# Keep changed survey questions out of an unqualified trend

Record a wording and recall-period change before comparing survey waves as the same measure.

If nothing was unclear or difficult, say so. Skip questions about steps you did not experience; use not applicable where appropriate.

## 1. Which dates or experiences does this question ask you to consider?

Answer: ____________________

## 2. What would you leave out under the seven-day wording?

Answer: ____________________

## 3. What would the thirty-day wording include that the shorter version would not?

Answer: ____________________

## 4. Which time-period explanation would make your answer less ambiguous?

Answer: ____________________

## Author notes / illustrative keys where applicable

1. Test the actual recall boundary without supplying an answer.
2. Identify the shorter window’s exclusions.
3. Expose the changed experience set rather than attributing a trend.
4. Find a wording repair to retest with the chosen period.

Interpretation caution: Changed recall wording changes the instrument. Sample percentages under different versions are not an unchanged-measure trend.


## Illustrative finding and follow-up

Observation: The two invented percentages refer to different recall windows.

Action to test: Create a version note and choose the next wave’s question before collection.

Follow-up: Pilot the chosen recall window and preserve its exact wording with later cohort counts.


## Collection protocol

When and whom to ask: Before the next wave, show both wording versions with fabricated date examples.

Decision owner: The survey owner approves the recall window; the reporting reviewer records the instrument break.

Follow-up: Pilot the chosen recall window and preserve its exact wording with later cohort counts.

Response handling: Do not ask volunteers to disclose private incidents to demonstrate a date window; fabricated examples suffice.

## Decision workshop

Synthetic examples only. Manual worksheet; no automatic import or product writes.

### Synthetic pilot and reporting records

Wave | Recall period | Synthetic favourable answers | Instrument
--- | --- | --- | ---
A | Last 7 days | 8/10 | Version A
B | Last 30 days | 6/10 | Version B

### Version changes

Boundary | Draft | Revised
--- | --- | ---
Recall | Changed without disclosure | Versioned 7-day and 30-day windows
Trend | Same measure assumed | Instrument break recorded

### Draft or earlier version (review its limits)

Satisfaction declined from 80% to 60% on the same measure.

### Revised example

The synthetic samples are 8/10 and 6/10 under different recall windows. Mark the wording-version break; do not describe the difference as an unchanged-measure trend.

### Decision situations

- Wording or recall changed: Keep versions separate and document why comparison is limited.
- Instrument is unchanged: Review sample and collection context before interpreting movement.
- Earlier wording cannot be recovered: Hold the same-measure claim.

### Method references

- Pew Research Center question design: https://www.pewresearch.org/writing-survey-questions/ — Wording and response-option consistency.