Federal regulations mandate that children with reading problems are eligible for special education services only if there is a significant discrepancy (one standard deviation or at least 15 points) between their IQ and reading-achievement scores.
Researchers Karla K. Stuebing, Jack M. Fletcher and Josette M. LeDoux, University of Texas Health Science Center/Houston; G. Reid Lyon, National Institute of Child Health and Human Development; and Sally E. Shaywitz and Bennett A. Shaywitz, Yale University, report that there is little evidence supporting the validity of the IQ-discrepancy classification and doubt the need for IQ tests in identifying reading-disabled students. Although they stress that learning disabilities are real, they believe that use of an IQ discrepancy can be harmful to children because it often requires that children fail before they get help.
Discrepancy between IQ and achievement as the standard
Public Law 94-142 states that a severe discrepancy between achievement and intellectual ability must exist for a child to be labeled learning disabled. This discrepancy is described as a specific learning disability that is distinct from underachievement and requires different instruction. Children who are poor readers but who do not display this discrepancy are not eligible for special education.
However, a meta-analysis carried out by Stuebing et al. reveals a substantial overlap between IQ-discrepant and IQ-consistent children (those with less than a 15-point discrepancy between their IQ and reading-achievement scores).
Two large meta-analyses previously conducted reached contradictory conclusions. Stuebing et al. state that this was due largely to differences in the research questions and sampling procedures. Stuebing et al. carried out their meta-analysis because of their concern that federal regulations defining learning disabilities may be inhibiting early and appropriate help for reading problems and that the definition requiring a discrepancy between IQ and achievement may not be based on rigorous research findings.
Discrepancy often identified after second grade
Previous research suggests that reading difficulties are chronic and persistent when identified after second grade, and that remedial interventions after second grade are less effective than early interventions. Many students appear to be identified after second grade because their achievement needs to fall far enough behind to create a large enough discrepancy between their achievement and ability. Only then can they be labeled learning disabled and qualify for special education.
This study showed that the IQ-discrepant group had higher IQ scores than the IQ-consistent group and lower reading achievement scores. There was little difference on the reading, writing and math tests between the two groups, but spelling, oral reading, real-word decoding and pseudoword decoding showed small but significant differences, suggesting lower performance by poor readers who are IQ-discrepant.
Overall, there was significant overlap between the two groups. IQ discrepancy did not significantly discriminate or provide important diagnostic information between the two groups, in these researchers’ opinion. These groups were not appreciably different on the four variables most closely linked to reading proficiency and poor reading: phonological awareness, rapid naming, verbal short-term memory and vocabulary skills.
There were statistically significant differences in nonlinguistic factors, but these differences appear to have little to do with reading problems. Differences in cognitive abilities amount to about three-tenths of a standard deviation.
It is not surprising that groups formed on the basis of IQ discrepancy show relatively strong development of some cognitive skills, since by definition this group will have more students with higher IQ scores. This does not mean, however, that IQ scores are highly relevant in explaining a reading disability. Stuebing et al. conclude that any classification of poor readers based on IQ discrepancy is an “antifactual distinction based on arbitrary subdivisions of the normal distribution.”
Early intervention for all children with reading difficulties
Since there is a lot of evidence that early intervention is effective for addressing reading difficulties and because later remediation is often less effective, these researchers call for early intervention for all children with reading-skill deficits. Stuebing et al. state that the types of pre-reading items used to assess early reading skills on achievement tests are not necessarily difficult for many children at risk for reading problems.
Research data also indicates that IQ-achievement discrepancies are not reliable in first grade and that many schools do not assess children for LD before second grade because of the difficulty in documenting discrepancies. Therefore, many children fail before they receive services.
These researchers believe that the IQ-discrepancy model does more harm than good because of the evidence supporting the greater efficacy of early intervention in kindergarten through second grade relative to remedial services provided from third grade on. Because of the need for a large discrepancy, children with higher IQ scores are more likely to be identified early. This practice might be justified if IQ or IQ discrepancy were related to prognosis or intervention outcomes, but Stuebing et al. conclude that there islittle evidence that this is so.
Assessing skills predictive of reading ability
The correlation of IQ and reading achievement does not validate the two-group classification, nor does it indicate an essential role for IQ tests in the decision-making process, especially in light of the time required and the high costs. Stuebing et al. assert that alternative approaches to definition that eliminate IQ discrepancy and promote early identification are possible.
They state that inclusionary definitions that specify criteria that represent attributes that are positive indicators of learning disabilities are possible, not only for reading but for math also. An advantage of inclusionary definitions is that they clearly indicate approaches to assessment that may be directly relevant for intervention.
For example, components of the reading process such as word recognition, fluency and comprehension as well as cognitive processes directly related to the type of reading disorder can be used in assessment. Specifically, for word recognition difficulties the assessment would include measures of phonological awareness, rapid naming, working memory, and vocabulary.
By focusing on the component processes, early identification is possible as the tasks do not require reading but predict reading development. Stuebing et al. stress that they are not recommending that low achievement definitions be substituted for IQ discrepancy. They are suggesting that students who appear to be struggling in beginning reading skills should be carefully assessed with skills-based assessments.
Stuebing et al. conclude that the weak evidence for the IQ-discrepant classification found in this meta-analysis does not mean that the concept of learning disabilities is invalid. There is evidence supporting the idea that people identified by this definition are different from typically achieving children neurobiologically, cognitively and socially.
However, emerging research knowledge is not being used for classifications of LD in federal regulations. There is a substantial overlap between poor readers identified as LD and those not so identified, and in the past 25 years much research has shown that IQ-discrepant classifications have weak validity and are potentially harmful because they delay services for many children.
The solution is to revise the methods of identification and to link them to research so children can receive early identification and effective remedial instruction and educators can prevent more reading failure.
“Validity of IQ-Discrepancy Classification of Reading Disabilities: A Meta-Analysis” American Educational Research Journal Volume 39, Number 2, Summer 2002 Pp. 469.
Published in ERN October 2002 Volume 15 Number 7