Using Digital Traces to Measure Digital Gender Inequality in Real-Time
Tracking progress on gender inequalities in internet and mobile access and use – digital gender gaps – is very important to ensure that women benefit from the opportunities afforded by the digital revolution. But measuring and sizing this gender gap has proven to be challenging due to significant gender data gaps. There are limited data from surveys or population-level data sources that examine digital inequalities at the population level in a harmonised way.
How can we better measure digital gender gaps in real-time at national and subnational levels? In which countries are women most “invisible” online and less likely to own mobile phones?
Our team has been exploring how big data innovations can help us measure women’s participation in the digital revolution in real-time, and our web app shows some of the indicators we have been developing.
Digital Gender Gap Indicators
We leverage social media data from the Facebook Marketing Application Programming Interface (API) in combination with different population and development indicators to estimate global digital gender gaps (GG) in internet use and mobile phone ownership and gender-specific digital adoption levels. We provide these indicators at both national and subnational (first administrative level, GADM 1) geographies for the period from 2015-current.
We train models to predict digital access in contexts where we have data on key predictors of digital access – for example, measures such as Facebook penetration by gender, the Facebook gender gap, education, and income levels in a country or subnational region – but don’t have direct measures from a survey on digital access. We train and validate our predictive models against data from nationally-representative surveys from the Demographic and Health Surveys, Multiple Indicator Cluster Surveys and in some cases in high-adoption settings, ITU data. Our subnational model estimates are trained on Demographic and Health Surveys (available for 33 countries), as this provides sub nationally representative ground truth.
The best-performing models are those that combine social media indicators (i.e. Facebook penetration or the Facebook gender gap) together with population and development indicators such as the Human Development Index and its sub-components. Through this approach, we can improve the geographical and temporal resolution of estimates of digital access by gender, as our method covers a broader set of countries and more recent time periods than available survey data.
The details of the methods used to generate the estimates shown in the maps on this website are described in “Using Facebook ad data to track the global digital gender gap” (Fatehkia et al. 2018) and “Mapping Subnational Gender Gaps in Internet and Mobile Adoption Using Social Media Data” (Breen et al. 2025).
Note: in app updates (v2, March 2025) we have switched in our national estimates to a model trained largely on DHS/MICS (with some additional estimates from ITU for select high-adoption only settings).
Internet Indicators
The Internet Gender Gap Indicator shows the ratio of female-to-male of internet use, and lie in the range of 0 to 1. Values of 1 or close to 1 show that the gender gap has closed. For example, a value of 0.75 could be interpreted as a 25% gap between male and female internet use, with 75 women online for roughly 100 men who are. These data are updated with a monthly frequency at the national level and subnational first-administrative level.
The Internet Men Adoption Indicator shows the proportion of men aged 15-49 who have used the internet in the past 12 months, and lie in the range of 0 to 1. These data are updated with a monthly frequency at the national level and subnational first-administrative level. For example, a value of 0.75 could be interpreted as 75% of men aged 15-49 have used the internet in the past 12 months.
The Internet Women Adoption Indicator shows the proportion of women aged 15-49 who have used the internet in the past 12 months, and lie in the range of 0 to 1. These data are updated with a monthly frequency at the national level and subnational first-administrative level.
Mobile Indicators
The Mobile Gender Gap Indicator shows the ratio of female-to-male mobile phone ownership, and lie in the range of 0 to 1. Values of 1 or close to 1 show that the gender gap has closed. These data are updated with a monthly frequency at the national level and first-administrative level.
The Mobile Men Adoption Indicator shows the proportion of men aged 15-49 who own a mobile phone. These data are updated with a monthly frequency at the national level and first-administrative level.
The Mobile Women Adoption Indicator shows the proportion of women aged 15-49 who own a mobile phone. These data are updated with a monthly frequency at the national level and first-administrative level.
Data and Technical Notes
Facebook Gender Gap and Adoption Level Index
We collect data for countries globally on the aggregate numbers of monthly active female and male users of Facebook available through the marketing API. Using these aggregate numbers, we generate indicators such as the “Facebook Gender Gap Index”, an indicator of the number of female-to-male Facebook users in a country weighted by the gender ratio of the population, as well as the Facebook penetration by gender. We compute these indices monthly.
While the Facebook Gender Gap Index reflects gender gaps in who has Facebook accounts and not internet use per se, we have found the Facebook Gender Gap Index is highly correlated with estimates on internet use and mobile phone ownership and gender gaps (from the DHS/MICS/ITU) collected via surveys, for the countries for which these data are available. The correlation between the Facebook measures and internet/mobile indicators, however, are weaker in some regions with low Facebook penetration, e.g. post-Soviet states. In these cases, the model predictions may underestimate internet and mobile adoption, and discrepancies between subnational and national estimates are possible.
Offline Indicators and Ground Truth Surveys
Our national models use country-level development indicators (e.g., Human Development Index, Gross Domestic Product (GDP) per capita and global gender gap indicators). Our subnational models use similar population and development indicators, but at subnational (admin-1) geographies. Where possible, we prioritise, for consistency and standardisation, for training our models at both subnational and national levels, ground truth survey data from DHS and MICS. To improve coverage in our national models in some high adoption settings (e.g. Europe, North America, Central Asia), we rely on ITU data.
Suggested Citation
When using the data, please acknowledge our website (digitalgendergaps.org), cite related academic publications and/or our web app, as appropriate.
- Fatehkia M, Kashyap R, Weber I. 2018. Using Facebook ad data to track the global digital gender gap. World Development 107:189-209. https://doi.org/10.1016/j.worlddev.2018.03.007
- Breen, Casey F., Masoomali Fatehkia, Jiani Yan, Xinyi Zhao, Douglas R. Leasure, Ingmar Weber, and Ridhi Kashyap. 2024. Mapping Subnational Gender Gaps in Internet and Mobile Adoption Using Social Media Data. Pre-print. https://doi.org/10.31235/osf.io/qnzsw
- Kashyap R, Fatehkia M, Al Tamime R, Weber I. 2020. Monitoring global digital gender inequality using the online populations of Facebook and Google. Demographic Research 43: 779-816. https://www.jstor.org/stable/26967824
- Leasure DR, Yan J, Bondarenko M, Kerr D, Fatehkia M, Weber I, Kashyap R. 2023. Digital Gender Gaps Web Application, v1.0.0. Zenodo, GitHub. doi:10.5281/zenodo.7897491. https://github.com/OxfordDemSci/dgg-www.
Source Code
The source code for this web application is openly available from OxfordDemSci on GitHub.
Our contact information
Contact us by reaching out to Principal Investigator Prof Ridhi Kashyap ridhi.kashyap@nuffield.ox.ac.uk if you have any questions or data enquiries.