59% of Americans want Army out of Iraq completely by August 2008

[Erik Mesoy]

Also, the "if" in "if you take the data" does not appear to close with a "then". Lots of incomprehensibility.

 

[CartesianFart]

This is what you posted:

What i am saying is that,if you take the data of all the people who have been questioned (1500)and inductively concur that it is probable by the questions that was given that all answers of a probable number of people(300,000,000) as being a percentage of answers can be misleading not only the polsters but the public as being a general one based on percentage of possiblity of the rest being the same as the ones that was polled.

I don't understand the use of the highlighted "as".

Isolating 'as' and telling me that you don't understand my use of it is making me confused.

I don't understand what the highlighted "general one" refers to.

The pollers(polsters)or the people who do the polls and the public that reads it.Sorry about that.

I think this sentence lacks a point in the middle, or at least commas.

Kinda wrote that too fast and neglected of editing it.

I'm under the impression that a lot of verbs are not in the proper form (like singular when the subject is plural, and vice-versa) and it doesn't help comprehension.

What are you referring to?

I do not understand the expression "Percentage of possibility".

What i mean is that to a favorable combination of circumstances that is likely to occur.

And generally the logic seems hazy.

Not logic but using indefinite words and phrases that i erroneously led to people thinking that i was using them idiomatically. Plus some grammatical errors of not properly arranging what i want to say on that you can't use polls and then interpret them in a inductive manner since it leaves out the rest of the population.

It looks like you're saying that we should not assume answers given in a poll are representative of answers of the general population?

Yes.

 

[skadistic]

How can a a poll of 1500 be representative of 350 million? Did they poll the 350 million after the 1500 to see if the numbers matched up?

 

[Masquerouge]

How can a a poll of 1500 be representative of 350 million? Did they poll the 350 million after the 1500 to see if the numbers matched up?

You can mathematically prove that you do not have to.
I know, it is completely counter-intuitive, but it's been mathematically proven.
EDIT: just to make it clear, it does not mean that the answers of the sample will exactly match the answers of the population. It means that there is a (usually) 95% confidence that the answers will match up to (usually) 5 percentage points.

 

[Desmond Hawkins]

How can a a poll of 1500 be representative of 350 million? Did they poll the 350 million after the 1500 to see if the numbers matched up?

Why is it that polling and statistical theory has to be explained to someone in every thread involving a representative poll.....Jesus.....Freaking......Christ......Esq.

 

[skadistic]

Why is it that polling and statistical theory has to be explained to someone in every thread involving a representative poll.....Jesus.....Freaking......Christ......Esq.

Because some of use disn't waste our parents money going to univerisities.

So why not explain it so even an idiot like me can understand how statistical theory works and just how people can be so sure that 1500 can represent 350 million.

 

[CartesianFart]

Another critique is that some of the questions in the poll consists of optional answers such as:Excellent,Good,Only fair,Poor,Don’t know/Refused (VOL. DO NOT READ), A lot, Some,Not much,None at all and etc. which is vague since it does not question the knowledge of the individual who understand the questions.A dumb-ass can answer these questions arbitrarily.

 

[MobBoss]

You can mathematically prove that you do not have to.
I know, it is completely counter-intuitive, but it's been mathematically proven.
EDIT: just to make it clear, it does not mean that the answers of the sample will exactly match the answers of the population. It means that there is a (usually) 95% confidence that the answers will match up to (usually) 5 percentage points.

Again...thats assuming that the pollsters themselves do not have a political agenda. And in my opinion, in these days and times, no such creature exists.

 

[Masquerouge]

Okay Skad and CartesianFart, please, read this:
http://www.publicagenda.org/polling/polling_error.cfm

Another critique is that some of the questions in the poll consists of optional answers such as:Excellent,Good,Only fair,Poor,Don’t know/Refused (VOL. DO NOT READ), A lot, Some,Not much,None at all and etc. which is vague since it does not question the knowledge of the individual who understand the questions.A dumb-ass can answer these questions arbitrarily.

Again...thats assuming that the pollsters themselves do not have a political agenda. And in my opinion, in these days and times, no such creature exists.

I think we're only talking about how polling 1,500 people is enough to have an idea of what 300,000,000 people think. Of course this assume random selection of people polled, and non-biased questions.

 

[CartesianFart]

Another is that what strikes my eye is that did the questionaire that did the questioning to the person who is polled,ask the same question again in a different time to that very same person who was being subjected by the questionaire in the past to see the change of the opinion?

This is one of the misleading one as an example:

QUESTIONS 27 THROUGH 44 HELD FOR FUTURE RELEASE
Turning to the subject of Iraq …
Q.45 Do you think the U.S. made the right decision or the wrong decision in using military force against Iraq?
Right Wrong
decision decision DK/ Ref
March, 2007 43 49 8=100
February, 2007 40 54 6=100
Mid-January, 2007 40 51 9=100
Early January, 2007 40 53 7=100
December, 2006 42 51 7=100
Mid-November, 2006 41 51 8=100
Late October, 2006 43 47 10=100
Early October, 2006 45 47 8=100
Early September, 2006 49 43 8=100
August, 2006 45 46 9=100
July, 2006 44 50 6=100
June, 2006 49 44 7=100
April, 2006 47 46 7=100
March, 2006 45 49 6=100
February, 2006 51 44 5=100
January, 2006 45 47 8=100
December, 2005 47 48 5=100
Late October, 2005 48 45 7=100
Early October, 2005 44 50 6=100
Mid-September, 2005 49 44 7=100
July, 2005 49 44 7=100
June, 2005 47 45 8=100
February, 2005 47 47 6=100
January, 2005 51 44 5=100
December, 2004 49 44 7=100
November, 2004 (RVs) 48 41 11=100

 

[CartesianFart]

I think we're only talking about how polling 1,500 people is enough to have an idea of what 300,000,000 people think. Of course this assume random selection of people polled, and non-biased questions.

