Reliability analysis of the motivation assessment scale: A failure to replicate

Abstract

The motivation Assessment Scale (MAS) has been proposed as an efficient questionnaire for identifying the source of reinforcement for an individual's self-injurious behavior (SIB). A previous reliability analysis of the MAS (Durand & Crimmins, 1988) reported interrater correlation coefficients ranging from .66 to .92, based on a comparison of responses provided by classroom teachers. In this study, the reliability of the MAS was reexamined with two independent groups of developmentally disabled individuals who exhibited SIB (N = 55). For the institutional sample (n = 39), the MAS was given to two staff members (a supervisor and therapy aide) who work with the individual daily. For the school sample (n = 16), the MAS was given to the teacher and teacher's aide who taught the student. The correlational analyses completed by Durand and Crimmins (1988) were repeated; in addition, a more precise analysis of interrater reliability was calculated based on the actual number of scoring agreements between the two raters. Results showed that only 16 of the 55 raters agreed on the category of reinforcement maintaining their client's or student's SIB, that only 15% of the correlation coefficients obtained were above .80, and that none of the reliability scores based on percent agreement between raters was above 80%.