The Relation between Intelligence and Adaptive Behavior: A Meta-Analysis

Very important meta-analysis of AB IQ relation. Primary finding on target with prior informal synthesis by McGrew (2015)

The Relation between Intelligence and Adaptive Behavior: A Meta-Analysis   
 
Ryan M. Alexander 
 
ABSTRACT 
 
Intelligence tests and adaptive behavior scales measure vital aspects of the multidimensional nature of human functioning. Assessment of each is a required component in the diagnosis or identification of intellectual disability, and both are frequently used conjointly in the assessment and identification of other developmental disabilities. The present study investigated the population correlation between intelligence and adaptive behavior using psychometric meta-analysis. The main analysis included 148 samples with 16,468 participants overall. Following correction for sampling error, measurement error, and range departure, analysis resulted in an estimated population correlation of ρ = .51. Moderator analyses indicated that the relation between intelligence and adaptive behavior tended to decrease as IQ increased, was strongest for very young children, and varied by disability type, adaptive measure respondent, and IQ measure used. Additionally, curvilinear regression analysis of adaptive behavior composite scores onto full scale IQ scores from datasets used to report the correlation between the Wechsler Intelligence Scales for Children- Fifth edition and Vineland-II scores in the WISC-V manuals indicated a curvilinear relation—adaptive behavior scores had little relation with IQ scores below 50 (WISC-V scores do not go below 45), from which there was positive relation up until an IQ of approximately 100, at which point and beyond the relation flattened out. Practical implications of varying correlation magnitudes between intelligence and adaptive behavior are discussed (viz., how the size of the correlation affects eligibility rates for intellectual disability).
 
Other Key Findings Reported
 
McGrew (2012) augmented Harrison's data-set and conducted an informal analysis including a total of 60 correlations, describing the distributional characteristics observed in the literature regarding the relation. He concluded that a reasonable estimate of the correlation is approximately .50, but made no attempt to explore factors potentially influencing the strength of the relation.
 
Results from the present study corroborate the conclusions of Harrison (1987) and McGrew (2012) that the IQ/adaptive behavior relation is moderate, indicating distinct yet related constructs. The results showed indeed that the correlation is likely to be stronger at lower IQ levels—a trend that spans the entire ID range, not just the severe range. The estimated true mean population is .51, and study artifacts such as sampling error, measurement error, and range departure resulted in somewhat attenuated findings in individual studies (a difference of about .05 between observed and estimated true correlations overall).
 
 
The present study found the estimated true population mean correlation to be .51, meaning that adaptive behavior and intelligence share 26% common variance. In practical terms, this magnitude of relation suggests that an individual's IQ score and adaptive behavior composite score will not always be commensurate and will frequently diverge, and not by a trivial amount. Using the formula Ŷ = Ȳ + ρ (X - X ̅ ), where Ŷ is the predicted adaptive behavior composite score, Ȳ  is the mean adaptive behavior score in the population, ρ  is the correlation between adaptive behavior and intelligence, X is the observed IQ score for an individual, and X ̅ is the mean IQ score, and accounting for regression to the mean, the predicted adaptive behavior composite score corresponding to an IQ score of 70, given a correlation of .51, would be 85 —a score that is a full standard deviation above an adaptive behavior composite score of 70, the cut score recommended by some entities to meet ID eligibility requirements. With a correlation of .51, and accounting for regression to the mean, an IQ score of 41 would be needed in order to have a predicted adaptive behavior composite score of 70. Considering that approximately 85% of individuals with ID have reported IQ scores between 55 and 70±5 (Heflinger et al., 1987; Reschly, 1981), the eligibility implications, especially for those with less severe intellectual impairment, are alarming. In fact, derived from calculations by Lohman and Korb (2006), only 17% of individuals obtaining an IQ score of 70 or below would be expected to also obtain an adaptive behavior composite score of 70 or below when the correlation between the two is .50. 
 
 
The purpose of this study was to investigate the relation between IQ and adaptive behavior and variables moderating the relation using psychometric meta-analysis. The findings contributed in several ways to the current literature with regard to IQ and adaptive behavior. First, the estimated true mean population correlation between intelligence and adaptive behavior following correction for sampling error, measurement error, and range departure is moderate, indicating that intelligence and adaptive behavior are distinct, yet related, constructs. Second, IQ level has a moderating effect on the relation between IQ and adaptive behavior. The correlation is likely to be stronger at lower IQ levels, and weaker as IQ increases. Third, while not linear, age has an effect on the IQ/adaptive behavior relation. The population correlation is highest for very young children, and lowest for children between the ages of five and 12. Fourth, the magnitude of IQ/adaptive behavior correlations varies by disability type. The correlation is weakest for those without disability, and strongest for very young children with developmental delays. IQ/adaptive behavior correlations for those with ID are comparable to those with autism when not matched on IQ level. Fifth, the IQ/adaptive correlation when parents/caregivers serve as adaptive behavior respondents is comparable to when teachers act as respondents, but direct assessment of adaptive behavior results in a stronger correlation. Sixth, an individual's race does not significantly alter the correlation between IQ and adaptive behavior, but future research should evaluate the influence of race of the rater on adaptive behavior ratings. Seventh, the correlation between IQ and adaptive behavior varies depending on IQ measure used—the population correlation when Stanford-Binet scales are employed is significantly higher than when Wechsler scales are employed. And eighth, the correlation between IQ and adaptive behavior is not significantly different between adaptive behavior composite scores obtained from the Vineland, SIB, and ABAS families of adaptive behavior measures, which are among those that have been deemed appropriate for disability identification. Limitations of this study notwithstanding, it is the first to employ meta-analysis procedures and techniques to examine the correlation between intelligence and adaptive behavior and how moderators alter this relation. The results of this study provide information that can help guide practitioners, researchers, and policy makers with regard to the diagnosis or identification of intellectual and developmental disabilities.


