Wednesday, August 22, 2007
The Problem with Survey USA Polls
Raising Kaine wonders what's going on with Warner's poll numbers going up and Webb's numbers coming down in the latest Survey USA approval poll.
Howling Latina has an explanation for the gyrations in approval rate numbers for the Virginia senators. The monthly sample size is simply too small and should be examined over the long run in conjunction with prior polls.
Think about it: No matter how scientific the model, does one truly believe that a sample size of 600 adults with 108 blacks and 24 Hispanics is meaningful, especially when the margin of error is as high as 4.1 percent???
The swings from month to month support HL's hypothesis and the numbers are interesting and great conversation fodder but should be taken in stride.
Howling Latina has an explanation for the gyrations in approval rate numbers for the Virginia senators. The monthly sample size is simply too small and should be examined over the long run in conjunction with prior polls.
Think about it: No matter how scientific the model, does one truly believe that a sample size of 600 adults with 108 blacks and 24 Hispanics is meaningful, especially when the margin of error is as high as 4.1 percent???
The swings from month to month support HL's hypothesis and the numbers are interesting and great conversation fodder but should be taken in stride.
// posted by Howling Latina @ 12:20 PM 
