Example catalog

AOI boundary example

A complete synthetic example of probabilistic AOI membership near a shared boundary.

Reuse

Uncertainty workflow

Expected output

Probabilistic AOI membership and uncertainty-aware summaries

Evidence boundary

Synthetic geometry and error values are not empirical calibration evidence.

Open example →

All structural-QC issues in one synthetic table

A fully synthetic GazeAudit example exercising all five implemented structural-QC issue families and all ten current diagnostic detail codes, with interpretation, decision, reporting, and limitation guidance.

Reuse

Complete structural-QC diagnostic and triage pattern

Expected output

One review report plus row/group diagnostics covering all current issue families and detail-code types

Evidence boundary

Synthetic flags demonstrate software behavior; they are not default repair, retention, exclusion, or quality rules.

Open example →

Analysis-readiness example

Reuse

Readiness workflow

Expected output

Readiness diagnostics and cohort-impact summary

Evidence boundary

Teaching workflow; readiness rules remain study-owned.

Open example →

Choose the right example

A documentation-only walkthrough showing how to choose a GazeAudit example by learning need, data context, and output without treating catalog filtering as scientific method selection.

Reuse

Example-selection pattern

Expected output

A chosen learning route with an explicit reason and boundary

Evidence boundary

Catalog filtering chooses learning material, not scientific methods or study parameters.

Open example →

Conclusion-rule edge cases

Deterministic synthetic examples for GazeAudit ConclusionRule covering missing tolerances, invalid thresholds, dual-tolerance conjunction, sign recovery, zero reference effects, threshold equality, circular-reference boundaries, and bounded reporting.

Reuse

Reference / tolerance / sign / recovery-threshold edge-case pattern

Expected output

Deterministic branch-recovery tables and summary classifications

Evidence boundary

Teaching reference effects and tolerances are not recommended empirical thresholds or generic robustness rules.

Open example →

Decision-to-report worked example

Use the deterministic synthetic robustness demo to practice declaring analytical choices, reading the complete 12-specification pattern, and translating it into bounded reporting language.

Reuse

Interpretation workflow

Expected output

Bounded reporting language tied to a complete synthetic audit

Evidence boundary

Synthetic result patterns are not empirical evidence for another study.

Open example →

Declared → valid → successful

A fully synthetic GazeAudit specification-space exercise that separates the Cartesian declaration, pre-execution scientific validity, and technical execution outcomes in an 8 declared → 6 valid → 5 successful audit.

Reuse

Declared → valid → successful denominator and failure-ledger pattern

Expected output

An eight-combination declaration with six valid specifications, five successful endpoints, two predeclared-invalid combinations, and one preserved valid technical failure

Evidence boundary

The factors, levels, validity rule, failure, and estimates are teaching choices; they are not recommended detector, QC, AOI, or robustness decisions.

Open example →

Diagnostic interpretation walkthrough

A deterministic synthetic robustness example that moves from eight finite specification estimates through specification ordering, effect stability, marginal and pairwise sensitivity, plot routes, bounded interpretation, reporting, and an explicit incomplete-execution contrast.

Reuse

Finite-estimate → curve → stability → marginal/pairwise sensitivity → bounded reporting sequence

Expected output

Eight-specification diagnostic tables, exact descriptive summaries, visual routes, and reporting language

Evidence boundary

The factors, estimates, diagnostic values, and interpretation thresholds are synthetic teaching material rather than empirical validation evidence.

Open example →

Documentation intent routing

A fully synthetic exercise showing how one research question moves through learning, task guidance, exact reference, and explanation without collapsing those purposes together.

Reuse

Navigation pattern

Expected output

Intent-routed documentation path

Evidence boundary

Navigation routing does not select a scientific method.

Open example →

End-to-end robustness audit

Run a complete synthetic GazeAudit workflow from canonical study construction through specification execution, stability summaries, and sensitivity diagnostics.

Reuse

Complete audit workflow

Expected output

Specification results, stability summaries, and sensitivity diagnostics

Evidence boundary

Teaching specifications and endpoints must be rebuilt for a real study.

Open example →

Endpoint drift audit

A fully synthetic worked example distinguishing acceptable specification variation from scientific endpoint drift in GazeAudit, including contrast reversal, unit mismatch, aggregation changes, different outcome families, and non-finite endpoint handling.

Reuse

Common-endpoint invariance and drift-detection pattern

Expected output

A branch-by-branch endpoint compatibility audit with corrected handoff rules

Evidence boundary

Synthetic endpoint scenarios teach comparability checks; they are not recommended eye-tracking endpoints or effect definitions.

Open example →

Example → study handoff

A fully synthetic walkthrough showing how to reuse the structure of a GazeAudit robustness example while replacing teaching values with explicitly declared study decisions.

Reuse

Example-to-study adaptation pattern

Expected output

Study-owned decision inventory and handoff record

Evidence boundary

The walkthrough intentionally does not supply replacement scientific values.

Open example →

Failed-audit recovery walkthrough

A fully synthetic break-diagnose-repair example that preserves invalid combinations, valid technical failures, rerun history, and execution denominators.

