Gender, Global Goals, and Big Data.
Harnessing the data revolution for a social revolution.
As a fiercely independent woman living in the UK, I admit that I sometimes take the right to vote, drive, and manage my own money for granted. Everytime I read a story about a girl being forced to marry her rapist, or remember that 49 countries have no laws specifically protecting women from domestic violence, I feel a fresh wave of disgust that parts of our society are still living in the dark ages.
The injustices and inequality women suffer are being addressed by Goal 5 of the Sustainable Development Goals (SDGs). Gender equality is strongly linked to education, health, and poverty, and every step we take that improves the life of a woman has a positive impact on everyone around her.
Since the SDGs were launched in 2015, nations have gathered annually to review their progress towards creating a happier, healthier and more equitable world. These occasions are a great opportunity to take stock of what’s been achieved and highlight areas where progress still needs to be made. However, to achieve these goals, we need good data to inform our decisions, and women are not yet fully represented by the statistics. We can’t develop solutions that improve women’s lives if we don’t fully understand the people we’re trying to help and what they really need.
To ensure women are fully represented, we need to fill the data-gaps and we’ve written previously about novel data sources that could be harnessed. Therefore, I was interested to read a recent report by Data2x reviewing how satellite data, digital transaction logs, and records of internet activity can be used to predict literacy rates, anticipate financial pressures, and detect the prevalence of mental illness. Their research demonstrates how the quality, availability and use of gender-disaggregated data could be improved. Let’s take a look at some of the examples in the report.
Geospatial data can tell us something about literacy rates.
Traditional surveys, such as a census, don’t always reach the people who need help the most. Whether it’s due to geographical isolation or a lack of infrastructure to conduct national surveys, the result is that people are unrepresented when important decisions that affect them are made.
To counteract this, the correlation between social and health indicators and geospatial data is being tested to improve the spatial resolution of data on girls’ stunting, women’s literacy, and access to modern contraception. Using geospatial data derived from satellite images, Flowminder Foundation and WorldPop had mixed success at forecasting the situation in places that haven’t yet been surveyed in traditional ways. Their models for girls’ stunting were inadequate, and models for modern contraceptive use worked in some countries but not others. However, in another example, there were positive signs that it’s possible to extrapolate a limited number of literacy data points into maps that accurately predict literacy rates across an entire country.
As high-resolution satellite imagery becomes more readily available, the next challenge will be ensuring everyone can access the data that’s relevant to their mission. Platforms are being developed that will make it easier to access and use satellite imagery, such as Radiant.Earth, opening up new possibilities to produce and use gender-disaggregated data.
Credit card transactions tell a story, but chapters are missing.
Reviewing anonymous credit card transaction data can reveal how people commute, how far they travel to purchase groceries, and how they spend money whilst on holiday. With this kind of information, it’s possible to understand how spending is influenced by economic pressures, and policymakers can more easily predict the impact of new policies. For example, raising the cost of fuel could disproportionately impact people who commute for work, but investing in public transport infrastructure would open up new employment opportunities for people living in rural areas.
However, this kind of analysis automatically excludes anyone who doesn’t have access to a credit card. When male and female credit card spending habitats were reviewed, it appeared that women were spending less, but this could be because they are paying for goods and services in other ways — such as mobile payments or exchange of goods. Financial payments are made in many ways and visualising the data can help us understand how women access sources of credit and highlight the differences between men and women.
Oversharing is caring.
We all have that one friend who shares too many baby pictures, but the status updates we share via social media offer a glimpse into our wellbeing. Across the world, the methods for assessing mental health are inconsistent and don’t always include gender-disaggregated information, which means women and girls may not be fully represented by healthcare policies. APIs have made it possible to easily read and write twitter data, and given that we share so much of ourselves online, it would be fantastic if we could put it to better use than creating a giant collection of avocado toast photos.
In a study led by Georgia Tech University, nearly 1.5 million posts from India, South Africa, the UK and USA were examined with machine learning techniques to identify genuine self-disclosures of mental illness from public social media posts. Consultations with medical professionals indicated that the method used can accurately identify mental illness, meaning social media data could be an important source of information that complements the records we already have.
What’s next?
Our access to novel data will only increase as more satellites are launched into orbit and more people start using smartphones. By 2020, nearly three billion people worldwide will own smartphones — presenting a remarkable mine of data that could be used for social good. What we need to do now is anticipate what data we might get access to, and plan how we could use it. Research has demonstrated the feasibility of using mobile data to track disease outbreaks and respond to humanitarian crises: but to gain access to that data, governments and businesses have to work together in bold, new ways. Making Every Woman and Girl Count is one those partnerships that’s improving the production, accessibility and use of gender statistics.
Improving the production of gender-disaggregated data begins by removing systematic gender biases in the way data is collected. International standards should be established and followed to ensure the data is comparable — and data must be collected regularly, so we can track progress as it’s being made. Most importantly, if we want to achieve gender equality, the data we collect must be used. This means delivering it in a format that’s tailored to the needs of the people who’ll be using it, and designing data visualisations that are representative of all people, no matter their gender identity.
As engaged citizens, we have the power to hold our governments to account and encourage pioneering companies to share their data for the benefit of society. We must speak up for the women and girls whose voices are not being heard and we can do this by sharing powerful, compelling, data-based stories that highlight the change we want to see in the world. Let’s consign inequality to the past and stride towards a future where every human is valued for their character and contribution to society above all else.
