Single Case Design and Growth Curve Analysis of An Intervention to Promote Joint Attention for Toddlers with ASD

Background:  For infants and toddlers with ASD, joint attention and social communication with caregivers is often delayed or nonexistent.  The Joint Attention Mediated Learning Intervention (JAML, Schertz & Odom, 2007) has been developed to support family members’ use of strategies that promote joint attention.  In a sequential fashion, caregivers and intervention coordinators plan activities that promote two foundational behaviors (Focusing on Faces and Turn Taking) that lead to subsequent intervention emphasis on Responding to and Initiating Joint attention (RJA and IJA respectively).  The purpose of this study was to examine the systematic replication of slight variations of JAML through multiple single case design (SCD) demonstrations and to explore an analytic strategy that would use growth curve analysis to examine aggregated SCD data across participants.

Objectives:  To examine (a) the application of growth curve modeling to the aggregation of SCD replications of JAML and (b) the efficacy of the JAML for promoting initiating joint attention for toddlers with ASD.

Methods:  Seventeen toddlers with ASD between the ages of 15-30 months and their caregivers participated in this study.  Intervention coordinators delivered the JAML intervention in the families’ homes.  For the first eight families, the intervention consisted for four phases that corresponded to the four target behaviors noted previously (FF, TT, RJA, IJA).  For the subsequent 9 children and families, the RJA and IJA phases were combined, so that the intervention consisted of three components (FF, TT, and RJA/IJA).  Children and families were video recorded during weekly 15-minute play sessions in their homes, and the four target behaviors noted previously were coded.  For the first set of children/families, a multiple baseline across four behaviors was conducted, and for the second set of 9 children/families a multiple baseline design across three behaviors (FF, TT, and RJA/IJA) was conducted. 

Results:  For the eight children in the first set, data were plotted on a four-tiered multiple baseline design graph.  For each tier of the graph, four growth curve models were run (linear, binary, cubic, and quadratic), with the cubic model accounting for the most variance.  While variability existed in data, curves revealed acceleration in target behaviors that approximated to the range of onset dates of the interventions.  Importantly, IJA displayed a marked acceleration at the end of the intervention.  Similar patterns of growth curves occurred for the second set of 9 children with ASD.

Conclusions:  A problem with conducting a large number of systematic replications of SCD demonstrations is concisely summarizing the results. This study reviewed that growth curve modeling may be a feasible procedures for displaying summary data of SCD replications.  The replications, as a group, revealed that the intervention produced increases in IJA for very young children with ASD.