cBS poll the left is trying to tout

[ShadesOfGrey]

Originally posted by: Tab

Originally posted by: ShadesOfGrey

Originally posted by: Bowfinger

Originally posted by: ShadesOfGrey

Originally posted by: Bowfinger

Originally posted by: ntdz
No, we are talking about polling methods, NOT the results of the polls. Clearly CBS is skewing the poll to get the results it wants (bad numbers for Bush). That's not to say that Bush's numbers aren't bad already, they are. Nobody is denying that.

"Do you have any statistically qualified basis for rejecting their weighting? I'm guessing not, that you're rejecting it solely because it contradicts your beliefs."

Lots of sour grapes in this thread. You know sour grapes makes sour whine, don't you?

BS.

TIME uses the following sample rates:
Likely voters, 34% Republican, 35% Democrat, 22% Independent. Registered voters 31% Republican, 32% Democrat, 26% Independent

This cBS poll doesn't tell us which they called but either way their sample is WAY different from TIME.

Did cBS really think people wouldn't notice how they skewed the numbers?

Time's weighting for predicting elections is irrelevant. This isn't about an election. This wasn't the Time population sample. While it's legitimate to question their adjustments, you have no factual basis for asserting as fact that CBS "fabricated" anything or did anything "outrageous". (Perhaps you're thinking of BushCo's fabrications re. Iraq's WMDs. You know, the ones in the thread you keep "ignoring". ROFL.)

I'll ask again, "Do you have any statistically qualified basis for rejecting their weighting?" All you've shown so far is your opinion that it's wrong. You've offered no factual evidence based on the actual demographics of this poll to show the CBS adjustment is improper. Nothing. I know you have trouble separating your opinion from fact, but that's your problem. You claim CBS is wrong. Prove it.

Yes, Bowfinger, we know you like to live in your little alter-world but even considering that, it is unbelievable you continue with this crap.

You want me to prove cBS didn't weigh it correctly? Well, first off, common sense tells you that 41% is too high for Independent and 24% is way too low for Republicans. As ntdz pointed out, last year's election breakdown was 37, 37, and 26. 26 being Independent. Those two things coupled with another media outlets demographic sampling numbers is more than enough evidence to show cBS has some bad data.

Oh, and one more thing - Read this. Still want to play stupid and claim you need more proof cBS has bad data? ROFL! :laugh:

LAST YEAR THATS WHAT IT WAS DO YOU UNDERSTAND? OVER TIME PEOPLE CHANGE THEIR POLITICAL AFFLIATION.

Not that drastically. There have been studies done about swings in voter affiliation. There is no possible way that would be the case here, and besides, cBS would have to prove that.

 

[ShadesOfGrey]

Originally posted by: totalcommand

Originally posted by: ShadesOfGrey
cBS link to their "story"

They make a lot of claims about why and what it means but it seems they have a little something to hide.

PDF of cBS poll

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!

Unweighted Total Respondents: 936
Republicans = 259 (27.67%)
Democrats = 326 (34.83%)
Independents = 351 (37.5%)

Weighted Total Respondents: 937
Republicans: 223 (23.80%)
Democrats: 326 (34.79%)
Independents: 388 (41.4%)

lowering the Republican sample from an already low 28%? As John Stossel says - Give me a break!


Talk about fabricating a story or guiding things towards the outcome you want to report or make a story about! I could care less what Bush's real poll ratings are but the fact that the media thinks they can get away with this sort of twisting of reality is outrageous.

Do Our Respondents Look Like The American Public?

At the end of our surveys, we find sometimes that we have questioned too many people from one group or another. Older people, for example, tend to be at home to answer the phone more than younger people, so there is often a greater percentage of older people in our surveys than exists in the American public.

When that happens, we take great pains to adjust our data so that I accurately reflects the whole population. That process is called ?weighting.? We make sure that our final figures match U.S. Census Bureau breakdowns on age, sex, race, education, and region of the country. We also ?weight? to adjust for the fact that people who share a phone with others have less chance to be contacted than people who live alone and have their own phones, and that households with more than one telephone number have more chances to be called than households with only one phone number.

So when we add up all the answers to our questions, we know that no one?s opinion counts for more than it should. When you see one of our poll results on TV or in the newspaper, you know that it does not show the opinions of only one or two groups of Americans.

