SciCAR 2019

Data journalism shares much with science. It's a discipline designed to generate knowledge from empirical observations, and borrows a great deal of quantitative methodology from science. It also suffers from some of the same problems as science: reproducibility, causal identification, measurement problems, and mistakes. But data journalism has additional challenges: reporters are not often trained in science, and the results must be effectively communicated to the public. In this talk I’ll show Workbench, a tool built at Columbia to try to make data work more reproducible and accessible for journalists without technical training, and discuss the future potential of AI for data analysis.