Marcelo Tyszler works at KIT Royal Tropical Institute and is part of the CGIAR Collaborative Platform for Gender Research team. He focuses on the collaboration between the Gender Platform and the CGIAR Platform for Big Data in Agriculture. In this interview, Marcelo shares insights about the start of this initiative a year ago, his excitement and questions at the prospect of deeper collaboration in 2019 and beyond.
Is there interest in collaborating across ‘big data’ and gender domains and platforms?
There is plenty of interest! CGIAR Platform for Big Data in Agriculture sees a great potential in addressing gender for agricultural research for development with the use of big data. They are really keen to understand how gender research fits into that picture and how it can bring additional insights. Within the CGIAR Collaborative Platform for Gender Research, people are trying to understand where big data (methods) can deliver on their promises for gender research.
Who is involved in the current projects?
The people directly involved at this stage are Brian King (Coordinator of the CGIAR Platform for Big Data), Gideon Kruseman (Coordinator of the community of practice on socio-economic data in the Big Data Platform) and Rhiannon Pyburn (Coordinator of the CGIAR Collaborative Platform for Gender Research). Other researchers that are more directly involved are colleagues working on the Women’s Empowerment in Agriculture Index (WEAI) and/or from the International Food Policy Research Institute (e.g. Hazel Malapit, Elena Martinez, Jessica Heckert, Agnes Quisumbing, Ruth Meinzen-Dick), Sophia Huyer from the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) and some members from the community of practice on geographic information systems in the Big Data Platform (Jawoo Koo and others etc.). Some other colleagues are involved, albeit more distantly.
What is the short term potential of such collaboration? And long-term prospects?
The immediate gain is to make gender data more FAIR (Findable, Accessible, Interoperable and Reusable), with emphasis on Findable to begin with. Gender researchers should be able to easily find available data and studies. Meanwhile non-gender researchers should be able to use that gender research to complement their own research. The ‘Big Data‘ mindset pushes for more structurally organized (and thus more reusable) data. This systematic approach is potentially very useful also for Gender researchers. In parallel, gender makes Big Data researchers think from a wider angle. It helps them understand socio-economic barriers and opportunities and realize how these factors can lead ‘purely technological solutions’ to fail.
In the longer run, this kind of collaboration may help us change the way we look at agricultural research for development and gender research. Nowadays, most studies tend to have a direct scope, with specific research questions and findings. Big Data makes use of high volume and non-traditional information flows (such as mobile network metadata) and thus changes the way we think about (gender) research.
What does this ‘inter-platform collaboration’ look like concretely?
Right now we have a seed grant (from the socio-economic data community of practice of the Big Data Platform), which focuses on improving the findability of gender datasets, and understanding minimal requirements for gender data sets that are publicly available. This means, for example, making sure that specific keywords are used in a consistent way. In that project we are now considering our next steps. We might decide to improve our inventory and its usability, and support the development of a generic core set of gender household survey questions. A more ambitious plan is to develop a world map based on WEAI scores and see how this connects with the research on mobile network metadata, for example to understand how empowerment relates to mobile phone usage patterns.
What’s most exciting and most challenging about this cooperation?
The potential is the most exciting part! It makes you think about research with different boundaries, something that was not possible only 10 years ago. For example, this paper shows how mobile phone metadata can be used to predict poverty wealth. This uses an approach around high volume and high frequency data that is quite different from the more traditional structure of household surveys with limited sample sizes. A similar approach could be used to predict empowerment from mobile phone metadata, for example. This approach to ‘learn’ from an environment and try to apply these ‘learnings’ in similar but different environments is essential in machine learning – a typical Big Data technique.
The challenge is that this is (relatively) so new that our scope remains very broad. Researchers from both the big data and gender communities are struggling to understand what can be realistically done, within what time frame, and at what cost. Articulating a very specific and clear joint research project that is of mutual interesting is our biggest short-term challenge for the moment.
See also the questions and answers video run by the CGIAR Platform for Big Data in Agriculture during the October 2018 Convention in Nairobi: