Intro to Applied Statistics for Evaluation
This session is part of the Introduction to Political Technology course at Newspeak House, open to faculty and fellowship candidates only.
Robust evaluation depends on understanding uncertainty, and statistics provides the language for doing so. Even simple questions — has this intervention made a difference, or is what we see just chance? — require clear reasoning about probability and sampling.
This session introduces those foundations, showing how statistical thinking helps distinguish real impact from noise, and how small errors in design or interpretation can lead to big mistakes.
Probability from first principles
Understanding samples and how they generalise
Applying statistical reasoning to real-world questions
Andreas Varotsis is a data scientist and AI engineer who works to improve operational delivery and services across government using technology, data, and evidence. He’s spent the previous decade in various roles in central government and front-line delivery, including the Metropolitan Police Service, the data-science team of 10 Downing Street, and the Incubator for AI.
He works to support a range of cross-government communities, including Evidence House, which works to improve the use of data and IT in government, and the Society of Evidence Based Policing, which champions research to enhance policing practices and reduce crime.