# Design a short feedback survey without writing the desired answer

Complete original constructed exercise. No real people, observations, reviewers, adopted policies, studies or business results.

Choose one learning decision, define who qualifies, and draft neutral questions with explicit response and skip meanings. Pretest with appropriate voluntary participants before actual use, then retain the questionnaire version and per-question denominators. An opt-in set describes its respondents; it is not automatically representative demand or a population preference estimate.

## Choose a learning question and appropriate participation

A useful objective might be deciding which source-checking instruction needs clarification for people who recently prepared a reporting handoff. AAPOR recommends choosing a method around the actual question, not producing predetermined results. Existing records or a conversation may be more suitable than another questionnaire.

Before real collection, explain the actual owner, purpose, voluntary participation, use, access, retention and withdrawal/skip process in an accessible format. Collect no unnecessary names or private client material. Do not promise anonymity unless the actual system and context support that statement. This guide neither recruits nor contacts anyone.

## Use a complete neutral three-item draft

Constructed draft introduction: “This optional questionnaire asks about a recent reporting handoff so we can choose which instruction to explain next. You may skip a question. Please include no names or private documents.” Supply the actual agreed data/consent information separately before use; the script alone is not a privacy process or validated questionnaire.

Original items: Q1, “In the past 30 days, did you prepare a reporting handoff?” with Yes/No/Cannot recall. Q2 for the intended eligible experience, “For your most recent handoff, how easy or difficult was identifying the data source?” with Very difficult/Somewhat difficult/Neither/Somewhat easy/Very easy, plus Not applicable/Cannot assess. Q3, optional: “What, if anything, made finding the source difficult?” No difficulty or explanation is presumed.

## Pretest the meaning and preserve the instrument version

Check whether intended respondents understand handoff, source and the reference period consistently. AAPOR recommends pretesting and piloting procedures, including programming, cleaning and reporting. An internal string-length check cannot replace a real interpretation check; no such participant pretest is claimed for this draft.

Retain exact question wording, choices, order, routing, language and mode. Keep ordinal choices in logical order and do not treat Not applicable as a low score. If wording or collection mode changes between rounds, record the change rather than attribute every resulting difference to audience opinion.

## Inspect routing and interpretation with a complete plan

Original constructed addition to the unchanged three-item draft: only Q1 Yes leads to Q2 and then optional Q3. Q3 remains optional after a scale answer, Not applicable or skipped Q2. Q1 No, Cannot recall or skipped ends this experience branch. Skipping is not recoded as No or low satisfaction; eligibility may remain unassessed.

Constructed pretest introduction: “Please first answer the question as you understand it. Afterwards I will ask how you arrived at your answer; there is no preferred response here.” Use the neutral probes below only after the initial answer. Do not supply the definition or correct option. Actual voluntary information, permission and secure handling must exist before a real rehearsal.

Record exact question/language version, expected route, actual observed route, the person’s own explanation and unresolved misunderstanding. Keep only necessary notes without names or private documents. Technical routing checks do not replace real interpretation research. These six declared paths and three probes are constructed; no pretest, additional respondent or change to the separate registered comparison study is claimed.

| Constructed Q1 state | Q2 | Next route |
| --- | --- | --- |
| Yes | Scale answered | Q3 optional |
| Yes | Not applicable | Q3 optional |
| Yes | Skipped | Q3 optional |
| No | Not shown | End |
| Cannot recall | Not shown | End |
| Skipped | Not shown | End; eligibility unassessed |

1. What does reporting handoff mean to you in this question?
2. Which event were you thinking about when answering?
3. How did you choose this response option?

## Resolve three constructed survey summaries

Q1 — Constructed list:20 listed eligible invitees and 8 submissions. Decision: report the declared list/submission counts and known collection scope; a formal response-rate label also needs appropriate disposition information. Outcome: not 8 representative readers or an estimate of all LinkedIn users merely because the fraction is 40%.

Q2 — Constructed eight submissions: six answer Q2 on the scale with counts 1/1/1/2/1; one marks Not applicable and one skips. Three of six scale responses are in the two easy categories:50%. Decision: preserve the full distribution and all bases. Outcome: not 50% of all 20 invitees and no forced zero for the non-scale responses.

Q3 — Constructed wording change: an earlier round asked how easy or difficult the task was; a later one asks “How much did our excellent guide help?” Decision: hold the trend comparison and retain both instrument versions. Outcome: a changed measurement, not proven improved satisfaction or actual preference over another guide.

## Report the sample and missingness honestly

Show eligibility, collection mode/dates, per-item counts, partial responses, exclusions and exact wording. Open text needs a declared coding approach if summarized numerically; one comment can contain several ideas and is not several people. Keep unnecessary identifying quotes out of the public result.

AAPOR’s disclosure standards require explicit methods and model-assumption treatment for precision claims in nonprobability samples. A simple binomial interval cannot certify representative opinion from a self-selected set. Keep task performance, stated ease, preference and purchase as separate outcomes; none can be inferred from another.

## Keep a versioned draft and reporting plan

The worksheet keeps purpose, eligibility, actual information/consent process, complete questions, routing, pretest status, item denominators, coding and allowed conclusion. The packet contains the full draft for editing privately. It neither publishes a survey nor records responses.

Primary guidance was read on 1 November 2026. Scripts and all counts are constructed. No participant, recruitment, survey, message or preference was collected. This methodological draft does not replace or modify the separate registered comparative reader study.

## Primary sources

- [AAPOR: survey purpose, wording, pretesting and reporting](https://aapor.org/standards-and-ethics/best-practices/)
- [AAPOR: sample, instrument and precision disclosure](https://aapor.org/standards-and-ethics/disclosure-standards/)