More tightly mapped trends in girls’ stunting and access to contraception in Bangladesh. A better understanding of women’s mobility in a Latin American city. Stronger insights into women’s mental health via social media in cities around the world.
All of these findings can be traced to big data — and to a three-year project spearheaded by the United Nations Foundation’s initiative Data2X to apply large data sets to help close the gaping gender data gap.
Big data generators such as credit card use, mobile phone calls, social media posts and satellite imagery are already being used to some extent to inform mainstream development efforts, and certainly to inform corporations about their consumers.
“But there really hadn’t been any [big data] work that we identified that was looking specifically at gender issues,” Rebecca Furst-Nichols, Data2X deputy director, told Devex.
In a new era of high-level advocates, pledged millions and trending Twitter hashtags, gender data seems to have "made it." Those who have been working in the once-shadowy field tell Devex it's time to take next steps. But first, five things to know about the gender data revolution.
Several years ago, when looking through Data2X’s initial research into 28 gender data gaps, or areas where sex disaggregated data is missing and vital, “we realized that all of them wouldn’t be able to be covered through traditional data collection efforts, like surveys or administrative data,” Furst-Nichols said. These gaps included everything from details on violence against women to comparable information on voter registration.
“We wanted to get out in front of the big data for development field and show that there are so many potentials to get information on women and girls’ lives,” she added.
At the time, the field of big data for development was nascent and growing, although many who considered themselves part of the “traditional statistics crowd” were skeptical of big data, Furst-Nichols noted. There will always be a place for conventional forms of data such as household surveys, national economic accounts and institutional records in closing the gender data gap, but there is also always room for new and useful data sources, she said.
“We didn’t say ‘[big data] is a panacea and it’s going to save women and girls’ lives,’” Furst-Nichols told us. “We said ‘big data needs to be complemented by other forms of traditional data and can amplify the findings and insights that we have from survey data.’”
With approximately $1 million in support from the William and Flora Hewlett Foundation and the Bill & Melinda Gates Foundation, Data2X hired lead economist Bapu Vaitla and spent two years piloting projects around the world to put big data skeptics’ fears to rest. The aim was to prove how big data could inform on key indicators for women and girls.
Now, the U.N. Foundation’s data arm is joining the digital data conversation by presenting results in three major data types: Geospatial, digital exhaust and internet data. Devex breaks it down here.
1. Geospatial data can predict women’s well-being.
Researchers at the Flowminder Foundation and WorldPop project used satellite imagery to improve existing data on women and girls from demographic and health surveys in Bangladesh, Haiti, Kenya, Nigeria and Tanzania.
Why? Many types of social and health data — such as child stunting, literacy and access to modern contraception, which are the indicators Data2X focused on in this case study — are correlated with geospatial factors such as accessibility, elevation and aridity that can be mapped across entire countries using satellite imagery.
When you combine well-being indicators available in regular demographic and health surveys, then look at different aspects of geospatial data, you’ll notice a pattern where geospatial variables are correlated with DHS variables.
Huh? Geospatial data is freely available almost everywhere, while DHS data isn’t — so using the information where both are available allows you to model out what the indicators will be in the areas where you only have satellite data. In short, geospatial modeling can transform a limited number of survey data points distributed unevenly across the country into a continuous landscape of information.
The challenge? The granularity of this data could allow for more localized decision-making, but it is not always certain that one geospatial variable is always relevant for predicting one of the DHS well-being outcomes, as it varies greatly by country: “Like anything else in development, the context matters for everything,” Furst-Nichols said.
2. Cell phone data can inform women’s social protection systems.
Researchers at the Massachusetts Institute of Technology, together with U.N. Global Pulse, utilized credit card and cell phone data to discover patterns of women’s spending and mobility.
Why? A combination of credit card and cell phone data can provide insights into women’s economic status and physical mobility. And economic status and mobility can tell you a lot about a person’s vulnerability to potential shocks.
Credit card spending patterns can create the basis for deducing economic lifestyle by noting the order in which individuals make purchases. Mobile phone data, meanwhile, can enrich this information by showing how evenly an individual splits travel time across the various locations he or she visits.
Huh? Using the combination of credit card and cell phone data, researchers were able to identify seven economic lifestyle clusters among women in the dataset — “commuters” and “homemakers,” for example. Once you identify a category, you can conduct interviews with those “types” in order to better target services.
If you consider that a tin roof versus a thatched roof is still one key indicator used today to identify someone as needing social protection, consider how much cell phone data — or even mobile money use data — would allow development professionals to understand about a person’s economic status.
“With this data you could potentially improve targeting methods because you can see much more than the kind of roof … you can see over time how vulnerable a woman is to economic downturn or shock,” Furst-Nichols said.
The challenge? The current project has a bias toward inclusion of well-off individuals with access to credit cards and cell phones. Also, mobile operators don’t always make call detail record data available for researchers, and privacy concerns are still top of mind. The identification of economic lifestyle clusters could be a vital input for policy formulation, especially regarding social protection — but institutions will need to continue working on data access and privacy, and options will need to be explored to better target the poorest.
3. Internet activity can help monitor women’s mental health.
Researchers at Georgia Tech University, supported by the University of Leiden and U.N. Global Pulse, located signals of depression in a large database of publicly available tweets from women and girls in India, South Africa, the United Kingdom and the United States.
Why? Sex disaggregation plays an important role in providing information about the disparities between women and men, but most of the publicly available data on mental health comes from infrequent exercises that rarely include sex-disaggregated information, especially in the developing world.
But internet use — and especially the expression of thoughts and emotions on social media platforms — can provide insight into women’s welfare, including mental health and political engagement.
Huh? Researchers employed machine-learning techniques to figure out whether public expressions of distress, anger or anxiety on social media revealed mental illness. They compared the posts with content shared on online mental health support communities, and consulted on a subset with mental health professionals. They found that the machine learning approach accurately identified genuine mental health disclosures in nearly all social media posts it examined.
The challenge? Overall, the findings suggest that discreetly gathered social media data can serve as an important source of mental health information, but social media monitoring can’t replace more formal approaches to mental health surveillance — and there are still plenty of barriers to using it in a way that will help the poorest. At that individual level, however, this data could help provide targeted tweets with information about access to counseling services in someone’s community or online, Hurst-Nichols said. At the national level, it could help with surveillance to understand the levels of mental health concerns and how they shift and change over time.
Now that Data2X has proven several varied and promising ways to use digital data to close the gender data gap, the group will enter the second phase of the project, which involves looking harder at what kinds of questions global development implementers or policymakers need to answer.
“What we want is different groups to come to us and say ‘we need information on xyz about women and girls,’ then we can put that out to our researcher network and ask whether it’s possible to answer those questions using digital data combined with other types of data,” Hurst-Nichols said.
Read more international development news online, and subscribe to The Development Newswire to receive the latest from the world’s leading donors and decision-makers — emailed to you free every business day.
