into his column by a small amount, if it's a consistent outperformance of McCain.

I think Donnelly was favored post Mourdock's comments regarding rape and abortion. I don't think there more then a poll or two done after his comments though.

As I noted elsewhere, looks like a very close race in the IN-Gov race as well.

It took a while for the first poll to come out on that race, the first one had Donnelly up 11, the next was Donnelly up around 5.

EDIT: There was also a poll that had Mourdock up 2, but that was one released by Mourdock's campaign.

True. That's why it needs to be measured more as how much he trails Romney's performance than how he does compared to the very limited polling (which, IIRC, skewed heavily Republican based on who polled it). I think it will be very close, and I think Mourdock will have a lead late, because the northwest corner, which has a strong Democratic lean, always seem to report late.

From 538 on Virginia

If I were an Obama supporter, I would start to feel pretty good based on Virginia results. Turnout is around 2008 level, and Romney is only slightly outperforming McCain. If I had to guess right now, with only a tiny bit of the information I need to make a call, I'd say Obama will wind up with a 2-point win in Virginia. Obviously, a long way to go, but it's not the turnout pattern Romney needed to see. Both Republican counties and Democratic counties (again, very early) are around the same as they were four years ago.

My sense of Allen/Kaine is that it's starting to look very good for Kaine. I'm going to move over to Ohio for a while, maybe check back in on Virginia in about an hour.

update on VA, I'm seeing reports that they are Suspending reporting votes because there's still long lines.

As a Virginian, the heavy emphasis on Virginia analysis is intriguing to me.

If turnout is similar, that's surprising based on my co-workers, facebook friends, etc - for many the polling lines were a disaster.

I think this state needs a LOT of work on streamlining that process.

Not sure where this goes, but the city where I am working as a teacher in, has a prop up for voting, in adding a 1cent tax for every ounce of sugary drinks

I'd think a Kaine win is a good sign for Obama. Not 100% sign though because there is some genuine hatred towards Allen whereas Kaine was more a "meh" Governor.

Funny that you post this right at the same time I was making my post ;).

I'm not very far into Ohio, but my sense of these very-early results is that Obama is actually outperforming how he did four years ago. Now, the big chunks of votes are coming in from his strongholds (like Cleveland), but the numbers are nowhere near what Romney needs to see.

So far, I'm thinking the fulcrum of this election is Florida and North Carolina, which, if true, means a rather significant Obama win.

Not among Jim's target races, but the exit polls are showing an interesting fallout from the Todd Akin episode.

The exits show a gender breakdown of 56/44 women to men. In the Senate and Governor race, men are breaking even but Claire McCaskill is up 56-39 over Akin among women. In the governor's race, men also are breaking even and Democrat Jay Nixon is up 54-45.

Last week, I heard a major GOP strategist in Missouri note that when the presidential race in Missouri is a blow out (more than 8 points), the winning party wins 88% of the other state races on the ballot. Exits suggest about a 10-point lead for Romney. In that case, Missouri Democrats should be lucky to win one race.

However, it looks like Democrats will certainly win the Senate, Governor and Attorney General. Granted, each of those races are incumbents. Another Democratic incumbent should hold his office. The other two races lean Republican (one is a Republican incumbent, the other is a Democratic open office).

In what the numbers for president should signal a wave election for Republicans in Missouri, it looks like the Democrats will hold all but maybe one office. And if they do, it looks like women will be the sole margin of victory. And women turned out and voted Democratic because of Akin.

It's a little harder to be precise about Ohio, because the %precincts reporting measure is seriously off. It reports Logan as 100% in with 30% of the 2008 vote in.

Missouri is also voting on Prop B. This is a 90 cent tax on tobacco products. It is losing right now, 61% to 49%.
And to top it off, the money by law goes to education.

So our citizens are defeating a sin tax that will put anywhere from $239 million to $438 million (If I remember the numbers right from voting today) into education. GO MISSOURI!

Romney seemed to be about +3-4 on Allen in pre-election polling. Based on the scattered early returns I think Romney can still win Virginia while Allen loses, but Romney needed to win Virginia by enough to put Allen/Kaine in play if he wanted to really put Ohio+ into play and have a chance nationally, and it doesn't look like that's happened. Especially if there are still long lines, which tend to be in cities, which tend to go Democratic.

If Mourdock is beating Donnelly in the portion of state where Romney is +13, then Mourdock should win comfortably (~10%). That means the blue parts of the state like Indy aren't reporting in yet.

SI

Anyone know if there has been a noticeable Virgil Goode effect in VA yet? I know there was supposed to be a specific part of the state where he might have more of an impact (I think SW VA), so wasn't sure if that area is reporting yet.

I don't see it. I think it's going to be very close. Without an Indiana model, though, I can't be precise at all.

Don't react too much yet. The first 90 minutes or so of Missouri results are always from outstate, which are always going to go Republican and are anti-tax. Once Columbia/Jefferson City comes in you get your first Democratic results, and Kansas City and St. Louis don't come in until after 10 or later.

Six years ago, Claire McCaskill was trailing with the City of St. Louis the last place to report results, and that city alone delivered her the winning margin.

That said, 60/40 is a larger margin than I would like to overcome.

Without any turnout numbers, or knowing if there's going to be an early-voting dump with a different skew, Ohio looks very favorable for Obama - almost on the level of calling it early favorable.

If I had to guess, again, with little to go on, I'd say Obama is +7-8 on Romney in Ohio, taken as a whole, down the road. This has nothing to do with the fact that the 200k lead right now is based mostly on 100k from Cleveland and 70k from Columbus in the reported votes.

If the tobacco tax money is going 61-49, it better be going to math education.

I'm just going on Pres vs Senate projections. Romney's not going to win Indiana by 14%. That means a lot of the more Democratic sections aren't reporting yet. One has to think Mourdock has an advantage in those.

SI

:D

SI

Still quite early, of course, but with a lot more data in Virginia I'm still not getting the numbers Romney needs to move this. It's projecting to about a 3-point Obama win in Virginia. Turnout seems about 1-2% over 2008 levels, almost uniformly.

Fox calls Donnely over Mourdock in IN-Sen

This evening is turning out rather anti-climactic, isn't it? Some have already called Wisconsin for Obama (and Baldwin). I'll take a quick look over there.

Very early in Missouri, but while Romney is up 58-40 McCaskill is up 51-42.

HAHA. I just realized I was swapping the names between Mourdock and Donnelly

I re-read what I said and it made no sense (d'oh).

SI

It's so early in Wisconsin that the numbers aren't meaningful, but what I'm seeing is very much in Romney's favor. If someone called Wisconsin, it's completely based on exit polling. It looks like Wisconsin will be very close. But nowhere near enough data to do more than guess.

:)

What type of voter does the legalization of marijuana issue bring out?
Will that sway a state to Obama or Romney?