fix(lpdid): report survey PSU count as headline G for only_event survey fits

CI-codex P2: under a survey design the effective variance cluster is the PSU
(cluster_name reports the PSU column), but for only_event=True fits (pooled is
None) headline_n_clusters fell back to the panel unit count -- so an explicit
PSU design with n_psu != n_units could display the unit count mislabeled as
G=<psu>. Per-row event-study n_clusters and inference were already computed on
the realized survey design, so this was a metadata/labeling issue only, not a
wrong SE/p-value. Fix: when a survey design is active, seed headline_n_clusters
from the panel-level effective PSU count (the pooled-post override still prefers
the realized survey-sample count when available). Regression test added
(only_event=True, explicit PSU, n_psu != n_units).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: igerber

diff_diff/lpdid.py

@@ -1499,9 +1499,16 @@ def fit(
             )
 
         # Headline G = realized clusters in the pooled-post (headline ATT) sample
-        # when computed; otherwise the panel-level unit-cluster count. Per-horizon
-        # rows carry their own realized n_clusters in the event_study/pooled tables.
+        # when computed; otherwise the panel-level cluster count. Per-horizon rows
+        # carry their own realized n_clusters in the event_study/pooled tables.
         headline_n_clusters = int(panel["_cluster"].nunique())
+        if survey_n_psu is not None:
+            # Under a survey design the effective variance cluster is the PSU, and
+            # cluster_name reports the PSU column. Use the panel-level PSU count so the
+            # headline G matches that label even for only_event fits (pooled is None);
+            # an explicit PSU design can have n_psu != n_units. The pooled-post override
+            # below still prefers the realized survey-sample count when available.
+            headline_n_clusters = survey_n_psu
         if pooled is not None:
             _post = pooled.loc[pooled["window"] == "post", "n_clusters"]
             if not _post.empty and pd.notna(_post.iloc[0]):

tests/test_lpdid.py

@@ -1864,6 +1864,22 @@ def test_get_params_excludes_survey_design(self):
         c = clone.fit(_survey_panel(), survey_design=_full_design(), **_FIT_KW)
         assert c.vcov_type == "survey_tsl"  # survey applies because passed to fit()
 
+    def test_only_event_survey_headline_g_is_psu_count(self):
+        # only_event=True (pooled is None) must still report the survey PSU count as the
+        # headline G, matching cluster_name=PSU -- not the unit count. The panel design
+        # has 10 PSUs across ~80 units (n_psu != n_units).
+        df = _survey_panel()
+        n_units = df["unit"].nunique()
+        res = LPDiD(pre_window=2, post_window=2).fit(
+            df, survey_design=_full_design(), only_event=True, **_FIT_KW
+        )
+        assert res.pooled is None
+        assert res.cluster_name == "psu"
+        assert res.n_psu == 10
+        assert n_units != 10
+        assert res.n_clusters == 10  # headline G == PSU count, NOT n_units
+        assert "G=10" in res.summary()
+
     # --- NaN-consistency: every stratum singleton + lonely_psu="remove" ---
     def test_all_singleton_strata_remove_is_nan(self):
         base = _survey_panel().drop(columns=["stratum", "psu", "fpc", "weight"])