Overview

This article describes a data-integrity limitation that can occur in Discover XI / EFS when you combine a Standard Matrix 1 (Type 311) question with Most/Least (multi-scale / multi-dimension) behavior inside Loops, while Randomization and/or item-level visibility logic (for example, hiding conditions) is also in use.

In affected setups, the survey UI can look correct during completion, but the export can contain an incorrect stored value for the Least selection.

Typical symptom

• Your Most result variable exports as expected, but your Least result variable does not match what was selected.
• There is usually no error message; the issue is only visible after exporting results.

Affected configuration (all/most apply)

• The question is Standard Matrix 1 (Type 311) (matrix questions have Scale options and Dimensions).
• The question uses a Most/Least style with multiple scales/dimensions (two stored result variables).
• The question is placed inside a Loop (loop pages and variables are calculated dynamically for each respondent).
• The set/order of visible dimensions can change per loop cycle (for example via dynamic answers and/or item-level hiding conditions).
• Randomization is enabled for one or more relevant answer options (or answer groups).

Solution

1) Confirm you are hitting this limitation

1. In the Questionnaire editor, locate the Most/Least matrix question that is inside the loop.
2. Run several test completes across multiple loop cycles and write down exactly what you selected for Most and Least (screenshots help).
3. Export the results via Projects → {Selected project} → Export.
4. Compare the exported values for the Most vs Least variables against your notes. If the Least values do not match, proceed with the workarounds below.

2) Workaround A (recommended): disable randomization for the affected matrix answer options

1. Edit the affected Standard Matrix 1 (Type 311) question.
2. Scroll to the Answer options section.
3. Find the Randomization column and clear the randomization checkboxes for the relevant answer options. If you are using the Grouping extension, disable group randomization as well.
4. Save changes and re-test (repeat the confirmation steps above).

Important notes

• In the question detail view, the Answer options table includes a Randomization column. When enabled, the order of the selected answer options is changed randomly so respondents see different arrangements.
• Loops have special behavior and limitations because loop pages and variables are calculated dynamically.

3) Workaround B: keep randomization, but redesign the question

• If you must randomize to reduce bias, avoid using a single multi-scale Most/Least matrix inside a loop.
• Instead, redesign as separate questions (one question for Most and one question for Least). Then re-test and export.

4) Workaround C: simplify item-level visibility logic when randomization is required

• If you use hiding conditions at the answer-option level, test carefully when randomization is enabled. If possible, prefer a stable (non-changing) set of visible options within the randomized question.

5) What to do about data already collected

• If your export does not match what was selected on-screen, treat the affected Least variable as potentially unreliable.
• If you need help assessing whether any recovery is possible for your specific survey design, contact Support and provide: project ID, question ID(s), whether randomization is enabled (and where), whether the question is inside a loop, and a small set of test completes with screenshots + matching export.

<supportagent>

Internal context: This limitation was escalated to Engineering and tracked under GHI 16663. Investigation indicated the issue can be triggered by combining multi-scale Type 311 + randomization + loop with dynamic list and (often) item-level hiding, producing incorrect stored/exported values for the Least scale.

</supportagent>

References

1. Question types
2. Questions (Detail view)
3. Questionnaire editor
4. Export
5. What you need to know about scales questions