... They manipulate the questions to get the result they want ...
That is SO true... poll results are worthless unless you also analyze both the sample set and the questions asked. Simple (but exaggerated) example - let's say I make ice cream and want to reach customers who are trying to eat a little healthier. I pay a company to take a poll so I can use the results in national advertising. The pollsters come up with the following question:
Which of the following choices do you think is the healthiest:
a) double meat cheeseburger,
b) extra large order of french fries,
c) 3 ounce serving of ice cream
When the results come in and I find that 80% of the respondents chose (c) then I can say in my advertising, "According to a recent national survey, 4 out of 5 people think eating ice cream is healthy." The numbers back it up but it hides the fact that the question was rigged.
Dishonest pollsters should be tarred and feathered.