Are polls an accurate reflection of public opinion? - Tilted Forum Project Discussion Community

onetime2 · 11-24-2003, 08:37 AM

Almost never. Too many organizations doing polls that have no business doing them.

Creating a non-biased poll is part science and a lot of art. You have to create questions that are clearly understood by the vast majority of those that you poll, you then need to be consistent in the coding of the answers, and correctly calculate the results.

Now, to make it more complicated, you need to create a panel. A true random panel of the population you want to measure is the most valid but almost no one goes to the trouble of attempting one. The other more valid alternative would be to create a sample that is representative (in terms of demographics, opinions, etc) of the population as a whole. This method is not ideal but is better than what you typically get which is a narrowly focused self selected sample with a lot of biases.

Most polls contain misleading (or outright biased) questions, large gaps in both sample size and representation, and the assumptions that the publishers make based on the "results" are more often than not leaps of faith with no real statistical footing.

But other than that, they're great.

Elegant Holmes · 11-24-2003, 08:53 AM

kudos onetime2, you should be a pollster. I agree that the questions are often biased, and every in which I have participated has had at least one question requiring more than a clear yes or no/ agree or disagree answer. Abortion, for instance. It would be foolish to immediatley answer yes always or no always without considering the individual case. (in my opinion)

txlovely · 11-24-2003, 12:38 PM

Scientific polls are typically constructed with more forethought as they are generally part of a funded study. There will be questions that will seem biased to the person being interviewed, but they are strategically placed within the poll to ensure overall, a more accurate result. I have been part of putting together research polls and if conducted properly, they can be very accurate and very informative. Now the ones we see here on the TFP, television, in magazines, etc. are a different story. That is why you see that margin of error percentage - to account for poll placement and those who may have not answered truthfully or who may have felt pressured to give a certain answer. For instance, if you post a poll about beef on a site frequented by vegetarians, your outcome will be much different than had you posted that same poll on a website geared toward the hamburger buying public. Research - grants, studies and soforth - depends upon poll accuracy, so they can't all be crap.

CSflim · 11-24-2003, 02:01 PM

The current answer is no, which means you shouldn't take this result seriously...

"This statement is false"

onetime2 · 11-24-2003, 05:09 PM

Quote:

Originally posted by txlovely
Scientific polls are typically constructed with more forethought as they are generally part of a funded study. There will be questions that will seem biased to the person being interviewed, but they are strategically placed within the poll to ensure overall, a more accurate result. I have been part of putting together research polls and if conducted properly, they can be very accurate and very informative. Now the ones we see here on the TFP, television, in magazines, etc. are a different story. That is why you see that margin of error percentage - to account for poll placement and those who may have not answered truthfully or who may have felt pressured to give a certain answer. For instance, if you post a poll about beef on a site frequented by vegetarians, your outcome will be much different than had you posted that same poll on a website geared toward the hamburger buying public. Research - grants, studies and soforth - depends upon poll accuracy, so they can't all be crap.

Margin of error will be in every poll as it has to do with variance from the mean. Even "scientific" polls are typically crap. You can place some sanity check questions into them to say something about the consistency of answers but they are far from being exacting. I would say less than 10% of polls done today are created with enough rigor to be considered successful measures of a population. This is in part due to the ridiculous number of polls done to gauge the latest public opinion but also due to the time consuming nature of creating a quality survey.

txlovely · 11-24-2003, 07:46 PM

I don't disagree. I do think polls are necessary and can be useful if conducted with care and attention.

ForgottenKnight · 11-24-2003, 08:10 PM

No. No way. The opinions that matter on the subject are often the ones that the poll doesn't include in the group of people they poll.

absorbentishe · 11-24-2003, 08:16 PM

No way. The polls are set up (usually by the media) using language that entices the respondent to answer the way the ask it. I work in market research, and some of the things we do seem to skew the data. You can not ask an unbiased question, one way or the other. You can get close, but there are no unbiased questions in the way they are asked.

txlovely · 11-25-2003, 04:43 PM

You are all a cynical bunch, aren't you? Perhaps you should participate in more polls.

Giant Hamburger · 11-25-2003, 09:03 PM

Hail Citizens!

These poles accurately reflect

my opinion of the public.
-GH

Redlemon · 11-26-2003, 06:37 AM

Thank you, GH and CSFilm. I almost forgot this was posted in Nonsense.

Buzz · 11-26-2003, 10:52 AM

Poll's are worded so that they get the answer they want most of the time..

Elegant Holmes · 12-04-2003, 07:26 AM

And the results of the polls can also be worded so as to prove an invalid hypothesis.

