Abstract
Cancer screening and diagnostic tests often are classified using a binary outcome such as diseased or not diseased. Recently large-scale studies have been conducted to assess agreement between many raters. Measures of agreement using the class of generalized linear mixed models were implemented efficiently in four recently introduced R and SAS packages in large-scale agreement studies incorporating binary classifications. Simulation studies were conducted to compare the performance across the packages and apply the agreement methods to two cancer studies.
DOI
10.22237/jmasm/1509495300
Recommended Citation
Mitani, A. A., & Nelson, K. P. (2017). Modeling Agreement between Binary Classifications of Multiple Raters in R and SAS. Journal of Modern Applied Statistical Methods, 16(2), 277-309. doi: 10.22237/jmasm/1509495300
Supplementary Tables
bcdata.csv (1 kB)
Example data set
appendixcodeSAS.txt (3 kB)
SAS Code (proc glimmix)
appendixcodeR.txt (7 kB)
R Code (ordinal, lme4, mcmcglmm packages)
Included in
Applied Statistics Commons, Social and Behavioral Sciences Commons, Statistical Theory Commons
