Something the Dog Said wrote: I find it interesting that the very same posters who were complaining about the "skew" against Romney in the polls now jump onboard those very same polls. If you happen to review the crosstabs on the Pew poll you would find that their sampling was "skewed" in favor of Republicans.
Do the cross tabs represent the weighted numbers or simply the respondents to the poll Dog? The point raised last week was that the weighting model being used was flawed and that this flawed weighting model was responsible for skewing.
I just want to keep it apples to apples. If the crosstabs represent the weighted numbers then I would agree with you on many points. If all they represent are the raw data, then I think you are overlooking a few points regarding your analysis.
The cross tabs referring to the number of hispanic voters is "weighted" as there were 70 hispanic voters out of 1200 total sampled, so Pew weighed their influence as zero. Kind of tells you the (lack) of validity of this poll, in that according to this poll, no hispanic voters will vote. If you remove 24 million voters (estimate of the number of latino voters in the US), particularly when about 2/3rds of those will vote for Obama, then you have a skewed poll.
The cross tabs relating to age and regional sampling are unweighted, as this is intended to show the source of the raw data. The weighting will not change the skew in the poll, if the raw data is already skewed, then weighting it will only magnify the skew.
Every polling organization uses a different weighting process, so there is no true apple to apple comparison.
"Remember to always be yourself. Unless you can be batman. Then always be batman." Unknown
There's also the fact that 9 in 10 phone calls don't end with a poll being done. 9 in 10 hang up on the pollster. I don't have any faith in the polls any way that they lean.
Yeah, I'm not so sure about the polls either. But I don't study them as much as others do here anyway.
Instead of worrying about all the intricate details about each poll, I'd just look at past history to decide who has the most accurate poll.
So who had to most accurate poll for the 2008 presidential/congressional elections, and who had the most accurate poll for the 2010 congressional elections?
I'd pay the most attenion to the most accurate polls in 2008/2010.
THis guy has been the most accurate through the last 8 plus years.... he uses most of the major polls, adjust for bias and adds certain factors such as the economy etc..
towermonkey wrote: There's also the fact that 9 in 10 phone calls don't end with a poll being done. 9 in 10 hang up on the pollster. I don't have any faith in the polls any way that they lean.
Great point.
Thomas Sowell: There are no solutions, just trade-offs.
