Cheek, C. L., Garcia, J. L., Mehta, P. D., Francis, D. J., & Grigorenko, E. L. (2023). The Exceptionality of twice-exceptionality: Examining combined prevalence of giftedness and disability using multivariate statistical simulation. Exceptional Children. https://doi.org/10.1177/00144029221150929
Twice-exceptionality (2e), broadly defined as the co-occurrence of a gift or talent and a disability, is a popular idea in special educational circles. The term has been around for about 30 years, and there are numerous publications advising best strategies to meet the unique learning needs of twice-exceptional students. However, most of the literature surrounding this phenomenon is theoretical, and the little empirical evidence available is problematic: methodology issues like low sample sizes and sampling bias, along with discrepancies in criteria used to identify students as twice-exceptional, gifted, or disabled, paint a very inconsistent picture of what exactly twice-exceptionality is, and what it looks like in the general population.
The article was designed to serve two main purposes. First, by taking a comprehensive look at the existing empirical literature, we illustrate the wide variety of metrics used to measure these phenomena. Because giftedness and disability broadly encompass many unique diagnostic labels, we conducted searches on combinations of three broad categories of giftedness (i.e., visual-spatial, music, and cognitive/intellectual) and three broad categories of disability (i.e., Attention-Deficit Hyperactivity Disorder, Specific Learning Disability, and Autism Spectrum Disorder/Asperger's Syndrome) for a total of nine different forms of twice-exceptionality. The second purpose was to examine systematically, using multivariate statistical simulation, the prevalence rates of twice-exceptionality using the identification criteria utilized in the literature.
We observed that definitions of giftedness and disability varied widely between studies. For instance, when examining intellectual giftedness, cut-offs ranged from an IQ of 110 or greater to an IQ of 130 or greater. There were also differences in the types of intelligence tests used and whether giftedness was defined using full-scale IQ or other forms (e.g., nonverbal) of intelligence. Another common problem was that many publications employed single case designs or projects with small samples, often less than 10 participants, which makes it difficult to generalize the results to the greater twice-exceptional subpopulation of children.
Since there is such variability in twice-exceptional identification criteria, designing a simulation study that adequately characterizes the prevalence rates of dual exceptionality is challenging. In many cases, assessment of giftedness or disability relies on standardized assessments of ability, with thresholds applied in order to separate students who are average from those who are exceptional. In the literature, the exact number of tests along with the thresholds used differ study to study. Furthermore, when simulating prevalence rates, certain assumptions must be made about the correlational structures between gifted and disabled constructs; these relationships depend on the ability’s context, and often is unknown a priori. The reliability of the assessment to accurately quantify the trait in question must also be accounted. For example, standardized assessments may adequately capture a student’s ability to decode words, but poorly capture their musical ability.
Following this, we created a list of parameters to adjust for our statistical simulation.
The assumed profile of a twice-exceptional student is one who exhibits high achievement or ability on measures of giftedness, and low achievement or ability on measures of disability. We created multivariate normal distributions under each of the above parameter combinations and measured the population-level twice-exceptional prevalence rates they generated. Using this approach, we capture the prevalence rates at two ends of a spectrum: where diagnostic criteria are very lax and impractical and where they are highly specified.
Our results find that, under the most relaxed and imprecise circumstances—only one of three tests are required to meet a threshold of one standard deviation above or below mean, the correlation between constructs is set to be low, and factor loadings are poor—the twice-exceptional prevalence rates were 14.8%. However, prevalence rates vary greatly by parameter choice. Keeping the aforementioned parameters the same but changing the identification threshold to two standard deviations instead of one, this prevalence rate drops from 14.8% to 0.39%. In general, the prevalence rates under all criteria were exceedingly low. Most observed prevalence rates quickly drop below 1% and approach 0 as the simulation parameters are made slightly stricter. That prevalence rates should drop as precision increases is not surprising—it is their magnitude that is striking. The observed prevalence rates of twice-exceptionality are much smaller than the relevant literature would indicate.
These results call into the question the utility of such a label in diagnostic settings. Twice-exceptionality represents a unity of two broad and ill-defined constructs, and the use of such a label to provide broad generalizations about the learning needs of supposedly twice-exceptional students is impractical. Although the prevalence rate of true twice-exceptionality is very low, children with both manifested high abilities and disabilities—regardless of whether they meet the technical definition of being twice-exceptional—are just as deserving of having their unique learning needs met.
Educators working with twice-exceptional students should ask themselves the following questions.
Academic accommodations for twice-exceptional students should be individually tailored based on their unique learning needs.