Leaving the community

"Newps" wrote in message
...
Polls are facts about statistics.

A poll isn't a fact about anything except the people who participated.

The poll itself is a fact about the statistical sample taken. Which is
exactly what I said (though apparently not in a verbose enough way for some
of you).

If you feel you have some good reason to dispite the Gallup poll results,
I'm all ears. If all you can come up with is "well, there's a 0.000000001%
chance that the poll is incorrect", then while that may be perfectly true,
it's a pretty useless statement.

The FACT remains that there's a much larger chance that the poll correctly
describes the overall electorate than that it doesn't.

Pete

In article , "Peter Duniho"
wrote:

The FACT remains that there's a much larger chance that the poll
correctly
describes the overall electorate than that it doesn't.

you have much more faith in polls than I do.

--
Bob Noel
Seen on Kerry's campaign airplane: "the real deal"
oh yeah baby.

"Peter Duniho" wrote in message
...
"Newps" wrote in message
...
Polls are facts about statistics.

A poll isn't a fact about anything except the people who participated.

The poll itself is a fact about the statistical sample taken. Which is
exactly what I said (though apparently not in a verbose enough way for
some
of you).

If you feel you have some good reason to dispite the Gallup poll results,
I'm all ears. If all you can come up with is "well, there's a
0.000000001%
chance that the poll is incorrect", then while that may be perfectly true,
it's a pretty useless statement.

The FACT remains that there's a much larger chance that the poll correctly
describes the overall electorate than that it doesn't.

You would be hard pressed to prove that. Polls are at best one step above a
WAG.

Pete

Dave Stadt wrote:



You would be hard pressed to prove that. Polls are at best one step above a
WAG.

Science proves it. But, everything has to go right for the poll to
achieve that margin of error. First you must get a represenative random
sample. This rarely happens, there's always a little error here.
Second the questions must not be skewed one way or the other. Third,
the people must tell the truth. This also never happens. They always
give the margin of error when you see a poll, this is a theoretical
number that cannot be reached because no poll will ever be truly random,
somebody always lies, or says they're someone their not, etc. One of
the pollsters on TV this week said that to get the 850+ responses for a
+-3% poll they had to call over 10,000 people. With those kinds of
problems no way can a poll be anymore than a guess.

On Fri, 05 Nov 2004 08:38:26 -0700, Newps wrote:



Dave Stadt wrote:



You would be hard pressed to prove that. Polls are at best one step above a
WAG.

Science proves it. But, everything has to go right for the poll to
achieve that margin of error. First you must get a represenative random
sample. This rarely happens, there's always a little error here.
Second the questions must not be skewed one way or the other. Third,
the people must tell the truth. This also never happens. They always
give the margin of error when you see a poll, this is a theoretical
number that cannot be reached because no poll will ever be truly random,
somebody always lies, or says they're someone their not, etc. One of
the pollsters on TV this week said that to get the 850+ responses for a
+-3% poll they had to call over 10,000 people. With those kinds of
problems no way can a poll be anymore than a guess.

This is really a hoot. We wouldn't be talking about this at all if
the exit polls hadn't been so wrong. There's the proof. As to why
this happened, my theory is that there is a systematic bias error
because people who voted for Bush had better things to do with their
time than talk to the pollster. Same thing with the phone polls,
9,150 people were too busy to talk to the pollster and there is a
bias that affects the results in that.

Klein

Peter Duniho wrote:


The FACT remains that there's a much larger chance that the poll correctly
describes the overall electorate than that it doesn't.

This statement is correct. There is a chance the poll represents the
actual fact. Depending on how accurate you want to be you can also say
the poll never correctly describes the actual fact. The poll will
always get you close, how close depends on the sample size. The same
science that tells you how close also tells you it will never be exactly
right.

"Peter Duniho" writes:
"Newps" wrote:

Polls are facts about statistics.

A poll isn't a fact about anything except the people who participated.

The poll itself is a fact about the statistical sample taken. Which is
exactly what I said (though apparently not in a verbose enough way for some
of you).

If you feel you have some good reason to dispite the Gallup poll results,
I'm all ears. If all you can come up with is "well, there's a 0.000000001%
chance that the poll is incorrect", then while that may be perfectly true,
it's a pretty useless statement.

The FACT remains that there's a much larger chance that the poll correctly
describes the overall electorate than that it doesn't.

My favorite statistics story: I was reading an article about
weather prediction in which NOAA claimed about 75% accuracy
in their predictions. You can say that tomorrow's weather
will be the same as today's and be about 90% accurate in most
parts of the world.