Why you can't trust polls

A poll is only as good as the questions.
You can control the results by the way you word the questions.

That's also why I'm against referendums.

I must credit IMRA for this one, the perfect example of a convoluted question to drag out the results the pollster wants.

The type of question to get the most accurate results are simple. People just can't follow the tangents. I know from my experience answering those telephone calls. Here's the introduction of the IMRA article:

Dr. Aaron Lerner - IMRA:
You really have to give credit to Dr. Mina Tzemach.
When she discussed how to put together this poll with the folks at Yedioth Ahronoth they were no doubt struggling to come up with a way to best support their political agenda (= withdrawal).
So how about this question:

"Could Minister Ehud Olmert achieve a public mandate allowing him to change Jerusalem's status as part of a permanent peace agreement with the Palestinians if 80% of the ministers were behind it?"

Try thinking that one through while the kids are hollering for dinner and you want to sound PC to the anonymous pollster on the line!