We may all wind up agreeing to disagree on this one. Taking the easy part first, the extraordinary discrepency is probably due to the fact that people are extraordinarily complex. No one feels any great need to be consistent. I suspect a lot of people are responding with an attitude of: yeah, the war sucks, wish we weren't there, now lets roll up our sleeves and get the unplesent chore done.

As to the questions, my read is that the poll adopted the language of the different parties to present the choices. As an example, and whether we agree or not, one side is calling for "immediate" withdrawal, and the other proposes staying until we restore order.

My point is that using the language chosen by the various policy advocates to present their position can not be termed push polling. Neither can framing a question in semantically neutral terms: "Restore order" is semantically neutral, since it only requires agreement that there is disorder in Iraq, and requires no agreement on its composition, causes, etc. "... stop the civil war between Sunni and Shiite Iraqis" is not neutral, because it imposes assumptions about the nature of the disorder.