/assess-publication-bias

name: assess-publication-bias
description: "Assess or plan tests for publication/reporting/time-lag/citation/language bias using appropriate diagnostics."

assess-publication-bias

Purpose

Assess publication, reporting, time-lag, citation, and language bias using diagnostics appropriate to the evidence design.

Input contract

required: [evidence_records, outcome_schema, inclusion_frame]
optional: [registry_records, search_log, effect_estimates]
constraints: [bias claims require an observable missingness or selection pattern]

Procedure

1. Define the eligible study universe and the observed evidence subset. 2. Compare availability, reporting, citation, timing, and language patterns across the declared dimensions. 3. Apply design-appropriate diagnostics and record assumptions. 4. Distinguish detected bias, plausible bias, and unassessed domains.

If the conclusion may change under plausible missing-study or selection-bias scenarios, consider `assess-sensitivity` as the next tactic.

Output contract

produces: [bias_domain_assessments, diagnostic_results, sensitivity_implications, unassessed_domains]
delta_fields: [findings, evidence_updates, uncertainties, decisions, open_questions]

Quality gates

• The denominator for each bias domain is explicit.
• Diagnostics are not used outside their design assumptions.

Failure and counterexamples

Do not equate small samples with publication bias or treat absence from one index as non-publication.

Provenance map

• `resolved: knowledge-acquisition-publication-bias-assessment`
• `resolved: bias-detection`