Resources · 109
Community polls: interpret who responds and who is missing
Use answers as a documented signal without automatically treating them as the opinion of an entire audience.
· 3 min
What this guide helps achieve
- Define the target population
- Test the question
- Document recruitment
- Publish counts
- Choose further verification
Quick check
- Are many responses enough?
- Can platforms be added together?
Step-by-step method
- 01
Define the target population
Separate followers, people exposed to the poll and actual respondents. Record the decision being informed; a community question does not automatically represent all customers.
Deliverable: population and decision.
- 02
Test the question
Check neutrality, understanding and mutually exclusive choices. Offer unknown or not-applicable states when relevant; avoid combining two subjects in one option.
Deliverable: question and comprehension test.
- 03
Document recruitment
Retain channel, period, invitations and participation rules. Qualify repeated answers and access mechanisms without collecting unnecessary identifiers; inspect lightly exposed groups.
Deliverable: participation protocol.
- 04
Publish counts
Show response count, percentage base and missing or excluded answers. Separate observed answers from generalisation; a large voluntary volume does not remove selection bias.
Deliverable: scoped results.
- 05
Choose further verification
When a decision affects a wider audience, prepare other recruitment or targeted interviews. Preserve contradictions and limitations; do not present an engagement poll as representative research.
Deliverable: next steps and inference limits.
Bound conclusions by observed evidence
Check the observed state before choosing next steps.
Respondent preference
Publish count, base and recruitment.
Lightly exposed group
Qualify the gap before generalising.
Audience-wide decision
Prepare appropriate complementary verification.
Acceptance case to reproduce
Fictional example: these inputs describe no customer or observed result.
View case inputs
{
"community_members": 1000,
"voluntary_responses": 20,
"positive_responses": 18,
"positive_among_respondents": 0.9
}Expected decision
90% describes the 20 voluntary respondents, not automatically all 1,000 community members.
Fictional variations to test
20 respondents select A 12 times and B 10 times.
With multiple choices allowed, 60% + 50% = 110% can be consistent. Name the base: respondents or selections.
Among 1,000 members: 18 yes, two no, 980 absent.
Assuming binary answers and no duplicates, total yes can range from 1.8% to 99.8%. These are logical bounds, not a confidence interval.
40 records come from 25 participating members.
Distinguish records from people under the stated rule. Resolve repeats before percentages per respondent, without collecting unnecessary identifiers.
Evidence to retain
- Question, permitted choices and counting unit.
- Invited people, respondents, absences and exclusions.
- Repeat rule and explicitly qualified bounds.
Verification procedure
- Define invited population and participation method.
- Compare voluntary respondents with absent members.
- Keep counts, wording and known selection biases.
Reusable worksheet
Complete with your authorised observations. These fields are a working template, not observed results.
| Field | Information to record |
|---|---|
| Question | Version, options and test |
| Participation | Channel, period and rules |
| Result | Count, denominator and limitations |
Worked example
Illustrative situation
Illustrative case: highly active members answer a feature question. Occasional users rarely see the invitation.
Decision and expected evidence
Describe respondents’ preference and check less-active users’ needs before a general decision.
Distinguish the mechanisms
| Mechanism | Purpose | Check or limitation |
|---|---|---|
| Voluntary poll | Gather a community signal | Qualify self-selection |
| Interviews | Understand reasons | Do not infer a proportion |
| Defined sample | Study a specified population | Inspect recruitment and non-response |
Management indicators
| Indicator | What it measures | First action |
|---|---|---|
| Scope | Results with an explicit base | Correct percentages without bases |
| Coverage | Targeted and actually exposed groups | Plan complementary checks |
Common pitfalls
- Confuse volume with representativeness
- Publish percentages without counts or denominator
- Add platforms without compatible methods
Frequently asked questions
Are many responses enough?
Volume alone does not correct participant selection. Examine who could see the question and who chose to answer.
Can platforms be added together?
Check question, period, population and repeat participation. Without compatible methods, present results separately.
Official references
References consulted: . The method and worksheet propose checks to adapt to your context; they do not constitute certification.
