The New Bailout

Classicman...with all due respect, your post suggest that you dont understand what is meant by the "data is weighted" after the sampling.

That is how pollsters account for the discrepancies you mentioned. If more Rs responded to a poll then Ds (there are more registered Ds in the country than Rs) or more from the NE than the SW, they would "weight" the results to more accurately reflect the national profile. I dont claim to be an expert, but I do remember that from Polling 101.

A 95% confidence level within maximum margin of sampling error is ±3 percentage points is pretty damn high by any statistical standards.

And no, I like polls because they tell me something about the what the American public is thinking about a political issue of the day.

Lately, they have reinforced my own opinions. For much of the early 00s, that was not the case, but I found them to be equally meaningful.

But, hey, if you dont think they have much value...thats fine with me.

[classicman]

Originally Posted by Redux

Classicman...with all due respect, your post suggest that you dont understand what is meant by the "data is weighted" after the sampling.

What I am saying is that neither do you or anyone else - They weight the polls to even out what is not even. That is my point.

Originally Posted by Redux

I dont claim to be an expert, but I do remember that from Polling 101.

I didn't take polling 101 - I don't recall it being offered where I got my degrees. Sounds like an interesting course though.
What I do remember about polling from college is how extremely difficult it is to get accurate information and how skewed the data therefore can be. They attempt to weigh/modify/alter the data to make it credible and make a prediction. Sometimes they are right and sometimes they are wrong.

Originally Posted by classicman

...
What I do remember about polling from college is how extremely difficult it is to get accurate information and how skewed the data therefore can be. They attempt to weigh/modify/alter the data to make it credible and make a prediction. Sometimes they are right and sometimes they are wrong.

Yep..I agree it is hard for you and me (with my one course 20+ years ago) or even self-proclaimed experts like Merc.

Its not that hard for those with the proper educational training and the knowledge and experience of applying widely accepted statistical procedures and anti-bias protocols.

And that's why they can report with a 95% confidence level within maximum margin of sampling error of ±3 percentage points.....far greater than "sometimes right and sometimes wrong."

I dont need 100% confidence with zero margin of error to find polls useful to understand public opinion on an issue.

In slightly different polling, one only need look at the 08 election results and final pre-election polls (by state and nationally) to see how close the pollsters were to the final results. Aggregating the major national polls predicted Obama -52.0%, McCain - 44.4 and the final results Obama- 52.9%, McCain -45.6

[TheMercenary]

Originally Posted by Redux

Yep..I agree it is hard for you and me (with my one course 20+ years ago) or even self-proclaimed experts like Merc.

Where did I make that statement? You say that because we disagree?