>>> dhenwood at panix.com 08/16/00 09:09AM >>>
Rob Schaap wrote:
>These polls don't mean jack-shit, do they? A day after Dubya's non-speech,
>he was riding seventeen points clear. By the beginning of this convention
>the gap was about eleven.
The American Association of Public Opinion Research list is discussing a piece that appeared in the Wash Post the other day by the paper's polling director. His argument was that the polling industry is using extremely small samples (ca. 500 people) to generate quick overnight polls, with extremely unreliable and volatile results. A lot of the movement from day-to-day is just statistical noise that gets interpreted as trend shifts.
Polls at this stage of the game are fairly meaningless anyway.
CB: Sorry for the conspiracy theory edge to this comment, but I feel even worse than Doug about polls, that is that they are worse than that they are meaningless at this stage of the game. I cannot put it past some sectors to use/influence ( buy) early polls with skewed results to influence the formation of public opinion. The small samples, may also be consciously biased. This does not address the issue of herding answers by questionnaire grammar, range of questions etc.
The final polls close to actual votes are less manipulable , because if they start miscalling elections, they will discredit themselves. But early on there is no actual vote to test the figures they put out, and they can create a herd or groundswell or bandwagon effect. The more intermittent and critically spotted the manipulation, the less detectable and more effective.