BUG: Fix __ne__ comparison for Categorical

doc/source/whatsnew/v1.1.0.rst

@@ -198,6 +198,7 @@ Categorical
 
 - Bug where :func:`merge` was unable to join on non-unique categorical indices (:issue:`28189`)
 - Bug when passing categorical data to :class:`Index` constructor along with ``dtype=object`` incorrectly returning a :class:`CategoricalIndex` instead of object-dtype :class:`Index` (:issue:`32167`)
+- Bug where :class:`Categorical` comparison operator ``__ne__`` would incorrectly evaluate to ``False`` when either element was missing (:issue:`32276`)
 -
 
 Datetimelike

pandas/core/arrays/categorical.py

@@ -103,7 +103,10 @@ def func(self, other):
             mask = (self._codes == -1) | (other_codes == -1)
             if mask.any():
                 # In other series, the leads to False, so do that here too
-                ret[mask] = False
+                if opname == "__ne__":
+                    ret[(self._codes == -1) & (other_codes == -1)] = True
+                else:
+                    ret[mask] = False
             return ret
 
         if is_scalar(other):

pandas/tests/extension/test_categorical.py

@@ -282,6 +282,19 @@ def _compare_other(self, s, data, op_name, other):
             with pytest.raises(TypeError, match=msg):
                 op(data, other)
 
+    @pytest.mark.parametrize(
+        "categories",
+        [["a", "b"], [0, 1], [pd.Timestamp("2019"), pd.Timestamp("2020")]],
+    )
+    def test_not_equal_with_na(self, categories):
+        # https://github.com/pandas-dev/pandas/issues/32276
+        c1 = Categorical.from_codes([-1, 0], categories=categories)
+        c2 = Categorical.from_codes([0, 1], categories=categories)
+
+        result = c1 != c2
+
+        assert result.all()
+
 
 class TestParsing(base.BaseParsingTests):
     pass