The number of domestic violence cases reported to the South African police between March and April dropped by 69.4%. This figure makes it tempting to believe that in South Africa, unlike many other countries, lockdown in response to the Covid-19 pandemic reduced family and domestic violence.

However police statistics, call numbers to hotlines and counts of women accessing services during lockdown tell us little about the number of women and families actually experiencing violence at the hands of men.

Many countries have documented an increase in reported cases of violence against women and children under lockdown. This may be because of the high levels of anxiety and tension in households where people are living together in close proximity. Unemployment, hunger, isolation and uncertainty also probably play a role. These are stresses that many, if not most, South African families experience.

Police statistics of reported cases and the number of calls to hotlines or non-governmental organisations are not sources of information on the extent of violence experienced by women. But they can tell us who was accessing services and reporting cases, and they raise important questions about whether it was possible for women to access services under lockdown.

What data should we be using to understand levels of violence against women during Covid-19? And what does the reduction in reporting in South Africa tell us about the nature of abuse, and women’s trust in the systems intended to support them?

It isn’t surprising that the police, NGOs and shelters reported a drop in cases during lockdown. Could women confined with their abusers find a safe time, a private space, and means to make contact? Did they trust support systems enough to believe they would get help if they risked asking? Did they know who to contact? And in the face of highly publicised abuse by some security force members, could women risk trying to access shelters or courts?

It is important to understand the impact of lockdown and the alcohol ban on levels of violence in our homes. But the available information, i.e. service level data, cannot demonstrate these trends. That being the case, can usable and actionable statistics on domestic violence during the pandemic be collected ethically and safely?

Rigorous studies on the impact of pandemics on levels of violence against women and children are scarce. It’s difficult to ensure the safety of women who are asked to respond to questions about their experiences of abuse, so the accuracy of the information gathered is likely to be questionable.

Social distancing and lockdown limit our ability to conduct face-to-face interviews and do research according to accepted ethical standards. This applies particularly to keeping respondents safe and their answers confidential during data collection.

And while reliable information during a pandemic is important, international organisations agree that the priority must be to ensure that survivors have access to good-quality support and services.

Data about the scale of violence against women and children can most reliably be captured through large national specialised surveys that use standard measures and highly trained enumerators. These include the World Health Organization Multi-Country Study on Women’s Health and Domestic Violence against Women, the United Nations Multi-Country Study on Men and Violence in Asia and the Pacific, and the Violence against Children Surveys.

Administrative or service data (e.g. from police, justice, or health and social services) are poor measures of domestic violence levels. This violence is largely under-reported and many survivors don’t seek help. When they do, cases may not always be properly recorded.

The research is clear about the factors that drive intimate partner violence. These include gender inequality, childhood emotional abuse and neglect, frequent arguing with a partner, depression, and gender norms that support men’s use of violence.

The image below shows findings from the UN Multi-Country Study on Men and Violence in Asia and the Pacific. It aimed to generate knowledge about the factors that contribute to men’s perpetration of intimate partner violence against women.

The circle sizes indicate the relative impact of each factor. The larger the circle the greater contribution the factor has on levels of intimate partner violence. The colours indicate thematic groups. This is a clear visual representation of how complex the drivers and risk factors of intimate partner violence are.

South African research also points to the significant role that alcohol and drugs, gender norms and beliefs about the roles of women in society, and childhood trauma play in increasing the likelihood that men will perpetrate violence against women.

The important questions to ask during this pandemic are not whether the violence has increased or decreased (which we cannot answer). We should rather be asking about the impact of the virus and social distancing measures on women and children, and whether they’ve been able to access services.

This question is best answered by those working with women and children on the frontline. We also need to learn from others elsewhere, and build a global set of measures to inform current responses to and future management of pandemics.

What we know suggests that social distancing regulations and lockdowns are unlikely to reduce violence in the home. As we emerge from the Covid-19 crisis, it will be important to undertake research to better understand how women and families experienced the lockdown and what informed the reduction in reporting.

Meanwhile the Department of Social Development and National Treasury should ensure that high-quality services are available to victims of domestic violence, and that preventive services aren’t compromised by the shift in state spending to address the pandemic. Finally, we need to be asking what can be done to ensure that women are not cut off from services when social distancing is required.

Written by Elizabeth Dartnall, Sexual Violence Research Initiative (SVRI); Anik Gevers, SVRI; Chandré Gould, Institute for Security Studies; and Angelica Pino, SVRI