Early Voting Estimates In Seven States

by: Chris Bowers

Wed Oct 22, 2008 at 14:00


I just whipped up my best estimates for current results in seven key states where early voting is well under way:

Early Return Estimates From Key States
State Reporting Obama % McCain %
Colorado 16% 49.2% 49.8%
Florida 11% 45.2% 51.2%
Georgia 23% 46.7% 52.2%
Iowa 12% 64.1% 35.4%
New Mexcio 28% 53.7% 46.1%
N Carolina 20% 55.4% 44.1%
Ohio ??% 60.5% 38.7%

I arrived at these numbers as follows:

  • Information gathered from news articles here and here, plus quick hits here and here.
  • Estimated turnout percentage = ((total number of votes cast in 2008) / (total number of vast cast in 2004 * 1.1)). I am assuming a 10% increase in total turnout. It probably won't be that high, but at least I am not over-estimating the percentage of votes cast so far.
  • I multiplied the current partisan breakdown of returns according to how they broke according to 2004 exit polls. For example, in 2004, Kerry won 93% of the vote among Colorado Democrats, so I am assuming Obama will so the same.
  • In Georgia, I multiplied numbers according to racial breakdown. Everywhere else, it was according to partisan breakdown.
These numbers tell me that things are going OK, but that the campaign is far from won. In particular, Colorado stands out as problematic, since McCain still holds a very slight lead there according to these estimates. Obama only leads in two "victory states," North Carolina and Ohio. I think we should all consider victory in either of those states to be very dicey, especially North Carolina. And I doubt anyone would be happy about the election coming down to Ohio.

One positive is that there are no numbers from Virginia, as I believe there is no partisan registration in the state. If Obama is winning there, then his clearly advantages in Iowa and New Mexico give him a relatively easy path to the White House. However, there is no information from Virginia right now that I know of, so that is purely speculation, nothing more.

Update: Florida numbers were reversed, I fixed them now. Obama is gaining fast in the state, but I estimate that he still trails by about 55,000 votes. Also, Obama could win Georgia if he wins African-Americans by a net 92% (say, 96%-4% or 95%-3%), and if turnout remains at current levels (roughly 36% of the electorate as African-American). The former seems doable. The latter, however, would be the greatest turnout surge I have ever seen, since African-Americans normally make up 28-29% of the Georgia electorate.

Update 2: Yes, I am counting absentee ballots requested, but not returned, in Florida. I know that tilts things toward McCain, but I feel it is the safest, most conservative estimate to make.

Also, check out this great early voting monitoring site.

Chris Bowers :: Early Voting Estimates In Seven States

Tags: , , , (All Tags)
Print Friendly View Send As Email
Early Voting Estimates In Seven States | 30 comments
Huh? FL is chopped liver? (0.00 / 0)
Looks like he's leading there too, unless your numbers are off.
Iowa 12% 64.1% 55.4% (4.00 / 2)
Iowa: 64.1 + 55.4 = 119.5.  Does not compute.
Didn't you know? (4.00 / 1)
Iowa gets an extra 20% for being "first in the nation . . . "
[ Parent ]
Fixed (0.00 / 0)
Lots of typos this afternoon, like most afternoons when I am typing, which is most afternoons. :)
[ Parent ]
Colorado (0.00 / 0)
Chris,

From what I understand the NY Times numbers on Colorado and Florida are questionable.

Here is a cite that has been tracking pretty thoroughly the early voting numbers.

http://elections.gmu.edu/early...

Thanks for the link (0.00 / 0)
Great site. I'll look that over and fix the numbers.
[ Parent ]
Colorado % reporting updated (0.00 / 0)
But I haven't found anything to contradict the NYT numbers on partisan reporting so far.
[ Parent ]
A couple notes... (0.00 / 0)
Is Florida not one of the victory states? You've got Obama up there.

Also, you've got a total of 119.5% between McCain and Obama in Iowa. I assume McCain is supposed to be 35.4%...

My own math for Georgia, using the SOS numbers combined with racial cross-tabs from the recent R2K poll gives Obama a narrow lead (47.7-45.8%) among early voters in Georgia.

Conduct your own interview of Sarah Palin!

