A/HRC/7/6 page 12 approaches to estimating prevalence in diasporic communities.16 The World Health Organization (WHO) multi-country study on intimate partner violence contains a series of questions addressing forced, child and early marriage. 42. Prevalence rates for intimate partner violence are calculated from answering “yes” to any one of a list of behaviours: to “count�� as a victim, therefore, one needs only to have been pushed, for example, on a single occasion. While accuracy of measurement is important methodologically, and headline figures have undoubtedly raised awareness, it is an open question whether all single incidents should be understood as violence against women under the definition of the Declaration on the Elimination of Violence against Women. Within the NGO sector, domestic violence has long been understood as a “pattern of coercive control”. Some studies have begun to recognize these complexities and use data analysis to explore more complex patterns. An Irish study17 defined “severe domestic abuse” as a “pattern of physical, emotional or sexual behaviour between partners in an intimate relationship that causes, or risks causing, significant negative consequences”; minor incidents, in contrast, were defined as isolated events with little or no impact. Applying these analyses not only reduced overall prevalence findings but also the spurious finding that men are as likely to be victims of intimate partner violence as women. 43. Specialist surveys that address multiple forms of violence against women ensure that the continuum of violence can be traced with implications for protection, prevention and limiting harm. Methodological guidelines 44. Bearing in mind the caveats outlined above, it is nonetheless possible to identify consensus positions on good practices in measuring violence against women. Normalization makes asking “have you ever experienced violence/assault?” problematic. Therefore, the wording of questions is something of an art in research. The number, format and content of questions limit or facilitate disclosure. Single questions are less effective than multiple probes, and the most effective are behaviourally specific lists which avoid using terms such as “rape”, “assault” or “violence”. 45. Surveys will produce more accurate and useful findings where the following standards are observed: • Multiple opportunities to disclose through behaviourally specific questions • Inclusion of a range of forms of violence against women 16 See A Statistical Study to Estimate the Prevalence of Female Genital Mutilation in England and Wales (www.forwarduk.org.uk/download/96). 17 Watson, D. & Parsons, S. (2005), Domestic Abuse of Women and Men in Ireland, National Crime Council.