Is it possible to link specific cognitive deficits with specific deficits in reading skills?

  

Feifer, S., Nader, R. G., Flanagan, D., Fitzer, K., & Hicks, K. (2014). Identifying specific reading disability subtypes for effective educational remediation. Learning Disabilities: A Multidisciplinary Journal, 20(1).  https://ldaamerica.org/wp-content/uploads/2013/10/LDMJ_free-article.pdf

"Overall results suggest that the specific cognitive subtests that are predictive of Letter-Word Identification, Reading Fluency, and Passage Comprehension vary depending on the subtype of Specific Learning Disability".

I'm not sure that this conclusion is the right one to derive from the data presented in this study.

The need to prove a linkage between a poor cognitive ability (one or more of the seven broad abilities Comprehension Knowledge, Fluid Ability, Short Term Memory, Long Term Storage and Retrieval, Processing Speed, Visual Processing and Auditory Processing) and poor specific performance in reading/writing/math (for example, poor decoding, poor reading fluency and/or poor reading comprehension) lies at the heart of Flanagan's approach to learning disability definition.  In this study, Feifer, Flanagan and their colleagues try to demonstrate the existence of such connections between cognition and performance in reading.

Two hundred and eighty-three students aged 6-16, studying in grades 2-12 participated in this study.  They were referred for evaluation due to learning and/or behavior problems.  Most of them (194) were boys.  They all had an IQ score of above 75 (usually a child who meets Flanagan's definition will have a higher IQ, since most of his cognitive abilities are supposed to be at least average.  But theoretically a child who scores 85 on most of the cognitive abilities, and has a very low score on one ability, will have an IQ score lower than 85).

All students were given tests from the WJ3 battery.  They were given 14 tests out of the cognitive battery (two tests for each cognitive ability) and three tests from the achievement battery:  Letter Word Identification, Reading Fluency and Passage Comprehension.
 

The students were divided into six groups according to a theoretical conceptualization:  an Associative Learning group (which was supposed to be poor at Long Term Storage and Retrieval); a Fluid Ability – Visual Processing group (which was supposed to be poor at those abilities); a Comprehension Knowledge group (which was supposed to be poor at Comprehension Knowledge); a Learning Efficiency group (which was supposed to be poor at Processing Speed), an Executive Functions group (which was supposed to be poor at Short Term Memory) and a control group (with no cognitive deficits).  The rational for this grouping is not clear to me and is not explained in the paper.  Perhaps it wasn't possible to group the participants according to the seven CHC broad abilities.  For instance, I see in the data that scores in Auditory Processing tests were average in all groups. This is odd since poor auditory processing is an indicator of poor reading.

In all groups except the control group, children had similarly poor scores in the three reading tests.  This is also odd.  The main point in Flanagan's definition is that specific poor reading skills are caused by specific poor cognitive abilities.  So we might have expected to find different "profiles" of reading test scores in each group.  For instance, it might have been expected that in the Associative Learning group, Letter Word Identification and Reading Fluency would be poor but Passage Comprehension would not necessarily be poor.  The results were the opposite (the lowest score in this group was in passage comprehension.  The scores in the three reading tests were in fact very similar.  It is not reported if the small differences between scores are significant).

All children had at least average scores on most cognitive abilities.

I prepared this table out of the data:

