# Choose an inclusion rule for survey resubmissions

Separate repeated submissions from distinct respondents using an explicit eligibility and revision rule.

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

## 1. What would you expect to happen if you submitted a correction?

Answer: ____________________

## 2. Where should the invitation explain whether another submission replaces the first?

Answer: ____________________

## 3. What would you need to know before allowing a submission to be linked to an earlier one?

Answer: ____________________

## 4. What would you do if you were unsure whether your correction was received?

Answer: ____________________

## Author notes / illustrative keys where applicable

1. Test the intended replacement rule without treating every repeat as abuse.
2. Locate the disclosure needed before participation, not after a report dispute.
3. Check understanding of linkage without collecting an identifying value in this pilot.
4. Find uncertainty about the process rather than asking for real response content.

Interpretation caution: Similar answers do not prove duplicate respondents. Identity, permission to link, eligibility and counted unit must be established separately.


## Illustrative finding and follow-up

Observation: Four rows reconcile to two included known respondent records, one superseded row and one unresolved row.

Action to test: Write the unit and permitted linkage rule before counting; document unresolved cases separately.

Follow-up: Use fabricated repeat cases in the next pilot and ask whether readers can explain replacement and unresolved identity.


## Collection protocol

When and whom to ask: Before collection, show volunteers a fabricated correction scenario and the intended replacement notice.

Decision owner: The survey owner defines the unit; the data reviewer checks the version and cutoff ledger.

Follow-up: Use fabricated repeat cases in the next pilot and ask whether readers can explain replacement and unresolved identity.

Response handling: Keep real identifiers out of this exercise. Explain any real linkage separately and retain optional participation.

## Decision workshop

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

### Synthetic pilot and reporting records

Synthetic respondent | Submission | Timing | Treatment
--- | --- | --- | ---
A | v1 | Before cutoff | Superseded
A | v2 | Before cutoff | Include
B | v1 | Before cutoff | Include
Unknown | u1 | Before cutoff | Unresolved identity

### Version changes

Boundary | Draft | Revised
--- | --- | ---
Unit | Submission row | Known respondent with eligible version
Unknown identity | Assumed new person | Separate unresolved row

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

Four rows are four distinct respondents.

### Revised example

Under the stated latest-eligible-version rule, two known respondent records contribute; one earlier version is superseded and one unknown identity remains unresolved.

### Decision situations

- Identity, eligibility and cutoff are known: Apply the stated latest-version rule and retain the superseded row in the audit.
- Repeated participation is an intended separate event: Define the event unit instead of deduplicating by person.
- Identity or permission to link is unavailable: Do not infer duplication from similar answers.

### Method references

- AAPOR survey best practices: https://aapor.org/standards-and-ethics/best-practices/ — Pretesting and transparent sample context.