Use of video modeling to increase generalization of social play by children with autism

Journal of Speech-Language Pathology and Applied Behavior Analysis, The, Summer, 2007 by Christos K. Nikopoulos

Empirical Validation of Video Modeling to Support Generalization of Social Play

Study 1 (Nikopoulos and Keenan, 2004a)

As noted above, only two studies have examined the ability of video modeling to increase generalization of social play in children with autism (Nikopoulos & Keenan, 2004a, 2004b). The overall objective of the first study (Nikopoulos & Keenan, 2004a) was to examine a) whether a video modeling intervention could be effective in promoting social play skills (i.e., social initiation & reciprocal play) in a simplified environment (i.e., in the presence of only one toy at a time) and b) whether potential success with one toy during video modeling training could increase the probability of success with new toys in the absence of video (i.e., generalization of social play).

Three children with autism, ages 7.5 to 10.5 years, served as participants in the study. All had scored within the range of mild-moderate autism on the Childhood Autism Rating Scale (Schopler, Reichler, & Renner, 2002). All participants had some speech and displayed limited interactions with other children or adults. Other aspects of their behavior included limited interest in toys, preferring solitary activities, marked impairments in the use of non-verbal behaviors and persistent desire to follow set patterns of behavior during their interactions with others. The study took place in three different classrooms (Rooms 1, 2, & 3) of the participants' school and a multiple baseline design across subjects was employed.

Initially, the participants were taken (individually) to Room 1 to view a 35-s video of a typically developing peer engaged in a simple social activity using a wooden train while interacting with the researcher. The video showed the researcher entering a room with the peer model. The researcher then sat on a chair close to the toy while the model spent a few seconds wandering around the room. Then, the model performed a sequence of behaviors to demonstrate social initiation. Specifically, the model approached the researcher, took him by the hand, said "Let's play", and led him to the toy. Together the model and the researcher then continued to interact for 15 seconds to demonstrate reciprocal play.

After watching this video sequence once, each participant was taken (individually) into Room 2, which was different from the room displayed in the video. In Room 2, each participant was presented with a number of conditions. In each condition, the researcher engaged in the same behavior shown on the video. No instructions were provided to the participant regarding the video that he or she had just watched. The transition from one condition to another was based on each participant's performance data. (Please see Figure 1, below, which provides a graphical representation of all conditions). Thus, in the first condition (T1), when a participant succeeded in emitting the modeled sequence of behaviors within the first 25 seconds over three consecutive sessions, s/he was transitioned to the next condition (T2). In that condition, a different toy (a ball) was provided, and it was assessed whether the behavioral changes (i.e., social play) that had occurred in the preceding condition would be generalized without the benefit of video modeling in the presence of this different toy (a ball). The three following conditions (T3, T4, T5) were identical to T2 except that the toy was different in each condition ("Hungry Frog" game, set of tambourines, and trampoline, respectively). In other words, the participants were assessed to determine whether they could engage in social play with the researcher without the benefit of video modeling and in the presence of another toy. In essence, Conditions T2, T3, T4, and T5 were exactly the same as Condition T1 except that no video model was provided and a different toy was used in each condition. If a participant emitted the modeled response in three consecutive sessions, then he/she experienced the next respective condition; otherwise, he or she experienced the preceding condition, just once. This process (of a participant transferring from one condition to another) continued until he or she was assessed with the fifth generalization toy in a row in Condition T5. Each session was scheduled to last a maximum of 5 minutes.