If a pollster is GENUINELY trying to gauge 'public opinion', said pollster must generate a polling sample REPRESENTATIVE OF THE US PUBLIC. Accordingly, said pollster MUST concern themselves with issues of gender, geography, party affiliation, and so on.
FYI: Not only are today's pollsters using UNREPRESENTATIVE samples, they are also producing results based on ridiculously low response rates (ranging from 5% - 25%) -- at it's most relevant, Gallup was getting response rates above 60%!
For these pollsters to hide behind the facade of 'random sampling' is beyond bogus! Why do you think they were SHOCKED by the election outcomes in 2002 and 2004 (and many special elections in between)? Why do you think they were SHOCKED by the inaccuracy of the exit polls in 2004 (and in MANY state/local races in 2002 and 2004)? Why indeed!
BOTTOMLINE: IT'S THE SAMPLE STUPID!
BTW: With 4 RED STATES essentially 'off line' (AL, LA, MS, and TX) where do you think these pollsters are getting their 'replacement' respondents?! Hmmmmmmmmmmmmm?!
Should read "its" rather than "it's" -- I obviously can't watch a football game and type at the same time!
[NOTE: The plural themselves is now accepted usage.]
The entire states are off-line? What percentage of the population do you think you're talking about here?
Folks,
This went on last year. Come on, we all remember the obsession with polls during the campaign and how we trumpeted those that showed Bush slippage as having an over sampling of Dems. Well, the DU people did the opposite, and the opposite did occur on occasion.
Polling methods have become dated and the polling companies are trying to address it. They are not doing a good job. Rasmussen has pointed out that most money that pollsters make does not come from politics. It is from market research. They have a financial interest in accurate surveys. They aren't doing this on purpose.
Rasmussen is ahead of them. He is sampling with an insistence that his result conform to a pre-specified partisan mix. There is debate about this because it does indeed render his sample non random.
The poll showing Bush slippage is because of who was sampled. The polls showing anything always show that because of who was sampled. Odds are if the poll in question was meaningful Rasmussen would have detected it.
Odds are, however, that high gas prices do have an effect on that few % in the middle that don't care about politics at all and just see their own pocketbook. Conversely, given that most people in the US don't have illegal immigration on their top 5 list of concerns, it is unlikely Bush's immigration stand is relevant.
That's a good point I hadn't thought of before. :-)