cBS poll the left is trying to tout

[shira]

Originally posted by: ShadesOfGrey
When cBS weighted the poll for their typical stuff, they didn't consider party affiliation. What that did was weigh out more Republicans because of their other demographic stats.

Something you don't seem to understand is that in any random poll the profile of the polled group will inevitably NOT match the profile of the general population.

For example, if CBS had weighted responders so that 35% were Republicans, they might have ended up with much too high a percentage of evangelicals, whites, the high-school educated, and/or those living in the South (as compared with the general population), and that might have led to results that are less accurate than those actually obtained. ANY deviation in the polled group from the population they're intended to represent skews results.

You seem to be implying that the deviation of the percentage of Republications in the actual, weighted poll group from that in the general population is due to some insidious CBS agenda intended to skew results in some anti-Bush way. But unless you have a deep understanding as to how the weighting process was in fact performed and what its assumptions were, I don't see how you can come to any valid conclusions about motives.

Breakdown by party affiliation is just one demographic out of many, all of which correlate with political beliefs. And the actual weighting used in this CBS poll might well be the very best structuring of the polled group to match the general population.

 

[ShadesOfGrey]

Originally posted by: Todd33
Please oh wise kid of the Internet, ShadesOfGrey, explain how they are skewed. Then when you are done, explain why they published their numbers for public scrutiny...

I already have shown how they are skewed. Try reading the thread.

They didn't publish their numbers in their story.

 

[Darkhawk28]

Originally posted by: ShadesOfGrey

Originally posted by: Todd33
Please oh wise kid of the Internet, ShadesOfGrey, explain how they are skewed. Then when you are done, explain why they published their numbers for public scrutiny...

I already have shown how they are skewed. Try reading the thread.

They didn't publish their numbers in their story.

Gallup does the same thing, but in the other direction.

 

[ShadesOfGrey]

Originally posted by: shira

Originally posted by: ShadesOfGrey
When cBS weighted the poll for their typical stuff, they didn't consider party affiliation. What that did was weigh out more Republicans because of their other demographic stats.

Something you don't seem to understand is that in any random poll the profile of the polled group will inevitably NOT match the profile of the general population.

For example, if CBS had weighted responders so that 35% were Republicans, they might have ended up with much too high a percentage of evangelicals, whites, the high-school educated, and/or those living in the South (as compared with the general population), and that might have led to results that are less accurate than those actually obtained. ANY deviation in the polled group from the population they're intended to represent skews results.

You seem to be implying that the deviation of the percentage of Republications in the actual, weighted poll group from that in the general population is due to some insidious CBS agenda intended to skew results in some anti-Bush way. But unless you have a deep understanding as to how the weighting process was in fact performed and what its assumptions were, I don't see how you can come to any valid conclusions about motives.

Breakdown by party affiliation is just one demographic out of many, all of which correlate with political beliefs. And the actual weighting used in this CBS poll might well be the very best structuring of the polled group to match the general population.

Exactly, you say they could have oversampled "evangelicals" - or they might have undersampled them. It wasn't part of their weighing scheme. However they do use age, sex, race and others -but when their poll is specifically about POLITICAL opinions, don't you think it might just make sense to weigh for political affiliation just like if they were polling on religion - they should make sure their sample accurately reflects the overal breakdown of religious affiliation in the country?

The only thing I'm contending is that cBS has used a flawed demographic to get their story. They know their sample is flawed, or would if they even looked at the raw numbers so for them to write the story anyway without correcting for the flaw - it's intent is obvious- -unless of course you believe that cBS too incompetent to know better.

No, the actual weighing used could not be the best since it ignored a clear and very important demographic stat - political affiliation.

 

[ShadesOfGrey]

Originally posted by: Darkhawk28

Originally posted by: ShadesOfGrey

Originally posted by: Todd33
Please oh wise kid of the Internet, ShadesOfGrey, explain how they are skewed. Then when you are done, explain why they published their numbers for public scrutiny...

I already have shown how they are skewed. Try reading the thread.

They didn't publish their numbers in their story.

Gallup does the same thing, but in the other direction.

Yes, Gallup does the same as cBS. They don't weigh party affiliation and sometimes oversamples or undersamples party affiliation.

