Teaching Social Communication Skills Using a Cool Versus (PDF)

EDUCATION AND TREATMENT OF CHILDREN Vol. 39, No. 1, 2016 Teaching Social Communication Skills Using a Cool Versus Not Cool Procedure Plus Role-Playing and a Social Skills Taxonomy Justin B. Leaf Mitchell Taubman Christine Milne Stephanie Dale Jeremy Leaf Donna Townley-Cochran Kathleen Tsuji Alyne Kassardjian Aditt Alcalay Ronald Leaf John McEachin Autism Partnership Foundation Abstract We utilized a cool versus not cool procedure plus role-playing to teach social communication skills to three individuals diagnosed with autism spectrum disorder. The cool versus not cool procedure plus role-playing consisted of the researcher randomly demonstrating the behavior correctly (cool) two times and the behavior incorrectly (not cool) two times, followed by the participants role-playing the behavior. We utilized a multiple baseline design across par­ ticipants to evaluate the effectiveness of the cool versus not cool procedure plus role-playing. The results of the study indicated that all three participants were able to increase their social communication skills following implemen­ tation of the cool versus not cool procedure; however, maintenance varied across the three participants. Keywords: autism; cool vs. not cool, pragmatics, social communication, social discrimination, social skills Address correspondence to: Justin B. Leaf, 200 Marina Drive, Seal Beach, CA 90740. Email: [email protected] Pages 44-63  COOL VERSUS NOT COOL 45 A hallmark characteristic of individuals diagnosed with an Autism Spectrum Disorder (ASD) is an inability to use communication to socially connect with other people within the individual's environ­ ment (Rogers, 2000; Scattone, 2007), often referred to as social com­ munication. Social communication is the communicative behavior (verbal and nonverbal) exchanged between two or more people, also referred to as pragmatic language (Taubman, Leaf, & McEachin, 2011). Although it is important to teach social communicative behaviors to individuals with autism, there are two challenges that professionals and parents may face. First, it may be difficult for parents and clini­ cians to determine what skills to teach individuals diagnosed with ASD. Second, it may be difficult for professionals and parents to de­ termine what procedures to implement when teaching social behav­ iors to individuals diagnosed with ASD. In 2011, Taubman and colleagues attempted to assist parents and professionals by creating a social skills taxonomy, which identi­ fied what social skills to teach individuals with autism. The social skills taxonomy consisted of five domains: (a) social awareness, (b) social interaction, (c) social learning, (d) social relatedness, and (e) so­ cial communication. Social behaviors under the social communica­ tion domain can be defined as both functional verbal and nonverbal expressions that are directed towards another person with the inten­ tion of engaging in a reciprocal interaction with that other person. Within social communication, a parent or professional may look at "what is said and how it is said; direct and implied conversation; figu­ rative and literal speech; serious, sarcastic, and humorous statements; tone, inflection, emphasis, and style" (Taubman, 2011, p. 97). Therefore, skills in the social communication domain are meant to improve the communicative and social exchanges between students diagnosed with ASD and their peers. Taubman et al. divided social communica­ tion behaviors into basic, intermediate, and advanced specific social communicative behaviors, so that professionals can teach social com­ munication to a wide variety of learners. Although the social skills taxonomy has been utilized as part of clinical intervention for hun­ dreds of individuals diagnosed with ASD in both home and school settings (e.g., R. B. Leaf, Taubman, McEachin, Leaf, & Tsuji, 2011), to date, there have been no empirically based studies that have utilized the social skills taxonomy when selecting social behavior(s) for indi­ viduals diagnosed with ASD. Therefore, it is not known if researchers can utilize the social skills taxonomy to select skills to be taught. The second challenge that professionals and parents face when teaching social behaviors to individuals diagnosed with ASD is select­ ing an appropriate intervention. Today, there are several interventions  46 LEAF et al. that have been implemented to teach social communication to indi­ viduals diagnosed with ASD. These interventions include discrete trial teaching (e.g., Matson, Sevin, Fridley, & Love, 1990), video modeling (e.g., Charlop-Christy, Le, & Freeman, 2000), and script fading (e.g., Krantz & McClannahan, 1993). Despite the implementation of several interventions to individuals diagnosed with autism, it is important for researchers to continue to develop and evaluate new interventions to teach social behavior and social communication skills to identify the most effective and efficient procedures. One intervention that has been implemented clinically to indi­ viduals diagnosed with ASD (R. B. Leaf et al., 2011; J. B. Leaf, Dotson, Oppenheim-Leaf, Sherman, & Sheldon, 2012) and only recently has been evaluated in the research is the cool versus not cool procedure (J. B. Leaf, Tsuji et al., 2012). The cool versus not cool procedure is a