>I've just been reading Richard Lewontin's article, "Sex, Lies and Social
>Science," and I think his critique there of surveys of sexual behavior
>apply also to all attempts to survey opinion on any question on which
>complex reactions are possible. "Would you vote for X?" can (more or
>less) be legitimately answered with a yes or no. "Do you support the
>(patriotic) war?" is as little apt to be answered accurately as "What do
>you think about your wife masturbating?"
>
>Carrol
the problem you have, though, is that this phenom is well-documented. it's well-known that a president can get hugely high approval ratings because there is a crisis. even Nixon got high approval ratings after Watergate Scandal broke. there are complex reasons for this phenom, which may have nothing to do with they _agree_ with him. some of it is reaction against an attacking press. some of it is a belief that people should unite around "their man". some of it is just human compassion.
even so, when it comes to behavior (rather than opinion), despite the high approval ratings, people don't necessarily vote for someone they approve of under what they perceive of as crisis conditions.
why? because there's a difference between the population surveyed and the population that votes. it depends on the methodology, of course. there are plenty of people in this country who probably wouldn't know the name of the president were there no "war on terr" (not a typho!). when an opinion survey comes round about approval ratings, it's generally from the gen'l population. when a "would you vote for ____" opinion poll comes round it's usually based on the population of registered voters.
kelley