 

[homercles337]

Originally posted by: ShadesOfGrey
BTW, this thread isn't about the results of the poll or what the poll contained as far as it's "findings" - it's about their badly skewed sample.

Your ignorance of statistics hurts my brain. Not all samples are equal. If you think you have a biased sample (or if you KNOW you have a biased sample) you CORRECT for it--as i have said before CBS must have done. If they did not do this then their statisticians are complete bafoons. Or mayebe they applied some nonparametric stats. Others have provided evidence that they are aware of their biased sampling thus the CORRECTION. I suspect you have NO statistical training/knowledge what so ever.

 

[Todd33]

Originally posted by: ShadesOfGrey

Originally posted by: Todd33
Please oh wise kid of the Internet, ShadesOfGrey, explain how they are skewed. Then when you are done, explain why they published their numbers for public scrutiny...

I already have shown how they are skewed. Try reading the thread.

They didn't publish their numbers in their story.

A. You showed nothing, you just quoted their number from the PDF. There is no info on how they used them.

B. They did publish the numbers, you pulled them from their PDF.

Again, explain and prove your claim...

Take a look at the end of the pdf. Not only was their sample off from the start, they weighted the poll results to skew them even more!

Their sample was off? What did they skew and how was it wrong? We want proof, not your finger pointing at their own numbers and you screaming bias. Show us how the 35% approval rating was calculated using incorrect numbers, please.

I know conservatives are not that fond of science, but we can't take your claims on faith, much the way you take the claim of the blog you got this from (PS who was it? I doubt you dug through the PDF and made this bold and incorrect claim all by your lonesome).

 

[ShadesOfGrey]

Originally posted by: homercles337

Originally posted by: ShadesOfGrey
BTW, this thread isn't about the results of the poll or what the poll contained as far as it's "findings" - it's about their badly skewed sample.

Your ignorance of statistics hurts my brain. Not all samples are equal. If you think you have a biased sample (or if you KNOW you have a biased sample) you CORRECT for it--as i have said before CBS must have done. If they did not do this then their statisticians are complete bafoons. Or mayebe they applied some nonparametric stats. Others have provided evidence that they are aware of their biased sampling thus the CORRECTION. I suspect you have NO statistical training/knowledge what so ever.

Ah yes, my ignorance. :roll: The problem with that is, it's not true. You see, cBS DOESN'T WEIGHT THEIR POLLS ON POLITICAL AFFILIATION! - thus there was no correction. That's what this whole thing is about. Their demographics skewed the political affiliation results because they didn't use them in their weighing.
So I guess you consider cBS a bunch of "bafoons". What does that mean for their reporting if they use "bafoons" for statisticians?

 

[BBond]

Thirty-five percent approval rating.

Looks like America found out what I always knew was the truth.

Bush sux.

 

[ShadesOfGrey]

Originally posted by: Todd33

Originally posted by: ShadesOfGrey

Originally posted by: Todd33
Please oh wise kid of the Internet, ShadesOfGrey, explain how they are skewed. Then when you are done, explain why they published their numbers for public scrutiny...

I already have shown how they are skewed. Try reading the thread.

They didn't publish their numbers in their story.

A. You showed nothing, you just quoted their number from the PDF. There is no info on how they used them.

B. They did publish the numbers, you pulled them from their PDF.

Again, explain and prove your claim...

Take a look at the end of the pdf. Not only was their sample off from the start, they weighted the poll results to skew them even more!

Their sample was off? What did they skew and how was it wrong? We want proof, not your finger pointing at their own numbers and you screaming bias. Show us how the 35% approval rating was calculated using incorrect numbers, please.

I know conservatives are not that fond of science, but we can't take your claims on faith, much the way you take the claim of the blog you got this from (PS who was it? I doubt you dug through the PDF and made this bold and incorrect claim all by your lonesome).

Yes, I did show it was skewed. Try reading the thread.

The pdf is from RCP. the cBS article does not provide the party affiliation numbers, nor does it link to them.


I've already explained why their sample is bad. Try reading the thread instead of whining about blogs.
My bold and ACCURATE claim is that cBS used a bad sample which is evidenced by the raw numbers. The numbers are right there in black and white which have nothing to do with faith.
BTW, if you had half a clue, you'd follow the pdf

 

[shira]

Originally posted by: ShadesOfGrey

Originally posted by: shira

Originally posted by: ShadesOfGrey
When cBS weighted the poll for their typical stuff, they didn't consider party affiliation. What that did was weigh out more Republicans because of their other demographic stats.

Something you don't seem to understand is that in any random poll the profile of the polled group will inevitably NOT match the profile of the general population.

For example, if CBS had weighted responders so that 35% were Republicans, they might have ended up with much too high a percentage of evangelicals, whites, the high-school educated, and/or those living in the South (as compared with the general population), and that might have led to results that are less accurate than those actually obtained. ANY deviation in the polled group from the population they're intended to represent skews results.

You seem to be implying that the deviation of the percentage of Republications in the actual, weighted poll group from that in the general population is due to some insidious CBS agenda intended to skew results in some anti-Bush way. But unless you have a deep understanding as to how the weighting process was in fact performed and what its assumptions were, I don't see how you can come to any valid conclusions about motives.

Breakdown by party affiliation is just one demographic out of many, all of which correlate with political beliefs. And the actual weighting used in this CBS poll might well be the very best structuring of the polled group to match the general population.

Exactly, you say they could have oversampled "evangelicals" - or they might have undersampled them. It wasn't part of their weighing scheme. However they do use age, sex, race and others -but when their poll is specifically about POLITICAL opinions, don't you think it might just make sense to weigh for political affiliation just like if they were polling on religion - they should make sure their sample accurately reflects the overal breakdown of religious affiliation in the country?

The only thing I'm contending is that cBS has used a flawed demographic to get their story. They know their sample is flawed, or would if they even looked at the raw numbers so for them to write the story anyway without correcting for the flaw - it's intent is obvious- -unless of course you believe that cBS too incompetent to know better.

No, the actual weighing used could not be the best since it ignored a clear and very important demographic stat - political affiliation.

The main objective of this and most other polls is to assess the OVERALL level of approval/disapproval of the President. Understanding the component percentages is a secondary objective. The two objectives are at odds, unless you use a VERY large sample.

Note that even had CBS polled 39% Dems, 35% Repubs, and 26% Indies, the component results (for example, that 77% of Republicans approve of Bush) might well have been LESS accurate that the ones we're seeing here, since within "Republicans" there's also a complex demographic mix. And unless the "mix" at all levels is close to the actual percentages in the general poplulation, the results obtained will be skewed. So pollsters have to carefully weight their samples to achieve the best possible numbers from a sample group that unavoidably is NOT fully representive of the general population.

Suppose, for example, that there were 350 Republicans (about 35%) in the group originally chosen. But suppose that 40% of that 350 was evangelical Christians, whereas the actual percentage Republicans that are evangelicals is 30% (I'm making these numbers up, but the point remains the same). If one wished to get a more accurate idea of what Republicans believe about Bush, then one would want to randomly reduce the number of evangelicals in the Republican group by 35. That would help to get a more accurate result for Republicans, and also might help the accuracy of the overall result. But note that a consequence of this reduction is that the percentage of Republicans in the poll would be below 35%.

The point is, the objective of any poll is to obtain meaningful results, not to blindly choose a percentage of this polical group and a percentage of that, and assume that the results (absent weighting) will be acccurate.

 

[piasabird]

America is against anything that hurts their pocketbook. If you studied polls and politics you would realize that Americans follow the money.

 

[ShadesOfGrey]

Originally posted by: shira

Originally posted by: ShadesOfGrey

Originally posted by: shira

Originally posted by: ShadesOfGrey
When cBS weighted the poll for their typical stuff, they didn't consider party affiliation. What that did was weigh out more Republicans because of their other demographic stats.

Something you don't seem to understand is that in any random poll the profile of the polled group will inevitably NOT match the profile of the general population.

For example, if CBS had weighted responders so that 35% were Republicans, they might have ended up with much too high a percentage of evangelicals, whites, the high-school educated, and/or those living in the South (as compared with the general population), and that might have led to results that are less accurate than those actually obtained. ANY deviation in the polled group from the population they're intended to represent skews results.

You seem to be implying that the deviation of the percentage of Republications in the actual, weighted poll group from that in the general population is due to some insidious CBS agenda intended to skew results in some anti-Bush way. But unless you have a deep understanding as to how the weighting process was in fact performed and what its assumptions were, I don't see how you can come to any valid conclusions about motives.

