DaSSWeb | Estimating error rates in binary decisions with inconclusive outcomes

18 October | 4pm (WEST) | online

Data Science and Statistics Webinar
Estimating error rates in binary decisions with inconclusive outcomes

JOIN US HERE

Speaker
Karen Kafadar
Department of Statistics, University of Virginia, Charlottesville VA, USA

Abstract
Binary decision-making occurs in many areas of science and policy; e.g., medicine (tumor present or absent), forensics (ID or exclusion), finance (good or bad credit risk), agriculture (healthy or diseased plant). Lab or field studies may be conducted to assess the error rates in such binary decision-making processes (e.g., proficiency tests for radiologists or latent print examiners). In such tests, a true outcome is known (e.g., latent print and file print did or did not come from the same source), but study outcomes allow three responses (e.g., "same," "different," "inconclusive"). Many articles in forensic science report results of such studies by completely ignoring "inconclusive" responses, which can artificially increase the apparent accuracy rate. In this talk, I will discuss ways of estimating error rates in such studies that more fairly account for "inconclusive" decisions and enable fair comparisons of results across studies.
This work is joint with Jordan Rodu and Sydney Campbell at University of Virginia.

About the speaker
Karen Kafadar is Commonwealth Professor and former Chair of Statistics at University of Virginia. Her research focuses on statistical methods and data analysis in the physical, chemical, biological, and engineering sciences. She received her BS and MS from Stanford and her PhD from Princeton and has held positions at NIST, Hewlett Packard, National Cancer Institute, University of Colorado-Denver and Indiana University. She co-authored several reports for the National Academy of Sciences, including Strengthening Forensic Science in the U.S. (2009) and the Anthrax Investigation (2011). Her recent work concerns statistical methodology for randomized cancer screening trials, and classification and estimating error rates in eyewitness identification and forensic science. She is a co-PI on the NIST-funded Center for Statistical Applications in Forensic Science and was former Editor of JASA Reviews, Technometrics, and The Annals of Applied Statistics, 2012 IASC President, and 2019 ASA President.

Presentation