This has now happened, so I think our team will need to manage as best we can with the data collected, but I’m keen to explore how we got to this point.
We ran two community surveys in Kobo, one of which is fine and complete. The second, only half the desired questions were asked. We only noticed this after data collection has happened.
In this specific survey, let’s say there’s 40 questions. I’ve looked at the downloaded spreadsheet of all data collected. Questions 1 to 20, they look fine. Questions 21-40, they are in the spreadsheet but columns underneath the column header are empty.
I went back to look at the multiple draft versions in form history. Those questions appear to have been lost/deleted in some way during the revisions, such that the final version did not have 21-40 in the survey and thus were not asked.
A colleague who developed the other (complete) survey, he reported several issues within Kobo where questions would be deleted without any obvious reason and he had to be vigilant in replacing them. We are uncertain as to why that was happening.
Clearly lessons for us to learn about how to avoid this on future occasions. But, for now, I guess my questions here are
- Are these known issues (e.g. it being too easy to accidentally delete questions within Kobo without the user obviously noticing?)
- Given questions 21-40 were not asked and seemingly deleted during the draft version, why are they appearing as headers and empty columns in the final downloaded spreadsheet of collected data?
Any input/insight very much appreciated.