This is a very strong statistical basis for the weighting. The poll is NOT weighted in order to lower Repubs and raise Independents (or correct for political affiliation), but is weighted for age, sex, race, etc.

In effect, the poll is also polling how the American public describes themselves politically. With an 95% confidence in +/- 3%, the poll is probably accurate.

If what you are trying to claim was true, then why do the numbers turn out so skewed? Why would cBS knowingly use this bad data?(I think we know the answer to that).

 

[ShadesOfGrey]

Originally posted by: Bowfinger

Originally posted by: totalcommand

Originally posted by: ShadesOfGrey
cBS link to their "story"

They make a lot of claims about why and what it means but it seems they have a little something to hide.

PDF of cBS poll

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!

Unweighted Total Respondents: 936
Republicans = 259 (27.67%)
Democrats = 326 (34.83%)
Independents = 351 (37.5%)

Weighted Total Respondents: 937
Republicans: 223 (23.80%)
Democrats: 326 (34.79%)
Independents: 388 (41.4%)

lowering the Republican sample from an already low 28%? As John Stossel says - Give me a break!


Talk about fabricating a story or guiding things towards the outcome you want to report or make a story about! I could care less what Bush's real poll ratings are but the fact that the media thinks they can get away with this sort of twisting of reality is outrageous.

Do Our Respondents Look Like The American Public?

At the end of our surveys, we find sometimes that we have questioned too many people from one group or another. Older people, for example, tend to be at home to answer the phone more than younger people, so there is often a greater percentage of older people in our surveys than exists in the American public.

When that happens, we take great pains to adjust our data so that I accurately reflects the whole population. That process is called ?weighting.? We make sure that our final figures match U.S. Census Bureau breakdowns on age, sex, race, education, and region of the country. We also ?weight? to adjust for the fact that people who share a phone with others have less chance to be contacted than people who live alone and have their own phones, and that households with more than one telephone number have more chances to be called than households with only one phone number.

So when we add up all the answers to our questions, we know that no one?s opinion counts for more than it should. When you see one of our poll results on TV or in the newspaper, you know that it does not show the opinions of only one or two groups of Americans.

This is a very strong statistical basis for the weighting. The poll is NOT weighted in order to lower Repubs and raise Independents (or correct for political affiliation), but is weighted for age, sex, race, etc.

In effect, the poll is also polling how the American public describes themselves politically. With an 95% confidence in +/- 3%, the poll is probably accurate.

Hush. Don't you be introducing fact and statistical science into the whine-fest. SoG and ntdz don't cotton much to that libuhrul science mumbo-jumbo. Like their hero Reagan, they know they're right in their heart, even though the facts say otherwise.

So you think the sample was good even though it doesn't come close to reflecting real party affilitation figures? It was a political poll, party affiliation is a big factor and cBS ignored the fact that they used bad numbers.

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.

 

[Bowfinger]

Originally posted by: ShadesOfGrey
So you think the sample was good even though it doesn't come close to reflecting real party affilitation figures? It was a political poll, party affiliation is a big factor and cBS ignored the fact that they used bad numbers.

In your opinion. You may even be right. Statistics is a sciene, however, not beholden to your emotional claims of how things should be. CBS claims to have used a statistically valid sampling, adjusted it according to sound scientific principles, and produced a result they say can be scientifically shown to have a 3% margin of error. You have offered no relevant evidence to challenge this except your emotional, clearly partisan opinion. It's much the same as your response to the fundamental economics tenet that scarcity increases value, another scientific principle you've attacked with emotional opinion.

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.

That's part of it. It is equally about your ignorance of statistical methods and a nice example of the faith vs. science arguments that so often pit the left against the right these days.

 

[ShadesOfGrey]

Originally posted by: Bowfinger

Originally posted by: ShadesOfGrey
So you think the sample was good even though it doesn't come close to reflecting real party affilitation figures? It was a political poll, party affiliation is a big factor and cBS ignored the fact that they used bad numbers.

In your opinion. You may even be right. Statistics is a sciene, however, not beholden to your emotional claims of how things should be. CBS claims to have used a statistically valid sampling, adjusted it according to sound scientific principles, and produced a result they say can be scientifically shown to have a 3% margin of error. You have offered no relevant evidence to challenge this except your emotional, clearly partisan opinion. It's much the same as your response to the fundamental economics tenet that scarcity increases value, another scientific principle you've attacked with emotional opinion.

