Survey Shenanigans

This survey seems to be really popular with the far right, seemingly proving that the American public is really against surrendering in Iraq.

I studied statistics extensively in college, and I took an entire 3-credit course on how to (and how not to) ask questions on a survey.

The questions in this survey are designed to elicit only one type of answer, which may (or may not) be contrary to what the person responding actually believes.

To put it in stark contrast as a way of example, let's rephrase this survey in a similar but more extreme context: Your mother is in the hospital dying of terminal cancer, and has only weeks to live. Question: "How important is it to you that your mother beats her cancer?" (Your choices are anywhere from "very important" to "not at all important".)

Next question: "How hopeful are you that your mother will be cured of her cancer?"

Now, do you see how statistics can be used (by asking the... ahem... right questions) to show any result that is desired? Obviously the question should not be one of "hope", and the question should not be one of "importance". Of course we "hope". Of course it is "important".

So, when you hear people touting this particular survey as some great sign that Americans don't want to lose the war in Iraq, tell them "No shit. That's what they want... but the more pertinent question is: What do they expect?"