[Revolting cat!]

That’s a very good explanation.

My own is that people lie, people equivocate, people don’t commit, people say what they think others want to hear, people say whatever comes to their minds when no cost to themselves is involved.

And, I just don’t believe the very premise of polling (and the “science” of statistics), which states that a poll of a careful selection of, say, 1024 respondents can accurately reflect opinions of millions. And the 3% or 4% error margin that the pollsters cite is of course their own self serving number pulled out of a hat. (Lovers of statistics will call me an ignoramus, I expect.)

[muawiyah]

The confidence interval is essentially the same as the 'margin of error', but you can get some incredible arguments out of statisticians on why it's really not ~ even though it is. Some statisticians average out the confidence intervals in all the constituent parts of a multifactor poll. others use the largest interval as the margin of error. some of them just reach up and pull one out of the air ~ based on experience!

All a poll is telling you is that it's likely the case that if you continued to poll the target population you would likely continue to get the same old answer with little variation.

[UltraV]

And remember, only a tiny minority actually answer the poll calls...Pew said only 9% were responding...down from 38% (according to a UW poll down in 2008.)

And they actually believe those who have the wherewithal to answer are the same demographically as the voting population of the US.