dis­ crimination program (J. B. Leaf, Sheldon, & Sherman, 2010) where stu­ dents have to discriminate whether a demonstrated behavior was either cool (socially appropriate) or not cool (socially inappropriate). There are five general components within the cool versus not cool pro­ cedure. First, the teacher demonstrates the behavior in a manner that corresponds with either the cool or not cool way of behaving. Second, the teacher asks the student(s) to discriminate if the demonstration was cool or not cool. Third, the teacher provides the student with reinforce­ ment for correct discriminations or provides corrective feedback for incorrect discriminations of whether the demonstration was cool or not cool. Fourth, the teacher asks the student(s) to state why the demon­ stration was cool or not cool. Finally, the teacher provides the student with reinforcement for correct explanations or corrective feedback for incorrect explanations of why the demonstration was cool or not cool. In addition, an optional sixth component may include the student ap­ propriately role-playing the social behavior (i.e., the "cool" way). The cool versus not cool procedure is a discrimination program that has been described in curriculum materials for individuals diag­ nosed with autism (e.g., R. B. Leaf, McEachin, Taubman, 2012c; Taub- man et al., 2011) and has been implemented clinically to teach social behaviors to hundreds of children in home behavioral programs (R. B. Leaf et al., 2011), social skills groups (J. B. Leaf, Dotson et al., 2012), and school programs (e.g., Au et al., 2015; J. B. Leaf, 2010a, 2010b). Despite its widespread use and discussions in non-empirically based literature, there remains little empirical evidence on the cool versus not cool procedure's effectiveness for individuals diagnosed with au­ tism (J. B. Leaf, Tsuji et al., 2012). J. B. Leaf, Tsuji et al. (2012) were the first to evaluate the cool ver­ sus not cool procedure empirically. The researchers taught a variety  COOL VERSUS NOT COOL 47 of social behaviors to three individuals diagnosed with an autism spectrum disorder (ASD). The researchers first evaluated the cool ver­ sus not cool procedure without participant role-plays; if the participant was unable to reach mastery criterion (i.e., 80% of the steps across three consecutive probes), then a role-play component was added. Re­ sults of this study showed that participants reached mastery criterion on 50% of skills without the addition of role-playing and were able to reach mastery criterion on an additional 38% of the skills when role- playing was added. There could be several behavioral explanations for why the cool versus not cool procedure has been found to be effective both in the empirical research (i.e., J. B. Leaf, Tsuji et al., 2012) and in clinical inter­ vention (e.g., R. B. Leaf et al, 2011; J. B. Leaf, Dotson et al., 2012). First, the cool versus not cool procedure incorporates the teacher modeling the correct behavior, which informs the student of what behaviors he or she should be engaging in and has been found to be an important component in skill acquisition (e.g., Charlop-Christy et al., 2000). Ad­ ditionally, when implementing the cool versus not cool procedure teachers model the behavior incorrectly. Given that social communi­ cation can be complex and highly nuanced, it may be possible that a student does not understand what behaviors they are displaying in­ accurately; providing a not cool demonstration will allow them to observe which behaviors they are demonstrating incorrectly and, hopefully, they will then be able to change these behaviors. Third, the cool versus not cool procedure utilizes student role-playing of the de­ sired targeted behavior. This allows the student to receive hands on practice in displaying the desired social behavior and has been a component of many effective procedures (e.g., J. B. Leaf et al., 2009). Finally, the cool versus not cool procedure utilizes both reinforce­ ment and punishment (e.g., corrective feedback), both of which are essential components of changing behavior. Despite the positive findings of the J. B. Leaf, Tsuji et al. (2012) study, there are several questions that still need to be answered. First, it is not known what effects the cool versus not cool procedure will have on different children and on different skills. Second, in the original J. B. Leaf, Tsuji et al. study, student role-playing was added only if the par­ ticipant was unable to acquire the targeted social behaviors; there­ fore, it is not known what the effects would be if student role-playing is added as a mandatory step from the beginning of intervention. Third, no teaching data was presented in the J. B. Leaf, Tsuji et al. study; therefore, it is not known how participants responded during actual teaching. Finally, it is not known if utilizing the social skills taxon­ omy to select skills based upon participants' strengths and deficits  48 LEAF et al. would be an effective way to determine target behaviors. Therefore, the purpose of this study was to expand the research on the cool ver­ sus not cool procedure by addressing the areas that w arrant research (described above). Method Participant and setting Brady was a four-year-old boy diagnosed w ith Autistic Dis­ order. Brady had a Wechsler Preschool and Primary Scale of Intelli- gence-IV (WPPSI-III) full scale IQ score of 110, a Vineland Adaptive Behavior Scales (VABS) composite score of 84, and a Social Skills Im­ provement System-Parent Version (SSIS-P) standard score of 70 (4th percentile). Sally was a three-year-old girl diagnosed with Autistic Disorder. Sally had a WPPSI-III full scale IQ score of 125, a VABS com­ posite score of 81, and a SSIS-P standard score of 81 (10th percentile). Hank was a six-year-old boy diagnosed with Autistic Disorder. Hank had a WPPSI-III full scale IQ score of 85, a VABS composite score of 72, and a SSIS-P standard score of 74 (4th percentile). All research ses­ sions took place in a clinic room as part of a private agency that pro­ vides behavioral intervention to individuals diagnosed w ith ASD. Targeted Skills Each participant was taught one social communicative skill. Each of these behaviors fell under the umbrella of the social commu­ nication dom ain as part of the social taxonomy created by Taubman and colleagues (2011). Each participant's parents were asked to fill out the Social Skills Improvement System (Gresham & Elliott, 2008) to evaluate if they had a deficit in social communicative behaviors and in overall social behavior. The results of this assessment showed that all participants were under the 15th percentile in overall social behav­ ior (see above) and had below average scores on social communica­ tion; thus, all children showed a deficit in social communication. To select which specific behaviors would be targeted, the re­ searchers and the participants' clinical supervisors utilized the social skills taxonomy as a guide to select the specific social communicative behaviors. The clinical supervisor was a behavioral therapist who had at least six years of experience working in the field of Applied Behavior  COOL VERSUS NOT COOL 49 Analysis and its implementation for individuals diagnosed with ASD, and was in charge of program development and staff training. Using the social taxonomy, each participant's clinical supervisor and the researcher identified which social communicative behaviors the participant was not currently displaying. Next, the clinical super­ visor removed skills that the participant had already received inter­ vention on or that the supervisor had planned to intervene on in the next month. At this point, a final list of possible social communicative behaviors was identified and the researcher asked the participant's parents which skills they would prefer be addressed; the skills identi­ fied by the parents were the skills targeted during this study. Finally, each of the social skills was then task analyzed and broken down into smaller behavioral steps. Brady was taught to provide verbal support to his friend when his friend was completing a fluency task (e.g., putting blocks in a bucket). Providing verbal support was defined as Brady providing two different positive verbal comments by the time the friend com­ pleted the task. Sally was taught how to "chat" with a friend while watching a short YouTube video. Chatting was defined as: (a) Sally making at least one comment about something occurring in the video while the video was playing, (b) Sally making at least one different comment about what happened in the video within 30 seconds of the video ending, and (c) Sally making at least one subjective comment of a qualitative nature (e.g., "I like . ..," "The girl's dress was pretty," or "That person was funny") either during the video or up to 30 seconds after the completion of the video. Hank was taught how to interrupt a conversation appropriately. Interrupting appropriately consisted of five behavioral steps: (a) approaching the person he wanted to talk to, (b) not saying anything to the person until within 3 feet of the person he had a question for, (c) saying "excuse me," (d) asking the person the question only after saying "excuse me" and waiting for a verbal cue (e.g., the person saying "yes") or a nonverbal cue (e.g., the person looking at him) from the person he or she was interrupting, and (e) saying thank you or acknowledging that he was leaving only after the person answered his question. Naturalistic Probes The researchers measured mastery of each social skill through naturalistic probes, which were opportunities for the participant to display the targeted skill. For both Brady and Sally, a peer engaged in  a behavior that set the occasion for the participant to display the tar­ geted social behavior. For Hank, three different adults engaged in a behavior that set the occasion for Hank to display the appropriate so­ cial skill. The researcher did not prompt or provide reinforcement to the participant during naturalistic probes. Naturalistic probes assessed generalization of the target skills to novel settings and persons, as the peers (for Brady and Sally) and adults (for Hank) utilized during nat­ uralistic probes were not present during the teaching procedure. For the skill of verbally supporting a friend, Brady and an un­ trained peer were told that they were going to see how fast the peer could engage in a task (e.g., "We are going to watch how fast it takes for Sam to put all the blocks in the bucket"). Next, the researcher in­ structed the peer to start the task and scored which steps of the tar­ geted skill Brady successfully demonstrated. Multiple exemplars of the task