Using Data to Shed Light on the Shadow Pandemic of Domestic Violence in Mexico

Violence against women and girls (VaWG) is a worldwide phenomenon that occurs in various forms and spaces, unfortunately including where one should feel safest: at home with one's partner or family. In the context of the COVID-19 pandemic, as gender inequality and domestic violence have worsened, it is becoming increasingly clear how much we are still in the dark when it comes to identifying and supporting potential victims, including potential ones. The so-called “shadow pandemic” is characterized by a lack of information, incomplete data, and a culture of silence. How are national governments and support services supposed to respond to such an intimate and private, but also urgent, issue?

The GIZ Data Lab, GIZ Mexico, and the Data-Pop Alliance have joined forces to address the dilemma of missing information, data gaps, and the need for data-driven solutions. As a result, the team is developing an experimental risk model to map geographical areas in Mexico where women could be at most risk of experiencing domestic violence by leveraging traditional (administrative) and non-traditional (e.g. mobility data) sources of data. Read ahead to learn more about the project’s approach and progress.

Globally, it is estimated that 35% of women have experienced physical and/or sexual violence by their partner or another person at some point in their lives (WHO, 2013). Latin America shows the highest percentage of sexual violence (UNDP, 2017), and in Mexico, 66% of women age 15 years and over report to have suffered at least one incident of emotional, economic, physical, or sexual violence throughout their lives in at least one sphere (e.g. public, private) (INEGI, 2016). Reducing and eliminating VaWG can take many approaches, yet it is undeniable that it requires the creation of gender data that can be used to “develop and implement policies and monitor results, delivering on commitments to achieve equality and opportunities for women” (Data2x, 2020).

The pandemic, however, has not only exacerbated domestic violence in all corners of the world (as women have been forced to spend more time at home with their aggressor due to social distance and confinement measures), it has also highlighted the existing gaps and limitations of gender data. For instance, the lack and/or unavailability of it (partly because it’s not efficiently collected and is underreported); the inadequate temporal and spatial granularity; the problems in its articulation between institutions; the insufficient mechanisms used to leverage data insights into public policy; and the difficulties of collecting such intimate information directly from the women. The latter challenge is highly relevant amidst mobility restrictions and the possibility of contagion.

These issues prevail in Latin America, the second region in the world with the most missing data on VaWG (Yu-Chieh Hsu et al., 2017). The UN Women’s project “Global Database on Violence Against Women" which accounts for all measures (institutional and otherwise) undertaken by countries to address all forms of violence against women, found a total of 57 measures available in Mexico, a number much lower than that of Canada (146) or Australia (152). Despite the lack of quality gender data, we also believe more projects that take an innovative approach to leverage available data are sorely needed to produce evidence that can help mitigate the crisis.

Amid the pandemic, the Mexican government,  especially the local administration in Mexico City, are open and invested in finding data-driven approaches to already existing aid efforts. Our project places itself at this junction to generate policy and advocacy insights that move us closer to providing better strategies and support services for women who have the most risk of experiencing violence.

Our approach involves developing a risk map using data from Mexico City,  then extrapolating those results to other urban areas of the country for which data is insufficient or lacking. This shall be achieved through the following three stages of analysis:

• Mapping prevalence of domestic violence against women in Mexico City based on calls to Línea Mujeres and on the crime records from the Attorney General’s office, while controlling for underreporting. This is the greatest challenge, as an estimated 78% of women do not seek institutional help or report violence to the authorities in Mexico (INEGI, 2020). Yet, Mexico City has the largest amount of quality data available regarding reports of domestic violence. 
   
• Analyzing and measuring the impact of risk factors in Mexico City using traditional and non-traditional data sources, including the impact of social-distancing measures, sociodemographic indicators, and others. The challenge here lies in understanding which data can serve as proxies for which risk factors with a convenient space and temporal granularity, as well as to develop adequate algorithms for measuring impact.
   
• Extrapolating the results to other urban cities across Mexico for which open data regarding reports of domestic violence is unavailable. The challenge for this last stage is to transfer the behavior of Mexico City to other regions of the country through the information gathered by the risk factors in the previous stage.

The results from the risk model will be visualized in an interactive, color-coded map, where municipalities will be assigned a risk index value. The risk factors with the greatest impact in each area will also be highlighted.

Administrative records of domestic violence against women are usually collected by the Security, Health, and/or Women’s Ministries. In Mexico City’s case, two publicly available databases have allowed us to map the prevalence of domestic violence against women and girls:

• The Línea Mujeres helpline
• The crime records from the Attorney General’s Office

The Línea Mujeres helpline provides orientation and counseling services under three categories: medical, legal, and psychological. They also classify each call under up to seven different thematics. As part of the filtering methodology, the following thematics were selected for analysis: gender violence, dating violence, IPV, family violence, and penal Crimes. The crime reports database filings contain a large number of complaint categories. We selected the ones related to domestic violence against women (eg. sexual abuse, threats, family violence, against sexual intimacy). Overall, calls to Línea Mujeres spiked in 2020. In fact, there were  58% more calls from January to September 2020 than in the same period for 2019. The opposite seems to be true for crime reports (all categories), which show a 20% reduction during this year.

