Estimating Interrater Reliability in Incomplete Designs.
Abstract
Estimates of interrater reliability are often needed for incomplete designs in which raters (e.g., employees) are nested within targets (e.g., organizations). It is shown that the popular use of estimates based on between-group ANOVAs accompanied by intraclass correlations can be seriously misleading if low variation exists among target means. An alternative based on a within-group procedure is proposed and shown to be superior to the intraclass correlation in the condition of low variation among group means, accompanied by low within-group variation. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 31, 1981
- Accession Number
- ADA104041
Entities
People
- Gerrit Wolf
- Lawrence R. James
- Robert G. Demaree
Organizations
- Texas Christian University