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.

That's part of it. It is equally about your ignorance of statistical methods and a nice example of the faith vs. science arguments that so often pit the left against the right these days.

And so do you believe their sampling was correct? While they might claim to have used good methods, the numbers show just how bad their method is -if it was even followed(which you are taking on faith).
The numbers don't lie - they have an over-sample of so-called Indepenedents and democrats.
This is not about faith, it's about the numbers which I've pointed out. The numbers just don't lie. Their sample is flawed.

 

[Czar]

Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

 

[ShadesOfGrey]

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

 

[shira]

Originally posted by: ShadesOfGrey
I could care less what Bush's real poll ratings are but the fact that the media thinks they can get away with this sort of twisting of reality is outrageous.

So you're saying you COULD care less. That means you care.

Did you perhaps mean to say you COULDN'T care less?

 

[ShadesOfGrey]

Originally posted by: shira

Originally posted by: ShadesOfGrey
I could care less what Bush's real poll ratings are but the fact that the media thinks they can get away with this sort of twisting of reality is outrageous.

So you're saying you COULD care less. That means you care.

Did you perhaps mean to say you COULDN'T care less?

No it doesn't mean I care. It means I don't care, but I could care less if need be. Just because there is the option of more opposition, does not mean support of.

Now this little exchage fits into Czar's thread. Worthless arguing for the sake of arguing.

 

[Czar]

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

how is it known political demographics? how do the determine the general population's political stand? polls?

 

[feralkid]

Originally posted by: Pabster


And CBS didn't care, because the fools who use CBS as their "news source" deserve exactly what they get. (And are delighted by it.)

This from the guy whose favorite media source is USA Today?

 

[totalcommand]

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

There is a 3% margin of error with a 95% confidence on those numbers. There is no reason to weight by political demographics except during an election year, especially when political demographics are in a state of flux.

 

[ShadesOfGrey]

Originally posted by: Czar

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

how is it known political demographics? how do the determine the general population's political stand? polls?

They have done studies on it. I'm trying to find a link I had about those types of studies.


Read #5 and #6.
The NCPP also uses a breakdown of 39-D /35 - R /26 - I from what I've read. which is a year 2000 breakdown and might not take into account the continuation of a trend towards more Republicans.

 

[ShadesOfGrey]

Zogby
"My polls use a party weight of 39% Democrat, 35% Republican and 26% Independent."

 

[Bitek]

Originally posted by: Czar

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

how is it known political demographics? how do the determine the general population's political stand? polls?

How would OP know that the "party" sampling was not an indication of the lowering poll numbers for the GOP, resulting in fewer people IDing themselves with it. In other words, how can you prove your methods, demanding sampling #'s on KNOWN (where you you get this data btw?...) affliations would not skew the results higher, as you are possibly over-representing a diminishing party size?

Eg. Greens nearly voted 5% in 2000, but I don't see them specifically represented in the sample. Would you institutue a quota for them as well? What % do you give them, and based on what? Please enlighten us....

 

[ShadesOfGrey]

Originally posted by: totalcommand

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

There is a 3% margin of error with a 95% confidence on those numbers. There is no reason to weight by political demographics except during an election year, especially when political demographics are in a state of flux.

Wrong wrong wrong.

That is a statistical error for what they have weighed for and tried to sample at. However since they did not include party affiliation - their error was not captured in that percentage. It was however captured in their raw numbers.

Also, you can keep trying to claim party affiliation is constantly fluxauating, and you'd be right to a small degree, but that doesn't explain this drastic under-sampling of Republicans. Party affiliation doesn't swing that wildly. It's been tracked by the NAES and doesn't support your apparant contention that the fluxaion could produce these results.

 

[ShadesOfGrey]

Originally posted by: Hafen

Originally posted by: Czar

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

how is it known political demographics? how do the determine the general population's political stand? polls?

How would OP know that the "party" sampling was not an indication of the lowering poll numbers for the GOP, resulting in fewer people IDing themselves with it. In other words, how can you prove your methods, demanding sampling #'s on KNOWN (where you you get this data btw?...) affliations would not skew the results higher, as you are possibly over-representing a diminishing party size?

