Keep a filtered fieldwork choice tied to the reference list that allowed it

State the allowed relationship independently of the form

Define which downstream choices belong under each upstream category. Use stable nonpersonal identifiers and meaningful labels for the reference items. A label may be repeated legitimately in different contexts, so visible text alone may not establish the selected item. Record the source and revision of the approved reference list without assuming that a familiar filename contains the same data.

Prepare a small acceptance set containing a typical branch, a branch with few choices, an empty allowed set and similar labels with different meanings. Identify what should happen when the reference list does not contain a legitimate reported item. That may require an approved other response or escalation, but it should not force the collector to choose an inaccurate category simply to continue.

Compare available choices and saved values together

ODK describes external CSV choice lists with separate stored values and displayed labels; additional columns can support choice filters. This supports checking both the visible set and the recorded identity, rather than treating menu appearance as the complete acceptance test.

Run each fixture on the intended collection route and compare the offered options with the approved relationship. Select an item and inspect the resulting test record. Confirm that the stored value represents the chosen reference item, not merely its position or an ambiguous label. Keep the fixture’s expected result independent of the form logic so the same faulty filter does not generate both the test and its supposed answer.

Change earlier answers and inspect the resulting state

In a synthetic record, select a downstream item and then change the upstream category. Observe what the deployed instrument does with the earlier selection. Does the collector receive a clear opportunity to correct it, and does the resulting record satisfy the approved relationship? Do not assume that hiding a formerly available choice necessarily clears its stored value; verify the actual behavior before making that claim.

Also test a reference item that is retired or relabeled in an approved new list. Decide with the research and system owners how existing work should retain its original meaning. A new label should not silently make an older response appear to refer to a different item. Keep reference-list changes separate from question-definition changes so a defect can be traced to the correct owner.

Release a demonstrated relationship, not just a valid file

Give the implementation owner the accepted fixtures, reference revision and observed output. Record any permitted exceptions and unresolved branches. A syntactically valid CSV can still contain the wrong associations, duplicate values or misleading labels. If a list update fixes one branch, repeat affected neighboring cases rather than assuming the change is isolated because only one row was edited.

Carry the reference context into the research handover sufficiently to interpret the collected values. Do not expand this into unnecessary personal linkage or a new participant roster. Reopen acceptance when a filter condition, value mapping or list revision changes. The result should establish that each collected selection was available under the intended relationship, while preserving uncertainty when the instrument cannot faithfully represent a reported business task.

Make the reasoning concrete.

A fictional workplace-tools study asks a respondent to choose a process family and then a task. Two families contain an option labeled review, but the saved codes describe different activities. A tester changes the family after choosing the first review option.

The team chooses relationship testing and gives the form owner a reproducible fixture rather than silently correcting the output in a spreadsheet.

Take the questions into your next brief.

Use these prompts in your own notes. Nothing is submitted from this page.

1. Relationship

   Which downstream values are allowed for each upstream answer?

2. Reference

   Which approved dataset revision supplies those choices?

3. Mapping

   Does the saved value match the item the tester selected?

4. Change

   What happens after an upstream answer or list item changes?

5. Exception

   How can a legitimate unlisted task be recorded without substitution?

Watch for the wrong conclusion.

• Treating a populated menu as proof of correct filtering.
• Assuming a hidden old choice has been cleared.
• Replacing an unknown task with the nearest convenient category.

Sources & scope

Workflows, worksheets and fictional examples are original editorial analysis. Sources support only specifically attributed principles. Platform guidance can change; legal requirements depend on jurisdiction and context. No measured results or named human reviewer are claimed.

• ODK Docs: Form Datasets →Primary collection reference-data documentation · Supports: Supports distinct stored values, displayed labels and filter columns in ODK external choice lists. The acceptance fixtures are original analysis.