That isn't logically valid. It's possible for a single person to abuse more than one woman. Therefore, the percentage of abusers in the population is likely lower than the percentage of abused. I don't know how much lower that is, but "less than 10%" is entirely plausible.

There's also a potential confounding effect if a higher percentage of abusers and abused remain in the same set of relationships, e.g. an abused person moves from one abuser to the next. It looks like sociopathy/psychopathy has a higher prevalence in the abuser population, about 15% to 30% for "batterers", from Domestic violence and psychopathic traits: distinguishing the antisocial batterer from other antisocial offenders., with a non-paywall version here. Most of the studies those results are based on seem to be relatively small (N < 100) and essentially self-selected, but I couldn't find anything better.