Premium
A review of statistical methods in the analysis of data arising from observer reliability studies (Part II) *
Author(s) -
Landis J. Richard,
Koch Gary G.
Publication year - 1975
Publication title -
statistica neerlandica
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.52
H-Index - 39
eISSN - 1467-9574
pISSN - 0039-0402
DOI - 10.1111/j.1467-9574.1975.tb00259.x
Subject(s) - biostatistics , chapel , public health , library science , citation , operations research , mathematics , medicine , history , computer science , art history , nursing
Many research designs in studies of observer reliability give rise to categorical data via nomial scales(e.g., states of mental health such as normal, neurosis, and depression) or ordinal scales (e.g., stages of disease such as mild, moderate, and severe). In these situations, each of the d observers classifies each subject once into exactly one of a fixed set of L categories. As such, these designs are directly analogous to those giving rise to the standard ANOVA models in (2.1), (2.5), and (2.10) when the measurement scale is assumed to be quantitative. However, standard ANOVA procedures are rarely appropriate for the analysis of nominal and ordinal scaled data. As a result, these data are usually cross-classified into an Ld contingency table, and can then be analyzed by techniques developed for multidimensional contingency tables.