Clustering of External Representations In Young People with Autism Spectrum Disorder

Background: There is a growing need to develop software, especially user interfaces, which can address the needs of individuals affected by Autism Spectrum Disorder (ASD). Effective user interface design for those with ASD necessitates an understanding of preferences for how visual material should be organised and represented (upon the screen). Previous research has suggested that external representations (ERs), such as textual and visual information, are utilised in a different way by those with ASD when compared to controls.  

Objectives: In order to inform the design criteria for effective and efficient user interfaces for those with ASD, particularly their textual and visual aspects, we investigated the clustering of ERs by young people with and without ASD.

Methods: The study involved 29 high-functioning young people with ASD (25 male; 4 female) aged between 11-15, and 28 young people without ASD (18 male,10 female) aged between 11-14. The two groups (ASD and non-ASD) were matched on age, mathematical ability and verbal ability. Each group was enrolled in specialist and non-specialist schools (respectively), in years 7, 8 and 9 of the UK curriculum.

Results: From participants’ card sorts a similarity matrix for each group was created, which was then input to the SPSS CLUSTER procedure to produce a hierarchical cluster analysis. The major ASD clusters were: 1. Maps; 2. Text, 3. Drawings; and 4. Graphs, charts, network diagrams and tree diagrams. In contrast the non-ASD clusters were: 1. Maps, 2. Drawings; and 3. Text, graphs, charts, network diagrams and tree diagrams. Also, the distance between the map and text clusters in the ASD group was less then half that for the non-ASD group.

Conclusions: Both groups identified ‘Maps’ as the most salient cluster, suggesting that incorporating map-like structure within user interface design would be the most intuitively appealing for all potential users. The differences in how text is clustered may also be of importance. The data suggest that whilst the non-ASD participants cluster text with other ERs such as graphs, this is not the case for ASD participants who cluster text separately from other ERs. Combined with the difference in distance between clusters, this would suggest that the way text is integrated within user-interface design may be very different for those with and without ASD.