What thick data are and how to use them

What thick data are and how to use them

We are collecting data on a global, national, regional, municipal scale, we have it all, what could possibly go wrong? Lots of things if we don’t look more closely. Thick data may represent a solution.

Data and technology play a fundamental role in the perception we have of human rights. We are used to thinking that data are neutral and objective and they can inform us about the world. There is nothing more wrong.

As the mirror of society, they represent its hierarchies and inequalities. Even the very exercise of collecting data is the product of an opinion and a choice. The risk is to translate into code, and therefore automate, the reality. This includes racism and sexism that characterize our societies. For black women and black men, for women and men of color, for marginalized communities, and for whoever is more vulnerable in society, this means that technology, instead of liberating, poses a risk of perpetuating inequalities.

It is not surprising that algorithms discriminate – but is it true? Artificial intelligence gives power but with power comes responsibility. The question is how we make sure that the use of AI and data are the result of fair choices.

Thick data as a guide

The limit could be mitigated by ethnographic observation and the collection of qualitative data: we bring thick data into normal analysis and research procedures. I heard about “thick data” for the first time during Tricia Wang’s TED talk in 2017, the first woman talking about them. She said that “using only big data increases the possibility of missing something while giving us the impression of knowing everything”.

Thick data investigate the context – the “why” behind the “what”, helping to reveal people’s emotions and stories. Big data certainly have a place of honor in specific environments, but sometimes they are not enough. Alongside the official statistics, there is a need for observation in the field, collection of testimonies, analysis at the local level.

We need to make sense of the data, in a human understanding. Would you agree?