Developing criteria for establishing interrater reliability of specific items: applications to assessment of adaptive behavior.

Abstract

A set of criteria based upon biostatistical considerations for determining the interrater reliability of specific adaptive behavior items in a given setting was presented. The advantages and limitations of extant statistical assessment procedures were discussed. Also, a set of guidelines for differentiating type of adaptive behavior that are statistically reliable from those that are reliable in a clinical or practical sense was delineated. Data sets were presented throughout in order to illustrate the advantages of recommended statistical procedures over other available ones.