It's just as likely that she heard that statistic many times and assumed that this must be because there were many studies with that result.
"She said that the research was replicated many times, and that no matter which country, or which year -- or even which definition of domestic violence was used! -- the results are always 20%" -- that's doesn't sound like a vague assumption to me. That sounds like she's being very specific.
I've started a podcast called Future Strategist which will focus on decision making and futurism. I have created seven shows so far: interviews of computer scientist Roman Yampolskiy, LW contributor Gleb Tsipursky, and artist/free speech activist Rachel Haywire, and monologues on game theory and Greek Mythology, the Prisoners' Dilemma, the sunk cost fallacy, and the Map and Territory.
If you enjoy the show and use iTunes I would be grateful if you left a positive review at iTunes. I would also be grateful for any feedback you might have including suggestions for future shows. I'm not used to interviewing people and I know that I need to work on being more articulate in my interviews.