were significantly more likely (53%) to come from rural as opposed to urban (35%) households”.90 Rural residents were also significantly more likely than urban respondents to experience all types of abuse.91 In addition to the statistics being wrong due to erroneous data entry, it also highlighted a flawed approach to data analysis that continued through to the SFSS 2017. In the SFHSS 2000 analysis, it was concluded that the majority of abused respondents come from rural areas. This is to be expected given 82% of the total population in Samoa live in rural areas.92 However, the report then concludes that women in rural areas are more likely to experience abuse than those living in urban areas. The report is now confusing prevalence with rate, or risk. To come to this conclusion accurately, the analysis needed to go one step further and look at the prevalence of violence in rural areas and urban areas, and compare both to the national average to see if those in rural areas are more at risk. The Inquiry was able to use the data available from the SFHSS 2000 and reverse engineer the pure numbers and make these calculations for all demographics apart from rural versus urban due to the basic error that had been made in entering the numbers into the table. This confusion between prevalence and rate/risk was a mistake carried over to the SFSS 2017 whereby only prevalence was analysed. Whilst it is important to identify the demographics among which family violence occurs the most, it is also important to identify the rate, within specific demographics to identify particular risk factors. For example, it is evident that the education level of the biggest group of respondents who had experienced abuse was secondary level education. But, using the analysis the Inquiry was able to perform using the SFHSS 2000, the rate of violence within that demographic was roughly the same as the national average, whilst the rate of violence among women who had finished primary school was much higher. The SFSS 2017 goes on to make a number of other mistakes which undermine its findings, including drawing conclusions from sample sizes of 22 (persons with disabilities) and using an erroneous definition of abuse for elderly persons, which significantly inflated the findings.93 Furthermore, the SFSS 2017 did not replicate the methodology of the SFHSS 2000, meaning the data is incomparable. It appears that the SFSS 2017 did not benefit from the same level of statistical technical expertise as the SFHSS 2000 but that both made significant mistakes and included poor analysis which could significantly impact the effectiveness of prevention programmes and strategies based upon this data. Although the SFHSS 2000 and SFSS 2017 were carried out in good faith and sheds light on the prevalence of family violence, it is the view of the Commissioners that higher standards should be demanded by both the implementing agencies and the funding partners when carrying out similar exercises in the future. 90 SPC and UNFPA. (2006). Samoa Family Health and Safety Study 2000. Document 83 in Inquiry anthology, accessible at www.ombudsman.gov.ws. 91 SPC and UNFPA. (2006). Samoa Family Health and Safety Study 2000. Document 83 in Inquiry anthology, accessible at www.ombudsman.gov.ws. 92 Samoa Bureau of Statistics. (2016). Samoa Census 2016. Retrieved from www.sbs.gov.ws/index.php/new-documentlibrary?view=download&fileId=2114. 93 This erroneous definition may also have been applied to other demographics with the same effect but an Inquiry request for clarification went unanswered.