« Marist Poll: Statistical Tie | Main | Marist Poll: Statistical Tie »

CBS Poll: Bush Ahead by 8

George Bush leads John Kerry among nation-wide RV's 49-41 percent in a head-to-head CBS News Poll conducted Sept. 20-22.

Comments

I got to say that the CBS poll of registered voters makes zero sense given the NY Times article about the trend with registered voters at least in the battelground states. I mean seriously a 250 percent increase in Democrats to 25 percent increase in Republicans (which amounts to tens of thousands and maybe 100,000 new voters in Democrat heavy portions of these battleground states- remember most elections in these states with close margins will not be decided by more than a few ten thousand votes, and not several hundred thousand), and their numbers are at 8 percent more registered (not even likely voters) in favor of Bush. Does this sound off to anyone else? I know they say that polls undercount new registrants, but this seems out of wack with whats happening on the ground. Can someone explain the difference?

If there is one thing history has taught us, it is that modern technologic changes often catch old disciplines with their pants down.

Polling is seeing one such blind spot come to fruition.

The data appear to show a major disconnect between what is really happening versus what is being sampled. A poll is a sample, and it matters WHAT it is a sample of.

The faulty premise is that pollsters are constructing valid models and carrying them out flawlessly. They are not even getting close in many instances.

The practice of calling voters and talking to those who will talk to you is inherently unreliable. It can only be made more reliable by sound methodology, and it isn't there for Gallup and others, whose strategies are as outdated as the French post WWI planning.

This modern day polling Maginot line will fail as surely as did the original. Campaign Kerry is coming through. We have the numbers, and the registrations in key battleground states prove it.

The fact is we lost in 2000 because too many Dems didn't register, didn't vote, didn't care enough. That is not happening this time, and any way you slice, we have at least 5 million more voters than they do.

We will win by 5 million.

This CBS poll appears to use a weighting that includes MORE Dems than Reps (See the end of the file), so I'm confused by all the Lib spin that suggests this poll is biased in favor of GWB because it oversamples Reps.

Of course, if the poll does not toe the Lib line that the race is a dead heat, then the poll is to be discarded - Sheeeeesh!!

The increase in voter registration is great, but I fear it will be essentially defeated by partisan secretary's of state in these various battleground states like OH and FL. There are many new legal technicalities which the (ironically named) Help America Vote Act of 2002 ("HAVA") introduced into federal elections law. For example, there are numerous byzantine technicalities contained in HAVA concerning the proper form of identification to be used on a state voter registration form. I assure you that it is not the goal of Republican secretaries of state in states with large numbers of minority voters to increase the rolls of registered voters! I predict enormous election difficulties in these states on Nov 2, with thousands of new voters who thought they had registered finding out that their registration form was thrown out on a technicality, with no notice to them by the helpful secretary of state-----those are the kind of tactics that our friendly Republican Party will be using in this election, and all future ones.

>This CBS poll appears to use a weighting that includes MORE Dems than Reps (See the end of the file), so I'm confused by all the Lib spin that suggests this poll is biased in favor of GWB because it oversamples Reps.


>Of course, if the poll does not toe the Lib line that the race is a dead heat, then the poll is to be discarded - Sheeeeesh!!


Hmm, looks to me like, unweighted at least, it still favors the Republicans (by one). I'm still not versed enough in polling techniques to understand what they mean by "weighted" vs. "unweighted" -- are the "weighted" responses the Likely Voters? How do they determine which responses are discounted? In any case, if the unweighted is Registered Voters, they're still oversampling Republicans.