America.co emblemAmerica.co

Democratic trust / Reform dossier

Election audits people can understand.

What evidence supports a reported election outcome?

Existing audit approaches; jurisdiction-specific implementation

The short answer

Publish the audit’s purpose, method, scope, evidence, findings, and escalation rule in language people can check. An audit label alone does not tell readers what was verified.

What the evidence establishes

The EAC’s audit overview describes different approaches and notes that there is no national auditing standard. Its practical risk-limiting audit guide explains a method intended to provide statistical evidence about a contest outcome and a high probability of correcting a wrong outcome. State methods and legal procedures vary.

EAC — Election Audits Across the United States ↗ · EAC — Risk-Limiting Audits: Practical Application ↗

What it does not establish

A risk limit is not a prediction that a particular share of ballots is wrong or a guarantee about every part of an election. These sources do not show that any specific election was inaccurate. Confirm current jurisdiction rules with its election office.

Original design / proposed

How it could work

Begin with the official report for one contest. Identify what was audited: tabulation, procedures, equipment, or another scope. For a risk-limiting audit, identify the contest, risk limit, sample design, underlying records, discrepancies, and the rule for expanding the audit.

Show the original report beside a short reading guide. Explain what would trigger more examination and who has authority to decide. Preserve chain-of-custody and privacy requirements; do not expose voter selections or invite unauthorized handling of election materials.

Tradeoffs & safeguards

Close margins can require more examination, staff time, and coordination. Suitable auditable records, secure custody, transparent procedures, and trained officials matter. No single audit method checks every possible failure.

Readers may confuse an observed discrepancy with a changed winner. Report the type, size, investigation, and official resolution rather than treating every difference as proof of manipulation. Public confidence should be built from a checkable process, not a promise of zero mistakes.

A pilot you could actually evaluate

Proposed public-information exercise: take an already-published audit report from one jurisdiction and ask readers to identify its scope, finding, and escalation rule. Have the election office review factual ambiguities before publishing a summary. This is not an unofficial recount.

Suggested measures—not reported results

  • Readers correctly identifying scope, finding, and remaining limits
  • Key method details linked to the original report
  • Unresolved ambiguities and time to correct a misleading summary
Find the official state election office ↗
Your next piece of useful work

Build an evaluation brief

Optional. Keep the unknowns visible. This is saved nowhere automatically and is not submitted to America.co or government.

Blank worksheet ↓

Questions worth asking

Does a 5% risk limit mean 5% of ballots are wrong?

No. It concerns the audit procedure’s risk of failing to correct an incorrect reported outcome under its assumptions, not a ballot-error rate.

Is an audit the same as a recount?

Not necessarily. Their purposes, scope, triggering rules, and legal procedures can differ. Read the jurisdiction’s official instructions.

Go to the original sources

AI-assisted synthesis and original evaluation questions, with linked source checks. No independent expert review, legal conclusion, funding commitment, or proven pilot outcome is claimed. People, methods, and corrections.