Originally posted by EladarAt least is rises to the level of "theory''. Creationism is not and never will be a theory. It is a fairy tale written down by men, conceived by men to make sense of a scary planet a few thousand years ago. AND it was plagiarized from Egyptian 7 day creation tales a few thousand years before the Jews got hold of it and said, sounds good to us. Must be true, it's written right here on this cartouche.
I like this one from the link:
[b]Fewer than half (48 percent) knew that human beings evolved from earlier species of animals.
Notice the statement of fact not that it is according to the Theory of Evolution.[/b]
6 edits
Originally posted by EladarHow exactly did you make this "observation" that "pollsters will attempt to take a sample which has the best chance of getting the desired results"?
It is my observation that pollsters will attempt to take a sample which has the best chance of getting the desired results.
Where you physically there looking over their shoulder to observe this while they conduct or analyzed opinion polls?
If not, did you read their minds at the distance?
If not that either, what exactly what form did this "observation" take?
Can you give any specific example of one of these "observation" for any specific example of a poll (please give web link ) and explain how that "observation" works?
And exactly HOW would the pollsters select samples for the OP poll so to "the best chance of getting the desired results"?
Did they go around and try and select those people that looked the most stupid such as the crossed-eyed and avoided anyone that looked intelligent so to get the result they "want" of a large percent of them not knowing that the Earth goes around the sun? If not, exactly how did they make the sample biased in this particular case? What was biased about the sample in the OP link?
And are you talking about ANY kind of opinion pole including ones to do with political opinions? or ones more to do with how much basic science knowledge people have? -because there wouldn't be the same credible motive for creating deliberate bias between the two.
6 edits
Originally posted by Soothfast"For a population of 300,000,000, if you want 95% confidence in being within 3 percentage points of the true percentage of the population that does not know..."
Not actually. For a population of 300,000,000, if you want 95% confidence in being within 3 percentage points of the true percentage of the population that does not know the Earth revolves around the Sun (a true-or-false question), you will need a sample size of 1067. Of course this is assuming your sample is truly random (and presumably there is a lower ...[text shortened]... y have a margin of error of plus/minus 3 points when the sample size is somewhere north of 1000.
That's not how I learned statistics in my senior level university course. The confidence interval encompasses the range of results that can be expected in 95% of the cases, if the sampling is repeated numerous times. If the sampling plan has a bias (such as might be introduced by calling only on land-lines, which I don't know could introduce a bias) then the interval might not overlap the result that would come from surveying 100% of the population.
So your assumption of randomness in the design extends beyond randomly choosing from the population you are asking the question (in this example, all people who can be reached by land line); it extends to whether that population represents the population of all Americans.
1 edit
Originally posted by JS357
"For a population of 300,000,000, if you want 95% confidence in being within 3 percentage points of the true percentage of the population that does not know..."
That's not how I learned statistics in my senior level university course. The confidence interval encompasses the range of results that can be expected in 95% of the cases, if the sampling is repeat ...[text shortened]... the interval might not overlap the result that would come from surveying 100% of the population
The confidence interval encompassed the range of results that can be expected in 95% of the cases, if the sampling is repeated numerous times.
The sampling was “repeated numerous times” in this case; specifically, 2,200 times. Thus the claim that we can have 95% confidence appears to stand. To back up the claim that we don't have 95% confidence, you have to show evidence of bias -which would be very difficult in this case because, even if there is bias in this case, there is not much detail about the poll given in the OP link.
1 edit
Originally posted by humyActually in polling, the burden is on the person reporting the results, and it is important to note that not all biases are conscious or intentional.The confidence interval encompassed the range of results that can be expected in 95% of the cases, if the sampling is repeated numerous times.
The sampling was “repeated numerous times” in this case; specifically, 2,200 times. Thus the claim that we can have 95% confidence appears to stand. To back up the claim that we don't have 9 ...[text shortened]... ven if there is bias in this case, there is not much detail about the poll given in the OP link.
Originally posted by JS357
Actually in polling, the burden is on the person reporting the results, and it is important to note that not all biases are conscious or intentional.
the burden is on the person reporting the results
The person reporting the results may not necessarily be responsible for a badly designed or badly run survey and may not necessarily be aware of the bad features of the design/running of it -it depends on how different tasks were allocated to different people.
Originally posted by humyThere have been multiple reports of political polls that were later found out to be rigged by intentionally asking certain people and ignoring others. If you want to see huge support numbers for Obama, do your poll and only ask poor blacks.
How exactly did you make this "[b]observation" that "pollsters will attempt to take a sample which has the best chance of getting the desired results"?
Where you physically there looking over their shoulder to observe this while they conduct or analyzed opinion polls?
