19347
A Meta-Analysis of the Wisconsin Card Sort Task in Autism

Thursday, May 14, 2015: 11:30 AM-1:30 PM
Imperial Ballroom (Grand America Hotel)
O. Landry1 and S. Al-Taie2, (1)La Trobe University, Bendigo, Australia, (2)McMaster University, Hamilton, ON, Canada
Background:   Despite 30 years of research, there remain inconsistencies in reports of the nature of the executive dysfunction of autism. One of the most widely used tests of executive function is the Wisconsin Card Sort Task. There are numerous conflicting reports of the nature and magnitude of purported impaired facets of performance, as well as presumed facets of intact performance. Executive function is a broad concept, and a more refined understanding of executive dysfunction in autism will inform both intervention programs and neurodevelopmental theory.

Objectives:   We conducted a meta-analysis of Wisconsin Card Sort Task performance in participants with autism to consolidate and clarify the inconsistencies in the literature, and examine potential explanatory factors such as participant sample characteristics and experimenter versus computerized administration.

Methods:   We were able to retrieve 31 studies reporting on a total of 793 participants with autism, published over a 30-year span. We also obtained five raw data sets, with a combined 191 participants with autism. We calculated Cohen’s d effect sizes for four measures of performance: sets completed, perseveration, failure-to-maintain-set, and non-perseverative errors. We examined effect sizes as a function of participant sample characteristics, year of publication and diagnostic criteria, and manual versus computerized administration. We also analysed relationships among participant age, verbal and performance IQ, and ADI scores with performance within the combined raw data sets.

Results:   We found the average weighted effect size ranged from 0.30 (failure to maintain set) to 0.82 (perseveration) for each measure, all of which were statistically greater than 0. As study authors more often than not concluded that failure to maintain set and non-perseverative errors, the failure to detect these small to medium effect sizes is likely due to small sample sizes. We also found a decrease in the magnitude of perseveration impairment in samples diagnosed under DSM-IV criteria relative to older criteria. We did not find any evidence for systematic variation as a function of mode of administration (experimenter versus computer). In the raw data, we found that age and IQ appear to influence overall performance via unique influences on perseverative and non-perseverative error rates. Age, ADI, and perseveration were associated with one-another. Performance and Verbal IQ were associated with non-perseverative errors and number of sets completed. We did not find any associations with failure to maintain set.

Conclusions:  

We conclude that failure-to-maintain-set and non-perseverative errors represent facets of impaired task performance in autism often overlooked due to small sample sizes and that the impairment is less pronounced than perseveration. We further conclude that perseveration decreases with age in autism, a relationship somehow tied also to symptomology, while non-perseverative errors are driven by cognitive abilities reflected in both verbal and performance IQ, independent of age and symptomology.