I was collecting user ratings in the following order using custom values – Great = 1, OK = .5 and Poor = 0. After collecting a bunch of responses, I wanted to allow for more nuance so I modified the range of answers to Great = 1, Good = .75, OK = .5, Weak = .25 and Poor = 0. As you can see, the original answers are preserved with their original values, although OK and Poor are no longer in their original ordinal locations, as the new values, Good and Weak are interleaved.
When I look at the responses from before I made my change, I am seeing that all answers lie in range of Great, Good and OK, that is ordinals 1, 2 and 3. But this is flat out wrong! The custom values which were (should have been?) saved should properly align all prior recorded answers to Great, OK and Poor. Is this not the way it works? Are you saying that if I modify a question I must preserve ordinality of pre-existing answers, otherwise I am going to be scrambling prior submitted responses?
Hmmm, not knowing exactly the manner in which you edited these answer options, it’s difficult to say what exactly happened here.
I know SurveyGizmo has some specific guidelines for editing surveys during data collection:
There are certain questions types that should not be edited after data collection. I’m not sure if any of this helps – you may need to reach out to SurveyGizmo directly and have them take a look.