When looking at the selected thematics and crime categories related to domestic violence, we found a similar increase in calls of 31%  to  Línea Mujeres when comparing the 69-day period previous to the official quarantine or Healthy Distance Initiative, which lasted the same number of days in Mexico (March 26 - May 30). Again, the opposite trend is observed in the crime reports database, which shows a 46% drop in crime reports related to domestic violence versus the pre-quarantine period. 

The results seem to be analogous to those in other countries where emergency calls increased (e.g. France, Argentina, Singapore) as lockdown measures were put in place, while legal reports of incidents of domestic violence decreased due to “restrictions on mobility, lack of information, increased isolation, and fear” (IRC, 2020).

Number of calls and number of crime reports by date of crime. Aggregated by week (for selected thematics/categories). Updated: September 20th

Mexico City is divided into 16 municipalities or delegations. The graph below shows that the ones with the largest number of calls and crime reports during 2019 and 2020 have been Cuahtemoc and Tlalpan, which are very different areas, socioeconomically and demographically.

Number of calls and number of crime reports by 100,000 inhabitants. Aggregated from  January 2019 to September 2020.

In terms of specific categories, calls cataloged as “gender-based violence” showed the biggest percentage increase from 2019 to 2020. However, “family or domestic violence” crimes had the most reports made this year in the Attorney General’s office for Mexico City, which is significant considering the downward trend in legal complaints.

Number of calls and number of crime reports.

The literature looking at risk factors impacting violence against women is vast. We first developed a framework divided into four main categories: personal, relational, communal, and institutional. Transversally, we considered factors related to COVID-19 (e.g. mobility restrictions, financial distress) and structural factors. Of course, finding a proxy variable to measure each one through available sources of data was not possible.

Some factors studied and included in the risk map are:

• Age
• Marital status
• Education level
• Power asymmetry
• Economic asymmetry
• Financial independence
• Social cohesion
• Access to institutional support services
• Trust in institutions
• Human mobility
• COVID-19 epidemiological curve

The pandemic has highlighted economic and social inequalities, pointing, in many cases, to the lack of health infrastructure or communication campaigns in some regions. Like COVID-19, the gender-based violence epidemic affects the most vulnerable—women who are isolated, who lack access to communication channels, and are accompanied by institutions/organizations in the event of violence. Therefore, it is important to study the behavior of COVID-19 cases, as shown in the following chart, where we can see a high number of cases in Xochimilco, but not necessarily many calls or crime records, taking into account the underreporting of domestic violence crimes.

Number of calls, crime reports, and Covid cases by 100,000 inhabitants. Aggregated from March 2020 to September 2020 (during the different levels of lockdown).

Analyzing the calls, the marital-status risk of violence is similar for married and single women, with a little difference for the delegation of Cuauhtémoc. Otherwise, the education level shows different behaviors according to the municipality. 

Calls by marital status and education level in Mexico City by municipalities by 100,000 inhabitants. Updated  July 2020.

The behavior of calls by age shows a different behavior by municipality, but in general, according to the results, the majority of the calls came from women between 30 and 50 years old.

Age in Mexico City by municipality by 100,000 inhabitants. Updated to July 2020. 

Data can help us shed light on important issues and understand their magnitude, but it should also help us solve some of the most pressing challenges facing the world today.  Despite the limitations in an experimental project such as this one (e.g. underreporting, lack of quality data, and limited availability of data), we believe it is fundamental to continue to push projects forward that are leveraging alternative uses of administrative and non-traditional data sources. Especially when the goal is to produce insights that contribute to the reduction and elimination of violence against women.

To learn more about the situation during the contingency, read Data-Pop Alliance’s previous blog post: “Domestic Violence in Times of COVID-19.”

The sensitive nature of the data and the topic at hand call for mechanisms that ensure the inclusion of independent members of the civil association sector and others who can nourish the project by sharing their expertise and experience. This is why we built CODE (Council for the Orientation of Development and Ethics), a panel of gender and/or data experts from 10 organizations: UN Women Mexico, UNODC Mexico, EQUIS Justicia para las mujeres, UNAM, INEGI, Secretariat of Women, among others.

During our periodic sessions, we share our progress, and they provide advice and counsel for the development and deployment of the experiment, within the scope of the following functions: relevance, ethics, data protection, community engagement, recommendations, and visibility. Their involvement has been crucial, for instance, in helping to narrow the scope of our conceptualization of violence against women and in filtering the databases selected for its analysis.

The project by the GIZ Data Lab, GIZ Mexico, and the Data-Pop Alliance was selected from over 800 submissions to be presented along with 100 other projects at this year's Paris Peace Forum, held from November 11 to 13, 2020. The event is considered a platform for exchange between heads of state, international organizations, civil society, and the private sector to discuss innovative ideas and approaches and to facilitate joint activities in all areas of global governance. In the context of the pandemic, such an exchange is more important than ever. Under the motto "bouncing back to a better planet together," we would like to invite you to join us at this forum to discuss the potential of our risk model, as well as further possibilities to make the lived reality of women and girls more visible and to make sure nobody is left behind when working on a better and more just future.