And the question was designed to get the answers they wanted.
The question was:
Pollsters know that most people answering a poll know little or nothing about the topic, so they will include an explanation of what they are asking. That's how they can get the answers they want.
As you may know, the United States and five other countries entered an agreement with Iran aimed at preventing Iran from developing nuclear weapons. Do you think the U.S. should or should not withdraw from that agreement?
The explanation is "the United States and five other countries entered an agreement with Iran aimed at preventing Iran from developing nuclear weapons." If that is the only thing in your mind, why wouldn't you support it?
A similar poll, widely reported, from Politico has the exact same methodological error, and a similar result (56-26.) The question there was:
As you may know, the United States and other countries made a deal in 2015 to lift economic sanctions against Iran in exchange for Iran agreeing not to manufacture nuclear weapons. This agreement is sometimes referred to as the Iran nuclear deal. Knowing this, do you support orAgain, no reason given to oppose something that sounds so great on paper, and critics have no say in the wording of the poll.
oppose this agreement?
A fair question would have either been "Do you think the US should or should not withdraw from the Iran nuclear deal" without any explanation whatsoever, to reveal that most people really don't know one way or another - or it should have added a single sentence: "Critics say that the deal only delays Iran's ability to build a nuclear bomb by a few years and it should be renegotiated."
The results from that poll would be radically different. But CNN and Politico doesn't want people to answer that question.
I believe that the ubiquitous wording that surveys use of "As you may know..." discourages people from answering honestly that they don't know about the topic, and it prompts them subconsciously to make a guess that would make them appear to be better informed than they are. Only people who are intimately aware of the details and who oppose the deal are going to say they are against it; the "don't knows" are prodded to lean towards support by the very wording.
These questions show that polls are now used not to inform, but to influence. It is a scandal that no one wants to talk about. And the media loves to report on polls that fit their biases.