Announcing the Data2X Big Data for Gender Challenge Awards
September 19, 2017
Data2X is pleased to announce the winners of the Big Data for Gender Challenge, ten projects that apply big data innovations to fill gender data gaps and improve understanding on key aspects of girls’ and women’s lives. These projects represent 29 researchers, from 20 different institutions, across 8 countries, and were selected from a pool of over 125 proposals.
The winning projects for the Big Data for Gender Challenge are:
1. Gender and Urban Mobility: Addressing Unequal Access to Urban Transportation for Women and Girls
2. Gender and Mobile Money Networks
3. Analyzing Big Data to Understand Uptake and Usage of Financial Services to Advance Women’s Financial Inclusion
4. Identifying Women in Mobile Phone Data to Further Financial Inclusion
5. Mining the Web for Insights on Violence Against Women and Conflict-Related Gender-Based Violence in the MENA Region and Arab States
6. Women in the Gig Economy: A Data Gap with Implications for Informal Work, Time Use and Poverty
7. Gender-Differentiated Credit Scoring Algorithms Using CDRs and Machine Learning
8. The Digital Traces of the Gender Digital Divide
9. Safety First: Perceived Risk of Street Harassment and Educational Choices of Women
10. Gender Disparity Signals: Analyzing Gender Disparities with Mobile Phone Metadata
Gender and Urban Mobility: Addressing Unequal Access to Urban Transportation for Women and Girls
The GovLab, UNICEF, ISI Foundation, Universidad del Desarrollo, Telefónica, DigitalGlobe
- Stefaan G. Verhulst, The GovLab, New York University
- Natalia Adler, UNICEF
- Ciro Cattuto, ISI Foundation
- Leo Ferres, Universidad del Desarrollo, Telefónica R&D Center
- Rhiannan Price, DigitalGlobe
Mobility, or the extent to which one can reach a desired destination, is one of our most basic needs. Access to mobility is also a prerequisite toward human development and having access to equal opportunities. As such mobility is a complex, gendered issue that requires a multidimensional, data-driven approach to fully unpack and offer insights on the way forward for decision-makers. Together, The GovLab (NYU), UNICEF, DigitalGlobe, IDS (UDD/Telefonica R&D), and the ISI Foundation will seek to embrace and decode the intersection of urban mobility and gender by uniquely combining a wide range of datasets, including commercial sources of call detail records and very high-resolution satellite data. The project seeks to define the potential and limitations of a Data Collaborative, a new form of public/private partnerships that combine data and expertise to create public value. Through this Data Collaborative on Gender and Urban Mobility, we want to know: does gender play a role in the way people move in a megacity such as Santiago, and, if so, how? Is there a mobility inequality from a gender perspective? What can be done to make transport planning more gender-sensitive and inclusive? And how can we replicate the analytic model to other places and contexts?
Gender and Mobile Money Networks
International Food Policy Research Institute, Drake University, Food and Agriculture Organization of the United Nations, Makerere University Business School
- Heath Henderson, Drake University
- Marya Hillesland, Food and Agriculture Organization of the United Nations
- Agnes Quisumbing, International Food Policy Research Institute
- Greg Seymour, CGIAR Research Program on Policies, Institutions, and Markets; International Food Policy Research Institute
- Benard Wabukala, Makerere University Business School
Gender and Mobile Money networks uses transaction-level data from mobile money users in Uganda to examine gender-related differences in mobile money access and usage. The project specifically focuses on how women are situated in the network of mobile money transactions and how locational differences influence access and usage. The objective is to identify constraints that women face in accessing resources and inform policy about how to alleviate those constraints.
Analyzing Big Data to Understand Uptake and Usage of Financial Services to Advance Women’s Financial Inclusion
Women’s World Banking
- Kate Hooper, Women’s World Banking
- Kathryn Glynn-Broderick, Women’s World Banking
- Tharun Jarugumalli, Women’s World Banking
Women’s World Banking will utilize individual-level data on women’s enrollment into and usage of digital financial services to analyze how women engage with digital financial services in Nigeria. Such an analysis is currently unprecedented in the financial inclusion industry. The findings will be able to inform the design of more women-friendly digital financial services. The policy relevance of such findings is substantial. Increasing women’s access to digital financial services creates pathways to formal financial services that can help break the cycle of poverty. A woman is still 20 percent less likely to own a bank account than a man is. By building a better understanding of women’s engagement and usage of digital financial services, we are able to uncover valuable insights about how women perceive, engage with, and interact with mobile banking, which could help drive financial inclusion for women around the globe.