Breakdown by party affiliation is just one demographic out of many, all of which correlate with political beliefs. And the actual weighting used in this CBS poll might well be the very best structuring of the polled group to match the general population.

Exactly, you say they could have oversampled "evangelicals" - or they might have undersampled them. It wasn't part of their weighing scheme. However they do use age, sex, race and others -but when their poll is specifically about POLITICAL opinions, don't you think it might just make sense to weigh for political affiliation just like if they were polling on religion - they should make sure their sample accurately reflects the overal breakdown of religious affiliation in the country?

The only thing I'm contending is that cBS has used a flawed demographic to get their story. They know their sample is flawed, or would if they even looked at the raw numbers so for them to write the story anyway without correcting for the flaw - it's intent is obvious- -unless of course you believe that cBS too incompetent to know better.

No, the actual weighing used could not be the best since it ignored a clear and very important demographic stat - political affiliation.

The main objective of this and most other polls is to assess the OVERALL level of approval/disapproval of the President. Understanding the component percentages is a secondary objective. The two objectives are at odds, unless you use a VERY large sample.

Note that even had CBS polled 39% Dems, 35% Repubs, and 26% Indies, the component results (for example, that 77% of Republicans approve of Bush) might well have been LESS accurate that the ones we're seeing here, since within "Republicans" there's also a complex demographic mix. And unless the "mix" at all levels is close to the actual percentages in the general poplulation, the results obtained will be skewed. So pollsters have to carefully weight their samples to achieve the best possible numbers from a sample group that unavoidably is NOT fully representive of the general population.

Suppose, for example, that there were 350 Republicans (about 35%) in the group originally chosen. But suppose that 40% of that 350 was evangelical Christians, whereas the actual percentage Republicans that are evangelicals is 30% (I'm making these numbers up, but the point remains the same). If one wished to get a more accurate idea of what Republicans believe about Bush, then one would want to randomly reduce the number of evangelicals in the Republican group by 35. That would help to get a more accurate result for Republicans, and also might help the accuracy of the overall result. But note that a consequence of this reduction is that the percentage of Republicans in the poll would be below 35%.

The point is, the objective of any poll is to obtain meaningful results, not to blindly choose a percentage of this polical group and a percentage of that, and assume that the results (absent weighting) will be acccurate.

Wrong wrong wrong again.

The main objective is overall level of the President, so don't you think they'd try to make their sample look like most of America?(yes) In this case while they adjusted thier sample to look like America using "normal" demographics, they neglected to add in the political weighing.

No, it wouldn't have been less accurate because they still would have adjusted for the other demographics.
Also it's funny that you want to claim that Republicans have a complex demographic mix, as if "Independents" or democrats don't :roll:

And you continue to prove my point. The existing sample is too heavy on democrats and Independents vs Republicans. Don't you think that just might have an affect on the poll? You want to claim a demographic within the Republican part might skew it, but undersampling them to begin with definately skews the data.

No one is blindly choosing a percentage. You obviously have zero clue what you are talking about. All the demographic stats have some sort of survey and data behind them, and that includes political affiliation.

 

[OrByte]

so CBS skewed the numbers...imagine that! they must have taken their cue from the LYING ADMINSTRATION in office!

again, imagine that....

so if we got the numbers right then that must mean Bush's ratings are STILL IN THE TOLIET with MOST OF AMERICANS.

:roll:

 

[smack Down]

Originally posted by: ShadesOfGrey

Originally posted by: shira

Originally posted by: ShadesOfGrey

Originally posted by: shira

Originally posted by: ShadesOfGrey
When cBS weighted the poll for their typical stuff, they didn't consider party affiliation. What that did was weigh out more Republicans because of their other demographic stats.

Something you don't seem to understand is that in any random poll the profile of the polled group will inevitably NOT match the profile of the general population.

For example, if CBS had weighted responders so that 35% were Republicans, they might have ended up with much too high a percentage of evangelicals, whites, the high-school educated, and/or those living in the South (as compared with the general population), and that might have led to results that are less accurate than those actually obtained. ANY deviation in the polled group from the population they're intended to represent skews results.