If not, did you read their minds at the distance?
If not that either, what exactly wha ...[text shortened]... because there wouldn't be the same credible motive for creating deliberate bias between the two.[/b]
Originally posted by sonhouseWow, you can't even admit that my point is valid. Pathetic.
At least is rises to the level of "theory''. Creationism is not and never will be a theory. It is a fairy tale written down by men, conceived by men to make sense of a scary planet a few thousand years ago. AND it was plagiarized from Egyptian 7 day creation tales a few thousand years before the Jews got hold of it and said, sounds good to us. Must be true, it's written right here on this cartouche.
8 edits
Originally posted by Eladar
There have been multiple reports of political polls that were later found out to be rigged by intentionally asking certain people and ignoring others. If you want to see huge support numbers for Obama, do your poll and only ask poor blacks.
There have been multiple reports of political polls that were later found ….
I wasn't talking specifically about “political polls”. The OP poll wasn't political, at least not directly, because it didn't ask questions like who you would vote for and why etc.
Although the poll did ask a couple of questions about what you think government policy on science should be, those questions were clearly not left wing or right wing questions in particular . And how is asking, for example, does the Earth go around the sun, “political” in particular? If THAT is “political”, then I guess the IQ test must also be “political”!
And, if you claim they selected in the OP pole people that would be more likely give the answer they 'want', exactly WHICH sort of people were selected for that? Black people? white people? poor people? rich people? I mean, can you give ANT specific example of a group or class of people that you would say this particular poll could have plausibly deliberately selected to give the answer they 'want' which I assume you would claim includes people that wouldn't know that the Earth goes around the sun? Are, for example, black people less likely to know the Earth goes around the sun than white people?
I completely fail to see exactly which group or kind of people they could have plausibly and deliberately selected to give the answer they 'wanted'.
But, even for what you said about political poles, -
There have been multiple reports of political polls that were later found out to be rigged by intentionally asking certain people and ignoring others. If you want to see huge support numbers for Obama, do your poll and only ask poor blacks
So can you give A specific example, ANY ONE example would do, of a pole that was done and then claimed that it shows that most voting US citizens, including both black and white ones, support the president (either a past one or the present one) but neglected to mention that only poor blacks were selected for that pole?
Originally posted by humyYou really aren't too bright are you? You asked for experience where polls have been biased.There have been multiple reports of political polls that were later found ….
I wasn't talking specifically about “political polls”. The OP poll wasn't political, at least not directly, because it didn't ask questions like who you would vote for and why etc.
Although the poll did ask a couple of questions about what you think gove ...[text shortened]... or the present one) but neglected to mention that only poor blacks were selected for that pole?
Originally posted by humyNonetheless, repetitions of a biased survey using a random sampling plan will report a confidence interval centered on a biased mean. The confidence interval on that mean may or may not encompass the result that would have been obtained from an unbiased 100% survey.the burden is on the person reporting the results
The person reporting the results may not necessarily be responsible for a badly designed or badly run survey and may not necessarily be aware of the bad features of the design/running of it -it depends on how different tasks were allocated to different people.
I'm not arguing against the findings of the study. Just making a point on statistics. Basically it boils down to the possibility that the population we are sampling might not be the population we think we are sampling. "All Americans over 17" might actually be "All Americans who have a land line, and say they are over 17, and have time to and/or like to answer surveys, and who understand the question." If that is who you get, the confidence interval will encompass THAT population's true mean.
Originally posted by joe shmoYou can get an accurate result from a small sampling. That's why we can project the winners of elections early.
26% of the 2,200 people...seems like a small sample to make a claim that 26% of the entire population would get it wrong.
Also, where was the sample taken?
I imagine the truth would be disappointing about how many actually do not know that information, but 26%...no way.
13 edits
Originally posted by JS357
Nonetheless, repetitions of a biased survey using a random sampling plan will report a confidence interval centered on a biased mean. The confidence interval on that mean may or may not encompass the result that would have been obtained from an unbiased 100% survey.
I'm not arguing against the findings of the study. Just making a point on statistics. Basica ...[text shortened]... on." If that is who you get, the confidence interval will encompass THAT population's true mean.
will report a confidence interval centered on a biased mean.
What is a “biased mean” as opposed to some other kind of mathematical “mean”? In mathematics, although you can have a “biased sample”, or even the "mean of a biased sample", there is no such thing as “biased mean”; only the “mean”. Thus, they cannot “report a confidence interval centered on a biased mean” because there is no such thing to report.
(I have studied mathematics at university level and am well disciplined in the field so you must excuse me for being rather pedantic but little trivial things like that just bother me! Please forgive. )