In case in which measurements are based on subjective classifications or ratings, Kappa techniques prove to be most appropriate for classifying nonquantitative data, while interclass correlation for ratings based on some type of scale. For example, to effectively classify objects as bad or good or to classify fabric flaws according to defects, Kappa techniques should be used for the reason that
... [Show full abstract] Kappa techniques treat all misqualifications equally. Designed for purely nominal classification, Kappa techniques require only that the units be independent, that the judges or raters make their classification independently and that the categories be mutually exclusive and exhaustive. For assessing the reliability of quantitative ratings however, interclass correlation should be used rather than the Kappa techniques. Interclass correlation is preferably used when data can be ordered and when the perception of distance between these ordered categories is roughly equal. Formulation for both techniques and sample measurements are presented.