Immediate Impact and Individual Differences: Using a “Micro-Dismantling” Approach to Elucidate Dissociable Effects of Knowledge- and Performance-Training Components

Background: While social skills interventions (SSIs) for ASD have been deemed “evidence-supported” (Reichow & Volkmar, 2010), little work has explored common  components of SSIs that may be responsible for outcomes, such as social knowledge (SK) training (didactically teaching appropriate behaviors) and social performance (SP) training (providing opportunities for engagement in prosocial behaviors, regardless of knowledge; Gresham, 1997). One approach to doing so is “dismantling,” where individuals are randomized to specific elements of treatment packages to identify “active ingredients” (Kiesler, 2004). A small study using this approach identified differences in peer engagement after a single session of SK- and SP-training, suggesting that effects on social behavior occur quickly (Lerner & Mikami, 2012). Thus, further exploration of immediate, component-specific mechanisms is plausible via brief “micro-intervention” sessions (Holtforth et al., 2004).  Likewise, by considering individual differences in baseline abilities, such a design can be used to identify “aptitude x treatment” (AxT) effects (Smith & Sechrest, 1991), which can answer crucial questions of how and for whoma treatment component may work (Kazdin, 2007).

Objectives: 1) To demonstrate the use and analysis of a “micro-dismantling” study to identify differences in effects of SK- and SP-training on peer engagement. 2) To elucidate AxT effects in SK- and SP-training.

Methods: Thirty-eight youth with confirmed ASD (30 male; M_age =12.92, SD_age= 2.09) completed a lab-based session in which they were randomly assigned, in dyads of previously-unacquainted participants, to complete 20 minutes of SK- or SP-training. Participants engaged in 10 minutes of free interaction immediately pre- and post-training. Peer engagement during training and free interaction was taped and coded by blinded, reliable coders for total and positive interaction (Bauminger, 2002). Participants completed baseline diagnostic (Lord et al., 2002), IQ (Wechsler, 2004), explicit SK (Michelson & Wood, 1982), and SP-linked (efficiency of processing social information, indexed by N170 ERP latency [Lerner et al., 2013]) measures.

Results: Generalized Estimating Equations with ANCOVA-of-change models (controlling for diagnostic and IQ scores) were used to account for nesting within dyads and estimate change in peer interaction. Analysis methods will be presented. Participants showed relative increases in total (B = .67; p < .001) and positive (B = .58; p < .001) peer interaction during SP- versus SK-training. For AxT, faster N170 predicted increased interaction regardless of condition (B = -.06, p = .021), but greater increases in positive interaction in SP-training (B = -.24, p = .043). Higher levels of social knowledge (B = -.04, p= .001) predicted greater increases in positive interaction during SK- versus SP-training.

Conclusions: The “dismantling” approach revealed relatively greater increases in positive peer interaction during SP-training, suggesting that theorized “opportunities for engagement” are uniquely presented in this component. The AxT analyses suggested that youth with faster N170s may capitalize more on positive interactions, but that they may be uniquely able to capitalize on SP-training to achieve richer peer interaction. Conversely, those with greater SK appear to benefit from didactic interaction training, suggesting it may act as a “learning style” in some youth with ASD.