Using Thin Slice Ratings to Measure Social Communication in Children with Autism Spectrum Disorder

Background: Behavioral observations are an important source of information regarding social communication behaviors in young children with ASD. The Brief Observation of Communication Change measure (BOSCC; Grzadzinski et al., 2016) was developed as a standardized observational coding system of social communication skills that can be used across intervention trials. Thin slice ratings (TSR), in which very short observations are rated by multiple untrained raters, have been shown to predict a variety of behaviors in children and adults (Baker et al., 2010; Murphy, 2005; Tackett et al., 2016). Recently, they have been used to rate social communication and autism symptomatology in young children with ASD (Hampton et al., 2018; Walton & Ingersoll, 2016), suggesting their potential as an alternative to more time intensive coding systems.

Objectives: The aim of the current study is to examine whether TSR of social communication skills during a parent-child interaction are psychometrically sound and capture similar information to the BOSCC for young children with ASD. We predicted that TSR would be highly correlated with BOSCC scores for the same observation and would show similar relationships to other measures of social communication and demographic variables.

Methods: TSR for 71 children with ASD were completed by 173 undergraduate college students from a psychology research pool. Fourteen groups of 11-15 raters viewed 2-minute clips of a 10-minute parent-child interaction for 4 to 7 different children. Raters used a 5-point likert-type scale to rate each clip on 7 items that were derived from the social communication items of the BOSCC (SC). Each group’s item-level ratings for an individual child were averaged to produce TSR for each item. All items were then averaged to produce an overall TSR for each child. Cronbach’s alpha based on the average item-level ratings was used to examine internal consistency of the coding scheme. Intraclass correlations (ICCs) based on overall TSR were examined to evaluate inter-rater reliability. Separate one-way random average measures ICCs were run for each group, given groups had different numbers of randomly-selected raters. Bivariate correlations were run to examine concurrent and discriminant validity between overall TSR and the BOSCC-SC, and other measures of social communication and demographic variables.

Results: Cronbach’s alpha for the TSR was 0.92. ICCs ranged from 0.75 to 0.98. Overall TSR were significantly correlated with BOSCC-SC (r = -.69, p < 0.01). In addition, overall TSR and BOSCC-SC scores showed similar relationships to other measures of social communication (i.e. MCDI words produced, VABS-2 communication and socialization age equivalents, and MSEL verbal age equivalents; p < 0.01), but not child sex or parent education.

Conclusions: Thin slice ratings of young children with ASD during a parent-child interaction using modified BOSCC-SC items are internally consistent and show a similar pattern of relationships with other measures. Findings suggest TSR may be used to measure child social communication as an alternative to more resource-intensive coding measures. Future studies should investigate whether TSR capture other aspects of autism symptomatology (e.g. restricted and repetitive behaviors) and are sensitive to change.