> eric 9.5 beck writes to wojtek:
>
> >Perhaps my calculator is broken, but it's telling me that if just over half
> >of the people support the war then just under half *don't* support it.
>
> plus or minus 3% even eric, so it could actually be over half.
Does the survey allow for or account for undecided? If I recall correctly, just over half responded that they supported the war, and thirty-odd percent responded that they opposed the war, with the remainder undecided, taking no answer, or responding in some other fashion (no opinion?).
Likewise, it's eminently possible (and unfortunately winked at by pollsters) to indulge in a push-poll - to give information biasing further questions (or even slanting public opinion) in the survey. For example, let's postulate an imaginary push-poll about LBO:
Question 3. Did you know that Doug Henwood, as a Marxist, advocates the bloody overthrow of the capitalist system?
Question 4. What do you think of the accuracy of Henwood's newsletter, Left Business Observer?
For this reason, reading the poll (they're frequently available on the pollster's website) in its entirety before considering it seriously. Most mainstream media outlets don't indulge in such monkey business overtly; it's entirely possible that subtle bias was introduced in the poll by prior questions or statements around american casualties, Serbian violence against innocent civilians, human shields, or the like.
marco
,--------------------------------------------------------------------------.
> | We know what causes violence: poverty, <
> Marco Anglesio | discrimination, the failure of the <
> mpa at the-wire.com | educational system. It's not the genes <
> http://www.the-wire.com/~mpa | that cause violence in our society. <
> | --Paul Billings. <
`--------------------------------------------------------------------------'