Investigation of the reliability of the A.A.M.D. adaptive behavior scale

The purpose of this Study was to provide reliability data on the American Association of Mental Deficiency Adaptive Behavior Scale. The subjects consisted of 80 adult mentally retarded residents in a community centered residence. Half of the subjects were mildly and half the subjects were moderately retarded men and women. The three methods of administration described in the test manual were all examined. For the First Person Assessment Method Raters trained in the standardized administration and scoring of the tests independently filled out the Adaptive Behavior Scale. In the Third Person Administration Method Examiners asked the Informants all questions on the Scale. The Interview Method utilized Examiners who asked Informants in a general way about each of the 24 subdomains on the Adaptive Behavior Scale. Two different reliability indices based upon inter rater agreement were employed. In order to obtain a measure of agreement between two raters in specifying each competency a Mean Number and Percent of Disagreement was calculated. Another index of reliability obtained was the Mean Range and Mean Percent Range of Disagreements for each Behavior Domain. In addition, for comparative purposes reliabilities for each Behavior Domain were found using the same statistical technique the Manual describes. A rank order correlation was done to compare each method of administration to the Adaptive Behavior Scale scores. Mild and moderate categories did not differ greatly in mean number or percent of rater disagreement. Nor were great differences found between the two categories of mental retardation for mean ranges and percent range of disagreement. An examination of the reliability by means of a correlational analysis revealed that the reliability for all methods of administration were much lower than that reported in the A.A.M.D. Manual. Reliability was however, highest for the Third Person Administration Method. An examination of the individual Behavior Domains revealed that some Behavior Domains were consistently more reliable and could be characterized as being categorically oriented and of infrequent occurance. Behavior Domains which were specific in nature were found to be less reliable. These results raise serious questions about the reliability of the scale and further provide information about the Behavior Domains which need much more examination before the instrument can be considered useful in the diagnosis of mental retardation