Identifying Women in Mobile Phone Data to Further Financial Inclusion
Dalberg Data Insights
- Kristyna Tomsu, Dalberg Data Insights
- Jerome Urbain, Dalberg Data Insights
This project will build on Dalberg Data Insights’ existing work and partnerships with telecom operators to identify women in mobile phone data and, in turn, produce comprehensive gender-disaggregated mobile money insights that answer key questions around female financial inclusion. We will use a detailed dataset of mobile phone and mobile money records from both MTN Uganda and Airtel Uganda, the two largest telecom operators in the country, in three stages of work First, we will strengthen our existing algorithms to identify women and aim to make the model more robust using ground-truth data; then we will apply our model to produce gender-disaggregated indicators for Mobile Money in Uganda. Finally, we will develop a case study with the UN Capital Development Fund to address the question: What are the usage patterns of women in terms of mobile money? This case study will include recommendations on ways to further the financial inclusion of women in Uganda.
Mining the Web for Insights on Violence Against Women and Conflict-Related Gender-Based Violence in the MENA Region and Arab States
Cadi Ayyad University
- Jihad Zahir, Cadi Ayyad University
Location: Middle East and North Africa
The web contains vast amounts of content that is continuously being updated. Social media provide new channels for people to express their ideas and encourage frequent user expressions of their thoughts, opinions and sometimes details of their daily personal information and activities. Analyzing User Generated Content (UGC) in the web, ranging from comments to broader interactions, can help in measuring different characteristics and indicators. Therefore, we see the web as a rich source of data to which data mining and Natural Languages Processing (NLP) can be applied to extract useful information about a specific topic of interest, in our case, violence against women and conflict related gender based violence in the MENA region and Arab states.
The main goal of this project is to get sense of feelings, perceptions and reactions of web users to content and messages that are directly or indirectly tackling violence against women and conflict related gender based violence. Upon completion of this project, a set of knowledge and insights will be available on violence against women and conflict-related gender-based violence in the MENA region, using web mining, machine learning and Natural Language Processing. The research team will have contributed to advancing the state-of-the-art web mining and natural language processing for Arabic content.
Women in the Gig Economy: A Data Gap with Implications for Informal Work, Time Use and Poverty
Overseas Development Institute, Ulula, Data-Pop Alliance
- Emma Samman, Overseas Development Institute & Data-Pop Alliance
- Abigail Hunt, Overseas Development Institute
- Antoine Heuty, Ulula
- Emmanuel Letouzé, Data-Pop Alliance
Location: Kenya, Mexico
This study aims to provide insights into women’s experiences of ‘gig work’ and factors conditioning their job satisfaction. Given growing media and policy attention to the ‘Uber-ization’ of the labour market, an improved evidence base on how the gig economy is developing, what it offers women workers and its consequences for them is urgently needed. It also holds promise to yield new data on earnings and on how this new form of work affects time spent on unpaid care and domestic work, two high-priority SDG data gaps relating to gender. This research has two inter-related elements. First, we will analyze the data from companies operating gig platforms in Kenya and in Mexico respectively to assess the quantity and quality of work that they offer, factors conditioning worker success and how this varies over time. Second, we will conduct a mobile-phone based longitudinal survey of women working through these platforms to understand better their experiences of the gig economy and how this affects (and is affected by) their involvement in unpaid care and domestic work. Through the analysis of platform data and of worker experiences, and in synthesizing their insights, this work holds the promise both to make methodological innovations to the study of the evolving employment landscape and to influence policies in a rapidly shifting context.