Reuse

Failure-recovery workflow

Expected output

Preserved failure record, repaired rerun, and reconciled denominator

Evidence boundary

Technical repair must not change the scientific branch after outcome inspection.

Open example →

Failure → repair → reconciliation

A fully synthetic GazeAudit execution-ledger example that preserves one invalid combination, one valid technical failure, one non-finite endpoint, one valid not-run branch, and successful repair reruns without changing the scientific specification denominator.

Reuse

Branch identity → attempt history → repair lineage → current-state reconciliation pattern

Expected output

A reconciled execution ledger separating 8 declared branches, 7 valid branches, historical attempt events, and current successful coverage

Evidence boundary

The failure types, repairs, endpoint values, and branch decisions are synthetic teaching material and are not recommended scientific rules.

Open example →

First real audit example

Run one practical GazeAudit script on deterministic demo data or a canonical eye-tracking CSV and save structural-QC provenance plus robustness tables.

Reuse

Project template

Expected output

Structural-QC provenance and robustness tables

Evidence boundary

Demo choices must be replaced or justified before use on a real study.

Open example →

Function to evidence walkthrough

A fully synthetic navigation exercise that starts from a GazeAudit public symbol and follows its governed path to a guide, runnable example, visual, and evidence boundary.

Reuse

Symbol-to-context navigation pattern

Expected output

Function-to-guide-example-visual-evidence route

Evidence boundary

Linked validation evidence remains protocol-bound.

Open example →

Generated API call contracts

A fully synthetic interface-reading exercise for required positional arguments, required keyword-only arguments, all-optional signatures, return annotations, and minimal call shapes.

Reuse

Interface-reading pattern

Expected output

Interpreted parameter, return, and minimal-call contracts

Evidence boundary

Software-call structure does not choose scientific values.

Open example →

Install, verify, first import

A software-only GazeAudit smoke check for a fresh virtual environment, optional extras, package identity, public imports, and console commands.

Reuse

Installation verification workflow

Expected output

Verified package identity, imports, CLI, and environment record

Evidence boundary

Successful installation does not establish scientific validity.

Open example →

Mapping change audit

A fully synthetic provenance exercise showing how source-column renames, source-version changes, unit conversions, coordinate transforms, grouping changes, and exclusions affect the data-mapping record.

Reuse

Mapping-change classification pattern

Expected output

A before/after mapping comparison with an explicit rerun decision

Evidence boundary

A mapping-provenance record improves reconstruction but does not certify preprocessing or scientific validity.

Open example →

Project lifecycle walkthrough

A fully synthetic five-stage GazeAudit walkthrough from planning and audit execution through interpretation, peer-review amendment, and final handoff.

Reuse

Project-stage workflow

Expected output

Five-stage project record from planning through final handoff

Evidence boundary

Synthetic project decisions are teaching choices, not a protocol.

Open example →

Publication and archive handoff example

Walk through an illustrative manuscript and archive handoff that connects a decision record, complete robustness evidence, scoped wording, provenance, and reproducibility metadata.

Reuse

Publication handoff pattern

Expected output

Manuscript/archive evidence map and reproducibility handoff

Evidence boundary

Archive completeness does not by itself establish scientific validity.

Open example →

Readiness policy design

A fully synthetic worked example for declaring a primary and alternative GazeAudit readiness policy, previewing trial- versus participant-scope cohort consequences, interpreting policy-relative status, and writing bounded Methods, Results, and limitations text.

Reuse

Readiness-policy declaration, comparison, denominator, and reporting pattern

Expected output

Primary/alternative readiness policies with policy-relative status and trial/participant cohort-impact interpretation

Evidence boundary

All threshold values are teaching choices for this synthetic fixture; they are not recommended eye-tracking cutoffs.

Open example →

Reference lookup walkthrough

A fully synthetic example showing how to answer exact API, denominator, failure, CLI, and frozen-label questions by moving through the GazeAudit reference layer.

Reuse

Reference lookup pattern

Expected output

Authoritative API, CLI, denominator, or evidence lookup route

Evidence boundary

Reference lookup states contracts; it does not choose research decisions.

Open example →

Reporting-language rewrite

A fully synthetic worked exercise that rewrites over-strong GazeAudit manuscript claims into evidence-bounded Methods, Results, and limitation language while preserving runtime status, denominator, execution completeness, and protocol scope.

Reuse

Evidence object → denominator → claim boundary → bounded rewrite pattern

Expected output

Seven repaired manuscript claims with explicit evidence and limitation boundaries

Evidence boundary

The sentences are synthetic teaching material; they are not manuscript claims about any real study.

Open example →

Result-pattern reporting example

Compare four synthetic robustness patterns to practise separating execution completeness, direction stability, magnitude sensitivity, and bounded manuscript wording.

Reuse

Result-pattern interpretation workflow

Expected output

Bounded wording for stable, sensitive, sign-changing, and incomplete audits

Evidence boundary

Synthetic patterns teach interpretation; they are not empirical priors.

