Question
Note: The following information pertains only to test scores that are "normally distributed" (i.e., form a bell-shaped curve), as is the case with tests that report standard scores.
As a middle school SLP I receive many reports from other S
Answer
The very short answer to these questions is "It depends". The long answers relate to the purpose of the testing: Is it to identify which students qualify for services, or is it to establish a profile of language/learning strengths and weaknesses for the student and therefore to determine a specific, useful diagnosis and a specific, useful treatment plan?
To identify which students qualify for services, most school systems (in my experience) use the standard deviation, a measure of how scores spread around the mean. The standard deviation is useful because it takes the variance of the population into account. If most people's scores are close to the mean, the standard deviation will be small; if scores are spread out, then the standard deviation will be larger. Statistically, the middle 68% of the population is considered to be "normal"; these are the scores that fall within one standard deviation of the mean. Another 27% of the scores fall between one and two standard deviations from the mean (half above and half below) and 2.3% fall more than two standard deviations from the mean (half above and half below).
Of course, we are primarily concerned with those that fall below the mean (unless we are assessing a child to determine whether a gifted/talented program is appropriate). Sixteen percent of children fall 1 standard deviation or more below the mean; 2.5% fall 2 standard deviations or more below the mean; 1.15% fall 3 standard deviations or more below the mean. In percentile terms, children whose scores fall at the 16th percentile are one standard deviation below the mean, and so on.
Rules vary from state to state and even from school district to school district (in some states) with respect to how far below the mean the child must fall in order to qualify for services. In my experience, most cutoffs range from 1 to 2 standard deviations, although some school systems do provide services to children with scores at the 25th percentile or below. From a statistical point of view, the 25th percentile is low average - below the scores that would be earned by 75% of children at that age, but still within the average range.
Whether these rules apply to overall test quotients or to individual subtests should depend upon the significance of the subtest for the child's functioning in the setting in question, but this is a tricky route to follow. It would seem that memory for following directions, for example, could have far more impact on the child's academic success than number memory backwards.
On the other hand, number memory backwards performance could be a reflection of the child's general ability to hold and manipulate abstract information in his/her mind without external visual or auditory cues, and could therefore presage significant academic difficulties even if actually recalling a list of numbers is never requested by the teacher. Many assessment tasks require skills that are not obvious from the title of the subtest. (See Stackhouse and Wells 1997 for an excellent analysis of the "hidden" requirements of many phonological awareness tasks.)
This leads to the second function of testing: establishing a communication or learning profile for a child, which is a much more functional and flexible procedure. This is vital to understanding whether and why the student will be able to perform adequately in a given classroom and also to giving a usefully specific diagnosis.
For example, a child with a statistically significant difference between his/her verbal and non-verbal IQ scores could be technically considered to be learning disabled even if both scores are within or even above normal limits (e.g., one at the 85th percentile and one at the 99th). Such a child might seem to be brilliant at math, for instance (assuming that the nonverbal score is the higher of the two), but have a difficult time with word problems that are supposedly at his/her level of quantitative ability.
Some assessment instruments provide procedures for weighing the requirements of various subtests in order to identify a specific performance profile, but many tests lack this feature and sometimes even those that have it are unrevealing with respect to particular children. Inconsistent performance across subtests -- some very strong scores and some very weak scores on the same assessment instrument -- can be washed out if only the overall test score, or only a summary quotient, is considered. Often it is the comparison of the lower subscores from a group of tests that have been administered to the higher subscores from the same group of tests that provides the key to understanding a child's academic or other communicative struggles. This requires more flexible, abstract thinking on the part of the SLP: to figure out what the common underlying skills are for those subtests on which the child does well versus those on which he/she does poorly, and to put those concepts together with the child's apparent strengths and weaknesses within the classroom in order to determine the appropriate diagnosis and the most efficacious intervention plan.
Stackhouse, J., & Wells, B. (1997). Children's speech and literacy difficulties: A psycholinguistic framework. London: International Thomson Publishers.
Shelley L. Velleman is an associate professor of Communication Disorders at the University of Massachusetts at Amherst. She is the author of many articles about normal and disordered phonological development (including childhood apraxia of speech), and also of "Making phonology functional" (Butterworth-Heinemann,1998) and the "Childhood apraxia of speech resource guide" (Delmar/Thomson, 2003).
