Atlas Forum

Election Archive => 2008 U.S. Presidential General Election Polls => Topic started by: Alcon on October 14, 2008, 11:55:58 pm



Title: The EVT -- Early Voting Tracker
Post by: Alcon on October 14, 2008, 11:55:58 pm
Early Voting Tracker
Methodology:

1. Raw vote totals are taken from the SurveyUSA polls for every state with early voting.  They are separated into likely voters who haven't voted yet (A), and early voters (B).  I also give statistics on overall likely voters (A+B).

2. These results are then weighted based on 2004 turnout.  For instance, although a North Carolina poll might be the same size as a North Dakota poll, North Carolina casts many more votes.  My model reflects this.  Instead of a ~1:1 ratio (poll sample), North Carolina's polling sample will "overwhelm" North Dakota's ~11:1 (2004 turnout).

3. These weighted results are combined together to give us a lower-MoE picture of early voters in these states vs. likely future voters.  This can help determine how many future votes is necessary for McCain (or Obama) to reach parity.  Unlike individual polls, this is not subject to ridiculously high MoEs.

Caveats:  This is SurveyUSA and SurveyUSA only.  People lie.  MoE is calculated unweighted for the sake of my sanity, and thus is subject to some technical error.  This is not all 50 states or nationally representative.

Any questions, please feel free to ask.

Early voters (MoE = +/-5.2%) Obama +19
Obama 58%
McCain 39%
Other 2%
Undecided/Total idiots 0%

Likely future voters (MoE = +/-1.9%) McCain +0
McCain 47%
Obama 47%
Other 3%
Undecided 2%

All voters (MoE = +/-1.8%) Obama +2
Obama 48%
McCain 46%
Other 3%
Undecided 2%

2004 results (for comparison) Bush +8
Bush 54%
Kerry 46%
Other 1%

Included polls
GA - SurveyUSA, 10/12
IA - SurveyUSA, 10/13
NM - SurveyUSA, 10/13
NC - SurveyUSA, 10/6
OH - SurveyUSA, 10/13


Title: Re: The EVT -- Early Voting Tracker
Post by: Alcon on October 15, 2008, 01:29:21 am
Bump

Now weighted, hope I explained it enough now to make it at least somewhat understandable


Title: Re: The EVT -- Early Voting Tracker
Post by: Lunar on October 15, 2008, 01:38:06 am
Is SUSA the only one that breaks this down?


Title: Re: The EVT -- Early Voting Tracker
Post by: Alcon on October 15, 2008, 01:43:37 am
Is SUSA the only one that breaks this down?

Unfortunately, so far, as far as I know.

If any other poll gives this and there's raw numbers available, I'll be more than happy to include them.  If multiple pollsters do this, maybe I can even split samples instead of just using the newest


Title: Re: The EVT -- Early Voting Tracker
Post by: tokar on October 15, 2008, 08:19:38 am
Undecided/Total idiots 1%

Bwaahahahaha.

I know, seriously.  How can you have voted early and still be undecided?  Either you voted for one guy, the other guy, or some independent guy.


Title: Re: The EVT -- Early Voting Tracker
Post by: emailking on October 15, 2008, 08:33:12 am
Perhaps the last category is for people who voted early but chose not to vote for the Presidential race? "Other" perhaps could mean that they voted for some Presidential candidate other than Obama or McCain.


Title: Re: The EVT -- Early Voting Tracker
Post by: Gustaf on October 15, 2008, 12:45:03 pm
Maybe they forgot...

Anyway, just so i get this straight, this is based on only these 4 states (I assume because they are the only ones for which data is available) so it says Obama is up 6 in the "region" of Ohio, North Carolina, New Mexico and Iowa and that this is an average of a 25% lead among early voters and a 2% lead among likely future voters? And the Bush number I presume is the GE result from 2004? 


Title: Re: The EVT -- Early Voting Tracker
Post by: Alcon on October 15, 2008, 02:33:10 pm
Added a Georgia poll I missed

Maybe they forgot...

Anyway, just so i get this straight, this is based on only these 4 states (I assume because they are the only ones for which data is available) so it says Obama is up 6 in the "region" of Ohio, North Carolina, New Mexico and Iowa and that this is an average of a 25% lead among early voters and a 2% lead among likely future voters? And the Bush number I presume is the GE result from 2004? 

You got it