Proportion of predicted variance	What was predicted	Tests in this group that predicted reading tests and the broad abilities they measure	The broad ability that was supposed to be poor in this group	Test scores in this group that were below 85 and the broad abilities they measure.	Group's name
0.449	Letter-Word ID	Sound Blending  (Ga) Numbers Reversed)  Gsm)	-	All scores were average and above	Control
0.239	Reading Fluency	Numbers Reversed)  Gsm)
0.493	Passage Comprehension	Numbers Reversed)  Gsm) Picture Recognition )Gv(
0.187	Letter-Word ID	Visual Matching) Gs( Numbers Reversed)  Gsm)	Glr	Visual Auditory Learning) Glr( Retrieval Fluency (Glr)	Associative learning
0.147	Passage Comprehension	Numbers Reversed)  Gsm)
0.401	Letter-Word ID	Sound Blending  (Ga) Auditory Attention (Ga)	Gv Gf	Visual Auditory Learning) Glr( Retrieval Fluency (Glr) Concept Formation)Gf(	Fluid/Visual Processing
0.409	Reading Fluency	Visual Matching) Gs(
0.406	Passage Comprehension	General Information)  Gc( Memory for Words )Gsm(
0.100	Letter-Word ID	Visual Matching) Gs(	Gc	Verbal Comprehension (Gc) General Information   (Gc( Visual Auditory Learning) Glr(	Crystallized
0.286	Reading Fluency	Visual Matching) Gs( Decision Speed (Gs)
0.210	Passage Comprehension	General Information)  Gc(
0.229	Reading Fluency	Numbers Reversed)  Gsm)	Gs	Visual Matching) Gs(	Learning Efficiency
0.391	Passage Comprehension	Numbers Reversed)  Gsm) Memory for Words )Gsm(
0.323	Letter-Word ID	Verbal Comprehension (Gc)	Gsm	Numbers Reversed   (Gsm) (the score was ( 84.53	Executive Subtype
0.142	Reading Fluency	Visual Auditory Learning  (Glr(
0.399	Passage Comprehension	Visual Auditory Learning   (Glr( Picture Recognition )Gv(

Table 1:  groups, low test scores in each group and tests predicting reading in each group

Glr long term storage and retrieval; Gsm short term memory; Gv visual processing; Gs processing speed; Gf fluid ability; Ga auditory ability; Gc comprehension knowledge

A close look at this table reveals two interesting things:

A.  The ability that was supposed to be low in a specific group and the test scores that were actually low in that group did not always fit.  For instance, in the "Fluid Ability/Visual Processing" group, Fluid Ability and Visual Processing tests were supposed to be low, but 2 of the tests that were actually low measured Long Term Storage and Retrieval.  The third low test score did measure Fluid Ability (but the score of the second Fluid Ability tests that was presented to the participants was average).  This means that in this instance the group's name did not reflect the poor abilities in this group.

B.  There were many incongruencies between tests that predicted reading in a group, abilities that were supposed to be poor in that group, and tests that were actually low in that group.  For example, in the "Associative Learning" group, Long Term Storage and Retrieval was supposed to be poor.  Low test scores in this group do indeed measure Long Term Storage and Retrieval.  But the tests that predicted reading in this group measure Processing Speed and Short Term Memory, not Long Term Storage and Retrieval.  This means that the classification of children to the "Associative Learning" group was not relevant to the prediction of reading.  The predictors of reading in this group were two tests in which children had average scores (Visual Matching (89.88) and Numbers Reversed (88.51)) and  that measured Processing Speed and Short Term Memory. 

I find the researchers' explanations for these incongruencies not entirely convincing.  For instance: "nearly 19% of the variance of their reading performance was accounted for by scores from the Visual Matching (perceiving the visual contour and shapes of numbers quickly) and Numbers Reversed (working memory) subtests. This finding suggests that these students may have struggled with the orthographical representation of print, coupled with the working memory demands inherent in print knowledge. Reading performance was likely weak for these students because reading requires the ability to detect the symbolic representation of letters as making up individual words  (i.e., orthographic processing), and holding and manipulating this information in the mind’s eye (i.e., working memory)".

What Feifer, Flanagan and their colleagues are doing here is to say:  among the children who are poor at A, B explains reading difficulties".  This does not explain why the poor performance in A in and of itself does not explain the reading difficulties.  Why the poor performance in Associative Learning tests in this group does not predict the poor reading scores.  In associative learning tests (like Visual Auditory Learning) the child learns to link sounds with symbols, which is the essence of reading.

Thus it seems that this study is not able to prove the existence of links between poor specific cognitive abilities and poor specific reading performance.  I'm sure such links exist, but this study does not show that.

What can be learned from this study?  Here are some of my conclusions (not the researchers'):

1.  Letter and Word Identification can be predicted by these WJ3 tests:  Sound Blending, Numbers Reversed, Visual ,Matching, Verbal Comprehension.

2.  Reading Fluency can be predicted by Numbers Reversed, Visual Matching, Decision Speed, Visual Auditory Learning.  Obviously Processing Speed predicts reading fluency.

