Method matters: differential reporting of gender-based violence by survey method, and potential for bias in treatment effect estimates

Abstract not yet added