I think you're both off on GA (0.00 / 0)
Firstly, I doubt 3rd parties are getting 6.5% of the vote.  Secondly, AA turnout is at 35.5% of the total electorate.  You can assign 90 or 95% of that to Obama.  Whites are almost entirely the rest.  Chris has given what looks like 90% of the AA vote to O, and 25% of the rest to O.  I think it's more realistic to assume 95% of the AA vote is Obama's. If you do that, and leave the white/other vote alone, Obama has 49.85% of the vote.  As long as third parties are taking at least 0.3%, then Obama is ahead.
[ Parent ]
Allocating polls (0.00 / 0)
My allocation method (because it's based on polling) assigns 2.4% to third parties and 4.2% to undecided. Obviously if they've voted, they're not undecided, so maybe I ought to allocate those voters somehow too...

Conduct your own interview of Sarah Palin!
[ Parent ]
I've removed the undecideds (0.00 / 0)
I've divided each of the decided vote parts by the total percentage of decided voters to factor out the undecided voters (basically assuming undecided voters haven't voted). That gives Obama a lead of 49.7-47.7 (2.5% to third parties) in Georgia. Jim Martin leads 50.7-44.9 with 4.4% going to third parties.

That probably slightly underestimates Obama and Martin's totals because the greatest part of the undecided population is African American.

Of course, caveats with margins of error and the accuracy of subsamples and polling and such apply.

Conduct your own interview of Sarah Palin!

[ Parent ]
sounds good (0.00 / 0)
and right in line with my assumptions based on demographics.  I wouldn't be surprised at all to see Obama take 93-97% of the AA vote in southern states.
[ Parent ]
Virginia (0.00 / 0)
At least asks what your political party is, as I have registered here just a month ago.  
Exit Polling on Election Day?? (4.00 / 1)
With a third of voters now voting early, what relevance to exit polls have at all? Do they have any predictive value? I doubt it.

Another point- there is an important distinction between absentee voting and early voting. Absentee ballots have always historically favored Republicans because of the demographic patterns. Early voting with people showing up at the polls tells us about voter enthusiasm.

And of course, in my state, Oregon, everyone votes early by mail. So none of these things apply.

Good point (0.00 / 0)
Good point about exit polling.  I think it's a horrible idea to do anyway.  All polling should finish before election day, and it was way off last time.

Let's hope the exit polling makes an exit.

[ Parent ]
exit polling isn't a terrible idea (0.00 / 0)
in fact, as I understand it it's the main method by which independent election monitors monitor voting in various countries, and is generally quite accurate. remove exit polls and you give even more opportunities to cover up voter suppression and make tampered results seem plausible.
[ Parent ]
I have often wondered this (0.00 / 0)
Given that Bush won early voting in 2004 by 20%, and that 22.5% of the population voted early, I think that explains the exit poll showing Kerry up 3% all by itself. Kerry proably did win election day by 2-3%, but he still lost the election because of early voting.
[ Parent ]
AP's Sidoti is at it again (0.00 / 0)
Can someone do a proper analysis/takedown. I don't have the skills and patience. She's basically claiming that it's neck and neck right now. These numbers are way out there:

AP presidential poll: All even in the homestretch
Wed Oct 22, 2008 1:16 PM EDT
Liz Sidoti, Associated Press Writer

The presidential race tightened after the final debate, with John McCain gaining among whites and people earning less than $50,000, according to an Associated Press-GfK poll that shows McCain and Barack Obama essentially running even among likely voters in the election homestretch.

The poll, which found Obama at 44 percent and McCain at 43 percent, supports what some Republicans and Democrats privately have said in recent days: that the race narrowed after the third debate as GOP-leaning voters drifted home to their party and McCain's "Joe the plumber" analogy struck a chord.

Three weeks ago, an AP-GfK survey found that Obama had surged to a seven-point lead over McCain, lifted by voters who thought the Democrat was better suited to lead the nation through its sudden economic crisis.



"Those who stand for nothing fall for anything...Mankind are forever destined to be the dupes of bold & cunning imposture" -- Alexander Hamilton
Yes, everyone just swooned with Joe the Plumber... (0.00 / 0)
Even the most of the other polls and all the focus groups thought it was pandering crap.
[ Parent ]
RV difference of 10 reduced to LV of 1 (0.00 / 0)
That's a pretty good trick. RVs + 10, LVs+1. That says everything about this wretched poll.
[ Parent ]
RVs is 5... (0.00 / 0)
All adults is 10.

Either way, it seems like a bit of a stretch.

