Skip to content

JSS Source-Audit Dossier

This dossier is the package-facing source audit shipped with the JSS submission archive. It is not a blanket validation claim: StatsPAI uses validation_status to separate certified or validated numerical evidence from API-stable breadth.

For the paper-specific reviewer path, start with Paper-JSS/README.md in the submission archive. That path runs Tier 1 without live R or Stata and points to the exact audit artifacts under Paper-JSS/replication/results/.

Project Status

Software Scope

StatsPAI exposes a unified Python interface for causal inference and applied econometrics. The live registry reports 1,020 registered public functions across 81 submodules:

python scripts/registry_stats.py --check

The registry and schema layer are part of the public surface. They support programmatic discovery through sp.list_functions(), sp.describe_function(), and sp.function_schema(). That schema breadth is useful for agents, but it does not mean every registered helper is numerically validated.

Validation Boundary

Current JSS source-snapshot audit counts: 52 certified, 25 validated, 940 api_stable, and 3 experimental registry symbols. The certified/validated surface is therefore 77 symbols, while 751 stable auto-registered symbols remain API-stable but not parity-backed.

The validated tier requires known-truth, reference-parity, external-parity, coverage, or explicit convention evidence. Unit and regression tests support API stability; they do not by themselves promote a function to validated. certified is reserved for entries in the main cross-language or published reference parity harness.

The source-snapshot evidence audit checks that all certified/validated symbols have registry-attached evidence notes and that those notes resolve to source files included in the JSS package. The current archive includes 133 registry-evidence source files. The current source snapshot also tracks 135 registry-evidence source files in the live validation-note inventory.

Reproducible Audit Artifacts

The submission archive carries the generated audit outputs:

  • Paper-JSS/replication/results/jss_full_audit.{json,md}
  • Paper-JSS/replication/results/claim_lint.{json,md}
  • Paper-JSS/replication/results/validation_evidence_audit.{json,md}
  • Paper-JSS/replication/results/source_snapshot_manifest.{json,md}
  • Paper-JSS/replication/results/reproduction_environment_audit.md
  • Paper-JSS/replication/results/methodological_gap_ledger.md

The headline short path is intentionally reviewer-bounded:

cd Paper-JSS
make reproduce-tier1
make verify-submission-package

Tier 1 exercises Python-only smoke, registry, schema, validation-claim, formal JSS, and package-coherence checks. The transcript explicitly states that the Tier 1 path intentionally does not require live R or Stata; parity evidence is shipped as source, lockfiles, scripts, and archived result artifacts.

Parity And Replication Anchors

StatsPAI includes validation fixtures for common teaching and replication benchmarks, including:

  • Card-style returns-to-schooling IV estimates.
  • LaLonde / Dehejia-Wahba job-training benchmarks.
  • Lee-style close-election regression discontinuity.
  • Callaway-Sant'Anna difference-in-differences examples.
  • California Proposition 99 synthetic-control examples.

Known convention differences are documented in parity reports rather than hidden. Bandwidth selectors, regularisation constants, small-sample standard-error conventions, fold-split randomness, and identification-dependent SCM disagreement are tracked in audit artifacts when they affect exact numerical matching.

Research Use

StatsPAI is being used in working-paper workflows connected to the Stanford Rural Education Action Program and related empirical policy evaluation work. No peer-reviewed research article using StatsPAI has yet been published. The current impact claim is therefore based on credible near-term research use, reproducible validation materials, public package distribution, and reviewer-verifiable examples rather than published downstream citations.

Public Distribution And Commercial Disclosure

StatsPAI is publicly distributed on PyPI and archived on Zenodo. Public stars, forks, issue templates, contribution instructions, release notes, and CI checks are community-readiness signals, not evidence of independent scholarly adoption.

StatsPAI Inc. is the legal entity associated with the project. CoPaper.AI is a commercial downstream product that may call the MIT-licensed StatsPAI package. The StatsPAI package itself is permanently open source under the MIT license.