Inter-rater score agreement between untrained and trained pairs of child care workers who rated a mentally retarded sample with the AAMD adaptive behavior scale.

Abstract

A study was undertaken to test the reported inter-rater reliabilities on Part Two of the American Association on Mental Deficiency Adaptive Behavior Scale. In addition, this study trained raters in order to determine if trained raters would obtain higher reliabilities than untrained raters, since the AAMD claimed that untrained persons could accurately administer the scale. From a state institution, a sample of 16 child care workers from three different residential units rated retardates on their respective units. The total number of retardates from each residential unit totaled 22 individuals. The results indicated that the trained raters achieved significant relationships less often than did the untrained raters' when their domain scores were compared. In the first unit, the scores of all raters were found to have a significant relationship in five of the 14 domains, while only three of the raters' domain scores in the second unit were found to have significant relationships, and none of the domain scores of the raters in the third unit were found to have a significant relationship on a consistent basis that considered untrained/trained, untrained and trained rater combinations. Although the correlation coefficients of each rater type varied from one unit to another, only two of the domain correlations consistently supported the findings of the reliability study cited by the AAMD on Part Two of the revision of this scale.