Delete dead staggered placebo helper, rewrite Step 6 prose

Remove the unused _placebo_step(staggered=True) branch that still
contained CS/SA-specific control_group/anticipation advice. The
function now only serves simple 2x2 designs.

Rewrite the Step 6 section in llms-practitioner.txt to give
estimator-family-specific falsification examples instead of generic
staggered advice that doesn't match all estimator APIs.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: igerber

diff_diff/practitioner.py

@@ -188,27 +188,8 @@ def _honest_did_step() -> Dict[str, Any]:
     )
 
 
-def _placebo_step(staggered: bool = False) -> Dict[str, Any]:
-    if staggered:
-        return _step(
-            baker_step=6,
-            label="Run sensitivity/falsification checks",
-            why=(
-                "For staggered designs, run_all_placebo_tests() is not "
-                "directly applicable (it refits a basic 2x2 DiD). Instead, "
-                "compare results across control group choices "
-                "(never_treated vs not_yet_treated), anticipation settings, "
-                "and subsample restrictions as falsification checks."
-            ),
-            code=(
-                "# Staggered falsification: compare across specifications\n"
-                "# - Re-estimate with control_group='never_treated' vs 'not_yet_treated'\n"
-                "# - Re-estimate with anticipation=1 to check no-anticipation\n"
-                "# - Drop one cohort at a time for leave-one-out sensitivity"
-            ),
-            priority="medium",
-            step_name="sensitivity",
-        )
+def _placebo_step() -> Dict[str, Any]:
+    """Placebo tests for simple 2x2 DiD designs only."""
     return _step(
         baker_step=6,
         label="Run placebo tests",

docs/llms-practitioner.txt

@@ -259,9 +259,15 @@ honest_sd = compute_honest_did(results, method='smoothness', M=0.5)
 
 ### Placebo tests (simple 2x2 designs only)
 `run_all_placebo_tests()` refits a basic 2x2 DiD — it is valid for simple
-designs but NOT for staggered adoption. For staggered designs, use
-specification-based falsification: compare results across control group
-choices, anticipation settings, and cohort subsets.
+designs but NOT for staggered, synthetic control, or other advanced estimators.
+
+**For non-2x2 estimators**, use specification-based falsification appropriate
+to your estimator's API. Examples:
+- **CS/SA**: compare `control_group='never_treated'` vs `'not_yet_treated'`
+- **StackedDiD**: vary `clean_control` definition
+- **EfficientDiD**: compare `control_group='never_treated'` vs `'last_cohort'`
+- **ImputationDiD/TwoStageDiD**: leave-one-cohort-out, cross-estimator comparison
+- **SyntheticDiD/TROP**: in-time or in-space placebo (fake treatment date, leave-one-unit-out)
 
 ```python
 from diff_diff import run_all_placebo_tests