BUG: does .any(axis=1, skipna=False) follow Kleene Logic?

[MarcoGorelli]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

In [16]: pd.DataFrame({
    ...:     'a': [True, False, None],
    ...:     'b': [None, None, None],
    ...: }, dtype='boolean[pyarrow]').any(axis=1, skipna=False)
Out[16]:
0     True  #
1    False  # <=== should be NA?
2    False  # <=== should be NA?
dtype: bool[pyarrow]


In [17]: pd.Series([False, None], dtype='boolean[pyarrow]').any(skipna=False)
Out[17]: <NA>

In [18]: pd.Series([None, None], dtype='boolean[pyarrow]').any(skipna=False)
Out[18]: <NA>

In [19]: pd.Series([True, None], dtype='boolean[pyarrow]').any(skipna=False)
Out[19]: True

Issue Description

I don't really understand what's happening, nor what's meant to be happening

The docs for Series.any say

If skipna is False, then NA are treated as True, because these are not equal to zero.

but that doesn't seem to be the case. If that was the case, then

pd.Series([False, None], dtype='boolean[pyarrow]').any(skipna=False)

would return True?

Instead, it looks like Series.any follows Kleene Logic. Fair nuff - but then, DataFrame.any(axis=1, skipna=False) doesn't?

Expected Behavior

I think I'd expect

0     True
1     <NA>
2     <NA>

from the DataFrame.any(axis=1, skipna=False) example

Installed Versions

INSTALLED VERSIONS

commit : fd3f571
python : 3.10.12.final.0
python-bits : 64
OS : Linux
OS-release : 5.15.133.1-microsoft-standard-WSL2
Version : #1 SMP Thu Oct 5 21:02:42 UTC 2023
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_GB.UTF-8
LOCALE : en_GB.UTF-8

pandas : 2.2.0
numpy : 1.26.3
pytz : 2023.3.post1
dateutil : 2.8.2
setuptools : 59.6.0
pip : 22.0.2
Cython : None
pytest : 7.4.4
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.1.2
IPython : 8.19.0
pandas_datareader : None
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : 4.12.2
bottleneck : None
dataframe-api-compat : None
fastparquet : None
fsspec : 2023.12.2
gcsfs : 2023.12.2post1
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 15.0.0
pyreadstat : None
python-calamine : None
pyxlsb : None
s3fs : None
scipy : 1.11.4
sqlalchemy : 2.0.25
tables : None
tabulate : 0.9.0
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.4
qtpy : None
pyqt5 : None

[phofl]

pretty sure that axis=1 casts to object, thus opting out of arrow dtypes and falling back to the default numpy behavior

[rhshadrach]

Not certain there is a cast to object, we go through _reduce_axis1 here:

pandas/pandas/core/frame.py

Line 11471 in c3014ab

def _reduce_axis1(self, name: str, func, skipna: bool) -> Series:

But we're using NumPy and/or ufuncs.

Edit: Doh, of course that has to cast to NumPy object dtype to work.

[qmarcou]

Hi!
Just ran into the same issue here. Regarding Series.all, is it a documentation issue or an implementation issue?
Regarding the DF case, could simply enforcing "boolean" dtype after calling _reduce_axis1 if all columns had "boolean" dtype solve the issue?