Polls relation to policy concerns?
If I seem like I'm piling on, it's because I know you're a lawyer, probably a very good and clever one and have argued a lot of cases. I sense a tendency to present "evidence", and shape the argument to fit that evidence while mitigating or even ignoring argument or evidence to the contrary. Common practice here, you can be more subtle about it than most though.
In the quoted statement, you would have us shape a conversation based on evidence (poll numbers) that may be accurate, but perhaps not truly representative with the idea that should another poll surface later that better explains it's demographics, the subject will be revisited. Well, it's doubtful that that will ever happen. So, we discuss the poll in question where a blind man can see that troops don't have a huge interest in being in Iraq, and that shapes the merits of the conversation. Anyone saying that the poll does NOT indicate that troops don't want to be in Iraq would be wrong, easily refuted by one piece of evidence. Stacked deck. Then, you tie it to policy implications (dangerous with ANY poll, but whatever).
Plus, sometimes it's simply fun to jerk your chain.