Gender-Differentiated Credit Scoring Algorithms Using Call Detail Records and Machine Learning
University of California, Berkeley
- Paul Gertler, UC Berkeley
- Joshua Blumenstock, UC Berkeley
- Laura Chioda, World Bank
- Sean Higgins, UC Berkeley
Location: Dominican Republic
Low-income women disproportionately lack access to credit, often because they lack credit histories, property rights, and formal earnings. In other words, a creditworthiness data gap exists that prevents lenders from being able to assess the credit-worthiness of low-income women in developing countries, and lend to those who would be most likely to repay and who would most benefit from credit. We are interested in understanding whether gender-differentiated credit scoring models using big data from call detail records (CDR) can increase women’s access to formal credit. Specifically, we plan to test a new approach to credit scoring that allows for men and women to have different determinants of loan eligibility. In close collaboration with the in-country partner bank Asociación La Nacional de Ahorros y Préstamos in the Dominican Republic, we are working to use call detail records from the Dominican Republic’s largest mobile network operator to predict creditworthiness for low-income women who lack credit histories. The credit scoring model will use machine learning algorithms to sift through a broad range of applicant characteristics revealed through mobile phone CDR to determine the best predictors of creditworthiness separately for men and for women.
The Digital Traces of the Gender Digital Divide
Oxford University, Qatar Computing Research Institute
- Ridhi Kashyap, University of Oxford
- Ingmar Weber, Qatar Computing Research Institute
- Masoomali Fatehkia, Princeton University
This project will leverage Facebook’s ad audience estimates, specifically data on aggregate numbers of Facebook users by gender, age and device type across different countries, as a type of digital census in combination with development indicators and data derived from official statistics to measure digital gender gaps in internet and mobile phone access for 140+ countries. Furthermore, it will seek to assess how online inequalities, as captured by Facebook-derived indicators such as female-to-male ratios of online visibility, link with different offline gender inequality measures, such as those in the sub-indices of the Global Gender Gap Index (GGI), including educational and economic opportunity for women and girls. By analyzing how online gender inequalities link with offline ones, we will generate an automated tracking system that makes “nowcasting” predictions for various indicators in real-time. Our project will seek to provide: i) an easy-to-use web application that provides downloadable indicators, visualizations and tracking of digital gender gaps, and online-offline gender inequalities; ii) research output documenting our idea, methods, and the indicators that we generate.
Safety First: Perceived Risk of Street Harassment and Educational Choices of Women
- Girija Borker, Brown University (PhD)
This project examines the impact of street harassment on college choice of women in Delhi, assessing whether psychological costs associated with sexual harassment while traveling make women choose worse quality colleges compared to men – a decision that affects women’s academic training, network of peers, access to labor opportunities, and lifetime earnings. Overall, the project aims to quantify the tradeoffs women make between college quality, travel safety, travel time, and travel costs while making higher education choices. To address this question, the researcher assembles a unique data set that combines student information from a primary survey of 4,000 students in the University of Delhi, mapping of their alternative travel routes using Google Maps, and crowd sourced mobile application safety data. The researcher relies on rigorous econometric techniques for her analysis using a combination of reduced form and structural estimation methods. In the first study to assess the long term economic consequences of street harassment, this paper provides a more holistic view of the factors that affect higher education choices of women in rapidly urbanizing countries.
Gender Disparity Signals: Analyzing Gender Disparities with Mobile Phone Metadata
- Muhammad Raza ur Rehman Khan, UC Berkeley (PhD)
One of the biggest problems for governments and development agencies around the world is the lack of accurate information about gender disparities. This project aims at modeling educational gender disparity at the district level in Pakistan using the network data extracted from the call detail records. Using survey-based educational gender disparity as the ground truth, we first intend to explore the prominent network metrics and patterns that can improve understanding of the gender disparities in the population. In the second half of this project, we intend to use these networks metrics to predict educational gender disparity at the district level. Our model would use advanced network features, extraction methods and prediction algorithms to model the educational gender disparities accurately. We will test our models on a massive data-set of network activity of more than 30 million customers. This research project would help in the understanding of the extent the social disparities can manifest themselves in the social networks. Furthermore, the predictive model would help in relatively low cost and faster estimation of the gender disparities in future. Lastly, our model can be applied to other countries provided the gender annotated call detail records are available.