|
Abstract
|
Sustainable Development Goal 5.2 urges countries to "eliminate all forms of violence against all women and girls". Indicator 5.2.1--the proportion of partnered women and girls subjected to intimate partner violence (IPV) in the previous 12 months--is recommended to monitor this goal. In their Article, Ning Ma and colleagues leverage data for 21 countries with two or more Demographic and Health Surveys to assess trends during 2000-21 in IPV for women aged 15-49 years.
We support this objective; however, our reading of their analysis and interpretations persuades us that a reliable, representative, and rigorous trend analysis of IPV against women is still needed.
First, the authors pool data across countries to find an overall annual negative change rate of 0·2% in combined IPV. However, since countries collect data in different years, pooled and regional trends might simply be a function of which countries have earlier or later data. Relatedly, the authors acknowledge that their sample includes countries willing to collect and to release data on IPV. This selectivity, alongside patchy coverage in the trend analysis (these data were only available in 21 of 134 low-income and middle-income countries), call into question what their pooled, regional, and income-group trends represent.
Second, the authors assume that IPV items are cross-country, cross-time measurement invariant, a requirement for comparative trend analysis. In our analyses of Demographic and Health Surveys data for 20 (mostly overlapping with those reported in Ma and colleagues' study) countries, physical IPV items approached approximate invariance, but controlling behaviours did not. If women's interpretation of or willingness to disclose the psychological IPV items they analyse changes over time, the suitability of these items for trend analysis is uncertain.
Third, variable content validity and composition of Demographic and Health Surveys IPV items might bias overall and type-specific trends. The Demographic and Health Surveys domestic violence module typically operationalises seven physical IPV items, two to three sexual IPV items emphasising acts of physical force, and two to three psychological IPV items. Yet, items overlook acts like non-contact sexual IPV and economic coercion. Moreover, the exact items asked, and their wording varies across countries and over time. If perpetrators adopt unmeasured acts of IPV, and the Demographic and Health Surveys items change over time, or both, trends in IPV could be misleading.
Finally, the clinical importance of a statistically significant 0·2% annual decline in IPV prevalence is questionable. Even at the high...
Language: en |