I am a general internist on the faculty at the University of Washington. I attend on the medicine wards at Harborview Medical Center anywhere from 4 to 8 weeks per year. The rest of the time I do research.

My initial niche was in measurement in general and psychometrics in particular as applied to medical settings. Psychometrics is the statistical machinery with which scores are generated from test items. Psychometrics is very advanced in educational testing settings, but in medicine we have been slow to move beyond approaches that have been superseded for half a century in educational testing. At the same time, tests used in medical settings have particular characteristics that are distinct from educational testing. While educational testing companies employ legions of test item developers to continually renew test content, in medicine we tend to develop the test and re-use it as is. Furthermore, while in educational settings items for high stakes tests are always carefully vetted before they are used to generate scores for people, typically in medicine items receive less attention before they are used very widely, resulting in important challenges for valid inference if there are measurement issues such as differential item functioning (DIF).