We also don't know who's being asked the questions. If they're doing polls by calling people who are home in the middle of the day on a Wednesday then maybe they should say that "60% of Americans who are home in the middle of the day on a Wednesday do not approve of Bush's handling of Iraq." People with jobs and busy lives don't talk to pollsters.
Also, given that there are over 300 million Americans and they interview maybe a thousand for a poll, the results are, well, statistically useless.
Also statistically useless is when they say the President's approval rating "drops" from 51% to 49% - because a change of 2% is well within the margin of error. Reporters treat the 50% approval rating mark in polls like it's something that actually exists.