occurred throughout all conditions of the study. For the skill of chatting, Sally and a trained confederate peer (trained to not speak until Sally did) were brought into a room and told that they were going to watch a short video (YouTube clip) on the IPad. The researcher then started the short video (i.e., no more than 3 minutes); once the video had completed the researcher told the children that he or she had to finish something and that they should sit until the researcher was done. The researcher then scored which steps Sally successfully demonstrated throughout the probe. Multiple exemplars of the video were utilized throughout all sessions and con­ ditions of the study. Three adults were required to run the naturalistic probe for in­ terrupting appropriately. During the naturalistic probe, one adult pulled Hank aside and told Hank to ask the second adult a question (e.g., "Ask Justin what time lunch is"). Meanwhile, the second adult and a third adult engaged in a conversation. The first adult scored which of the steps of the targeted skill Hank displayed correctly. The adults and questions varied throughout all sessions and conditions of the study. General Procedure and Design Research sessions ran five days a week and lasted approximately 20 minutes in length across the four conditions: baseline, intervention, maintenance, and a booster session condition. We utilized a multiple baseline across participants design. Naturalistic probes lasted ap­ proximately 60 seconds and the teaching procedure lasted 15 to 20 minutes.  COOL VERSUS NOT COOL 51 Baseline During the baseline condition for each participant, the researcher removed the participant from his or her clinical therapy session and set up one naturalistic probe (described above). After the naturalistic probe was completed the participant continued with his or her regu­ lar clinical therapy. Intervention The intervention condition consisted of two components: a nat­ uralistic probe followed by teaching using the cool versus not cool procedure plus role-playing. During the intervention condition for each participant, the researcher removed the participant from his or her clinical therapy and set up one naturalistic probe (described above) identical to baseline. After the naturalistic probe was completed the participant continued with his or her regular behavioral intervention for approximately ten minutes. Then, the researcher removed the par­ ticipant from his or her clinical therapy and implemented the cool versus not cool procedure plus role-play. The teaching procedure began with the researcher labeling the skill to be practiced (e.g., "We are going to practice interrupting ap­ propriately"). Next, the researcher demonstrated the target skill with a second researcher. The researcher demonstrated the behavior appro­ priately (cool) two times and inappropriately (not cool) two times, for a total of four demonstrations. All four demonstrations were randomly implemented; therefore, the order of the appropriate and inappropri­ ate demonstrations was always changing. The demonstrations were set up similar to naturalistic probes; however, the adults in the dem­ onstrations were not utilized during the naturalistic probes and no peers were utilized during teaching. During correct demonstrations, the researcher displayed all of the steps of the targeted social behav­ ior. During incorrect demonstrations, the researcher either omitted one of the steps or demonstrated one of the steps incorrectly. The step that the researcher omitted or completed incorrectly was based upon the participant's performance during the naturalistic probe earlier that research session. If the participant displayed 100% correct respond­ ing in the naturalistic probe that research session, then the researcher randomly selected a step to omit or display incorrectly. After each demonstration the researcher asked the participant to verbally discriminate if the demonstration was "cool" or "not cool." If the participant provided a correct discrimination, then the researcher provided a general praise statement (e.g., "Good Job," "Great," or "Way to Go") for correct verbal responding. The researchers did not  52 LEAF et al. provide specific praise as it was a two-choice discrimination and gen­ eral praise indicated that they discriminated the trial correctly. If the participants discriminated the trial incorrectly, then the researchers provided general corrective feedback (e.g., "That's not it," "That is not right," or "Nope"). The researchers did not provide specific corrective feedback or re-trial the discrimination trial, as each participant would be able to switch their response based on this general corrective feedback. The participant was then asked to verbally state one reason why the demonstration was either "cool" or "not cool". The researcher pro­ vided similar general praise for correct responding. For incorrect re­ sponding, the researcher provided general corrective feedback and provided a verbal statement of why the demonstration was "cool" or "not cool" (e.g., "Nope, that is not right; I did not say excuse me"). After the four demonstration trials, the researcher informed the participant that it was his or her turn to practice the skill the "cool" way; the participant practiced the skill with the second researcher. The re­ searcher set up the role-play