Measuring gender-based violence in context: considerations for the UniSAFE project

The UniSAFE project aims at providing an empirical measurement of the prevalence of gender-based violence in the context of universities and research organisations throughout Europe. A cross-cultural survey across 45+ research organisations in 15+ countries is planned for 2022, providing the first of its kind source of data on this topic. Besides prevalence, the survey also seeks to understand the mechanisms of gender-based violence, including its determinants and consequences. The survey itself will provide evidence about individuals (both staff and students),but be informed by broader empirical research looking at the institutional environment and national legal and policy frameworks. Measuring the prevalence of gender-based violence is not without its challenges, particularly within a comparative framework across a large number of countries and institutions. Previous large-scale surveys, though focusing on the general population rather than universities or research organisations, have been questioned about the validity of their measurement. One of the most robust sources of data on violence against women - rather than gender-based violence - is the EU-wide survey carried out by the European Agency for Fundamental Rights (FRA). While some researchers (e.g. Gracia et al., 2019) have scrutinised its data and established measurement validity (i.e. the idea that the questions about violence are understood in the same way by respondents in different countries), others have questioned how to make sense of the data in the presence of a so-called ‘Nordic Paradox’ (Gracia and Merlo, 2016; Ivert et al., 2020; Humbert et al., 2021): higher (disclosed) prevalence of violence against women in countries that rank as more gender equal, including the Nordic countries. Where researchers agree on the necessity to consider the measurement of the prevalence of gender-based violence in the context in which it takes place, be it at institutional or national level. In this paper, we examine how differences in prevalence of violence across institutions might also respond - or not - to measurement invariance. Prevalence of violence does not capture all incidents of violence, but only those that are disclosed by participants. As such, prevalence always under-estimates the actual prevalence of violence. The question of interest is of course by how much this might be underestimated, and what are the factors (institutional or national) that might be relevant to disclosing incidents of violence. The paper, rather than relying on empirical results, wants to set out a first discussion of the factors that need to be considered in such analysis, and initiate a discussion on the context-specific indicators to create and incorporate into further empirical assessment using the survey and complementary qualitative data obtained from the national and institutional mappings.