At the time I said I would run another projection at the precinct level instead of at the county level. My assumption was that Obama was registering new voters in Democratic friendly precincts in the Bush counties. By running the projections at a precinct level, I expected my projections to narrow considerably.
I have completed my analysis and the results and the exact opposite of what I expected. Running the projections at a precinct level (there are over 11,000 precincts in Ohio) and summing them up increases McCain's margin of victory to over 147,000 votes.
I used the same method as I did at the county level in my previous post. I took the current number of registered voters for each precinct, multiplied it by the turnout factor for that precinct in 2004 and then applied then gave McCain and Obama the same percentage of the vote that went Democratic and Republican in 2004. This netted McCain an additional 22,000 votes over doing the same thing at a county level.
It should be noted that approximately 1% of all precincts in Ohio appear to have been redefined since 2004. In cases where the precinct did not exist in 2004, I applied countywide factors to make the projections.
What this means is that the new voters did not just come in primarily Republican counties. They came from the most Republican precincts in those counties.
Even in Democratic leaning counties, precincts with some of the largest increases in registered voters went Republican. For example, Monclova Township Precinct 11 is in Lucas county, increased by over 1,000 registered voters in 2008. In 2004, Lucas county went for Kerry 61% - 39% but this particular precinct when for Bush 54% - 44%.
The opposite is also true, some of the largest decreases in registered voters came in precincts that lean heavily Democratic.
All of this would appear to be good news for McCain, but obviously the polls don't show this.
This means one of three things are happening:
- Obama went into the heart of the Republican strongholds and found Democratic voters and registered them.
- Voters who voted for Bush in 2004 are crossing over and voting for Obama this year or not voting at all.
- The polls are wrong and are assuming a larger turnout of Democratic voters than has occured in the past and will occur this year.
If it's 1, that's bad news for Republicans and Ohio could be blue for a long time.
If it's 2, it will be interesting to see if the voters who switch will stick or not in future elections.
If it's 3, hang on to your hats, it could be an interesting election night.
UPDATE: Geraghty at The Campaign Spot notices the same thing in Virginia.
|