When the hate preachers show up on campus with a bullhorn and try to tell everyone that they deserve to burn in hell, they don't want the rest of the world to tell them to STFU.

These open questions are always kinda crap to draw conclusions from because we don't know how the question was interpreted by the people answering. In this case we also don't know the sample size of the groups. Could also be a multitude of other variables at play like location because there are few places where you can find all these religions while getting a good sample size and controlling for other variables like income, education, age etc.

And does it apply only to verbal, podium speech, or also to written books and speech by people in [drag] costume.