Open example →

Reviewer reconstruction worked example

Walk through an illustrative GazeAudit manuscript archive as a reviewer and identify what is reconstructable, what is missing, and what must be repaired before submission.

Reuse

Reviewer reconstruction workflow

Expected output

Reconstructed claim path plus missing-evidence diagnosis

Evidence boundary

The synthetic archive is for review practice, not empirical validation.

Open example →

Reviewer-requested reanalysis worked example

Extend a synthetic submitted GazeAudit record with a reviewer-requested sensitivity analysis while preserving the original denominator, decision history, and manuscript provenance.

Reuse

Post-review amendment workflow

Expected output

Separate submitted and reviewer-requested evidence layers

Evidence boundary

Post-review analyses must not be relabelled as pre-specified.

Open example →

Revision reproducibility package worked example

A synthetic executable-style example that scaffolds a reviewer-revision package, records separate submitted and post-review evidence, demonstrates a missing-file failure, and validates the repaired structure.

Reuse

CLI package workflow

Expected output

Validated reviewer-revision package with failure-repair history

Evidence boundary

Structural package validity is not scientific validation.

Open example →

Revision response package worked example

A synthetic peer-review response package linking reviewer requests to amendment timing, separate denominators, manuscript edits, and archived revision evidence.

Reuse

Reviewer-response evidence pattern

Expected output

Linked response, amendment, manuscript-change, and archive records

Evidence boundary

Response structure cannot substitute for justified analysis decisions.

Open example →

Revision-package quickstart

Synthetic end-to-end quickstart for creating, populating, validating, and handing off a reviewer-revision provenance package with the real GazeAudit CLI.

Reuse

Revision-package workflow

Expected output

Created, populated, and structurally validated revision package

Evidence boundary

Package validity establishes provenance structure, not scientific validity.

Open example →

Revision-round routing scenarios

Synthetic peer-review scenarios showing clarification, correction, sensitivity amendments, incomplete post-review execution, endpoint amendments, response records, and revision-package handoff.

Reuse

Revision routing scenarios

Expected output

Scenario-specific clarification, correction, amendment, and handoff routes

Evidence boundary

Scenario routes are teaching cases, not editorial rules.

Open example →

Sampling sensitivity example

Downsample a synthetic GazeStudy to controlled target rates and inspect endpoint stability.

Reuse

Sampling sensitivity workflow

Expected output

Endpoint stability across controlled target sampling rates

Evidence boundary

Synthetic target rates do not define acceptable sampling for another study.

Open example →

Search to contract walkthrough

A fully synthetic example showing how grouped GazeAudit search results lead from an ambiguous query to the correct guide, example, reference, or evidence authority.

Reuse

Search-to-authority navigation pattern

Expected output

Grouped search path to the relevant documentation authority

Evidence boundary

Search ranking is navigation support, not method ranking.

Open example →

Sensitivity protocol worked example

Build a deterministic synthetic sampling and missingness sensitivity protocol, preserve the declared grid, plot complete curves, and write bounded interpretation.

Reuse

Predeclared perturbation protocol

Expected output

Sampling and missingness sensitivity tables with reproducible plots and bounded reporting

Evidence boundary

The synthetic ranges and endpoint are teaching values, not recommended thresholds or acquisition standards.

Open example →

Source-level API inspection

A fully synthetic documentation exercise showing how to verify GazeStudy, run_specs, and aoi_probabilities from generated signatures through source links and governed pathways.

Reuse

Source-inspection workflow

Expected output

Verified signatures, source links, and pathway context

Evidence boundary

Source inspection describes implementation; it does not justify study choices.

Open example →

Specification curve example

Summarise synthetic multiverse results with ordered estimates, sign stability, and factor sensitivity.

Reuse

Specification-curve workflow

Expected output

Ordered estimates, sign stability, and factor sensitivity summaries

Evidence boundary

Synthetic specifications illustrate mechanics; they are not recommended defaults.

Open example →

Study preflight example

Run structural QC, inspect row/group diagnostics, record decisions, and preserve deterministic provenance.

Reuse

Structural-preflight workflow

Expected output

Structural QC diagnostics, decision record, and provenance

Evidence boundary

Structural QC flags are not automatic scientific exclusions.

Open example →

Submission-to-accepted-record worked example

A synthetic end-to-end revision example that preserves submitted evidence, reviewer-requested amendments, response records, version changes, and the final editor-facing archive as separate temporal layers.

Reuse

Submission-to-handoff workflow

Expected output

Temporally separated submitted, amended, and final editor-facing records

Evidence boundary

Revision chronology preserves provenance; it does not certify scientific correctness.

Open example →

Valid, invalid, and reviewable tables

A fully synthetic walkthrough showing which table conditions prevent GazeStudy construction, which construct but require structural review, and which checks belong downstream.

Reuse

Schema-validation and preflight pattern

Expected output

A classified set of constructor failures and structural-review conditions

Evidence boundary

Passing the constructor or structural preflight does not establish measurement or scientific validity.

Open example →