feat(v4): PR 2 of the 4.0 cut - remove the 3.9 rename/wrapper/alias shims

.claude/memory.md

@@ -93,18 +93,23 @@ everywhere; `to_dict()` / `to_dataframe()` on most classes. See
 Target contract: `docs/v4-design.md` section 5; per-surface lifecycle:
 `docs/v4-deprecations.yaml` (CI-enforced by `tests/test_v4_matrix.py`).
 
-### Column Naming — CURRENT (3.x)
-- `unit` unit id (`unit_col` on HAD; `group` on dCDH — both slated for 4.0)
-- `time` calendar period, EXCEPT DifferenceInDifferences / TripleDifference /
-  static TwoWayFixedEffects, where `time` is the 0/1 post dummy (TWFE warns
-  on >2 unique values; 4.0 renames these to `post`, and TWFE's `time`
-  becomes the event-study calendar column)
+### Column Naming — CURRENT (4.0)
+- `unit` unit id everywhere (the 3.x `unit_col` on HAD and `group` on dCDH
+  were removed at 4.0)
+- `time` calendar period everywhere; DifferenceInDifferences /
+  TripleDifference take the 0/1 post dummy as `post` (the 3.x `time=`
+  spelling on DiD was removed at 4.0, M-030; on TripleDifference it is
+  still a warning shim, M-031, because `time` persists as the merged
+  class's staggered calendar column - its 4.0 enforcement is M-085);
+  static TwoWayFixedEffects takes `post` and its `time` is the event-study
+  calendar column (the static `time=` alias on TWFE is the same kind of
+  surviving shim, M-082, enforced by M-083)
 - `treatment` 0/1 treated-group indicator; `first_treat` cohort column
-  (`cohort` on WooldridgeDiD — slated for 4.0)
-- `covariates` covariate list (`controls` on dCDH — slated for 4.0)
+  everywhere (the 3.x `cohort` on WooldridgeDiD was removed at 4.0)
+- `covariates` covariate list everywhere (the 3.x `controls` on dCDH was
+  removed at 4.0)
 
-### Column Naming — 4.0 TARGET (do not write against this pre-4.0)
-Target vocabulary and rules: `docs/v4-design.md` section 8.
+Vocabulary and rules: `docs/v4-design.md` section 8.
 
 ## Session Notes
 

CONTRIBUTING.md

@@ -205,8 +205,8 @@ A new parameter is only complete when it is:
   forwards via `super().__init__(*args, **kwargs)` with an import-time
   `__signature__` mirror - so both inherit a new
   `DifferenceInDifferences` constructor parameter automatically;
-  `SyntheticDiD` defines its OWN signature (it forwards only
-  `robust`/`cluster`/`alpha` to `super().__init__`), so a new parent
+  `SyntheticDiD` defines its OWN signature (it forwards only `alpha` to
+  `super().__init__`, pinning `cluster=None`), so a new parent
   parameter must be explicitly mirrored — or explicitly rejected, like its
   `vcov_type`/`conley_*` guards — there, and `BaseEstimator` introspects the
   concrete signature, so an unmirrored parent param simply won't appear in

benchmarks/python/benchmark_basic.py

@@ -94,12 +94,12 @@ def main():
 
     if args.warmup:
         print("Warm-up fit (untimed)...")
-        DifferenceInDifferences(robust=True, cluster=args.cluster).fit(
+        DifferenceInDifferences(cluster=args.cluster).fit(
             data, formula="outcome ~ treated * post"
         )
 
     # Use DifferenceInDifferences with formula to match R's fixest::feols
-    did = DifferenceInDifferences(robust=True, cluster=args.cluster)
+    did = DifferenceInDifferences(cluster=args.cluster)
 
     with Timer() as timer:
         results = did.fit(

benchmarks/python/benchmark_multiperiod.py

@@ -102,7 +102,7 @@ def main():
 
     if args.warmup:
         print("Warm-up fit (untimed)...")
-        MultiPeriodDiD(robust=True, cluster=args.cluster).fit(
+        MultiPeriodDiD(cluster=args.cluster).fit(
             data,
             outcome="outcome",
             treatment="treated",
@@ -112,7 +112,7 @@ def main():
             absorb=["unit"],
         )
 
-    did = MultiPeriodDiD(robust=True, cluster=args.cluster)
+    did = MultiPeriodDiD(cluster=args.cluster)
 
     with Timer() as timer:
         results = did.fit(

benchmarks/python/benchmark_twfe.py

@@ -79,11 +79,11 @@ def main():
 
     if args.warmup:
         print("Warm-up fit (untimed)...")
-        TwoWayFixedEffects(robust=True).fit(
+        TwoWayFixedEffects().fit(
             data, outcome="outcome", treatment="treated", post="post", unit="unit"
         )
 