[ Parent ]
Well, if they manage to steal another election (0.00 / 0)
Then they're going to need favorable pre-election polls to point to, to show that it was predicted. Plausible deniability and all that. I can see this serving such a purpose. A nation of Joe the Plumbers emerge out of nowhere and save the Maverick!

But I'll let the experts go at it from a technical pov.

"Those who stand for nothing fall for anything...Mankind are forever destined to be the dupes of bold & cunning imposture" -- Alexander Hamilton

[ Parent ]
Using absentee ballots, Chris? (0.00 / 0)
I don't see how you arrive at your numbers otherwise.

What's the percentage of absentee ballots that actually turn into votes?  I'm sure it's a fairly high number, but is it worth adding them in before they're actually received?

More Florida questions... (0.00 / 0)
I assume then that you've included the absentee ballot totals from Florida? Those are ballots requested - not ballots returned (though I agree with counting them all, which is the most conservative assumption.)

Conduct your own interview of Sarah Palin!
Chris-- (0.00 / 0)
I would add that I am, in fact, very heartened by the Colorado numbers.  Colorado gets early voting started much later than other states--early voting in CO only started on the 20th, meaning that this party ID breakdown, which is essentially even, will factor in all of the absentee ballots in the state, plus only two days of early voting.  The absentee voters will, of course, tend to favor Republicans.  And even at that, we're dead even.

I think those numbers will reverse in a couple of days.

Also, early voting is off the charts in Nevada as well.  I'll get more details on that to you.

Hmmm (0.00 / 0)
So basically, identical to the situation in Florida then?

Conduct your own interview of Sarah Palin!
[ Parent ]
seems to me to be that way (0.00 / 0)
if, after only two days of early voting, we're just about even in Party ID in Colorado, I'm happy.  We have the better ground game, by all reports, and we'll start banking Denver, Boulder and Littleton.  It has already begun, in fact.
[ Parent ]
Counting FL Early Votes vs Absentees (0.00 / 0)
In order to handle the situation properly for assessing the vote counts, we really need to know how absentee ballots are handled, versus early in-person voting.

I early voted in FL just prior to the close of the polls, yesterday.  The count on the box was 1377, with only a half dozen folks left inside, and nobody in line waiting.  The site showing FL voting results indicated a total count of 1539 for my location after yesterday.  This is a difference of 162 votes.  There is no way that many more votes went in, unless absentee ballots received, and processed, that day were added to the ballot box.  

This implies that absentees are processed as they come in at this location, anyway.  I do not know whether handling absentee ballots is mandated at the state, or county, level.

It makes all the difference, though, whether absentees are held to the end of the process, or entered as they are received.  If entered as received, you'd need a model to reflect the return rate prior to election day, in order to have these calculations be any good.

My suggestion is only count the early votes, and leave absentees aside, otherwise you risk double counting the effects of the absentee ballots.

Thanks for the EV site reference, which is excellent.

statsguy  

Counting FL Early Votes vs Absentees (0.00 / 0)
In order to handle the situation properly for assessing the vote counts, we really need to know how absentee ballots are handled, versus early in-person voting.

I early voted in FL just prior to the close of the polls, yesterday.  The count on the box was 1377, with only a half dozen folks left inside, and nobody in line waiting.  The site showing FL voting results indicated a total count of 1539 for my location after yesterday.  This is a difference of 162 votes.  There is no way that many more votes went in, unless absentee ballots received, and processed, that day were added to the ballot box.  

This implies that absentees are processed as they come in at this location, anyway.  I do not know whether handling absentee ballots is mandated at the state, or county, level.

It makes all the difference, though, whether absentees are held to the end of the process, or entered as they are received.  If entered as received, you'd need a model to reflect the return rate prior to election day, in order to have these calculations be any good.

My suggestion is only count the early votes, and leave absentees aside, otherwise you risk double counting the effects of the absentee ballots.

Thanks for the EV site reference, which is excellent.

statsguy  

Numbers numbers (0.00 / 0)
There are a number of sources of error in your calculations (assigning votes to candidates based on how they voted in 2004, for example, or by race, or counting absentee ballots requested rather than recieved), but you still use three significant digits. This makes it look like you are pretty certain in your projections, instead of publishing wild-ass-guesses.  My guess is is that your error bars are something like +-5%, in which case you should at least round the decimal point. (if you think the error is any less than 5% then I will heavily discount your other electoral projections, 'cause that means you's crazy).
Early Voting Estimates In Seven States | 30 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