This kind of has been my experience as well in Statistics/EE/Machine Learning. While, IEEE/ACM have word style files, I have seen very few submissions that use word. In fact, I know reviewers who have had a significant bias (which whether it is right or not exists) against word submissions.