Release v3.1.0

[igerber]

Summary

• Bump version to 3.1.0 across all locations (pyproject.toml, init.py, Cargo.toml, llms-full.txt, CITATION.cff)
• Finalize CHANGELOG entry for v3.1.0

What's in v3.1.0

dCDH Phase 3 - complete feature set for ChaisemartinDHaultfoeuille:

• Phase 3a (PR dCDH Phase 3a: placebo SE, non-binary treatment, parity SE assertions #300): Placebo SE via multiplier bootstrap, non-binary treatment support with crossing-cell detection
• Phase 3b (PR Add Phase 3 PR B: covariates, trends, and extensions for dCDH #302): Covariate adjustment via controls parameter, group-specific linear trends via trends_linear=True, R parity tests
• Phase 3c (PR Add HonestDiD integration for dCDH, summary() Phase 3 blocks #303): HonestDiD sensitivity analysis integration with automatic event-study-to-sensitivity bridge, support trimming, Delta-RM/Delta-SD bounds

The de Chaisemartin-D'Haultfœuille estimator now ships the full feature set for reversible-treatment DiD - the only such implementation across R, Stata, and Python that includes covariates, trends, and HonestDiD in a single package.

Methodology references (required if estimator / math changes)

• Method name(s): Version bump only - all methodology changes were reviewed in PRs dCDH Phase 3a: placebo SE, non-binary treatment, parity SE assertions #300, Add Phase 3 PR B: covariates, trends, and extensions for dCDH #302, Add HonestDiD integration for dCDH, summary() Phase 3 blocks #303
• Paper / source link(s): N/A
• Any intentional deviations from the source (and why): None

Validation

• Tests added/updated: No test changes (version metadata only)
• All Phase 3 PRs (dCDH Phase 3a: placebo SE, non-binary treatment, parity SE assertions #300, Add Phase 3 PR B: covariates, trends, and extensions for dCDH #302, Add HonestDiD integration for dCDH, summary() Phase 3 blocks #303) passed CI individually

Security / privacy

• Confirm no secrets/PII in this PR: Yes

---

Generated with Claude Code

[github-actions]

Overall Assessment

✅ Looks good — no unmitigated P0/P1 findings. The diff is release metadata/docs only; I found one P3 documentation overstatement in the changelog.

Executive Summary

• The PR does not modify estimator implementations, weighting, variance/SE code, identification checks, or defaults; it updates release metadata/docs only in CHANGELOG.md:10-19, pyproject.toml:5-8, diff_diff/init.py:229-231, CITATION.cff:1-13, docs/llms-full.txt:1-9, and rust/Cargo.toml:1-7.
• Methodology cross-check: the Phase 3 release-note summary generally matches the documented dCDH/HonestDiD scope in docs/methodology/REGISTRY.md:460-546 and docs/methodology/REGISTRY.md:2137-2173.
• P3 only: the changelog says placebo SE “resolves Phase 1 deferral,” but the registry and TODO still document single-period DID_M^pl SE as unresolved; only multi-horizon placebos have valid SE CHANGELOG.md:14, docs/methodology/REGISTRY.md:546-588, TODO.md:59.
• Version 3.1.0 is internally consistent across the bumped metadata files.
• No security, performance, or maintainability regressions are introduced by this diff.

Methodology

• Severity: P3. Impact: CHANGELOG.md:14 says placebo SE via multiplier bootstrap “resolves Phase 1 deferral,” but the documented contract still says single-period DID_M^pl (L_max=None) has NaN SE, while only multi-horizon placebos (L_max >= 1) have analytical/bootstrap SE docs/methodology/REGISTRY.md:546-588. That limitation also remains tracked in TODO.md:59. Concrete fix: narrow the changelog wording to dynamic placebos / L_max >= 1, or explicitly note that single-period DID_M^pl SE remains deferred.
• No other findings. Severity: none. Impact: no estimator math, weighting, variance, assumption checks, or defaults changed in this PR. Concrete fix: none.

Code Quality

• No findings. Severity: none. Impact: the version bump is synchronized across package metadata, import surface, citation metadata, AI docs, and Rust crate metadata at pyproject.toml:5-8, diff_diff/init.py:229-231, CITATION.cff:1-13, docs/llms-full.txt:1-9, and rust/Cargo.toml:1-7. Concrete fix: none.

Performance

• No findings. Severity: none. Impact: no runtime paths or computational kernels changed. Concrete fix: none.

Maintainability

• No findings. Severity: none. Impact: the diff is narrowly scoped and keeps release metadata aligned across ecosystems. Concrete fix: none.

Tech Debt

• No findings. Severity: none. Impact: the underlying single-period placebo-SE limitation is already properly tracked in TODO.md:59, so it is not a blocker; only the new changelog wording needs alignment. Concrete fix: none beyond the P3 methodology note above.

Security

• No findings. Severity: none. Impact: metadata/docs-only diff; no new secrets, inputs, or attack surface. Concrete fix: none.

Documentation/Tests

• No additional findings beyond the changelog wording issue above. Severity: none. Impact: no executable logic changed, so the absence of test changes is acceptable for this release PR. Concrete fix: none.