- Published
- Updated
A visible language option does not demonstrate that a participant received the accepted questionnaire version. Similar screens may contain different wording or response options, changing the meaning of the resulting data despite identical question numbers.
Preserve the meaning of data
Manage instrument identity, language, text version and applicability together. Link technical language codes to the language materials accepted for the study. Translation suitability requires appropriate expertise, while correct deployment mapping is a separate technical check. Completion of one activity does not establish that the other has been performed correctly.
Worked example
For example, a participant selects Turkish but the application displays outdated Turkish response options while English is current. The team compares each deployed language with its accepted source package and tests how language-version context is stored with responses. It identifies the affected population and refers any analysis implications to the relevant scientific reviewers.
Review checks
- Compare each language screen with its accepted source version.
- Test mappings after language switching and application updates.
- Check which language and version are stored with each response.
Acceptance and boundaries
Acceptance means a response can be traced to the actual wording displayed to the participant. Technical mapping alone does not establish measurement equivalence of a translation. If a translation is missing, do not assume silent fallback to another language is appropriate. Address that behaviour in the study arrangements rather than leaving users with an ambiguous selection.
Frequently asked questions
- Is an identical question number sufficient?
- No. Wording or response options may differ, so language and version relationships also matter.
- Does a later language preference change earlier responses?
- Preserve their original language-version context rather than relabelling them using the current preference.