-    twfe = TwoWayFixedEffects(robust=True)  # auto-clusters at unit level
+    twfe = TwoWayFixedEffects()  # auto-clusters at unit level
 
     with Timer() as timer:
         results = twfe.fit(

benchmarks/python/benchmark_wooldridge.py

@@ -74,13 +74,13 @@ def main():
             outcome="lemp",
             unit="countyreal",
             time="year",
-            cohort="first_treat",
+            first_treat="first_treat",
         )
 
     estimation_time = estimation_timer.elapsed
 
     # Compute event study aggregation
-    results.aggregate("event")
+    results.aggregate("event_study")
     total_time = estimation_timer.elapsed
 
     # Store data info

benchmarks/speed_review/bench_brand_awareness_survey.py

@@ -73,13 +73,13 @@ def naive_fit():
         # accounted for the sampling structure. The SE-inflation story
         # only shows up if this step is as untreated-for-design as
         # practitioners actually start.
-        did = DifferenceInDifferences(robust=True)
+        did = DifferenceInDifferences()
         results["naive"] = did.fit(
             data, outcome="outcome", treatment="treat_unit", post="post",
         )
 
     def tsl_fit():
-        did = DifferenceInDifferences(robust=True)
+        did = DifferenceInDifferences()
         results["tsl"] = did.fit(
             data, outcome="outcome", treatment="treat_unit", post="post",
             survey_design=sd_tsl,
@@ -92,7 +92,7 @@ def replicate_fit():
             weights="weight", replicate_weights=rw_cols,
             replicate_method="JK1",
         )
-        did = DifferenceInDifferences(robust=True)
+        did = DifferenceInDifferences()
         results["replicate"] = did.fit(
             data, outcome="outcome", treatment="treat_unit", post="post",
             survey_design=sd,
@@ -101,7 +101,7 @@ def replicate_fit():
     def multi_outcome_loop():
         out = {}
         for y in ("outcome", "consideration", "purchase_intent"):
-            did = DifferenceInDifferences(robust=True)
+            did = DifferenceInDifferences()
             out[y] = did.fit(
                 data, outcome=y, treatment="treat_unit", post="post",
                 survey_design=sd_tsl,
@@ -118,10 +118,10 @@ def pretrends():
     def placebo_refit():
         pre = data[data["period"] < 7].copy()
         pre["placebo_post"] = (pre["period"] >= 4).astype(int)
-        did = DifferenceInDifferences(robust=True)
+        did = DifferenceInDifferences()
         results["placebo"] = did.fit(
             pre, outcome="outcome", treatment="treat_unit",
-            time="placebo_post", survey_design=sd_tsl,
+            post="placebo_post", survey_design=sd_tsl,
         )
 
     def honest_did_grid():

benchmarks/speed_review/bench_dose_response.py

@@ -84,7 +84,7 @@ def binarized_comparison():
         data_bin = data.copy()
         data_bin["treated_any"] = (data_bin["dose"] > 0).astype(int)
         data_bin["post"] = (data_bin["period"] >= treated_cohort).astype(int)
-        did = DifferenceInDifferences(robust=True)
+        did = DifferenceInDifferences()
         results["binarized"] = did.fit(
             data_bin, outcome="outcome", treatment="treated_any", post="post",
         )

benchmarks/speed_review/bench_memory_scaling.py

@@ -135,7 +135,7 @@ def _run_one(cfg):
         from diff_diff.twfe import TwoWayFixedEffects
 
         ncov = cfg.get("ncov", 6)
-        TwoWayFixedEffects(robust=True).fit(
+        TwoWayFixedEffects().fit(
             _twfe_panel(nu, ncov=ncov),
             outcome="outcome",
             treatment="treated",

benchmarks/speed_review/bench_reversible_dcdh.py

@@ -52,7 +52,7 @@ def main():
 
     results = {}
     fit_kwargs = dict(
-        data=data, outcome="outcome", group="group", time="period",
+        data=data, outcome="outcome", unit="group", time="period",
         treatment="treatment",
     )
     sd = SurveyDesign(

