Originally posted by BRussellBut they didn't address the primary allegation. The specific charge in that moveon ad (pdf file) is that something Gallup does in their calculation of likely voters biases the sample toward Republicans. It's not that they intentionally ask for more Republicans. They didn't address that issue at all.
Gallup's current poll shows Kerry getting more support among likely voters than it does among registered voters.
But it really seems like a straw man to me. How is being Republican or Democrat going to make you more likely to vote? I would think that the criteria that determines whether you are a likely voter would be completely independent of party. I would think that it would something along the lines of how many previous elections have you voted in, have you applied for an absentee ballot, have you moved and re-registered, or something of that nature.
Here is the Moveon contention...
Simply put, Gallups methodology
has predicted lately that Republican
turnout on Election Day is likely to
exceed Democrats by six to eight
percentage points. But exit polls show
otherwise: in each of the last two
Presidential elections, Democratic
turnout exceeded Republican by four to
five points. That discrepancy alone can
account for nearly all of Bushs phantom 14-point lead.
I think this goes hand in hand with more people indentifying themselves as Republican after already being asked who they are going to vote for in the election. I hate being a parse master but notice the charge is that Gallup has only being doing this "lately." This is because when you look at the Gallup chart through the entire year, there are times when the poll had up to 5% more people identifying themselves as Democrats. This was especially true through the Democratic primary season when the Democrats and their message were getting lots of coverage.
They addressed it by saying they don't use a party sample to alter their poll sample because it will double the margin of error. If you have a poll that tells you the country is X% Republican and Y% Democratic with a 3-4% margin of error and you use that to change the results of another poll with a 3-4% margin of error, you've just added huge amounts of error into your polling data.
Now I'm not arguing that Gallup is the best, most accurate poll or anything like that. But they have clearly explained their methods and they seem to be above politics. If they prove to be inaccurate, they will have to change them in order to continue to receive the types of customers they do receive. But until they are proven wrong, they get the benefit of the doubt, especially over Moveon.org who seems to be willing to do anything and question or smear anyone to get the results they desire.
Here's a nice link of some historical poll results.Polls
Note that in the last election, Rasmussen was off by over 9%. I know people cite that one here alot.