From Data to Action: Creating Safer Public Spaces for Women in Mexico City

At a glance




2021 – 2022


Women’s Secretariat of Mexico City Government, GIZ Data Lab, UNDP Accelerator Labs, GbV


Open Data, Data Integration, Data Analysis, Data Visualisation, Positive Deviance, Public Safety, City Planning​


Open Data Portal of Mexico City, National Institute of Statistics and Geography


SDG #5, SDG #11

The Challenge

In Mexico, most girls and women experience violence in public spaces. In addition to the direct harm caused, this is limiting their full participation in public life. With some areas in Mexico City demonstrating better safety for women than others, further action was needed to understand these differences.​

The Approach

To understand what makes some public spaces safer than others in Mexico City, GIZ and the UNDP Accelerator Lab adopted a “Data Powered Positive Deviance” approach to analyzing public safety in the city. Using a combination of satellite imagery, crime statistics and qualitative data, the project team created a statistical model that identified which neighbourhoods and public spaces of the cities were relatively safer for women and what it was that made them safer. ​

The Benefits

Through the innovative use of data, actionable recommendations for creating safer public spaces for women were identified. The project illustrated how close collaboration allows data from government agencies to be efficiently combined with new data sources to produce more insights about complex social issues – an approach which can be replicated elsewhere.

The use case contributes to both SDG 5 (Gender equality) and SDG 11 (Sustainable cities and communities) in several ways. Identifying safer public spaces and the characteristics that make them more secure supports policymakers to develop targeted interventions to reduce discrimination against women and girls in public and mitigate the risk of violent assaults. The project team's use of data and evidence-based analysis further helps to inform urban planning and design decisions, guiding the development of safe, inclusive and accessible public spaces that meet the needs of all members of society.



of women in Mexico have experienced some form of sexual violence in public spaces.​

The context​

In Mexico, two out of three girls and women over 15 have reported experiencing at least one incident of violence in their lifetime. On average, 11 women are murdered daily; in 2019, more than 50% of female homicides occurred in public spaces.

Violence against women at the community level occurs in streets, parks, and, to a lesser extent, on public transportation. Attacks in public are most often sexual and can take the form of catcalling, bullying, stalking, sexual abuse and sexual assault. Further, there is an underreporting of these incidents. This problem is significant; it limits women's freedom of movement and diminishes their rights, which are anchored in Mexico City's Constitution, to safety in public spaces.

Against this backdrop, the Women's Secretariat of the Mexico City government, the GIZ Data Lab and the UNDP Accelerator Lab launched a project to identify public spaces in Mexico City where women are relatively safer. The results were presented at a session chaired by Mexico City's Women's Secretariat, which was attended by representatives of 16 government agencies of Mexico City.

“The recommendations [from this research] serve to strengthen the work and strategies that government agencies already have underway to improve the security of public space for women”

Ingrid Gómez Saracíbar, Head of the Women's Secretariat

How it was implemented

The project team piloted the Data Powered Positive Deviance method for developing public policy recommendations.

The Data Powered Positive Deviance (DPPD) method focuses on outliers, or positive deviants, and seeks to discover why some data points perform better than others. In this case the method is applied in order to understand why some public spaces are safer for women and girls than others.

The process for discovering such public spaces began with mapping the relevant data sources, grouping data points with similar characteristics and defining performance measures to identify positive outliers. This was followed by both quantitative and qualitative fieldwork, to collect and analyze the positively deviant factors.

Figure 1: Overview of the 5-stage model
Source: GIZ/Mexico

In stage 1, the project team conducted a series of interviews with experts, including academics, urban planners and representatives of women’s rights groups. This exercise provided initial guidance on the factors that make public spaces safer for women: urban infrastructure, security infrastructure, people, usage of space and mobility. Furthermore, initial datasets for the analysis were identified.

Figure 2: Step-by-step process of the DPPD method. Currently working on the positive deviant's identification
Source: GIZ/Mexico

Then, the team extensively mapped public and non-public datasets, adding to their understanding of their data ecosystem. The initial mapping created a wish list of 67 datasets that included urban infrastructure, population, commuting patterns, socioeconomic index, security and justice.

Two primary data sources:

the Open Data Portal of Mexico City and the National Institute of Statistics and Geography. Non-public datasets owned by government and private entities, such as mobile data, emergency telephone reports and usage of panic buttons were also considered.

Key data from several datasets were identified based on their relevance, level of “aggregation” (where data is compiled from multiple sources) and their timeliness. The project team selected open data related to urban infrastructure (e.g., subway stations, bus stations), land usage, security infrastructure (e.g., location of panic buttons, cameras), census data and marginalisation indexes (which combines socioeconomic data with crime rate statistics) for analysis. Additionally, an Attorney General's Office dataset was used containing current information on crime victims in Mexico City’s investigation files.

It proved important to get information that was as granular as possible to make use of geographic units of analysis small and precise enough to uncover the underlying factors behind better performing areas. The project team chose Áreas GeoEstadística Básica (AGEBs), geostatistical units that are commonly used in Mexico.

Finding better-performing areas of the city

The AGEBs were then sorted into groups with similar characteristics based on population density, commuter rates and the marginalisation index. The resulting clusters of AGEBs are shown below:

Figure 3: Cluster analysis shows the diversity of Mexico City.
Source: GIZ/Mexico
Developing a useful dataset

To identify positively deviant AGEBs (AGEBs that recorded a lower number of crimes against women than expected), the project team had to find a reliable performance measure. The Attorney General’s dataset of victims in investigation files between 2019 and 2020 was chosen. The dataset included information on 1) the type of crime, 2) the day of the week the crime occurred, 3) the time of the day it occurred, 4) the age of the victim, 5) the gender of the victim, and 6) the geolocation of the occurrence. This last aspect is particularly important, as it allows crime figures to be assigned to their respective AGEBs.

This dataset was then adjusted to only include gender-based violence crimes that occur in public spaces (e.g., sexual assault, feminicide). These crimes were then categorised by severity and impact on women.

Statistical methods used​

The next step of the DPPD method was to identify the positive deviants through quantitative analysis.

Statistical modelling created a predicted performance measure and by comparing this with the observed data, they derived “residuals”, which represent the difference between predicted and observed values. Where the number of victims observed per AGEB was much lower than the number of victims predicted by the model, a positive deviant was identified.

This modelling was built through three types of regression analysis and supplemented by qualitative interviews with women from Mexico City, which expanded the picture of their experiences, needs and desires.

How can better data contribute to better policy? ​

All three regression analyses demonstrated that several key variables are highly relevant for predicting the number of victims in each area. These variables include population size, disposability of financial services, the number of restaurants and bars and the proximity to the closest Metrobús and metro station. As the distance to the nearest metro station decreases, the number of victims is expected to increase, suggesting that areas in closer proximity to metro stations tend to have a higher incidence of violence. These insights can help policymakers target their responses for maximum impact.

Another key finding was the importance of taking advantage of specific characteristics of green areas in city planning to increase the perception of security and promote greater use of these areas by women, children, families and older adults. The increased presence of members of these groups deters crime.

Where do we go from here?​

This analysis strengthens the strategies that government agencies in Mexico are already undertaking to improve the safety of women in public spaces. The next step is to transfer learning to technical teams within government departments so that they can incorporate them into their work, and thus build safer public spaces for women in Mexico City.

The use case further served as a reference for another UNDP project focusing on Mexican women in unpaid care work.

Case downloads

A step-by-step guide for development practitioners to apply the DPPD method

Further resources

explore more Use Cases