- Posted using BlogPress from my iPad

Stephen Greenspan on "Why DSM5 suggested a switch from adaptive behavior to adaptive reasoning": APA Div 33 featured conversation

 

My long time professional colleague Dr. Stephen Greenspan, is conducting a featured conversation hour for Division 33 at the forthcoming APA convention in Denver.  He has provided me an advanced copy of his outline and has graciously given me permission to make it available at the ICDP blog.  A copy can be obtained by clicking here.

Stephen is one of the great "thinkers" in the field of intellectual disabilities.  Our professional lives crossed long distance when I was a doctoral student.  My advisor, Dr. Robert Bruininks, put me in charge of a series of studies investigating the constructs of adaptive and maladaptive behavior.  These studies eventually led to my dissertation--which was a CFA validation study of Greenspan's Model of Personal Competence (see 1990 reference below).  To the best of my knowledge, this was the first published article validating Greenspan's model.

Below are links to the various articles (I simply grabbed them from my MindHub web page--please visit if you want additional information).  Consistent with Stephen's outline notes, in this validated model of personal competence, conceptual intelligence was operationalized as measured by intelligence tests, and was not considered a domain of adaptive behavior.

Of interest is the recent study by MaCann et al. that provides structural (CFA) evidence for a separate cognitively oriented social-emotional construct, distinct from the other cognitive domains in the CHC taxonomy of human intelligence.  Although MaCann et al. refer to the construct as emotional intelligence, a reading of the dimensions suggest it could easily be called social intelligence.  

Finally, as Bruininks and I were pulled away from our AB/PC program of research for different reasons, I continue to be perplexed why other researchers have not tried to extend and refine the research on the model of personal competence, particularly given its prominence (and disagreements) in definitions of ID.

Adaptive Behavior and Personal Competence Research (select articles)

• Thompson, J. R., McGrew, K. S., & Bruininks, R. H. (2002).  Pieces of the puzzle. Measuring the personal competence and support needs of persons with intellectual disabilities. Peabody Journal of Education, 77(2), 23-29.
• Thompson, J., McGrew, K. & Bruininks, R. (1999).  Adaptive and maladaptive behavior.  Functional and structural characteristics.  In R. L. Schalock & D Braddock (Eds), Adaptive behavior and its measurement: Implications for the field of mental retardation(pp. 15-42).  Washington, DC.  American Association on Mental Retardation.
• Widaman, K. & McGrew K., (1996).  The structure of adaptive behavior.  In J.W. Jacobson & J.A. Mulick (Eds.), Manual of diagnosis and professional practice in mental retardation (pp. 97-110).  Washington, D.C.  American Psychological Association.
• McGrew, K., Bruininks, R., & Johnson, D. (1997).  Confirmatory factor analysis investigation of Greenspan's model of personal competence.  American Journal on Mental Retardation, 100(5) 533-545.
• Greenspan, S., & McGrew, K. (1996).  Response to Mathias and Nettlebeck on the structure of competence.  Need for theory-based methods to test theory-based questions.  Research in Developmental Disabilities, 17, 145-152.
• Bruininks, R., Chen, T., Lakin, C., & McGrew, K. (1992).  Components of personal competence and community integration for persons with mental retardation in small residential programs. Research in Developmental Disabilities, 13, 463-479.
• Ittenbach, R., Spiegel, A., McGrew, K., & Bruininks, R.  (1992). A confirmatory factor analysis of early childhood ability measures within a model of personal competence.  Journal of School Psychology, 30, 307-323.
• McGrew, K., Bruininks, R., & Thurlow, M. (1992). Relationship between measures of adaptive functioning and community adjustment for adults with mental retardation. Exceptional Children, 58, 517-529.
• McGrew, K., Ittenbach, R., Bruininks, R., & Hill, B. (1991). Factor structure of maladaptive behavior across the lifespan for persons with mental retardation. Research in Developmental Disabilities, 12, 181-199.
• McGrew, K., & Bruininks, R. (1990).  Defining adaptive and maladaptive behavior within a model of personal competence. School Psychology Review, 19, 53-73.
• McGrew, K., & Bruininks, R. (1989).  The factor structure of adaptive behavior. School Psychology Review, l8, 64-8l.
• Bruininks, R., McGrew, K., & Maruyama, G. (1988).  Structure of adaptive behavior in samples with and without mental retardation.  American Journal on Mental Retardation, 95(3), 265-272.

More on Greenspan's model of personal competence: Relationship between IQ and social, practical, and conceptual abilities

I am pleased to see that, after a relatively long draught in published research, someone is again investigating the relations between general intelligence, and the primary domains of adaptive behavior, in models (that when examined closely) that are investigating aspects of Greenspan's' model of personal competence. The title, abstract, and key figure from this new research follow. The article can be read here. Kudos to these researchers

Click on images to enlarge





My primary criticism of this study is that it completely ignores the primary foundation research in this area that occurred between 1990 and 2000, some of which are the primary research studies cited in the AAIDD manuals to support the domains of practical, conceptual and social competence (Greenspan's model). I have provided a list of that research, and results from the most prominent article from that group of researchers, below.












Yes, my name is all over these MIA studies (in the current featured article) so some could see my comments as academic sour grapes for being overlooked. But I see their omission as a lack of scholarly rigor by the researchers and the journal who published the current article. All of the MIA studies can be found at the MindHub--scroll down until you see the list of studies shown above. Then click away and download and read. It would have been nice if the new study results would have been integrated with the extant personal competence research literature.

In the final analysis I am pleased that someone is conducting much needed research on these constructs given the pivotal role they play in the definition and assessment of MR/ID.


- Posted using BlogPress from my iPad