I am afraid that "is enough to have an idea" is simply too misleading to my saying "is not enough to have an overall idea" which i think is more practical and not lead one to blind faith.

 

[skadistic]

Having just made soup I'd like to point out if you taste test it and say a carrot wasn't in that spoon full your taste is off and so would your poll. Unless the pollies are a divers cross section of region/socialeconomic/religous/education level and a whole host of other variables that comes with 300 million individuals the poll will only be representative of those polled and people that think just like them. There is no way .0005% of the population can be indicitive of the mass. Some section or sections will not be factored in. Maybe I'm just not smart enough but it looks like there is to much probability alot will be left out, more then the miniscule margin of error.

Thats asuming the pollsters don't have an agenda with loaded questions and vauge options. And the pollies answered truthfuly.

 

[JollyRoger]

My guess is that if you took a poll right now, 28%-34% would say they didn't believe polls are accurate.

 

[bhsup]

Whether the poll numbers are correct or not is irrelevant. We, thankfully, do not govern by poll numbers. We govern by electing officials who will then act in (hopefully) the best interests of the nation, regardless of the fickle moods of the American public.

 

[MobBoss]

I think we're only talking about how polling 1,500 people is enough to have an idea of what 300,000,000 people think. Of course this assume random selection of people polled, and non-biased questions.

I have seen more than my fair share of biased questions and I have my doubts (very large doubts) about the 'random selection' of such polls.

For example. I am not sure if the pollsters break down their 'random selection' of people by demograph - ie only x number of whites, blacks, hispanics, asians to conform to real demograph of the USA. Or for that fact, are they taking 'random' people from blue states or red states, or from urban (more liberal) or rural (more conservative) areas?

While I do believe you wholeheartedly that the science behind it backs it up, I just dont trust the human element of such a thing. Our world has seen the power of propaganda and for what its worth, I think polling just another way of doing it while wearing a mask of legitimacy.

 

[Pontiuth Pilate]

Having just made soup I'd like to point out if you taste test it and say a carrot wasn't in that spoon full your taste is off and so would your poll.

Oh. My. God.

:dohsmiley: ointandlaughsmiley:

 

[JollyRoger]

Having just made soup I'd like to point out if you taste test it and say a carrot wasn't in that spoon full your taste is off and so would your poll.

Would you have to eat the whole bowl before you knew the soup had carrots?

 

[skadistic]

Would you have to eat the whole bowl before you knew the soup had carrots?

What does that have to do with Gitmo?

 

[Masquerouge]

I am afraid that "is enough to have an idea" is simply too misleading to my saying "is not enough to have an overall idea" which i think is more practical and not lead one to blind faith.

Having just made soup I'd like to point out if you taste test it and say a carrot wasn't in that spoon full your taste is off and so would your poll. Unless the pollies are a divers cross section of region/socialeconomic/religous/education level and a whole host of other variables that comes with 300 million individuals the poll will only be representative of those polled and people that think just like them. There is no way .0005% of the population can be indicitive of the mass. Some section or sections will not be factored in. Maybe I'm just not smart enough but it looks like there is to much probability alot will be left out, more then the miniscule margin of error.

Okay the basic idea is that if you poll 1,000 people, RANDOMLY chosen, you will have a 95% confidence that the answers will be within 3 points of the total population's answers.
Can you guys agree with that?

Now the thing is, polling, say 1,000,000 people, you will have a 95% confidence that the answers are going to be within, say, a half-point of the total population.
In other words, polling 1,000 times MORE people is not going to make your poll 1,000 more accurate. The returns are insanely degressive.
Thus 1,000 is a very good compromise between time, size and accuracy. 95% confidence within 3 points is nothing to sneeze at.

I'm afraid you will have to trust the maths here. If you took the appropriate classes this is something not very hard to prove and verify, but I completely agree that if advanced maths were not your thing, then it seems very dubious.

Thats asuming the pollsters don't have an agenda with loaded questions and vauge options. And the pollies answered truthfuly.

I have seen more than my fair share of biased questions and I have my doubts (very large doubts) about the 'random selection' of such polls.

For example. I am not sure if the pollsters break down their 'random selection' of people by demograph - ie only x number of whites, blacks, hispanics, asians to conform to real demograph of the USA. Or for that fact, are they taking 'random' people from blue states or red states, or from urban (more liberal) or rural (more conservative) areas?

While I do believe you wholeheartedly that the science behind it backs it up, I just dont trust the human element of such a thing. Our world has seen the power of propaganda and for what its worth, I think polling just another way of doing it while wearing a mask of legitimacy.

We're in complete agreement here. I'm talking about how the theory of polling vindicates 1,000-sized samples, not how biased polls can be due to human manipulations.

 

[MobBoss]

Whether the poll numbers are correct or not is irrelevant. We, thankfully, do not govern by poll numbers. We govern by electing officials who will then act in (hopefully) the best interests of the nation, regardless of the fickle moods of the American public.

Sigh. However, you left out the fact that many (dare I say a good majority?) of our 'elected' leaders do govern by poll numbers as it is seen as the method of keeping themselves in office. Ergo, in reality, we do often govern by poll numbers...