This is the reason we should be very sceptical of opinion polls where people talk about values, the results are almost entirely dependant on question wording. As such, they are generally meaningless and are best ignored. We love polls because they have nice % figures that make us think they are 'scientific', but they are mostly worthless.
For example, do you believe that colleges should be able to punish someone for calling a black student a *n word*? If so, you agree with the latter. Or does free speech restriction make you think of 'snowflakes' and 'safe spaces', in which case you might be against restrictions.
Even with precise questions your wording makes a huge difference, with vague, fuzzy statements it means next to nothing. Also it's not just about the question, but the context. You could easily make this 57% change significantly based on the wording of the previous questions and how these 'prime' the responders.
The actual questions were listed at the end of the survey. They seem pretty clean to me.