I'd agree with you conjecture, and would totally imagine myself reverting opinions on past decisions depending on the degree of interest I have on the matter.
The 53% number in the article was about reversing position on at least one question. It should mean most of the subject retracted their aledged answer on the other questions, probably the one or two were the faked answer was embraced are questions with no strong positionning.
The article doesn't give details on the strength of the reversed opinions, although the survey sheet seems to have a 7 or 8 point scale for each question. I'd guess this detail was brished out for dramatic effect?