3. Passage Comprehension can be predicted by Numbers Reversed, Picture Identification, General Knowledge, Memory for Words and Visual Auditory Learning.  It's interesting that tests that measure Fluid Ability do not predict reading comprehension, and that only one test out of the two that measure Comprehension Knowledge predict Passage Comprehension.

4.  Numbers Reversed test is a predictor of all reading skills.

Does IQ predict future achievement in reading, writing and math?

Watkins, M. W., Lei, P. W., & Canivez, G. L. (2007). Psychometric intelligence and achievement: A cross-lagged panel analysis. Intelligence, 35(1), 59-68.‏  http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.397.3155&rep=rep1&type=pdf

https://pdfs.semanticscholar.org/4bd0/28fb097c364b5572ac0680ea92ff0f0bf558.pdf

http://edpsychassociates.com/Papers/IQachieveCrossLag(2007).pdf

 Fletcher and Miciak (2017) claim that "there is substantial evidence showing little difference between IQ-discrepant and low achieving children in achievement, behavior, or cognitive skills, prognosis, intervention outcomes, and neuroimaging markers of brain function".

IQ-DISCREPANT are children who have a discrepancy between their IQ score and their reading/writing/ math scores. The term IQ-DISCREPANT usually refers to children with poor reading/writing/math and (at least) average intelligence.  LOW ACHIEVING in Fletcher and Miciak's paper refers to children who have poor reading/writing/math and lower than average IQ scores.   These children do not have a discrepancy between their IQ and achievement scores.

Assertions like Fletcher and Miciak's were one of the reasons for abandoning the discrepancy definition of learning disability (learning disability as a discrepancy

between at least average IQ and poor reading/writing/math that cannot be explained by exclusionary factors) in DSM5.

This article by Watkins, Lei  & Canivez presents a slightly different picture:

In current usage, intelligence tests are thought to measure general reasoning skills that are predictive of academic achievement. Indeed, concurrent IQ–achievement correlations are substantial and, consequently, comparisons of IQ and achievement scores constitute one of the primary methods of diagnosing learning disabilities (at least when this paper was written). However, intelligence tests often contain items

 or tasks that appear to access information that is taught in school (i.e., vocabulary, arithmetic) and there has been considerable debate regarding the separateness or

 distinctiveness of intelligence and academic achievement.  This apparent overlap in test coverage, among other factors, has led some to view intelligence and achievement as identical constructs. Some researchers have suggested that the relationship between intelligence test scores and educational achievement is reciprocal, mutually influencing each other.  According to this approach, children who read a lot develop their cognitive abilities and intelligence.  Children who do not read because of learning disabilities have less opportunity to develop these abilities.  Subsequently, special education researchers have suggested that only achievement tests should be used to identify children with learning  disabilities (as Fletcher suggests).  Other researchers assert that intelligence is causally related to achievement.

In order to determine whether and to what extent IQ affects achievement (or vice versa), children must be tested twice with an IQ test and twice with achievement tests over a period of a few years. If IQ affects achievement and causes it, the correlation between the IQ scores obtained in the first measurement (IQ 1) and the achievement scores obtained in the second measurement (achievement 2) should be higher  than the correlation between the achievement scores obtained in the first measurement (achievement 1) and the  IQ scores obtained in the second easurement (IQ2).

Two thousand school psychologists were randomly selected from the National Association of School Psychologists membership roster and invited via mail to participate in this study by providing test scores and demographic data obtained

 from recent special education triennial reevaluations. Data were voluntarily submitted on 667 cases by 145 school psychologists from 33 states. Of these cases, 289 contained scores for the requisite eight WISC-III and four academic

achievement subtests.

Special education diagnosis upon initial evaluation included 68.2% learning disability, 8.0% emotional disability and 8.0% mental retardation.  The rest of the students received other diagnoses.  The mean age of students at first testing was 9.25 years and the mean age of students at second testing was 12.08.

Contemporary versions of the Woodcock–Johnson Tests of Achievement, Wechsler Individual Achievement Test, and Kaufman Test of Educational Achievement were used in more than 90% of the cases. In reading, all achievement tests included separate basic word reading and reading comprehension subtests. In math, separate calculation and reasoning subtests were available for all academic achievement instruments

Here are some interesting correlations I found in the second testing (which took place when the child had already spent about three years in special education):

Basic reading skills were correlated 0.56 with Information, 0.42 with Similarities, 0.49 with Vocabulary.