Eg. Greens nearly voted 5% in 2000, but I don't see them specifically represented in the sample. Would you institutue a quota for them as well? What % do you give them, and based on what? Please enlighten us....

As noted, the NAES tracked party affiliation. While there are "fluxuations", there are not massive swings.

To ask you a question though, do you think that only 24% of America is Republican vs 35% Democrat? Ofcourse not, it's a wildly absurd suggestion. That's why when asking political questions in a poll, it's weighting must take into account party affiliation. This cBS poll is like taking a P&N poll and then weighting it by age, sex, etc but leaving out party affiliation. It'd skew the results towards a liberal result.

 

[Darkhawk28]

I agree the weightings are a bit strange, but no stranger than Gallup's and Rasmussen's.

 

[Bitek]

The screwy thing in that poll is that Bush's support has only increased among white evangelicals, for all other groups, incl Conservatives ad Reps, it has fallen somewhat.

The other important point is the very low support among independents. Not good going into '06.

 

[shira]

Originally posted by: ShadesOfGrey
Zogby
"My polls use a party weight of 39% Democrat, 35% Republican and 26% Independent."

Using Zogby's percentages, and applying the 11%, 77%, and 31% approval percentages reported in this poll for Democrats, Republicans, and Independents, respectively, one gets an overall approval rating for Bush of 39%.

Pretty pathetic.

 

[totalcommand]

Originally posted by: ShadesOfGrey

Originally posted by: totalcommand

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

There is a 3% margin of error with a 95% confidence on those numbers. There is no reason to weight by political demographics except during an election year, especially when political demographics are in a state of flux.

Wrong wrong wrong.

That is a statistical error for what they have weighed for and tried to sample at. However since they did not include party affiliation - their error was not captured in that percentage. It was however captured in their raw numbers.

Um, I think you need to take a stats class. raw numbers is the same thing as a percentage, just that percentage is normalized to 100.

Also, you can keep trying to claim party affiliation is constantly fluxauating, and you'd be right to a small degree, but that doesn't explain this drastic under-sampling of Republicans. Party affiliation doesn't swing that wildly. It's been tracked by the NAES and doesn't support your apparant contention that the fluxaion could produce these results.

It's a reflection of both fluxuations and the standard error of the survey, as well as the 95% confidence (it is not 100% confidence). Basically, you need to realize that this is just a poll, and needs to be repeated several times to confirm it.

 

[ShadesOfGrey]

Originally posted by: totalcommand

Originally posted by: ShadesOfGrey

Originally posted by: totalcommand

Originally posted by: ShadesOfGrey

Originally posted by: Czar
Here is how they did it so you can understand

Country has a population of 100

50 are white
20 are hispanic
20 are black
10 are other

50 are working age
20 are retired
30 are children

Now we do a poll of a sample of 10

We see we need to ask 5 white people, 2 hispanics, 2 blacks and 1 of other race. We also see that 30 are of non voting age so they are not included. So we need to ask 7 people of working age and 3 people who are retired.

Now we come to the questions

1. Do you approve or disaproof of the leader of the country?
2. Which party did you vote for in the last elections?
3. Do you like white bread or brown bread?

The poll results would then give us a result based on the census data. The poll is not and never is based on how people voted in the last elections because that number is not a constant and it is not registered unless you are registered in that party. And frankly we all know that not nearly everyone is registered in political parties.

How was their sampling wrong?

Their sampling is wrong because they asked political questions and didn't take into account KNOWN political demographics. I don't care about last election, those numbers just go to show how skewed these numbers are and I'm not saying they should weight based on election results but it would be alot more accurate than ignoring political affiliation all together like cBS did. There are plenty of demographic statistics available to cBS so they can weigh political polls using political demographics. TIME does it. But for some reason cBS decided not to. I wonder why that is.

There is a 3% margin of error with a 95% confidence on those numbers. There is no reason to weight by political demographics except during an election year, especially when political demographics are in a state of flux.

Wrong wrong wrong.

That is a statistical error for what they have weighed for and tried to sample at. However since they did not include party affiliation - their error was not captured in that percentage. It was however captured in their raw numbers.