similar to the naturalistic probe; however, the adults in the role-play were not utilized during the naturalistic probes and no peers were utilized during the role-play. After the role- play, the researcher asked the participant if he or she role-played the behavior correctly and provided either praise or corrective feedback (described above) based upon his or her verbal response. The researcher then asked the participant why the role-play was "cool" or "not cool" and provided praise or feedback based upon his or her verbal re­ sponse (described above). The participant was asked to role-play the skill until he or she demonstrated 100% of the steps on two consecu­ tive role-plays. If the participant role-played the skill incorrectly on two consecutive role-plays, the researcher verbally prompted the par­ ticipant on the next role-play. Maintenance During the maintenance condition for each participant, the re­ searcher removed the participant from his or her clinical therapy ses­ sion and set up one naturalistic probe (described above). After the naturalistic probe was completed the participant continued with his or her regular clinical therapy. Maintenance probes occurred 5, 6,13, and 17 days following intervention for Brady. Maintenance probes oc­ curred 3, 4, 8, and 24 days following intervention for Sally. Mainte­ nance probes occurred 7, 8, 10, and 14 days following intervention for Hank.  COOL VERSUS NOT COOL 53 Booster session Unfortunately, Brady and Hank demonstrated poor perfor­ mance during the maintenance condition; therefore, a booster session condition was added to ensure that Brady and Hank could appropri­ ately demonstrate their targeted social skills. During the booster session condition, the researcher removed Brady or Hank from their clinical therapy session and set up one naturalistic probe (described above). After the naturalistic probe, Brady or Hank had to role-play the behavior two times, regardless of how they performed during the naturalistic probe. During the role-play the teacher provided praise for correct demonstrations and corrective feedback for incorrect dem­ onstrations. The researchers elected to only implement the role-playing component, as opposed to role-playing and demonstration, for two reasons. First, the researchers wanted to work on the skill immediately following the naturalistic probe (e.g., in the moment), and a demonstra­ tion would require another adult (which was not available), whereas the role-play could be done with the researcher alone. Second, the re­ searchers wanted to systematically fade out the procedure by remov­ ing one component to assess whether the skills would maintain with lesser components. Dependent Variables Skill acquisition was the main dependent variable in this study. Skill acquisition was determined during naturalistic probes. Mastery criterion was set as the participant displaying 100% of the skill steps across three consecutive naturalistic probes. Naturalistic probes also assessed whether the participants would be able to generalize the skills taught to them to a peer or a different adult. The second mea­ sure was the percentage of correct responding during researcher demonstrations (i.e., discrimination and statement of why the demon­ stration was "cool" or "not cool"). The third measure was the average number of role-plays necessary during each teaching session for the participant to meet the criterion of demonstrating 100% of the skill steps across two consecutive role-plays. IOA and Treatment Fidelity A primary observer scored each naturalistic probe and a second­ ary observer was utilized for IOA. IOA was collected on the primary  54 LEAF et al. dependent variable during 33.3% of probe sessions (range, 27.3% to 38.9% across participants). IOA was calculated by totaling the number of agreements on the scoring of each skill step divided by the total number of agreements and disagreements. IOA for the prim ary de­ pendent variable was 98% (range, 80% to 100% across sessions) across all probes. To assess treatment fidelity, an independent observer recorded planned researcher behaviors during 34.7% of teaching sessions (range, 22% to 45.4% across participants). Planned researcher behav­ iors were: (a) the researcher demonstrating the behavior correctly twice and incorrectly twice, (b) the researcher asking the participant to discriminate whether each role-play was "cool" or "not cool" and why, (c) the researcher providing appropriate feedback based upon participant responding during each demonstration trial, (d) the par­ ticipant role-playing the behavior until the participant displayed the skill 100% correct across two consecutive probes, (e) the researcher asking the student to rate his or her own performance, (f) the researcher providing appropriate feedback after each role-play opportunity, and (g) the researcher providing prompting after two consecutive incor­ rect responses. Treatment fidelity was 100% across all sessions. Results Skill acquisition Participant performance during naturalistic probes is displayed in figure 1. During baseline, Brady displayed 0% of the steps for pro­ viding verbal support across three consecutive sessions. During in­ tervention, Brady demonstrated variable responding until he reached mastery criterion on the thirteenth naturalistic probe. During the as­ sessment of