Models of agreement can also satisfy the property of quasi-symmetry (Darroch & McCloud, 1986). Because there is no objectively precise definition of classification categories for most cases in the social sciences, the discrepancies between classifications by different methods are attributable to measurement error and to different perceptions or interpretations of what a category definition means. "The correct category for an object exists partially in the eye of the beholder" (Darroch & McCloud, 1986, p. 376). On the other hand, there are signals sent out by each object that partially conform to each of the categories to a certain degree. These signals are assumed to differ between objects. Thus, the classification of an object into a particular category depends on the signals sent out by the object and the rater-specific category definition. Agreement between methods is displayed in the joint distribution of the marginal distributions of objects:
Was this article helpful?