Um, I think you need to take a stats class. raw numbers is the same thing as a percentage, just that percentage is normalized to 100.

Also, you can keep trying to claim party affiliation is constantly fluxauating, and you'd be right to a small degree, but that doesn't explain this drastic under-sampling of Republicans. Party affiliation doesn't swing that wildly. It's been tracked by the NAES and doesn't support your apparant contention that the fluxaion could produce these results.

It's a reflection of both fluxuations and the standard error of the survey, as well as the 95% confidence (it is not 100% confidence). Basically, you need to realize that this is just a poll, and needs to be repeated several times to confirm it.

Uh, no, you brought up percentage error and percent certainty. If YOU would have ever taken a stats class you'd know those come from sampling and weighing demographics and their repeatability.

I know it's "just a poll" but it's also just a flawed poll that cBS used to write a story as news. They used a flawed sample and did not inform the reading public of their skewed sample.

 

[Todd33]

Originally posted by: ShadesOfGrey
cBS link to their "story"

They make a lot of claims about why and what it means but it seems they have a little something to hide.

PDF of cBS poll

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!

Unweighted Total Respondents: 936
Republicans = 259 (27.67%)
Democrats = 326 (34.83%)
Independents = 351 (37.5%)

Weighted Total Respondents: 937
Republicans: 223 (23.80%)
Democrats: 326 (34.79%)
Independents: 388 (41.4%)

lowering the Republican sample from an already low 28%? As John Stossel says - Give me a break!


Talk about fabricating a story or guiding things towards the outcome you want to report or make a story about! I could care less what Bush's real poll ratings are but the fact that the media thinks they can get away with this sort of twisting of reality is outrageous.

I know you and the blog you got this from our outraged, but I think your also ignorant. Weighting the by party ID is typical, but I think you are misinterpreting what the PDF says. The have party ID as given to them by the person, which is typically fluid and then they have historical numbers that the random sample should match. I think they just adjust the numbers, in this case moving identified Republicans into the Independent column. If you think about this it would not skew it in favor of the way you are suggesting, but the opposite. But I'm just guessing at this point as are you, but at least I didn't make a stupid trolling thread that accused a poll of blatantly being biased. You know it's very typical to leave party ID out completely, so the fact they have those numbers should suggest they didn't do harm with them. The PDF does not contain how they used them, but I suspect polling expert do know, which you and your blog friends are not - experts.

 

[ShadesOfGrey]

Originally posted by: Todd33

Originally posted by: ShadesOfGrey
cBS link to their "story"

They make a lot of claims about why and what it means but it seems they have a little something to hide.

PDF of cBS poll

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!

Unweighted Total Respondents: 936
Republicans = 259 (27.67%)
Democrats = 326 (34.83%)
Independents = 351 (37.5%)

Weighted Total Respondents: 937
Republicans: 223 (23.80%)
Democrats: 326 (34.79%)
Independents: 388 (41.4%)

lowering the Republican sample from an already low 28%? As John Stossel says - Give me a break!


Talk about fabricating a story or guiding things towards the outcome you want to report or make a story about! I could care less what Bush's real poll ratings are but the fact that the media thinks they can get away with this sort of twisting of reality is outrageous.

I know you and the blog you got this from our outraged, but I think your also ignorant. Weighting the by party ID is typical, but I think you are misinterpreting what the PDF says. The have party ID as given to them by the person, which is typically fluid and then they have historical numbers that the random sample should match. I think they just adjust the numbers, in this case moving identified Republicans into the Independent column. If you think about this it would not skew it in favor of the way you are suggesting, but the opposite. But I'm just guessing at this point as are you, but at least I didn't make a stupid trolling thread that accused a poll of blatantly being biased. You know it's very typical to leave party ID out completely, so the fact they have those numbers should suggest they didn't do harm with them. The PDF does not contain how they used them, but I suspect polling expert do know, which you and your blog friends are not - experts.

:roll: Yeah, you seem to be a real expert, did you even look at the raw numbers? 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. That certainly does skew the data opposite of what I have stated. I can't believe you are that ignorant, are you just playing stupid - hoping no one calls you on it?

Anyway, keep defending the obviously skewed sample poll and cBS for using it - we wouldn't expect otherwise from your ilk.

 

[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...