Nov 06 2012
Edited at 3:00 & 3:07 PM Eastern, Updated 3:40 PM
OK, I needed to get some confidence that Romney will win this thing, and so I took a look at the county-by-county tallies for Ohio’s early/absentee voting and did some analysis. Due to the complexity and size of the spreadsheet I am not posting the details, but here are the scenarios and results.
First off, I wanted to assume 2012 looked like 2008 in terms of D+whatever and just see if there is an enthusiasm edge out there for the GOP. And this should indicate Obama’s ceiling.
Here is the general process:
For each county I have the total absentee/early votes cast (not counted) for 2008 and 2012.
For each county I compute the change in votes from 2008.
Then I look at the 2008 county by county tallies for Obama and McCain and note which counties went blue or red. This was then used to determine the difference in voting from 2008 by red or blue county.
If you look at counties that went GOP in 2008, they gained 84,568 votes from 2008.
If you look at counties that went Dem
GOP in 2008, they lost -41,652 votes from 2008.
That right there shows a GOP momentum. Blue counties are down, red counties up. Problem is, counties are not monolithic or clearly on one side or the other. But as a rule of thumb, this is encouraging.
Update: To be clear because the math is complex and sometimes misleading. If you take the 2012 vote deltas in county votes, and run them through the McCain splits (which is not the same as what I computed below for the 100,000 change) the GOP gets a net 30,000 votes. What I did below was run the total 2012 vote through the 2008 turn out model, not just the delta’s. There are many ways to model a 2012 result, and many knobs to twirl that change the result. All are mathematically valid, the question is are the realistic! My engineering approach is not to determine a precise estimate out of noisy data, but to bound the likely outcomes based on positive and negative assumptions. To build a multidimensional box of results that span realistic outcomes. Please don’t take this as a given for Romney, but he has more positive indicators than Obama. And no, I don’t do Monte Carlos! - end uupdate
Now of course each county splits the vote based in their partisan make up. SO I decided to project the 2012 absentee voting using the 2008 county percentages. If we hold the county percentages constant between Dem and Rep but now project the 2012 absentee/early vote outward, Obama loses 100,000 votes in his hypothetical match up with Sen McCain. He still wins, but his lead is cut in half and indicates what Obama can expect today at best. If the counties enthusiasm is run through the 2008 returns, the race is 51-49. But of course, that is not what is happening this year at all.
We all know this is not a D+7 election like 2008. Not even close. So how much does the 2008 race have to switch to Romney for him to win Ohio?
The number even surprised me – because it was so low. If I take just 3%
2% from Obama in each county and give it to Romney, then compute the 2012 results based on absentee/early votes, Romney wins. Just 3%! That means a D+4 nationally can give Romney the win in Ohio by a tiny margin
I am seeing GOP counties, on average, increasing their voters by 11%, Dem counties by only 2% (assuming my math is right – always a challenge).
So, bottom line: Yes, I am confident Romney wins Ohio, and therefore the election. I am thinking, if it is really a tie in Ohio, then the state has moved 4% towards the GOP since 2008, and that would give Romney a 51-49 win