Gallup: It's a Tie Ballgame

Gallup has Obama up 49-47 with a margin of error of 2%. Have you heard that anywhere today? Of course not.

The media neeeeeds this to be over right now. Anything that contradicts that template is utterly disregarded. Gallup took some heat for changing their party weighting to include more Democrats, which accounted for Obama's inflated numbers over the last couple of weeks. So they decided to do three polls, one of Registered Voters (RV) (useless as a predictor), Likely Voters (LV), and a new "Traditional" Likely Voters (TLV), which reflects the, well, traditional weighting of the parties. So here are today's Daily Tracking Poll numbers.

I don’t think you have it right.

I think that you're not understanding what I asked. I understand these are three current poll results.

In past years, Gallup used the traditional model which usually was shown to overstate Democrat support when compared to actual results. Given that premise, why did Gallup shift to a model which seems likely to even more overstate Democrat support? That was my question. It seems they are adjusting in the wrong direction.

I don’t think you understand me, either.

If so, we’re even.

I was merely pointing out that their most up-to-date change is to go back to the traditional weighting.

This election year to date they’ve been using an expanded likely voter weighting. If you’re asking why did they go to that one in the first place, when the traditional one is already overstated, that doesn’t track.

Because the traditional weighting is based on actual votes that are cast in actual historical elections. They aren’t hypothetical or projections.

So by definition the traditional one wasn’t overstated, if it’s based on actual voting patterns in past elections.

Now why did they go to the new model? Before they just now started computing and releasing numbers based on the old one?

Simply, because they have been basing their projections - and that’s what they are, projections - on increased Democrat voter registration (ACORN and the like) and a belief that black people and young people are going to come out in droves for Obama.

Basically they’ve decided to punt the issue and list 3 categories (the third one is registered voters only and no projection based on increased black vote and youth vote, but is subject to false weighting based on voter registration fraud, people registering Dem under operation chaos etc.)

Take your pick.

The most accurate poll in the last Presidential Election was Investor’s Business Daily. I believe they came within 4 tenths of one percent of the final popular vote margin.

So by definition the traditional one wasn’t overstated, if it’s based on actual voting patterns in past elections.

Correct, if we are talking about the mix of Republican/Democrat/Independent voters but I'm not.

The main objective of the poll is to predict the percentage of votes each candidate gets, not to predict the percentage of party affliation of the electorate. The second is important only to help achieve the main objective.

If prior election results show that the previous years' polls were overstating the vote of the Democrat candidate, why switch to a party affliation model which would exacerbate that problem? Since polling is as much art as science, wouldn't common sense tell Gallup that they are adjusting in the wrong direction with their new weighting system.

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.