Let's say we have 20 persons and we ask them a similar question. Our results are as follows:

8 dislike apples

8 dislike pork

8 dislike chicken, and

8 dislike beef.

Following your logic we can conclude that 24/20 or 120% of the people asked dislike meat. The data aren't necesarilly wrong, but the conclusion is obviously untrue, hence there must be something wrong with the reasoning. We can't just add the percentages together since there might be overlap between the members of the group that dislikes pork and members of the group that dislikes chicken. This can illustrated by an extreme example:

10 people are asked

40% disapprove of atheists

20% disapprove of jews

20% disapprove of muslims

20% disapprove of christians

Let's say that we also know that

2 persons disapprove of both atheists and religious (they disapprove of all) and

2 disapprove only of atheists

That would allow us to conclude that 6 don't care. Please check to see that this is consistent with the the percentages.

We know that there are 10 people. 4 of them will disapprove of atheists, while only 2 of them will disapprove of religions. So if we pick a person at random we would find a person that disapproves of atheists 40% of the time. We will not find a person that disapproves of religion 60% of the time. On the contrary, since only two persons account for all the disapproving of the religious we will only see find these persons 20% of the time.

So since we don't know anything about the overlap between the groups that disapprove muslims, jews and christians you simply can not add the percentages together like that.