If a figure is inserted before question text, then in the csv of the quiz data, the Q text field is blank. Now that I know this, I can put text in front of the figure to make sure this doesn't happen and I can identify these questions in the downloaded test data. But it seems like this problem could be easily fixed.
Also, question text edits don't back-update into the data from previous tests. (For example, if I run a quiz in which two questions having identical text are drawn randomly from a pool and I later insert question-specific codes into the text so they can be identified.) I can see where we might not want to do this if we want the data to indicate exactly what the student saw. But the option should be there. Now that I know this, of course I insert invisible question codes into each question.
I shouldn't have to do these workarounds because the LMS should be able to provide unique identifiers for questions in the downloaded data even if question text is identical and questions are drawn randomly from a pool. In the report within Brightspace, it has the aggregate data for each question, but if we are going to analyze individual student performance, we need the information in the downloaded data as well.
