I looked for information about the poll but the page you linked and the linked page within the relevant section both appear to lack that information. The 20% is an approximation, apparently.
It seems the question was:
Should "the executive order that freed all slaves in the states that were in rebellion against the federal government" have been made, or something similar. Was that worded badly on purpose, it appears to say "is freeing people who are rebelling against the government good"?
I find the statistic hard to believe, such a claim needs careful support which is lacking. The details that are given suggest an opaque question.
Such questions need follow-up questions to gauge understanding and depth of sentiment. Anyone who asks that and doesn't then ask 'so should the slaves have been freed at all?' as a follow-up is most likely being deceptive.