maintenance, Brady's performance continually declined until he displayed 0% of behavioral steps across two consecutive ses­ sions. Therefore, a booster session condition was implemented and within two sessions Brady was once again displaying 100% of the be­ havioral steps across three consecutive sessions. Based upon anecdotal information, incorrect responding during the intervention condition, maintenance condition, and first session of the booster condition was believed to be due to inattention. During baseline, Sally displayed 0% of the behavioral steps for chatting during nine out of the ten baseline sessions. Sally reached mastery criterion within nine naturalistic probes during the interven-  COOL VERSUS NOT COOL 55 tion condition. During the assessment of maintenance, Sally showed variable responding, ranging from 40 to 100% of the behavioral steps. Since Sally's responding was at criterion during half of the mainte­ nance probes and was 100% during the final probe, no booster sessions were implemented. During baseline, Hank displayed less than 40% of the behav­ ioral steps for interrupting during all baseline sessions. The only steps that Hank displayed correctly during baseline were the first two be­ havioral steps of interrupting (i.e., approaching the person who Hank had to ask a question to and not saying anything to the person until within 3 feet of the person). Hank reached mastery criterion on the 10th naturalistic probe during the intervention condition. During the assessment of maintenance, Hank displayed 100% of the steps during the first naturalistic probe in the maintenance condition and then dis­ played 40% of the steps on the following three naturalistic probes. Since Hank continued to display low performance during the assess­ ment of maintenance, a booster session condition was implemented; within four naturalistic probes Hank consistently demonstrated 100% of skill steps. Responding During Researcher Demonstration Table 1 represents the percentage of teacher demonstration trials where participants were able to correctly discriminate if a teacher demonstration was either cool or not cool. Table 2 displays the per­ centage of teacher demonstration trials where a participant was able to correctly state why a demonstration was cool or not cool. For Brady, there were a total of 52 demonstrations; overall, Brady was able to cor­ rectly discriminate whether the demonstration was "cool" or "not cool" during 98% of all researcher demonstrations and state why a demon­ stration was "cool" or "not cool" during 96% of all demonstrations. For Sally, there were a total of 36 demonstrations; overall, Sally was able to correctly discriminate whether the demonstration was "cool" or "not cool" during 97% of all researcher demonstrations and state why a demonstration was "cool" or "not cool" during 92% of all dem­ onstrations. For Hank, there were a total of 40 demonstrations; over­ all, Hank was able to correctly discriminate whether the demonstration was "cool" or "not cool" during 93% of all researcher demonstrations and state why a demonstration was "cool" or "not cool" during 85% of all demonstrations.  Percentage of Steps Correct 56 LEAF et al. BL INT MAINT BOOSTER Figure 1. Performance during Naturalistic Probes. Along the x-axis are num­ ber of sessions (number of naturalistic probes) and along the y-axis is the percentage of steps displayed correctly. Each panel represents a different participant.  COOL VERSUS NOT COOL 57 Table 1 Percentage of Correct Discrimination during Research Demonstration Session Brady Sally Hank 1 100 100 100 2 100 100 100 3 75 100 75 4 100 100 100 5 100 100 100 6 100 100 100 7 100 75 100 8 100 100 75 9 100 100 100 10 100 — — 11 100 — — 12 100 — — 13 100 _ _ Table 2 Percentage of Participants Correctly Stating Why a Demonstration was Cool or Not Cool Session Brady Sally Hank 1 75 100 25 2 100 100 100 3 75 75 100 4 100 100 100 5 100 100 100 6 100 75 100 7 100 75 100 8 100 100 75 9 100 100 75 10 100 — — 11 100 — — 12 100 — — 13 100 _ _  58 LEAF et al. Table 3 Number of Participant Role-Plays per Session Session Brady Sally Hank 1 5 5 3 2 3 2 2 3 2 2 2 4 2 2 2 5 2 2 2 6 6 2 2 7 7 3 8 8 2 2 2 9 6 2 2 10 2 — — 11 3 — — 12 5 — — 13 2 — — Participant Role-Playing During Teaching The number of role-plays each participant engaged in during a single teaching session is displayed in table 3. The number of role- plays varied from session to session for Brady. Anecdotally, sessions where Brady was paying better attention required fewer role-plays and sessions where Brady's attention was decreased required a greater amount of role-plays. Brady received a total of 47 role-plays, for an average of 3.62 role-plays per session during the intervention condi­ tion. Finally, Brady received a total of 6 role-plays during the booster session condition. For the majority of sessions, Sally required the minimal number of role-plays (two role-plays); overall, Sally completed a total of 22 role-plays, with an average of 3.4 role-plays per session during the intervention condition. Hank also required the minimal number of role-plays during the majority of teaching sessions. Overall, Hank completed a total of 31 role-plays, with an average of 2.8 role-plays per session during the intervention condition. In the booster session condition Hank completed a total of 6 role-plays.