Latent Profiles of Social Interaction and Classroom Educational Experiences in Students with Autism Spectrum Disorder

Background: Teachers report feeling underprepared to effectively program for students with Autism Spectrum Disorder (ASD) and less able to manage social communication and behavioral concerns characteristic of ASD (Locke et al., 2015). Difficulty providing effective programming for students with ASD has been identified as a significant source of stress, leading to low self-efficacy and a higher frequency of teacher burnout (Boujut et al., 2017)—which could place students with ASD at risk for educational failure. This could, in part, relate to the notable heterogeneity in social communication skills that students with ASD present in educational settings (Fleury et al., 2014).

Objectives: The purpose of this study was to identify subgroups of students with ASD based upon a set of social interaction features and examine if and how social interaction patterns influence the nature and quality of students’ classroom experiences.

Methods: Participants include 119 preschool–3^rd grade students with ASD and their teachers (n = 65) across 16 districts who were video-recorded in their classrooms at the beginning of the school year. Research assistants coded a 10-minute sample of the video observation using an adapted version of the Classroom Measure of Active Engagement (Sparapani et al., 2016) and Noldus Observer® Video-Pro Software (XT 14; 2017). Inter-rater agreement between the coders indicated overall good reliability, with coefficients ranging between 82–98%. We utilized latent profile analysis (LPA) to identify subgroups of interaction among students with ASD and compared models with two, three, and four profiles based on students’ communicative intentions (protesting, regulating others’ behaviors, securing attention, commenting and asking questions) and the frequency they initiated communication with their peers and looked at others’ faces.

Results: Preliminary findings (n = 63; 74% male) indicated that four patterns of social interaction best described the sample (Entropy =0.98), showing excellent overall model fit, the best fit in comparison to competing models, and strong membership probability (0.96 to 0.99). Students in subgroup 1 (5% of the sample) showed the highest frequency of comments and questions (2 SD above the sample mean) and initiated communication with peers (1 SD above). Subgroup 2 (10%) was characterized by a high frequency of protesting (3.5 SD above). Subgroup 3, the largest group (80%), was characterized by limited communication and eye gaze overall (0.5 SD below the mean). Students in subgroup 4 frequently communicated to gain a social partner’s attention (4 SD above), exhibited a higher rate of communication and eye gaze overall, and frequently communicated with peers.

Conclusions: These preliminary findings indicate that patterns of social interaction form constellations that vary among students with ASD. We will run further analyses to explore the dynamic relations among profile membership, teacher discourse moves, and the quality of instructional practices. Examining the influence that profile membership may have on the nature and quality of students’ educational experiences within classroom activities will provide insight into malleable features that support student engagement and help to guide educational programming for learners with ASD.