Deconstructing the Play Impairment in Autism: An Analysis of Interdependent Variables

Background: Play impairments in autism have been observed for over three decades and are central to its definition and manifestation; previous research has found symbolic impairments (Baron-Cohen, 1987), deficits under certain conditions (Jarrold et al, 1996) and play in its entirety affected (Libby et al, 1998). It is also still far from clarified whether play in children with autism is delayed or displayed differently to typically development inconsistent definitions, mixed control samples and limited longitudinal research make conclusions difficult. Defining this impairment is also reliant upon the analysis of interdependent variables, often ignored by pre-existing studies, therefore possibly leading to misleading findings. Due to the over-reliance of traditional statistics in the field, which assume play behaviours are independent, and a reliance on cross sectional methods, the play impairment in autism remains unclear.

Objectives: This study sought to clarify the play impairment in autism through a short-term longitudinal study. The study explored differences relative to neurotypical controls cross-sectionally whilst mapping development over time. A novel statistical technique was utilised, outlined previously by Pennington et al (2009), to overcome the interdependence of the play variables.

Methods: Play behaviours (coded using Noldus Observer) in preschool children with autism (n = 49) were compared over three timepoints within 13 months to typically developing children matched on gender and non-verbal development (n = 44). Definitions were based on previous research, however the definition of symbolic play was tightened and functional play was divided into simple and advanced behaviours. The statistical method entailed testing differences in proportions of play behaviours between the two samples with the non-significant behaviours then merged to form a reference category. The remaining significant behaviours were then transformed by log ratio to the reference category to ensure each ‘level’ of play was on the same scale, independent and  normally distributed and subsequent analysis performed using the transformed variables.

Results: At all timepoints, children with autism demonstrated more simple exploratory play and less advanced functional play. Only at time three did a difference emerge in symbolic play. Autism symptoms did predict differences in simple play behaviours but failed to predict more advanced behaviours despite the group differences found. Play development between time one and two was slower in children with autism but reflected that of typical development. Between time two and three, development slowed and a reduction in symbolic play was evident.

Conclusions: By utilizing a new statistical technique, this study sheds new light on the presumed symbolic-specific impairment in autism. Symbolic play impairments were not evident until the final timepoint and were still accompanied by heightened simple play and reduced functional play, mirroring the findings of Jarrold et al (1996) and Williams et al (2001). The apparent loss of symbolic skill has been reported anecdotally by caregivers and shown here empirically.  Autism symptomatology also failed to shed light on the differences found, despite the presence of group differences. The findings do not support the notion of a symbolic-specific impairment and lend themselves to an overall delay in play development compared to neurotypical progression.