I have a small doubt here...
is the category name wrong in the raw data or the category assigned to the question is wrong?
if the category name is wrong then you can edit them in alias of that dimension as it would show only distinct values.
if the category assignment is wrong then blending is the only option In my knowledge.
Well, the original file has 5 different categories.
1 additional category needs to be added, so a number of different questions will be assigned a new category. Looks like blending might be the only option here?