Latent Structure Agreement Analysis
Abstract
How do we know how many opinions are required to make a diagnosis with necessary accuracy? One way is by examining how often physicians agree on the diagnosis. This Note discusses statistical techniques that can be used to analyze agreement data to address this and related questions. Specifically, these methods make it possible to determine from the opinions of panels of diagnosticians in an agreement study the following: (1) the probable accuracy of an individual diagnosis; (2) the probability of disease presence or absence given unanimous or conflicting opinions by several diagnosticians; and (3) how many opinions should be required to make the diagnosis. The methods discussed include two related techniques, which differ in assumptions about disease subtypes and associated differences among cases in their ability to be correctly diagnosed. These techniques have many applications in addition to that of medical diagnosis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1989
- Accession Number
- ADA258076
Entities
People
- John Uebersax
- Will Grove
Organizations
- RAND Corporation