Rates of violence vary widely from country to country, region to region, and community to community. Having good data on violence is important for designing effective programmes and advocating for change. The most reliable data on violence—specifically intimate partner violence, but also violence against children and other forms—comes from nationally or geographically representative surveys that have been specifically designed to collect data on violence. These surveys include extensive training of the interviewers and specific procedures to protect and support respondents, as well as to encourage disclosure (see Ethics and Best Practices).

Rates of violence captured by these surveys are much more accurate—and much higher—than those reported by administrative sources, such as police records, health clinics, or magistrates. Why is this the case? Only a small proportion of women who experience violence ever seek help from formal services. In a sample of low-income countries, for example, only 7 percent of women on average reported their experience to authorities, with regional variation between 2 to 14 percent (learn more here). Those who report to formal sources generally represent a more severe sub-set of all victims. Thus, this administrative data is not at all a reliable indicator of the actual proportion of women suffering violence.