Strategies for working around missing data in education
On January 30, data scientists from across the country spoke at the second Birds of a Feather symposium, which seeks to improve connections and knowledge-sharing across discrete teams working on common issues. This symposium focused on the best practices and new methods for using datasets with missing values for labor and education data analysis.
Included among the speakers were Caro Haensch, a statistics lecturer from Ludwig Maximilian University in Munich, Tian Lou, a policy researcher from Ohio State University, and Angela Tombari, senior research analyst at the Kentucky Center for Statistics.
Speakers shared the various causes of missing data, methods for handling each case, and their own experience working with real-world datasets containing missing values. Haensch went through synthetic data in a Jupyter notebook as a demo of methods for dealing with missing data.
“Please do not assume that missingness is random in administrative data. It is probably not,” Haensch said.
Watch the full video here.