Originally Posted by
notEnuf
There are some real sampling issues with the polling. The wife is a marketing research analyst and sees this all the time. There are also question construction issues. Bias is impossible to avoid in questioning, but it should be minimized by not asking leading questions and the use of open ended questions. Coding and data analysis is also some what subjective when open ended questions are used. All this said, if done right it has a statistical correlation to the larger group. The margin for error can only be quantified on direct A or B type answers.
All of this could be done right with a professional poll, my doubts are around the construction of the questions, the sampling, and the fiduciary relationship of a long term vendor. The wife has been asked several time to soften language in her report by an account executive who doesn't want to upset a client. She quit a job after hearing from a VP, "the client has already decided X, they're just looking for data to support that decision." TIFWIW
I haven't been polled so I couldn't say what I think of DALPA polling.
^This has 10 times more relevance than 'anything' else posted, imo.