You seem to be implying that the deviation of the percentage of Republications in the actual, weighted poll group from that in the general population is due to some insidious CBS agenda intended to skew results in some anti-Bush way. But unless you have a deep understanding as to how the weighting process was in fact performed and what its assumptions were, I don't see how you can come to any valid conclusions about motives.

Breakdown by party affiliation is just one demographic out of many, all of which correlate with political beliefs. And the actual weighting used in this CBS poll might well be the very best structuring of the polled group to match the general population.

Exactly, you say they could have oversampled "evangelicals" - or they might have undersampled them. It wasn't part of their weighing scheme. However they do use age, sex, race and others -but when their poll is specifically about POLITICAL opinions, don't you think it might just make sense to weigh for political affiliation just like if they were polling on religion - they should make sure their sample accurately reflects the overal breakdown of religious affiliation in the country?

The only thing I'm contending is that cBS has used a flawed demographic to get their story. They know their sample is flawed, or would if they even looked at the raw numbers so for them to write the story anyway without correcting for the flaw - it's intent is obvious- -unless of course you believe that cBS too incompetent to know better.

No, the actual weighing used could not be the best since it ignored a clear and very important demographic stat - political affiliation.

The main objective of this and most other polls is to assess the OVERALL level of approval/disapproval of the President. Understanding the component percentages is a secondary objective. The two objectives are at odds, unless you use a VERY large sample.

Note that even had CBS polled 39% Dems, 35% Repubs, and 26% Indies, the component results (for example, that 77% of Republicans approve of Bush) might well have been LESS accurate that the ones we're seeing here, since within "Republicans" there's also a complex demographic mix. And unless the "mix" at all levels is close to the actual percentages in the general poplulation, the results obtained will be skewed. So pollsters have to carefully weight their samples to achieve the best possible numbers from a sample group that unavoidably is NOT fully representive of the general population.

Suppose, for example, that there were 350 Republicans (about 35%) in the group originally chosen. But suppose that 40% of that 350 was evangelical Christians, whereas the actual percentage Republicans that are evangelicals is 30% (I'm making these numbers up, but the point remains the same). If one wished to get a more accurate idea of what Republicans believe about Bush, then one would want to randomly reduce the number of evangelicals in the Republican group by 35. That would help to get a more accurate result for Republicans, and also might help the accuracy of the overall result. But note that a consequence of this reduction is that the percentage of Republicans in the poll would be below 35%.

The point is, the objective of any poll is to obtain meaningful results, not to blindly choose a percentage of this polical group and a percentage of that, and assume that the results (absent weighting) will be acccurate.

Wrong wrong wrong again.

The main objective is overall level of the President, so don't you think they'd try to make their sample look like most of America?(yes) In this case while they adjusted thier sample to look like America using "normal" demographics, they neglected to add in the political weighing.

No, it wouldn't have been less accurate because they still would have adjusted for the other demographics.
Also it's funny that you want to claim that Republicans have a complex demographic mix, as if "Independents" or democrats don't :roll:

And you continue to prove my point. The existing sample is too heavy on democrats and Independents vs Republicans. Don't you think that just might have an affect on the poll? You want to claim a demographic within the Republican part might skew it, but undersampling them to begin with definately skews the data.

No one is blindly choosing a percentage. You obviously have zero clue what you are talking about. All the demographic stats have some sort of survey and data behind them, and that includes political affiliation.

Do you have a poll or some evidence that the given party affelations are skewed for the general public and not likely voters.

 

[ShadesOfGrey]

Originally posted by: smack Down
Do you have a poll or some evidence that the given party affelations are skewed for the general public and not likely voters.

The only data that suggests they are different is that Republicans tend to vote at a higher percentage, which is in one of the links I've provided already in this thread.

 

[Tab]

So, what is your point SoG? CBS is purposely skewing the results to make the current administration to look bad?

 

[Engineer]

Damn skewed polls anyay.

 

[ShadesOfGrey]

Originally posted by: Engineer
Damn skewed polls anyay.

It wasn't about the supposed results. This thread is about the way cBS in particular skewed their sample for their story. The debate can be had if they meant to or not but there is plenty of evidence that the data was bad.

Tab - it's about how cBS conducts itself. This isn't the first time they've put something out without double-checking it.

 

[homercles337]

Originally posted by: ShadesOfGrey

Originally posted by: Engineer
Damn skewed polls anyay.

