Handling submission of partially completed surveys in KoBo collect

We recently conducted a large-scale household survey involving 10,000 samples using KoBo Collect. During this process, we encountered a situation where some surveys were only partially completed and are currently saved as drafts. This partial completion occurred because some respondents chose to discontinue their participation midway through the survey. Due to numerous mandatory questions, we are unable to submit these surveys in their current, incomplete state. At the survey’s conclusion, there’s a question about survey status with three options: i) completed, ii) partially completed, and iii) rejected. We are able to submit surveys marked as ‘rejected’ because we have a consent question at the survey’s start. If a respondent does not consent, we can skip all sections and submit the survey. However, for surveys marked as ‘partially completed,’ respondents initially consented but at mid-way chose to withdraw. Is there a method for submitting these partially completed surveys?

Your form doesn’t allow status “rejected” together with consent (yes) - right?

Are there repeats among the incomplete mandatory questions?

  • Otherwise, as a (uncomfortable) workaround, you might fill the mandatory questions with special values, set a “marker” value at one place for these cases and submit them. On server level, you could use bulk edit (and filter) to empty all these values.
  • But if mandatory repeats are missing too, you could only create/deploy an adapted form version and edit the cases individually on server level.

If you have few incomplete cases on the local devices, you might even create/deploy an adapted form version and copy/reenter the values manually into the new form.

Hopefully, the core team or the community knows other, smarter solutions, e.g. with bulk upload. (@Kal_Lam et al.). See Search results for 'upload submissions to new form version' - KoboToolbox Community Forum.

Methodological hint: You would no more have a defined numbers of (mandatory) answers which would create challenges for cross-tabulations and analytics. Sometimes, it is better to remove those incomplete cases.

Always include incompleteness and (partly) drop-out issues in your systematic pre-testing, before data collection, please.

@jonaed, maybe you will require 2 different approaches to handle these type of scenarios:

  • The first approach is to use the custom logic instead of the mandatory response. You should be able to learn more about the same through these posts discussed previously:

And this one too …


Thank you so much @Kal_Lam @wroos

1 Like