Examination of item weighting on the Adaptive Behavior Scale Part II.

Abstract

The Adaptive Behavior Scale ( 5 ) has been used extensively to measure the social skills and maladaptive behavior of developmentally disabled individuals. Recently, the scale has been criticized because some of the items in Pam I1 vary widely in terms of perceived severity (1, 3, 4 ) . To correct this problem, Clements ( 1 ) published a table of weights for severity to be applied to the items of Parc 11. However, psychometric theory states that, in general, differential item weights are ncx useful because the correlation between weighted and unweighted scores is high. The purpose of this study was to examine the degree of association between weighted and unweighred domain scores and the Part 11 sum scores on the Adaptive Behavior Scale. The subjects, 211 institutionalized mentally retarded clients, were distributed over four levels of mental retardation as follows: 27 mild, 54 moderate, 89 severe, 41 profound. Consistent with theory, all weighted scores correlated .95 or more with their unweighted counterparts. For a discussion of possible reasons as to why weighting does not work the reader is referred to Gulliksen ( 2 ) . This does not, however, preclude the value of weighting for other purposes, such as the prioritization of habilitative, goals.