Dallas-Mavs.com Forums - View Single Post

jacktruth · 10-18-2004, 03:51 PM

Some poll data over teh last few days:

B45 K45 U8 Zogby
B47 K44 U9 TIPP
B52 K44 U3 Gallup
B47 K50 U2 Democracy Corps
B50 K46 U2 ABC/Wash Post
B50 K44 U5 Newsweek

I think it is interesting that the range across polls is: 45-52 for Bush and 44-50 for Kerry. That is 6 and 7 points difference, which is well outside the MOE claimed by any poll, so they are somewhat discredited anyway.

Also interesting, the undecided vote in these polls ranges from 2-8. This undecided number is fairly consistent within each poll (Gallup show consistenly 1-3 undecideds;Zogby shows consistently 7-8). This makes me wonder what constant is being used for each poll. The chances of an undecided vote must be more in Ohio than Texas, for example. And if there is a possible 8 pt swing based on undecided vote, there is considerably more error.

Can you safely project the same proportion onto the undecided vote? I don't know.

What scares me about this election is that it is going to be decided by people that can’t make up their minds. Think about it, what could an undecided voter possibly be still waiting for? If you don’t have a preference by now, you very possibly could be a flake. Are there any undecided voters on this board?

Also, there is a mixture of registered and likely voters accross the polls. It seems likely that likey voters is more accurate, since historically 46% of registered voters vote. Of course, there is no way to tell how many likely voters will vote.

In summary, if you really think about polling accuracy, your head will begin to spin.

10-18-2004, 03:51 PM	#130
jacktruth Golden Member Join Date: Jul 2001 Location: TX Posts: 1,868	RE:Official Political Poll Thread Some poll data over teh last few days: B45 K45 U8 Zogby B47 K44 U9 TIPP B52 K44 U3 Gallup B47 K50 U2 Democracy Corps B50 K46 U2 ABC/Wash Post B50 K44 U5 Newsweek I think it is interesting that the range across polls is: 45-52 for Bush and 44-50 for Kerry. That is 6 and 7 points difference, which is well outside the MOE claimed by any poll, so they are somewhat discredited anyway. Also interesting, the undecided vote in these polls ranges from 2-8. This undecided number is fairly consistent within each poll (Gallup show consistenly 1-3 undecideds;Zogby shows consistently 7-8). This makes me wonder what constant is being used for each poll. The chances of an undecided vote must be more in Ohio than Texas, for example. And if there is a possible 8 pt swing based on undecided vote, there is considerably more error. Can you safely project the same proportion onto the undecided vote? I don't know. What scares me about this election is that it is going to be decided by people that can’t make up their minds. Think about it, what could an undecided voter possibly be still waiting for? If you don’t have a preference by now, you very possibly could be a flake. Are there any undecided voters on this board? Also, there is a mixture of registered and likely voters accross the polls. It seems likely that likey voters is more accurate, since historically 46% of registered voters vote. Of course, there is no way to tell how many likely voters will vote. In summary, if you really think about polling accuracy, your head will begin to spin.