There are many ways to introduce bias into a poll. One is to use a small or unrepresentative sample. Wording the question can swing results wildly. Rephrasing the actual poll question for a news story is propaganda (and as it turns out is very relevant in this situation). Another is to survey anyone when it's more appropriate to survey registered voters. Geography is critical. A survey which relies on Berkeley, California will not match the results when polling Salt Lake City, Utah. Most importantly on political issues, the sample must reflect the partisan distribution in the entire population.
Or, you can just pick a number and call people. As it turns out, that's the 'scientific method' the USA Today / CNN / Gallup Poll relies on.
Polls listed chronologically. All data are from nationwide surveys of Americans 18 & older.
CNN/USA Today/Gallup Poll. Nov. 11-13, 2005. Adults nationwide.
"Do you approve or disapprove of the way George W. Bush is handling the situation in Iraq?"
N=1,006, MoE ± 3
Approve / Disapprove / Unsure
35% / 63% / 2%
Wait a minute. This poll doesn't even address whether we should be in Iraq. The negative opinion could be people who want to put more troops in Iraq to invade Syria or Iran. Who knows? The "disapprove" is lumped into one category. Whether you are more or less hawkish than GWB, you are in the same category.
Sometimes what they don't tell you is more important than what they tell you. Look at the poll below the poll being used to claim Americans want to cut and run.
As it turns out, American opinion on a question more related to withdrawal has been remarkably stable for months. There is no question which directly addresses withdrawl. This is the closest they get. The worst result came in September when the "did not make a mistake" registered 39%.
"In view of the developments since we first sent our troops to Iraq, do you think the United States made a mistake in sending troops to Iraq, or not?"
Form A (N=491, MoE ± 5)
Made a Mistake / Did Not Make a Mistake / Unsure
54 / 45 / 1
Given the lack of any information regarding breakdown of party, geography, registered or not, and sample size (among other things) the nation is split right down the middle, even if you accept their margin of error (which I wouldn't).
Now consider the phrasing of the question.
"make a mistake in sending troops to Iraq."
What percentage of people will consider any mistake in the conduct of the war to be "a mistake in sending troops to Iraq"?
If even five percent are fooled by the wording of the question, the results reverse.
What does the phrase "in view of the developments since" do to the question?
It becomes a referendum on all the horrible mosque and car bombings, troop and civilian deaths.
Is the phrase necessary in the question if the issue is withdrawal? No.
This is the question without forcing the surveyed to consider recent bloodshed:
do you think the United States made a mistake in sending troops to Iraq
What if we remove the word "in." "In" suggests, again, looking at the conduct of the war, not the war itself.
do you think the United States made a mistake sending troops to Iraq
That's the real issue, approaching a question on withdrawal, though not reaching it, isn't it? Now ask yourself why the question is so wordy. Perhaps I've already answered it.
Next time you hear the phrase "war growing increasingly unpopular with the American public" over the next few days recall this is the polling data the mainstream media is using to reach that conclusion.
UPDATE: It has come to my attention that I can't spell withdrawal. Apparently I think it's a word describing someone who can't shake a southern accent.