It wasn't about the supposed results. This thread is about the way cBS in particular skewed their sample for their story. The debate can be had if they meant to or not but there is plenty of evidence that the data was bad.

Tab - it's about how cBS conducts itself. This isn't the first time they've put something out without double-checking it.

You still havent shown how or IF they intentionally skewed results. Are you still upset at CBS for proving that DumbYa used his daddies influence to avoid the draft and join the NG even though he recently said that he never did?

 

[Tab]

Originally posted by: ShadesOfGrey

Originally posted by: Engineer
Damn skewed polls anyay.

It wasn't about the supposed results. This thread is about the way cBS in particular skewed their sample for their story. The debate can be had if they meant to or not but there is plenty of evidence that the data was bad.

Tab - it's about how cBS conducts itself. This isn't the first time they've put something out without double-checking it.

How did they skew it? Who skewed it? You haven't show any evidence that they are purposely changing polling information.

 

[Todd33]

Let be real here, he/she/it made this thread to preempt someone that would post the results, without screaming bias and attacking the messenger. This thread is baseless. There is no proof, there is nothing bad typical poll number obtained in typical poll methods, which includes not using party ID to normalize, which is typical. this is the typical "liberal media" echo that has been going on for 30 years in an attempt at discrediting journalism. Repugs hate a free and open press, they want Baghdad Bob style press.

 

[ShadesOfGrey]

Originally posted by: homercles337

Originally posted by: ShadesOfGrey

Originally posted by: Engineer
Damn skewed polls anyay.

It wasn't about the supposed results. This thread is about the way cBS in particular skewed their sample for their story. The debate can be had if they meant to or not but there is plenty of evidence that the data was bad.

Tab - it's about how cBS conducts itself. This isn't the first time they've put something out without double-checking it.

You still havent shown how or IF they intentionally skewed results. Are you still upset at CBS for proving that DumbYa used his daddies influence to avoid the draft and join the NG even though he recently said that he never did?

Whether it was intentional or whether it was incompetence - it doesn't change the fact that cBS has serious problems when it comes to credibility. Edward Murrow would not be pleased at the current state of affairs in his old house. If the media is supposed to be where people can go for truth, this certainly doesn't help that image.

 

[Tab]

Originally posted by: ShadesOfGrey

Originally posted by: homercles337

Originally posted by: ShadesOfGrey

Originally posted by: Engineer
Damn skewed polls anyay.

It wasn't about the supposed results. This thread is about the way cBS in particular skewed their sample for their story. The debate can be had if they meant to or not but there is plenty of evidence that the data was bad.

Tab - it's about how cBS conducts itself. This isn't the first time they've put something out without double-checking it.

You still havent shown how or IF they intentionally skewed results. Are you still upset at CBS for proving that DumbYa used his daddies influence to avoid the draft and join the NG even though he recently said that he never did?

Whether it was intentional or whether it was incompetence - it doesn't change the fact that cBS has serious problems when it comes to credibility. Edward Murrow would not be pleased at the current state of affairs in his old house. If the media is supposed to be where people can go for truth, this certainly doesn't help that image.

You don't know that at all, nor have you even proved it.

 

[ShadesOfGrey]

Originally posted by: Todd33
Let be real here, he/she/it made this thread to preempt someone that would post the results, without screaming bias and attacking the messenger. This thread is baseless. There is no proof, there is nothing bad typical poll number obtained in typical poll methods, which includes not using party ID to normalize, which is typical. this is the typical "liberal media" echo that has been going on for 30 years in an attempt at discrediting journalism. Repugs hate a free and open press, they want Baghdad Bob style press.

There was no preemption. It was already out and people were trying to use it. They bought it without even looking at it critically to see it's accuracy.
The only thing baseless are your claims and attacks on me and this thread. The poll sample is flawed, which is obvious to anyone who looks at the raw data.

:roll: "hate free and open press"? Nice try. I would love to see an open and free press, but we don't seem to have that. We have Fox on the right, and CNN, MSNBC, and the "big 3" on the left. Why don't we have a source that doesn't inject a bias? Why is that? Why do we as news consumers put up with this activist journalism? Why do you attack me for calling out this bad journalism instead of joining the critique to make "news" - News again?