Assessment of Intelligence in Young Autistic Children: A Comparison of Tests Available Under and over Age 3

Background: Very early autism diagnosis is often recommended, and DSM-5 autism spectrum diagnosis must specify the presence or absence of accompanying intellectual disability (APA, 2013). However, few tests are available to assess intellectual functioning in children under age 3 years. Further, developmental tests often used in this age range as measures of IQ may not accurately represent the potential of young autistic children, whose developmental paths are highly atypical. School-aged autistic children perform differently depending on the test used to assess their intelligence (Dawson et al., 2007) and present a discrepancy between intelligence and adaptive level (Charman et al., 2011). Thus it is important to document how the intellectual abilities of younger autistic children would be judged using different available tests.  

Objectives: To compare the performance of young autistic and typical children on Mullen Scales of Early Learning and Vineland Adaptive Behavior Scales to their performance on intelligence tests which provide norms for the 3-to-5 years age range  

Methods: The complete sample will include 25 autistic and 25 typical children aged 18 months to 5 years. To date, 12 autistic and 16 typical children have been tested on Mullen and Vineland. Those aged 3 or more were also tested on WPPSI-IV (autistic N=6, typical N=10) and Leiter-3 (autistic N=8, typical N=10). Those aged 4 or more (autistic N=4, typical N=8) were administered Raven’s Colored Progressive Matrices board form (RCPM). 

Results: Performance on Mullen was significantly lower than Vineland scores in autistic children (p<.05), whereas no significant difference was found for the typical group (p=.18). The autistic group’s lowest mean scores were on Mullen (IQ=61, SD=19.2; below the 1^st percentile), followed by Vineland (IQ=77, SD=12.3; 6^th percentile), WPPSI-IV (IQ=79, SD=19.0; 8^th percentile), Leiter-3 (IQ=95, SD=9.6; 37^th percentile) and RCPM (raw score=15.25/36, SD=6.29; 66^th percentile). For the typical children, lowest mean scores were on Vineland (IQ=102, SD=15.2; 55^th percentile), followed by Mullen (IQ=107, SD=18.3; 68^th percentile), Leiter-3 (IQ=113, SD=9.6; 81^st percentile), WPPSI-IV (IQ =117, SD=14.4; 87^th percentile), and RCPM (raw score=23.63/36, SD=3.34; 95^th percentile). 

Conclusions: Preliminary results indicate that the intellectual functioning of young autistic children, and therefore the presence or absence of intellectual disability in this population, may be judged very differently depending on the test administered. Autistic children would be judged as most intellectually impaired on Mullen, a commonly used test in autism research and practice. Data from a small number of children suggest dramatic discrepancies between Mullen and both Leiter-3 and RCPM scores in the autistic group, which in turn would suggest caution in judging autistic potential at very young ages.