Cognitive referencing is the practice of using IQ scores to establish eligibility for special education services, specifically in areas of language and learning disabilities. It’s often called by it’s gentler label, the “discrepancy model.” Many others disapprovingly call it the “wait to fail” model. Cognitive referencing has been denounced by groups such as the American Speech-Language-Hearing Association (link), the President’s Commission on Excellence in Special Education, 2002, and very explicitly, by the U.S. Department of Education (link, pg. 31). It has been eliminated in many states, but persists in many others. Even those who don’t come right out and denounce this practice (as they should), state that it should be only one component of a larger process used to determine eligibility (e.g. this CEC link). The problem is that wherever it is used, the IQ-Academic discrepancy becomes the sole method of determining eligibility in nearly all cases. In my state of Missouri, our state law very specifically mandates this discrepancy, unless a school district is willing to go through much expense and work to use other methods, such as RTI. My guess is that 99% of kids tested for LD and Language Impairment in our state use only IQ comparison to determine eligibility.
Despite its prevalence, cognitive referencing is wrong on many levels.
- It uses a single IQ score, ignoring standard deviation. A kid that scores 80, may actually have a “true” IQ of something like 85 or 90, but could have performed poorly on that one day, for various reasons. Tough luck for that kid. An IQ score of 80 usually means that your academic or language scores have to be 58 or lower, an extremely difficult thing to do.
- By even using IQ at all, the assumption is that this is as good as a kid can get. That was the initial rational for the discrepancy model way back before we knew better. Now we know that IQ can go up (or down) in relationship to environmental factors. (When IQ scores of large groups of children are studied, IQ scores do tend to remain stable, especially in older children. However, this skews the fact that a smaller percentage of children do show substantial IQ fluctuations over time. For more on this interesting topic, see Sigelman and Rider, 2008.)
- IQ and language are correlated. Vocabulary and IQ especially correlate well. This means that children with low language scores tend to have comparably low IQ scores. It is virtually impossible to obtain a low IQ score and say that language difficulties didn’t have something to do with that score.
- Kids with certain scores are especially difficult to qualify for special education under this model. Whenever a child scores in the 70s you can just about rest assured that the kid will not qualify, and you will be testing that kid again, perversely hoping that the academic and/or language scores have fallen enough to qualify the next time. In effect a child is punished for having an IQ score that just happens to be in that one certain range.
- IQ scores can set artificially low levels of expectation for kids, teachers, and parents. IQs describe obstacles, not limits. It may be harder for someone with a lower IQ to learn, but it is never impossible. Only comatose or dead people can’t learn, and IQ scores too often allow somebody to say, “Well he’s achieving close to his level.” IQs can provide a stimulus to somebody with a high IQ who is not motivated to learn, and can provide a bit of insight into why a particular student may be having trouble learning, but to withhold helping a child because of a lower then average IQ is at the least dishonest, and borders on unethical.