Reading comprehension was correlated 0.64 with Information, 0.54 with Similarities, 0.47 with Picture Arrangement, 0.50 with Block Design, 0.60 with Vocabulary, 0.50 with Comprehension subtest.

Mathematical calculations were correlated 0.62 with Information, 0.55 with Similarities, 0.52 with Picture  Arrangement, 0.53 with Block Design, 0.57 with Vocabulary and 0.55 with Comprehension.

Mathematical reasoning was correlated 0.70 with Information, 0.63 with Similarities, 0.52 with Picture Arrangement, 0.58 with Block Design, 0.67 with Vocabulary, 0.65 with Comprehension.

The relatively high correlation of Information and Vocabulary with all achievement tests stands out.

In the first testing (before the child entered special education) the highest correlations were found between those same IQ subtests and achievement tests, but correlations were generally lower. The reason for this is unclear to me and the   researchers do not explain it.

Another thing that stood out to me was that the mean of the group of children in the Verbal Comprehension and Perceptual Organization indices did not change between the first and the second testing. This may be an indication of the stability of intelligence.  On the other hand, this may mean that the intervention the children may have received in special education did not improve their crystallized knowledge.

Even more striking is the fact that the average scores in basic reading, reading comprehension, mathematical calculations, and mathematical reasoning have not changed during these two years and eight months. This means that the children did

not make progress in their skill level relative to the norm, but on the other hand, they also did not fall behind. Another interesting thing is that the children's average scores in the achievement domains were average (around 85), not lower.

Oh, Glutting, Watkins, Youngstrom, and McDermott (2004) demonstrated that both g (general intelligence) and Verbal Comprehension contributed to the prediction of academic achievement, although g was at least three times more  important than Verbal Comprehension.

In the present study, the average correlation between IQ1 and Achievement2 was 0.466 while the average correlation between Achievement1 and IQ2 was 0.398. This means that IQ predicts achievement, not the other way around.

IQ tests were built by Alfred Binet to measure (and predict) the ability of students to succeed at school. This basic feature of IQ tests has been empirically supported for more than 100 years and is also supported by this study.

The assertion that IQ predicts future achievement has been tested with students in regular education.  In this study, it was examined with special education students and has also been confirmed. Some researchers have suggested that correlations between reading and IQ tests may often be an artifact of language, which affects both reading and intelligence.  By this line of thinking, reading difficulties lower

 IQ scores over time, and cause them to be weak predictors of achievement in students with learning disabilities.

One of the most influential researchers in the field of reading, Linda Siegel, wrote in 1998: “low scores on the IQ tests are a consequence, not a cause, of … reading disability”. I can find some logic in this argument, but I would mitigate it and say:

poor scores in some IQ subtests may also be caused by learning disability.

 However, this position was not confirmed by the present results nor by those of Kline, Graham, and Lachar (1993), who found IQ scores to have comparable external validity for students of varying reading skill. Nor was such a conceptualization supported by the relatively high long-term stability of WISC-III IQ scores among more than 1000 students with disabilities. Further, IQ has been a protective factor in several studies. In a longitudinal analysis, Shaywitz et al. (2003) found that two groups of impaired readers began school with similar reading skills and socioeconomic characteristics, but those students with higher cognitive ability became significantly better readers as young adults. A meta-analysis of intervention research for adolescents with LD demonstrated that IQ exercised similar protective effects (Swanson, 2001). A New Zealand 25-year longitudinal study found strong relationships between IQ at age 7 and 8 and academic achievement at ages 18– 25 years, independent of childhood conduct problems as well as family and social circumstances (Fergusson, Horwood, & Ridder, 2005). In sum, considerable evidence contradicts the assertion that IQ has no predictive or seminal relationship with academic achievement.

 The present study provides evidence that psychometric intelligence is predictive of future achievement whereas achievement is not predictive of future psychometric intelligence.

In conclusion, Fletcher and Miciak argue that there is no difference between children with and without an IQ-Achievement discrepancy in achievement, behavior,

 cognitive abilities, prognosis, intervention outcomes, and neuroimaging markers of brain function.

This study suggests that there is a difference in prognosis between these two groups of children.