One can do polls to establish eg. "do you believe this story", "did you support Clinton before this", "did this change your voting preference"... While no poll

One can do polls to establish eg. "do you believe this story", "did you support Clinton before this", "did this change your voting preference"... While no poll is 100% accurate, as long as you make it statistically significant and introduce a good control (say multiple fake and real stories, with some of them positive too), you'll get good signal and ability to say exactly if it did affect the election.