Did attitudes reverse, or did students that had no real opinion on some topics make up answers to them, and when they re-read the questions and they were different, accepted that because they didn't feel strongly either way?
Or, did students have some opinion on the topic, but when faced with a reworded question, be embarrassed that they must have misread it initially - which means they had not done what they were asked to do, which is answer the survey properly? There can be a lot of pressure to participate correctly in psychological experiments, I know that in my undergrad you had a requirement to do some amount of experiments to graduate (and non-psych students would be paid, which also puts pressure on you to do your role correctly). To say "sorry, I wasted your time - I misread these questions" would not be easy, much simpler to just make up new answers and leave.
I find it hard to accept their interpretation that attitudes actually reversed. Yes, some experiments - like Milgram's - show that people can do surprising things, that we would not expect in advance, and would even say "I would never do that", but likely we would. But this isn't the same - in Milgram's study, behavior was all that mattered. In this one, it is the interpretation of changing attitudes. I agree it is likely people will voice different attitudes, but I doubt it is because they actually changed them, instead it is either social pressure to not admit you failed at your task to read the questions and answer them honestly, and/or that the student doesn't care either way about the topic.