In 2017, Data2X issued a call for research projects that sought to apply insights drawn from big data to the lives of women and girls.
Ridhi Kashyap (Oxford University), Ingmar Weber and Masoomali Fatehkia (Qatar Computing Research Institute), and Reham al Tamime (University of Southhampton) received one of ten Big Data for Gender challenge grants. Two years later, grantees are reporting on their findings. In this context, Data2X sat down with these researchers to discuss their motivations for this work, what their results mean for gender data more broadly, and where they plan to go from here.
Data2X: You were already working on big data research when you applied for a Data2X Big Data for Gender challenge grant. What made you interested in this challenge?
Ridhi Kashyap & Ingmar Weber: Data2X’s Big Data for Gender Challenge gave us the opportunity to explore how a novel kind of digital data source — large-scale online advertising data — could be used to measure and understand gender inequality.
Ad audience estimates are aggregate, anonymous measures of the how many users of certain demographic characteristics (e.g., gender, location) are on an online platform, such as Facebook. Ingmar had been exploring the use of this kind of data for studying migration and Ridhi had been working on different dimensions of gender inequality (e.g., health and mortality, family behaviours) in populations using both conventional data sources as well as exploring the potential of big data sources.
The Challenge allowed us to bring our two sets of interests together and explore how online advertising data from Facebook, and subsequently Google, could be used to monitor gender gaps in internet and mobile access. Technology use by gender is a domain in which conventional data from surveys are lacking significantly, and the Challenge gave us the impetus to explore if ad audience estimates could be valuable to help fill this data gap.
Data2X: What was your topline learning from this project? (More details are available in this project brief.)
RK & IW: Our topline finding is that online advertising data can be repurposed to track global digital gender inequalities in internet and mobile phone access, as shown on our website www.digitalgendergaps.org. We generated a simple indicator, the Facebook Gender Gap Index, a ratio of female-to-male users on Facebook, derived from Facebook’s advertising platform. In countries in South Asia and Sub-Saharan Africa, we found this Facebook Gender Gap Index showed significant levels of gender inequality.
By validating the Facebook indicator against survey data on internet access gender gaps, we found that the two were highly correlated. This suggests that the Facebook gender gaps are a good signal of wider gender gaps in internet access in these populations; i.e. when women are missing on Facebook in these countries, it is a sign that they are not online at all.
Using this approach, we have been able to expand the geographical coverage of data on gender gaps in internet access and mobile phone ownership to many more countries than would be possible with survey datasets. We have also explored how Google’s advertising data can be used to monitor global digital gender inequality. Our finding is that Google estimates independently are not as good as Facebook estimates; however, the best performing predictions combine data from both sources.
Data2X: What does your work show about how big data can be used to gather insights about the lives of women and girls?
RK & IW: Our work has shown that big data can help fill gaps in gender data in the domain of women’s technology access. This is important because while the digital revolution has meant important advances for individuals, societies, and economies, we risk losing out on the full transformative potential if women are unable to access internet and mobile technologies in many parts of the world.
The first step is being able to monitor digital gender gaps, and our work highlights how online advertising data can be a complement, not a substitute, to conventional data sources in this regard. By working on the project, we have been able to understand the promises but also the challenges associated with using these data sources.
For example, by regularly collecting the data we have found that the data can sometimes be erratic, or changes occur with them that we don’t fully understand. We aren’t able to go back in the past with online advertising data, which is a hindrance for measuring long-term trends.
Data2X: Where will your research go from here?
RK & IW: With the help of our Data2X grant, we have been able to collect data spanning over a year on www.digitalgendergaps.org and are now analyzing this data and making sense of any changes we’ve observed. Beyond measuring digital gender gaps, we are interested in exploring whether the digital data source we are using can track the effect of interventions that might affect women’s technology use/access.
We’ve had encouraging results monitoring sub-national gender gaps in internet access and are also exploring how online advertising data from other platforms can be useful for measuring gender gaps in other domains like education and occupation. We are eager to find collaborators for these workstreams.