Survey Phase 7: CS IPW/DR covariates, repeated cross-sections, HonestDiD survey variance#240

Open

igerber wants to merge 20 commits intomainfrom

survey-phase-seven

Commits on Mar 28, 2026

Survey Phase 7: CS IPW/DR covariates, repeated cross-sections, HonestDiD survey variance

igerber
and
claude
committed

Commits on Mar 29, 2026

Address CI review: RCS IF corrections, aggregation weights, replicate VCV, panel on results

igerber
and
claude
committed
Fix DR RC normalizer mismatch, holistic RCS cohort-mass weighting, unequal-count tests

igerber
and
claude
committed
Rewrite RC reg/DR to match DRDID::reg_did_rc and DRDID::drdid_rc formulas

igerber
and
claude
committed
Fix bootstrap RCS cohort-mass weighting, reset stale event-study VCV

igerber
and
claude
committed
Clear analytical event_study_vcov when bootstrap overwrites event-study SEs

igerber
and
claude
committed
Fix RC IF normalization scaling: M1 uses n_all denominator, PS M2 uses sum not mean

igerber
and
claude
committed
Document RCS IF phi=psi/n convention, add analytical-vs-bootstrap SE convergence test

igerber
and
claude
committed
Refactor RC IFs to R's psi convention, fix HonestDiD VCV subsetting

igerber
and
claude
committed
Resolve merge conflict, match R colMeans convention in panel IPW/DR M2 gradients

igerber
and
claude
committed
Match R's H/n, asy_rep/n, colMeans convention for panel PS corrections; fix VCV index subsetting

igerber
and
claude
committed
Fix VCV index alignment, add stationarity warning for panel=False

igerber
and
claude
committed
Document panel DR control-augmentation normalization deviation from DRDID

igerber
and
claude
committed
Warn on non-universal base period in HonestDiD CS path, update tests

igerber
and
claude
committed
Fix panel M2 full-sample colMeans, add HonestDiD consecutive event-time guard

igerber
and
claude
committed
Fix HonestDiD grid validator for reference-period gap, defensive bootstrap for zero-mass cells

igerber
and
claude
committed
HonestDiD: raise ValueError on non-consecutive event-time grid (was warning)

igerber
and
claude
committed
HonestDiD: validate full grid around reference period, not just within-block

igerber
and
claude
committed
Fix HonestDiD: reference-aware pre/post split, replicate df=0 sentinel

igerber
and
claude
committed
Fix _estimate_max_pre_violation to use reference-aware pre_periods

igerber
and
claude
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Survey Phase 7: CS IPW/DR covariates, repeated cross-sections, HonestDiD survey variance#240

Survey Phase 7: CS IPW/DR covariates, repeated cross-sections, HonestDiD survey variance#240
igerber wants to merge 20 commits intomainfrom
survey-phase-seven

Commits on Mar 28, 2026

Survey Phase 7: CS IPW/DR covariates, repeated cross-sections, HonestDiD survey variance

Commits on Mar 29, 2026

Address CI review: RCS IF corrections, aggregation weights, replicate VCV, panel on results

Fix DR RC normalizer mismatch, holistic RCS cohort-mass weighting, unequal-count tests

Rewrite RC reg/DR to match DRDID::reg_did_rc and DRDID::drdid_rc formulas

Fix bootstrap RCS cohort-mass weighting, reset stale event-study VCV

Clear analytical event_study_vcov when bootstrap overwrites event-study SEs

Fix RC IF normalization scaling: M1 uses n_all denominator, PS M2 uses sum not mean

Document RCS IF phi=psi/n convention, add analytical-vs-bootstrap SE convergence test

Refactor RC IFs to R's psi convention, fix HonestDiD VCV subsetting

Resolve merge conflict, match R colMeans convention in panel IPW/DR M2 gradients

Match R's H/n, asy_rep/n, colMeans convention for panel PS corrections; fix VCV index subsetting

Fix VCV index alignment, add stationarity warning for panel=False

Document panel DR control-augmentation normalization deviation from DRDID

Warn on non-universal base period in HonestDiD CS path, update tests

Fix panel M2 full-sample colMeans, add HonestDiD consecutive event-time guard

Fix HonestDiD grid validator for reference-period gap, defensive bootstrap for zero-mass cells

HonestDiD: raise ValueError on non-consecutive event-time grid (was warning)

HonestDiD: validate full grid around reference period, not just within-block

Fix HonestDiD: reference-aware pre/post split, replicate df=0 sentinel

Fix _estimate_max_pre_violation to use reference-aware pre_periods