Practical guide

Change a survey question without rewriting the past

Last materially reviewed 2026-10-01

Quick answerPreserve the version that produced each answer and mark where comparisons stop being like-for-like.
What to know

Treat wording as part of the data

An answer has meaning in relation to its question and options. If “Was collection clear?” becomes “Was collection easy and quick?”, the same numeric rating no longer represents the same judgment. Retain the old wording with its collection period rather than replacing the label in a report and pretending nothing changed.

What to know

A fictional revision record

Version one asks whether instructions were clear. Review identifies that customers interpret “instructions” differently. Version two explicitly names the collection email and separates entrance directions from opening hours. Record the change and why it was made. The new results may be more useful, but they do not automatically continue the old series.

What to know

Check the active destination

Confirm which version is published, where its links are used and whether an old embedded copy remains available. Do not assume that changing an editor updates every shared artifact. Test only in an authorized way and preserve a recovery route for the previous version where appropriate.

What to know

Report the boundary

A later report should name the question version and explain any material difference. Avoid joining incompatible answers to increase the apparent sample size. If historical reconstruction is impossible, state that limitation instead of inventing the wording that probably appeared.

What to know

A minimal version ledger

Keep a version label, exact wording and options, effective date, reason for change and the collection destinations affected. A fictional entry can state that version two separates entrance directions from opening hours. That record helps explain why two answer sets should not be merged without qualification. Do not rely only on an editor’s latest state to reconstruct history. If an old version cannot be recovered, make that gap explicit and limit any comparison accordingly.

Continue when useful

Next: Compare periods

A changed score may reflect a changed question, group or collection method.

Open Compare periods →

Sources used for this page

These records support the facts and comparisons above. Merchant-controlled records are labelled so you can separate product claims from independent evidence.

  1. AAPOR survey-research best practices — Standards and certification reference · aapor.org · Publisher independence not verified · checked 2026-10-01
  2. Yay! Forms response export — Merchant documentation · yayforms.com · Merchant-controlled · checked 2026-10-01