Fake Conclusions From Cherrypicked Data