That combination presents opportunities, but also risks: no data source is perfect.
The CLEAR STATS project is now working on a solution to make those risks more visible. In the assigned ERC Proof of Concept project by sociologist Dimitris Pavlopoulos, researchers are developing an open-source tool that helps statistical agencies detect measurement errors and biases in their data. Measurement errors occur when something is not measured quite correctly; bias means that certain groups are over- or under-represented in the figures.
By detecting these problems early, statistical agencies can improve their figures before they are published. This is important because inaccurate statistics can lead to wrong policies and uneven decision-making.
The tool is being developed as a freely accessible software package in the statistical program R and tested with data from the Italian statistical agency Istat and the Central Bureau of Statistics (CBS). The collaboration shows how international knowledge sharing can contribute to more reliable figures - and thus to better informed choices that affect society as a whole.