Nominal scale agreement among observers

摘要: Experiments are considered where each of a sample subjects is assigned to one C categories separately by fixed or varying group observers. Building on earlier publications, general procedures proposed analyze agreements and disagreements among In the case observers, it shown that not necessary demand constant number observers per subject. problem missing data considered. The illustrated within context two clinical diagnosis examples. first example investigated which relatively hard distinguish from another; new theorem applied shows useful property statistic kappa. second if subgroup can be found with significantly higher degree interobserver agreement.