There are a variety of ways this could be possible. For instance, feminism might have a PR problem, where everyone believes a stereotype that isn't true, even people in the group being stereotyped. Having direct knowledge might only be enough to partially overcome the stereotype.

Alternatively, feminists might speak and/or act negatively towards men--the study authors didn't measure what anyone actually did, only what people's attitudes were--far more freely than non-feminists. Since actions speak louder than words, as the saying goes, everyone may correctly be judging that feminism is relatively hostile to men in practice, but the hostility comes despite near-normal levels of internal positivity.

One can come up with other hypotheses also; the point is that you can't really tell if you don't measure it because humans are complicated.