>On this site U find alternative polls and a critique of usual opinion polls
>(like the factual one that about 70% of interviewed in any
>poll are "not respondent").
>Highly recommended.
>http://www.retropoll.org/
On looking at the actual question compared to the press release, this gang spins a lot more than Gallup.
Press release <http://www.retropoll.org/press_release_poll03.htm>: -------------
>At least one in three Americans believe that George W. Bush should
>face impeachment for misleading the public and Congress about
>Saddam's weapons of mass destruction to create support for war on
>Iraq. This is a new finding from a national survey conducted by the
>Retro Poll organization between October 29 and November 12. The
>actual proportion supporting impeachment was 40% but with a margin
>of error of plus or minus 8%, 1 in 3 remains a conservative
>population estimate. "We are seeing a rising tide of public anger
>that no one is paying attention to", said Dr. Marc Sapir, Retro
>Poll's Director.
Question <http://www.retropoll.org/results_poll_03.htm>: --------
>31. President Bush claimed that the US had to invade Iraq because
>Iraq had nuclear, biological and chemical weapons. None has been
>found. Do you think that misleading the U.S. public and Congress in
>order to take the country into war is grounds for impeachment?
>
>Yes (39.9%), No (42.5%), Don't Know (17.6%).
The press release would lead you to believe the question was direct: "Do you think President Bush should be impeached for misleading the public about WMDs in Iraq?" But the actual question is in two parts. People could think that Bush really believed his own propaganda, and so would not be eligible for impeachment under these criteria. The release also doesn't say that the "No" vote was larger than the "Yes" vote (though with a sample of 150, the margin is way smaller than the margin of error).
Just because something is spun the way you like doesn't make it more "objective" than "corporate" polls (and, by the way, Gallup is employee-owned).
The non-response issue is very controversial in the public opinion world, and most honest pollsters are aware of the difficulties. So Retropoll isn't accurately reporting on the state of mainstream pollster thinking either.
Doug