View Poll Results: Are polls an accurate reflection of public opinion.
Yes.	4	13.33%
No.	26	86.67%
Voters: 30. You may not vote on this poll

11-24-2003, 08:13 AM	#1 (permalink)
jwoody Shackle Me Not Location: Newcastle - England.	Are polls an accurate reflection of public opinion? ?

11-24-2003, 08:37 AM	#2 (permalink)
onetime2 Junkie Location: NJ	Almost never. Too many organizations doing polls that have no business doing them. Creating a non-biased poll is part science and a lot of art. You have to create questions that are clearly understood by the vast majority of those that you poll, you then need to be consistent in the coding of the answers, and correctly calculate the results. Now, to make it more complicated, you need to create a panel. A true random panel of the population you want to measure is the most valid but almost no one goes to the trouble of attempting one. The other more valid alternative would be to create a sample that is representative (in terms of demographics, opinions, etc) of the population as a whole. This method is not ideal but is better than what you typically get which is a narrowly focused self selected sample with a lot of biases. Most polls contain misleading (or outright biased) questions, large gaps in both sample size and representation, and the assumptions that the publishers make based on the "results" are more often than not leaps of faith with no real statistical footing. But other than that, they're great. __________________ Strive to be more curious than ignorant.

11-24-2003, 12:38 PM	#4 (permalink)
txlovely Insane Location: ...We have a problem.	Scientific polls are typically constructed with more forethought as they are generally part of a funded study. There will be questions that will seem biased to the person being interviewed, but they are strategically placed within the poll to ensure overall, a more accurate result. I have been part of putting together research polls and if conducted properly, they can be very accurate and very informative. Now the ones we see here on the TFP, television, in magazines, etc. are a different story. That is why you see that margin of error percentage - to account for poll placement and those who may have not answered truthfully or who may have felt pressured to give a certain answer. For instance, if you post a poll about beef on a site frequented by vegetarians, your outcome will be much different than had you posted that same poll on a website geared toward the hamburger buying public. Research - grants, studies and soforth - depends upon poll accuracy, so they can't all be crap. __________________ Cruel words erode self-esteem like the ocean eats away the shore.

11-24-2003, 02:01 PM	#5 (permalink)
CSflim Sky Piercer Location: Ireland	The current answer is no, which means you shouldn't take this result seriously... "This statement is false" __________________

11-24-2003, 08:53 AM	#3 (permalink)
Elegant Holmes Crazy Location: "TX"	kudos onetime2, you should be a pollster. I agree that the questions are often biased, and every in which I have participated has had at least one question requiring more than a clear yes or no/ agree or disagree answer. Abortion, for instance. It would be foolish to immediatley answer yes always or no always without considering the individual case. (in my opinion)

11-24-2003, 07:46 PM	#7 (permalink)
txlovely Insane Location: ...We have a problem.	I don't disagree. I do think polls are necessary and can be useful if conducted with care and attention. __________________ Cruel words erode self-esteem like the ocean eats away the shore.

11-24-2003, 08:10 PM	#8 (permalink)
ForgottenKnight Jesus Freak Location: Following the light...	No. No way. The opinions that matter on the subject are often the ones that the poll doesn't include in the group of people they poll. __________________ "People say I'm strange, does that make me a stranger?"

11-24-2003, 08:16 PM	#9 (permalink)
absorbentishe Boy am I horny today Location: T O L E D O, Toledo!!	No way. The polls are set up (usually by the media) using language that entices the respondent to answer the way the ask it. I work in market research, and some of the things we do seem to skew the data. You can not ask an unbiased question, one way or the other. You can get close, but there are no unbiased questions in the way they are asked.

11-25-2003, 04:43 PM	#10 (permalink)
txlovely Insane Location: ...We have a problem.	You are all a cynical bunch, aren't you? Perhaps you should participate in more polls. __________________ Cruel words erode self-esteem like the ocean eats away the shore.

11-25-2003, 09:03 PM	#11 (permalink)
Giant Hamburger Gastrolithuanian Location: low-velocity Earth orbit	Hail Citizens! These poles accurately reflect my opinion of the public. -GH

11-26-2003, 06:37 AM	#12 (permalink)
Redlemon Devoted Donor Location: New England	Thank you, GH and CSFilm. I almost forgot this was posted in Nonsense. __________________ I can't read your signature. Sorry.

11-26-2003, 10:52 AM	#13 (permalink)
Buzz Addict Location: Earth	Poll's are worded so that they get the answer they want most of the time.. __________________ The Only BEER is a Cold BEER. And a Single Malt will get ya every time.

12-04-2003, 07:26 AM	#14 (permalink)
Elegant Holmes Crazy Location: "TX"	And the results of the polls can also be worded so as to prove an invalid hypothesis.