changelog.d/20261018-v4-pr2-shim-removals.md

@@ -0,0 +1,59 @@
+### Removed
+- **The 3.9 rename shims are gone (ledger rows flipped to `removed`).** Old keyword spellings
+  now raise `TypeError`: `DifferenceInDifferences.fit(time=)` -> `post=` (`M-030`),
+  `permutation_test(time=)` / `leave_one_out_test(time=)` -> `post=` (`M-137`, `M-138`),
+  `WooldridgeDiD.fit(cohort=)` -> `first_treat=` (`M-032`), `ChaisemartinDHaultfoeuille.fit(group=,
+  controls=)` -> `unit=`, `covariates=` (`M-033`, `M-034`), `twowayfeweights(group=)` -> `unit=`
+  (`M-097`), the `*_col` spellings on `HeterogeneousAdoptionDiD.fit`, `RegressionDiscontinuity.fit`,
+  `RDPlot.fit`, `joint_pretrends_test`, `joint_homogeneity_test`, `did_had_pretest_workflow`
+  (`M-035`..`M-042`, `M-088`, `M-089`, `M-098`..`M-112`) and `trim_weights(weight_col=)` ->
+  `weights=` (`M-113`), `StackedDiD(clean_control=)` -> `control_group=` (`M-043`),
+  `WooldridgeDiDResults.to_dataframe(aggregation=)` -> `level=` (`M-044`) and
+  `summary(aggregation)` (`M-087`; `summary(alpha=None)` is now the positional signature),
+  `ContinuousDiD(covariates=)` -> `fit(covariates=)` (`M-084`). The `"event"` value spelling is
+  rejected with `ValueError` on `WooldridgeDiDResults.aggregate(type=)`, `.to_dataframe(level=)` and
+  `LPDiDResults.to_dataframe(level=)` - use `"event_study"` (`M-086`, `M-150`, `M-136`).
+- **`robust=` removed** from `DifferenceInDifferences` (and so `TwoWayFixedEffects`,
+  `MultiPeriodDiD`), `TripleDifference`, `HeterogeneousAdoptionDiD` and `LinearRegression`
+  (`M-045`..`M-047`, `M-115`); `vcov_type=` is the single selector. The legacy-CR1 remap that
+  turned `robust=False` + clustering into `hc1` went with it. Positional constructor slots after
+  the removed parameter shift left by one (DiD slot 1, LinearRegression/TripleDifference slot 2,
+  HeterogeneousAdoptionDiD slot 6); `SyntheticDiD`'s inert `lambda_reg`/`zeta` (`M-001`, `M-002`,
+  constructor slots 7-8) are removed too, shifting `vcov_type` and the `conley_*` parameters two
+  slots left. Pass these by keyword.
+- **Results-field aliases removed**: `SyntheticDiDResults.placebo_effects` (`M-003`; use
+  `variance_effects`), `RegressionDiscontinuityResults.treatment_col` (`M-094`; `takeup`),
+  `StackedDiDResults.clean_control` and `StackedDiD.clean_control` (`M-095`, `M-043`;
+  `control_group`), `ChaisemartinDHaultfoeuilleResults.groups` (`M-114`; `units`),
+  `ChangesInChangesResults.estimator` (`M-143`; `method`). Their serialized dual keys are gone from
+  `to_dict()` (`treatment_col`, `clean_control`, `estimator`, Wooldridge's
+  `aggregation_weights["event"]`); 3.x pickles still load (`__setstate__` key migrations kept).
+- **The eight module-level wrapper functions are removed** (`M-070`..`M-077`): `imputation_did`,
+  `two_stage_did`, `stacked_did`, `trop`, `synthetic_control`, `triple_difference`,
+  `bacon_decompose`, `chaisemartin_dhaultfoeuille`. Call the estimator classes. `diff_diff.trop`,
+  `diff_diff.synthetic_control`, `diff_diff.stacked_did` and `diff_diff.chaisemartin_dhaultfoeuille`
+  now name the submodules (as `diff_diff.bacon` always did).
+- **The dieted aliases are removed** (`M-132`..`M-134`): `CDiD` -> `ContinuousDiD`, `Stacked` ->
+  `StackedDiD`, `Gardner` -> `TwoStageDiD`; the module `__getattr__` warning shim (`M-135`) is gone.
+
+### Changed
+- **`WooldridgeDiDResults.summary(alpha=None)`** renders the headline (simple) row only; the
+  group / calendar / event-study tables it used to render on request - and their `[pre]`/`[antic]`
+  relative-period labels - are no longer part of `summary()`. Use `aggregate(...)` /
+  `to_dataframe(level=...)`; `anticipation` is still reported on the results object.
+  `to_dataframe(level=)` now rejects an unknown level with `ValueError` instead of returning an
+  empty frame.
+- `practitioner_next_steps` no longer recognises duck-typed results that carry only the legacy
+  `estimator` attribute; `ChangesInChangesResults.method` is the routing field.
+
+### Breaking Changes
+- **`BusinessReport.to_dict()` schema `2.0` -> `3.0`**: the StackedDiD `assumption` block no longer
+  carries the legacy `clean_control` key (read `control_group`). `DiagnosticReport`'s schema stays
+  `2.0`.
+
+### Internal
+- `tests/test_v4_matrix.py` removed-status probes: a function row that resolves to a submodule
+  counts as absent; a param row absent from the explicit signature counts as removed despite
+  `**kwargs`. Naming-guard allowlist extended for the kept body-local aliases, `__setstate__` keys
+  and shared vocabulary. The `diff_diff._deprecation` helpers survive for the remaining
+  M-031/M-082, PR-3 and 5.0-window shims.