In its next paragraph, the Division 33 discussion allows for part scores, but the criterion for eligibility is that “two or more of these scores lie two or more standard deviations below the mean” (p. 13). The Division 33 system also adopted the now rather standard criterion of intellectual functioning at two or more standard deviations below the mean. The consequences of these requirements on the number of people currently considered for a diagnosis of mental retardation or on the number of persons considered in the future have not been addressed. The results of previous studies had suggested that a stringent criterion for adaptive behavior plus the usual criterion for intellectual functioning led to a sharply reduced number of people eligible to be considered for a mental retardation diagnosis (Heflinger et al., 1987; Reschly, 1981a). However, these studies used a particular measure of adaptive behavior that had a very low correlation with measures of intellectual functioning.
The classification criteria governing diagnosis of mental retardation for special education services by state departments of education generally do not provide guidance regarding the use of adaptive behavior composites, part scores, or cutoff scores to determine eligibility. It is not surprising that the use of an adaptive functioning criterion in the schools is inconsistent and unpredictable (Reschly & Ward, 1991). Moreover, enormous variations exist across the states and, in some instances, across local school districts within states.
Diagnostic systems are either silent on the appropriate number of adaptive behavior domains, or they adopt widely varying schemes. The most recent classification system of the American Association on Mental Retardation (AAMR) specifies 10 adaptive skills areas without any explanation of how that number was determined or why some domains were included and others excluded (American Association on Mental Retardation, 1992). In contrast, the American Psychological Association Division 33 diagnostic system is generally consistent with the factor analytic literature (see Chapter 4 and later discussion in this