A Unified Model for Continuous and Categorical Data

Abstract

In this chapter, we generalize agreement assessment for continuous and categorical data to cover multiple raters (k ≥ 2), and each with multiple readings (m ≥ 1) from each of the n subjects. In Chapters 2 and 3, we discussed agreement statistics for continuous and categorical data, respectively, based on the basic model of two raters with a single measure each per subject. In the terminology of this chapter, those earlier chapters discussed primarily the case k = 2 and m = 1. We utilize the results from Barnhart, Song, and Haber (2005), who first proposed the within-rater CCC, between-rater CCC based on the average of replicates, and between-rater CCC based on individual replicate, and used GEE methodology for estimation and statistical inference. We then combine the GEE methodology with the knowledge gained from Robieson (1999) and Carrasco and Jover (2003), and propose a unified approach that is applicable to continuous and categorical data.