Feedback and survey design
Reviewing who answered your event survey
Your survey came back with a good number of responses, and you want to know how far those responses speak for everyone who attended.
Opens WhatsApp with a draft you can edit before sending. Nothing is sent automatically.
The short answer
Compare who responded with who attended, using only facts you already hold, such as registration type, session attended or region. If the two groups look different, say what that could mean for the findings and report the limit next to the result.
A high response count does not remove bias. It is the pattern of who is missing that tells you how far to trust a score.
The kinds of bias worth checking
| Type | What it looks like at an event | What to look at |
|---|---|---|
| Non-response | Quiet delegates and those who left early do not answer. | Response rate by registration type and by session. |
| Self-selection | People with strong views, good or bad, answer most readily. | Spread of scores and tone of comments. |
| Timing | A survey sent a week later reaches fewer people and memories fade. | Responses by day sent and by reminder. |
| Access and language | Some groups cannot easily use the link or the language. See language equivalence. | Response by language and by channel. |
| Question order and wording | Early questions shape later answers. | Order of items and any leading phrases. |
A review sequence
- List who attended, in groups you can count from records, such as member or non-member, speaker, sponsor representative, on-site or online.
- Count responses in the same groups, without asking for extra personal details.
- Put the two sets side by side and note any group that is much smaller among respondents than among attendees.
- Read the comments from the groups you did hear from and note whose views might be missing.
- Write one sentence per finding that states the limit, then decide what extra evidence, if any, you can gather to fill the gap.
Write the limits next to the finding
A statement like most respondents rated the programme as useful is incomplete without who responded. Add how many responded, from which groups, and what you cannot say about the rest.
Do not weight, extrapolate or give a precise-looking percentage for the whole audience unless someone with survey methods experience has agreed the approach.
Ways to hear from the missing groups
- A short, targeted follow-up to a group that is under-represented, with its own purpose stated.
- A few brief conversations or a coded sample of open comments from people who did not fill in the form.
- Evidence that is not self-reported, such as attendance at later sessions or repeat registration, labelled as such.
- A plain note in the report that this group was not heard.
What to do with the review
Carry the limits into the report and into your action priorities, so a point raised by a small group does not drive a decision it cannot support.
Keep the review sheet with the survey file so next year's team can improve the response mix.
Worked example · Fictional example
A seminar series where speakers answered most
Fictional organisation and figures, written to show how the review reads.
Persatuan Fiktif Perunding Pendidikan Dewasa held a one-day seminar for 160 registered participants and received 48 survey responses. The secretariat compares attendance records with responses by group.
Speakers and committee members make up 12 of the 48 responses, though they were a small share of attendance. Delegates who left before the afternoon session sent almost none. The report therefore gives the number of responses by group, describes the scores as the views of those who stayed to the end, and adds an action to ask early leavers one short question at registration next time.
| Group | Attended | Responded | Note |
|---|---|---|---|
| Members | 92 | 20 | Responded less often than their share of attendance |
| Non-members | 36 | 10 | Small group, read with care |
| Speakers and committee | 32 | 18 | Over-represented among responses |
Use this yourself
Respondent bias review worksheet
Fill in one row per group you can count from your records. Use real figures from your own event.
- Total attended and total responded, with the date range of the survey:
- Was the survey sent in more than one language or channel? Response by each:
- Which group is missing or very small:
- Did early leavers or last-day sessions get less coverage?
- One sentence that states who the findings do and do not speak for:
- Extra evidence you will gather to fill the biggest gap:
| Group | Attended | Responded | Over or under-represented? | What this could mean | Limit sentence for the report |
|---|---|---|---|---|---|
Handle it in-house, or bring in help?
Your team can usually handle this when
- You hold registration data by group and can compare it with responses.
- The results are used for internal learning, not as proof to a funder or board.
- The audience is small and you know most of the attendees.
Outside planning help earns its fee when
- The findings will go to a board or funder and need a clear statement of limits.
- Responses are low or heavily skewed to one group and the committee wants to act anyway.
- Several teams hold different parts of the attendance and survey data.
Want the response review built into your plan?
If your team wants the attendance groups, response checks and limit statements agreed before the survey opens, an Event Blueprint can set out the survey scope, who holds which data and how limits are reported. Interpretation of the results stays with your committee. The feedback questionnaire builder can help draft the questions and measure map first.
Questions organisers ask
What response rate is good enough?
There is no single number that makes a survey reliable. What matters is who responded and who is missing. Report the count and the groups, and state the limits plainly.
Can I weight the results to match attendance?
Weighting needs enough responses in each group and careful handling. Ask someone with survey methods experience before using it, and label any weighted figure clearly.
Should I discard a survey if the response is skewed?
Usually not. Report it as the views of those who responded, state who is missing, and avoid conclusions about the whole audience. It can still show useful issues to follow up.
Related resources
Content record: Draft. Written from the cited sources and checked by automated rules; not yet independently reviewed.