vary as to whether they consider adaptive behavior to be made up of a single factor or to have multiple factors or domains. In the definitions that imply a multifactor construct, deficits in adaptive behavior must be specified in a certain number of areas/domains. With regard to identifying decision-making criteria, Division 33 presents the only definition that employs a statistical cutoff based on standard norms. In contrast, the other definitions employ more qualitative terms, which are open to interpretation in describing deficits and limitations in adaptive behavior.
Multidimensional or Unidimensional? Answers to this question have been mixed. Meyers et al. (1979) concluded from their review of factor analytic studies that adaptive behavior was definitely multidimensional and that the use of a total score would be inappropriate to indicate a general level of adaptation. Their view has been both supported and disputed in the past two decades, and there are currently firm adherents on each side of this issue. McGrew and Bruininks (1989) and Thompson et al. (1999) have concluded, for example, that the number of factors emerging from factor analyses depends on whether data were analyzed at the item, parcel, or subscale level, with fewer factors found for subscale-level data than item- or parcel-level data.
They also found that it was not the selection of the instrument that determined the number of factors. This important finding has direct implications for definitions that require limitations to be observed in a specific number of areas. If there is actually one underlying domain that “causes” behaviors in all different conceptual domains, and there is relatively little unique variance found in each domain, then a total score with a single cutoff point could reliably distinguish those with and without significant limitations. If not, diagnosticians would have to consider a profile of adaptive behavior deficits that takes all domain scores into account. Widaman et al. (1991) and Widaman and