Maike Weiper is one of the current ODISSEI SoDa Fellows. A PhD candidate at the Sociology Department of Utrecht University, her background is in psychology – but after hearing how population datasets were used to reckon with the COVID-19 pandemic, she pursued a Master’s in Applied Data Science and eventually her PhD. Her project is about very timely topics: gender inequality and climate change. It was this research that first brought her to the ODISSEI SoDa team.
Maike joined one of the SoDa drop-in sessions, one of the low-threshold opportunities the team offers for researchers to explore whether a data-scientific approach might work for their project. There, she spoke with Dr. Javier Garcia-Bernardo, a computational scientist on the SoDa team, who walked her through what the fellowship could offer early-career researchers: expert input, a collaborative work environment, and the chance to be part of a team tackling socio-scientific problems using data science tools. She decided to apply.
During the SoDa fellowship, fellows work on a self-chosen project aimed at solving a data-related problem in the social sciences. The goal of Maike’s project is to estimate the carbon emissions of people living in the Netherlands based on their household expenditures, and to link those emissions to everyday activities through time-use data. “However,” she points out, “these data are collected in surveys with relatively small sample sizes that can not be combined into one dataset.” Therein lies the problem.
The solution? Training supervised machine-learning algorithms to predict household expenditures and individual time-uses, based on the users’ demographics, such as gender, age, and household size. First, Maike and her team used various techniques to combine and harmonize different datasets from the CBS data register. Then, they developed the pipeline that allows them to train the machine learning models and compare their performances in predicting household expenditure and individual time-use data at the population level. “Eventually, this knowledge could be used to understand how carbon emissions may be reduced without structurally burdening some individuals more than others,” Maike adds, highlighting the gendered dimension of her research.
“Eventually, this knowledge could be used to understand how carbon emissions may be reduced without structurally burdening some individuals more than others.”
The next step of her project will be to account for the uncertainty that such models naturally contain when predicting data. At the same time, research ethics are a big concern. Most of the data they use are very sensitive, and are therefore stored on CBS servers, which means full open access isn’t possible. Even so, a transparent and reproducible research process is always the aim. “For instance,” Maike notes, “our code can be found on the GitHub page of the SoDa Team, and we are constantly working on applying FAIR principles in our work.”
Her time with the SoDa team has only spurred her interest in how data scientific approaches can enrich sociological research. “There are so many different methods and approaches that can advance social scientific research and make it even more interesting and relevant for society,” Maike shares, while mentioning that she also found the collaborative culture of the SoDa team very rewarding. “Throughout my fellowship so far – for example, at the ODISSEI Conference – I have met so many interesting and helpful people who love to think along.”
“There are so many different methods and approaches that can advance social scientific research and make it even more interesting and relevant for society.”
After the fellowship, Maike hopes to share what she has learned – perhaps through a workshop – to help inspire other researchers and give an example of how hers and similar approaches can further increase our understanding of society. As for tips for any future SoDa fellows, Maike gets to the point: “Join the data drop-ins, attend the ODISSEI conference, and talk to people about your project. You do not need to have a full understanding of the data-scientific methods behind your idea,” she underlines, “as long as you are open to learning and collaborating with others.”
Photo by Myriam Jessier on Unsplash