Which gender data gaps are the most important to address first?

In Invisible Women, Caroline Criado illustrates that in multiple industries, vast amounts of data on women is incomplete or missing — despite a global population that is half female — resulting in a massive gender data gap. The implications of this assumption range from small inconveniences to matters of life and death.

Whether it be missing gender data in the areas of health, education, economic opportunities, politics, or security — we want to know: which gender data gaps are the most pressing, and why?

@Karan thanks for posing this question for discussion.

Data2x is an organization working to improve the quality, availability, and use of gender data in order to make a practical difference in the lives of women and girls worldwide.

In a recent study, they identified need, population, and policy relevance as three criteria when prioritizing data gaps to map.

Within each of the domains of health, education, economic opportunities, politics, and security, I’m curious to know what our online community of experts feels are the most critical gaps to close first, and why?

If we look within the domain of health, for example, is it more important to focus on closing data gaps in mental health, adolescent health, disease burden, or another area? Why?

@Karan, A pathbreaking move forward would be to create ways of collecting data (e.g. an algorithm) that gathers intersectional data on (gender, age, race, socioeconomic factors, and more) that can easily replicate across domains. However, to choose one area to start, selecting a topic like heart disease that disproportionately impacts women in low, middle and high-income countries as one of the top 10 causes of death, according to the World Health Organization, could produce life-saving results that have social, economic, and environmental impact.

@AndreaDr. I agree that heart disease is a great example of a pressing gender data gap. Also, as you mentioned, there is significant potential to leverage this data to improve the lives of a large swath of the population as it affects a diverse group of people. With that in mind, I would like to pivot over to another health-related area, pharmaceuticals.

The FDA acknowledges the efficacy and safety of drugs can vary between men and women (Understanding Sex Differences at FDA | FDA). I’m curious to know more about the gender data gap issues in pharmaceuticals. Thoughts?

This webinar might be interesting - Data 2x, UN Foundation is hosting a webinar on March 24th to share and discuss the findings from their new report, which maps the status of gender data gaps across economic opportunities, education, environment, health, human security, and public participation.

Register here: https://lnkd.in/e-iXRkk

Thanks @ElsaMarie for sharing the Data2x webinar information.
Looking forward to extract important insights from this webinar.

@ElsaMarie thank you for sharing the Data2x webinar with us! The team has registered and looks forward to attending and learning from the webinar.

In pharmaceuticals it’s also a question of prioritizing funding for research on diseases prevalent in women or affecting women only. We need to look at the data regarding funding.

@“ÅsaEkvall” that is a great point. We conducted an interview today where someone referred to the amount of funding allocated to “lifestyle” drugs and research (such as erectile dysfunction) while there is minimal (if any) funding available for endometriosis and other painful and quality of life disrupting conditions experienced by women.

@Aaron_Denham I’m unfortunately not surprised to hear that. This really needs to change. I have both endometriosis and myoma myself and know what it is like to be told by the GP “yeah women, have pain, that’s just how it is” when it is debilitating and handicapping to the point of preventing me from working on and off. So many women are like me. We won’t get help unless someone funds it.

Hi, all. I would like to add gender data gaps in government and in the measurement of the effects of different policies and reforms for women, e.g. gender budgeting in all facets of public policy.

Thanks @stellunak for sharing this interesting perspective.

ÅsaEkvall: “In pharmaceuticals it’s also a question of prioritizing funding for research on diseases prevalent in women or affecting women only. We need to look at the data regarding funding.”

I think so, too. Maybe this young EU project is relevant in this regard: Home – GRant AllocatioN Disparities

Let me add: I think public research funding is already observed to some degree. To have a look at business enterprise research funding - which is even higher than public - would be very promising.

Support to all the comments that highlight how funding for research and data collection efforts related to gender is needed.

While it is hard to prioritize what gaps to fill first, some broad principles that we believe should are important to consider when filling a gap are:

  1. Inclusiveness. We need to count people who are currently marginalized in order to understand their needs and develop and implement adequate programs and policies.
  2. Intersectionality. If it isn’t being adequately captured with pure data, shifting the conversation towards what we like to call “knowledge gaps” rather than “data gaps”. How does x affect y? (to capture the impact of various factors on women’s lives). Qualitative data is so important but is based on small samples (not generalization nor representative). How can we scale up collection of qual data that really gets to the lived realities of girls and women?
  3. Impact. In particular, gaps related to long-term impacts and cost-benefit analysis. So much of current data collected is cross-sectional. Evidence that builds the case for gender equality and shows how women’s full participation in socio-economic and political spheres reaps benefits not just for themselves, but their families and societies as a whole is really helpful.
  4. Scalability. What works to advance gender equality? There are so many smaller-scale studies of great practices or programs that have worked here and there, how can we move some of these to be acknowledged as high-impact practices worldwide?
  5. If we want to focus on gender in the SDGs, the lack of data on gender-specific indicators is alarming: “only 12 of the 54 gender-specific indicators on the SDGs, or 22%, have enough data that is regularly produced and available that can be used to monitor progress across all regions” according to UN Women. We need this data to keep governments accountable.

My research is on biases embedded in AI models and I focus on biases we can identify in natural language models. If we have more natural language data from women that is collected from online platforms or traditional media sources (transcribe spoken language), we can automatically identify gender biases represented in the models. Accordingly, we can focus on collecting data on critical issues from domains that have high gender bias. As a result, we can characterize and map the gender data gap landscape and prioritize based on data-driven findings.
When women are not accurately represented in datasets, for example the health domain, the analysis about them, especially if it is an algorithmic process usually ends up being inaccurate and in some cases completely wrong since different social groups have different patterns. AI algorithms are optimized to learn the patterns of the majority group unless they are explicitly presented data from different social groups. Consequently, for example an algorithm on cardiac disease would learn the symptoms and patterns associated with (white) men and potentially lead to catastrophic results for women as well as other non-majority groups.

@WD_Research I appreciate your perspective listed in #2 regarding “knowledge gaps.” This is perhaps a much more inclusive way of thinking through the full scope of the issues (and their intersections) at hand. I’m a qualitative researcher at heart. Always at the back of my mind is what the quant data (and big data resources) can’t tell us about experience and the “unknown unknowns.” Thanks for bringing this forward for us.

Thanks @aylin We are just starting to look deeper into the issues with AI and its potential to help address gender health data gaps. We might be in touch with questions. Are there any quality resources or examples that come to mind (or stand out) where people have used AI (or optimized algorithms) to confront gender data gaps?

@ElsaMarie I’ve only just joined so I missed the webinar on 24 March - but would love if there is a recording that we could access.

@Andrea I agree that the most disruptive, innovative and powerful thing we could do would be to develop something that would help data custodians understand, collect and analyse data with an intersectional lens.

A couple of other - off the top of head thoughts on gender data gaps and areas of need - although I will go and do some further thinking:

  • mobile technology - I understand that the impact of 5G has only been tested on men and not of its impact on women or children
  • violence against women - we have some public data on prevalence, incidence and response but very little on early intervention and recovery. It would be great to have this on the radar. Also it would be revolutionary to encourage or incentive private companies to share their sexual harassment and sexual assault data (de-identified of course).
  • we must also include in our thinking gender non-binary as well as male and female identified genders. There is virtually no data on gender-non binary
  • mental health - this is front of mind for me as we go through a global pandemic and there is a Royal Commission into Mental Health in Australia. The interim report on mental health had very little on gender and mental health even though it presents differently and at different rates depending on gender
  • AI - my understanding that biases have been built into algorithms that underpin AI (see the google translate example that I think is in Caroline’s book) as women are not part of the teams in the development and testing of AI

Thanks

@staceyo Here is the link to the recording of the Data2X webinar: Launch of Mapping Gender Data Gaps in the SDG Era - Data2X Data2X