Release 3.4.1: SpilloverDiD survey + Conley lifts, SunAbraham vcov_type, WLS-CR2 BM, methodology-tracker promotions

[igerber]

Summary

• Bumps version 3.4.0 → 3.4.1 across pyproject.toml, rust/Cargo.toml, diff_diff/__init__.py, diff_diff/guides/llms-full.txt, and CITATION.cff (date-released: 2026-05-21)
• Renames CHANGELOG.md [Unreleased] content to [3.4.1] - 2026-05-21, preserves an empty [Unreleased] placeholder, adds the [3.4.1]: .../compare/v3.4.0...v3.4.1 link

Release contents (8 PRs since 3.4.0, 2026-05-19)

Public-surface variance lifts (additive, defaults bit-equal):

• SpilloverDiD survey_design= on HC1/CR1 via Binder TSL (Wave E.1, SpilloverDiD survey_design= on HC1/CR1 via Binder TSL (Wave E.1) #468)
• SpilloverDiD vcov_type="conley" + survey_design= via panel-aware stratified-Conley on per-period PSU totals × Wave D GMM correction (Wave E.2, SpilloverDiD vcov_type='conley' + survey_design= via panel-aware stratified-Conley sandwich on per-period PSU totals (Wave E.2) #474)
• SpilloverDiD same composition with lag_cutoff > 0 via serial Bartlett panel-block (Wave E.2 follow-up, SpilloverDiD conley + survey + lag>0 via panel-block composition (Wave E.2 follow-up) #477)
• SunAbraham vcov_type ∈ {classical, hc1, hc2, hc2_bm} (Phase 1b 1/8, Thread vcov_type through SunAbraham (Phase 1b 1/8) #472); default "hc1" preserves prior behavior bit-equally
• WLS-CR2 Bell-McCaffrey gates lifted via clubSandwich port — vcov_type="hc2_bm" + weights (one-way and cluster-robust) at atol=1e-10 vs clubSandwich::vcovCR(..., type="CR2") (Lift hc2_bm + weights gates via clubSandwich WLS-CR2 port #475)

Methodology-review-tracker promotions (mostly docs/tests):

• PreTrendsPower R pretrends parity goldens (PR-C, PR-C: PreTrendsPower R pretrends parity goldens at commit 122731d082 #471) — tracker row → Complete
• HeterogeneousAdoptionDiD methodology-review-tracker promotion (HeterogeneousAdoptionDiD methodology-review-tracker promotion: In Progress -> Complete #473) — 36 new methodology tests, REGISTRY Deviations block
• ContinuousDiD methodology-review-tracker promotion (ContinuousDiD methodology-review-tracker promotion: In Progress -> Complete #476) — REGISTRY consolidation, no source changes

Methodology references (required if estimator / math changes)

• N/A - this PR only edits version-string surfaces and the CHANGELOG header. No methodology source code changes. All cited methodology already shipped in the 8 underlying PRs above (Conley 1999; Gerber 2026 arXiv:2605.04124; Butts 2021; Gardner 2022; Pustejovsky-Tipton 2018; clubSandwich; Roth 2022; de Chaisemartin-Ciccia-D'Haultfoeuille-Knau 2026 arXiv:2405.04465v6).

Validation

• Tests added/updated: none in this PR. All eight underlying PRs landed with their own test surfaces; the release commit itself touches no source or test files.
• Backtest / simulation / notebook evidence: N/A

Security / privacy

• Confirm no secrets/PII in this PR: Yes

---

Generated with Claude Code

[github-actions]

Overall Assessment

✅ Looks good

Executive Summary

• The diff is release-metadata only: changelog promotion plus version/date bumps in package metadata and guides (CHANGELOG.md:L8-L10, CITATION.cff:L10-L11, diff_diff/__init__.py:L296, diff_diff/guides/llms-full.txt:L5, pyproject.toml:L7, rust/Cargo.toml:L3).
• Affected methods: none. No estimator implementation, weighting, variance/SE logic, identification checks, or default behavior changed in this PR.
• The methodology-heavy text now under CHANGELOG.md:L12-L20 is release-note content, not new implementation code in this diff.
• Active version surfaces are internally consistent at 3.4.1; the only remaining 3.4.0 strings are expected historical changelog references (CHANGELOG.md:L22, CHANGELOG.md:L1495-L1496).
• No unmitigated P0/P1 issues were identified.

Methodology

• Severity: P3. Impact: No methodology delta is in scope for this PR. The only changed Python surface is the package version constant in diff_diff/__init__.py:L296, and the rest of the diff is release metadata/changelog text (CHANGELOG.md:L8-L20, pyproject.toml:L7, rust/Cargo.toml:L3). Concrete fix: None.

Code Quality

• Severity: P3. Impact: The version bump is mechanically consistent across runtime/package/citation/doc surfaces (CITATION.cff:L10-L11, diff_diff/__init__.py:L296, diff_diff/guides/llms-full.txt:L5, pyproject.toml:L7, rust/Cargo.toml:L3). Concrete fix: None.

Performance

• Severity: P3. Impact: No runtime code path, numerical routine, or dependency behavior changed; no performance regression is introduced by this diff. Concrete fix: None.

Maintainability

• Severity: P3. Impact: The changelog keeps the empty [Unreleased] placeholder and adds the 3.4.1 compare link, which preserves the existing release-management pattern (CHANGELOG.md:L8-L10, CHANGELOG.md:L1495-L1496). Concrete fix: None.

Tech Debt

• Severity: P3. Impact: No new deferred limitation is introduced, and this PR does not modify any TODO-tracked correctness item. Concrete fix: None.

Security

• Severity: P3. Impact: The changed files are metadata/documentation only; no secret-bearing config, executable surface, or dependency update is introduced. Concrete fix: None.

Documentation/Tests

• Severity: P3. Impact: No tests changed, but that is appropriate for a metadata-only release PR with no executable behavior change. Concrete fix: None.