Meta-Analysis of Imitation Abilities in Children with Autism Spectrum Disorders

Background: Imitation is essential for social and cognitive learning and typically arises early in development. Mixed results on the imitative abilities of children with autism spectrum disorders (ASD) and other developmental disorders demonstrate that it is not yet clear whether impairments in imitation are significant and specific to the condition of having ASD. Although several systematic reviews of imitation in autism (see Rogers, 1999, 2006; Smith & Bryson, 1994; Vanvuchelen, Roeyers & De Weerdt, 2011), and most recently a quantitative review of action imitation in ASD (Williams, Whiten & Singh, 2004) exist, this meta-analysis will serve to update prior reviews and is the first known meta-analysis to quantitatively synthesize studies on facial/emotional, body, and object-oriented imitation. 

Objectives: In this metanalytic study we seek to determine whether children with ASD show significant imitation deficits in comparison to typically developing children (TD), and children with non-ASD developmental disorders (DD). The current study also reports on the magnitude of any differences in imitative abilities found, and whether these are specific to children with ASD. 

Methods: An extensive literature search was conducted to identify studies relevant to the current analysis. Those studies that directly assessed imitative abilities, tested children with non-syndromic ASD, and contained at least one control group were included in the final analysis. Participants ranged in age from 20.3 months to 18.5 years. Using standard meta-analytic techniques in a random-effects model, performance on imitation tasks by children with ASD was compared to that of TD and DD children. Subgroup analyses were also conducted to assess the impact of study setting and imitation task type on the relative performance of these three groups.

Results: Preliminary results from a 10 percent random probability sample of the included studies (n_studies=9 ; n_subjects=153 ASD, 154 TD, 157 DD) suggest that children with ASD show deficits in imitation, performing on average 1.28 SDs below children without ASD on general (facial, body and object-oriented imitation) tests. Imitation deficits did not vary by age, and were specific to ASD, rather than a general feature of developmental delay. A mixed-effects moderator analysis of the effect of study setting on imitation in ASD suggests however, that imitation deficits are only observed in studies carried out in unfamiliar settings; studies that were conducted in familiar environments found that children with ASD imitated comparably to children without ASD. 

Conclusions: The generalizability of these results is limited, as the studies included in this analysis are a small subsample of the larger population of studies on imitation in ASD. Nonetheless, these findings suggest that impairments in imitation abilities found in laboratory studies are significant and specific to the condition of having ASD. The observed impact of study setting on the imitative deficits of children with ASD have important implications for the validity of studies of children with ASD, and they may call for a move to more familiar—rather than laboratory-based—study environments.