Introduction
This paper addresses the validity of alternative models to the identification of learning disabilities (LD). In preparing this paper, we assumed that the "alternatives" involved any approach other than a discrepancy between a composite IQ test and achievement. Variations on four different approaches were considered:
- models involving other forms of discrepancy;
- low achievement models;
- intra-individual differences models; and
- models that incorporate response to intervention.
Two sets of assumptions were made in evaluating these models. The first assumptions were that the construct of LD operates as a latent variable that is imperfectly measured by these different models (Fletcher & Morris, 1986; Francis et al., in press). The key aspect of the construct is the notion of "unexpected underachievement," representing children and adults who should be able to learn, yet don't attain levels that would be expected based on the absence of obstacles to learning, including adequate opportunities to learn. Thus, a key aspect of evaluating validity is determining how well the approach to identification produces a unique group of poor achievers.
The second group of assumptions involves the assessment of validity. We assume that identification methods stem from an overarching classification that indicates how persons with LD are similar to and different from those with other childhood disorders, other groups of poor achievers, and typically achieving children (Fletcher et al., 1993). A valid classification will reflect definitions and measurements that operationalize the construct of unexpected underachievement by showing that the group identified as LD differ from other subgroups on variables not used to create the classification (Morris & Fletcher, 1988; Skinner, 1981). Obviously, people defined as LD will differ on variables used to create the classification, unless it is completely unreliable. The extent to which differences emerge may reflect the correlation of the external variables with those used to create the classification (Stuebing et al., 2002), which usually involve some form of cognitive assessment, so it is important to look beyond comparisons of cognitive skills whenever possible. In many instances, validity is limited because of problems with reliability (Francis et al., in press), but the focus of this paper is on the external validity of the classifications that lead to identification.
Previous Page | Next Page
(Table of contents) | (Other forms of discrepancy)
