diff --git a/doubleml/plm/plpr.py b/doubleml/plm/plpr.py
index bf860b77..c5507f56 100644
--- a/doubleml/plm/plpr.py
+++ b/doubleml/plm/plpr.py
@@ -335,9 +335,9 @@ def _transform_data(self):
 
     def _set_d_mean(self):
         if self._approach in ["cre_general", "cre_normal"]:
-            data = self._original_dml_data.data
-            d_cols = self._original_dml_data.d_cols
-            id_col = self._original_dml_data.id_col
+            data = self._dml_data.data
+            d_cols = self._dml_data.d_cols
+            id_col = self._dml_data.id_col
             help_d_mean = data.loc[:, [id_col] + d_cols]
             d_mean = help_d_mean.groupby(id_col).transform("mean").values
             self._d_mean = d_mean
diff --git a/doubleml/plm/tests/test_plpr_transformations.py b/doubleml/plm/tests/test_plpr_transformations.py
index db162d0a..d6ea2d23 100644
--- a/doubleml/plm/tests/test_plpr_transformations.py
+++ b/doubleml/plm/tests/test_plpr_transformations.py
@@ -87,6 +87,7 @@ def test_plpr_approach_x_dim(approach, time_type):
         static_panel=True,
     )
     dml_plpr = dml.DoubleMLPLPR(obj_dml_data, ml_l, ml_m, approach=approach)
+    dml_plpr.fit()
     if approach == "wg_approx":
         assert len(dml_plpr._dml_data.x_cols) == dim_x
     else:
@@ -106,6 +107,7 @@ def test_plpr_approach_d_mean(approach, time_type):
         static_panel=True,
     )
     dml_plpr = dml.DoubleMLPLPR(obj_dml_data, ml_l, ml_m, approach=approach)
+    dml_plpr.fit()
     if approach in ["cre_general", "cre_normal"]:
         assert dml_plpr.d_mean is not None
     else:
@@ -145,6 +147,7 @@ def test_plpr_fd_exact_unbalanced(time_type):
     )
     with pytest.warns(UserWarning, match=msg_warn):
         obj_plpr = dml.DoubleMLPLPR(obj_dml_data, ml_l, ml_m, approach="fd_exact", n_folds=2)
+        obj_plpr.fit()
     # 4 rows after fd transformation as id 3 has no possible first difference
     assert obj_plpr.data_transform.data.shape[0] == 4
 
@@ -168,6 +171,7 @@ def test_plpr_one_id(approach, time_type):
     )
     with pytest.warns(UserWarning, match=msg_warn):
         obj_plpr = dml.DoubleMLPLPR(obj_dml_data, ml_l, ml_m, approach=approach, n_folds=2)
+        obj_plpr.fit()
     # 2 rows after fd transformation, 4 rows else
     if approach == "fd_exact":
         assert obj_plpr.data_transform.data.shape[0] == 2
@@ -196,6 +200,7 @@ def test_plpr_fd_exact_one_id_unbalanced(time_type):
     # capture warnings
     with pytest.warns(UserWarning) as record:
         obj_plpr = dml.DoubleMLPLPR(obj_dml_data, ml_l, ml_m, approach="fd_exact", n_folds=2)
+        obj_plpr.fit()
     # assert two warnings were raised and content
     assert len(record) == 2
     assert msg_warn_one_id in str(record[0].message)
@@ -215,6 +220,7 @@ def test_plpr_time_cre_transformation(cre_approach, data_time_type):
         static_panel=True,
     )
     dml_cre = dml.DoubleMLPLPR(obj_dml_data, ml_l, ml_m, approach=cre_approach, n_folds=2)
+    dml_cre.fit()
     assert dml_cre.transform_cols["y_col"] == "y"
     assert dml_cre.transform_cols["d_cols"] == ["d"]
     assert dml_cre.transform_cols["x_cols"] == ["x1", "x2", "x1_mean", "x2_mean"]