Thomas Gumpel, School of Education, The Hebrew University of Jerusalem, Israel 91905. TGumpel@vms.huji.ac.il.

Author notes: The authors wish to thank Ruth Butler for her insightful comments.

    We examined an alternative model to social skills and social competence training through conceptualizing social skills deficits according to a dichotomous model of either (a) not possessing the necessary skill in the behavioral repertoire, or (b) possessing the skill but not using it appropriately. We hypothesized that many skills deficits could be explained by the latter interpretation. Using an ABAC design, we compared the use of two behavioral monitoring systems: self-monitoring and group monitoring of social skills while playing board games. The data suggest substantial increases in prosocial behaviors along with concomitant reductions in anti-social behaviors as a result of the two monitoring procedures. Data also show generalization across gaming situations and maintenance over time. Suggestions are presented for further research to isolate the role of self-regulation in social skills research.
 

    A child's ability to interact socially with her or his peers is a basic skill, central to a wide variety of other measures of quality of life. Researchers have determined highly situation-specific and rule-governed behaviors, including both overt and covert skills (Gumpel, 1994) which, when correctly performed, will lead to positive or neutral environmental consequences (McFall, 1982). Despite the fact that differing operational definitions of social skills exist (e.g., Chadsey-Rusch, 1992; Gresham & Elliott, 1989; Gumpel, 1994; Storey, Rhodes, Sandow, Loewinger, & Petherbridge, 1991), the positive outcome of the appropriate use of these social skills is clear (Coie, 1990): peer acceptance and its concomitant social status.

    Social skills, as conceptualized by learning and social learning theories, are shaped by the environment, which continually reinforces and refines these behaviors. Theoretical understanding of social competence has developed along two principal lines: (a) molecular approaches, and (b) process approaches. Molecular models (Foxx, McMorrow, Bittle, & Ness, 1986; Storey & Gaylord Ross, 1987; Trower, Bryant, & Argyle, 1978) have delineated a finite set of overt behavioral components, which when performed adequately are reinforced by the environment. Such molecular skills include specific responses to social stimuli such as tone of voice, body posture and eye contact. The early 1980s witnessed a proliferation of social skills interventions based on molecular models. For example, Foxx and his colleagues (Foxx & McMorrow, 1985; Foxx, McMorrow, Bittle, & Fenlon, 1985) used a modified board game to teach a variety of social skills to adults with varying degrees of mental retardation. In these studies, stimuli were presented via game cards, with correct responses scored according to fixed criteria. In other words, social skills were viewed as specific responses to social discriminative stimuli (Wong, Morgan, Crowley, & Baker, 1996). Despite the large amount of applied research based on this approach, problems with generalization and maintenance of treatment goals were recognized early on (Bryan, 1997; DuPaul & Eckert, 1994; Wong et al., 1996). Process approaches are an attempt to breach problems with treatment generalization by extending the training paradigm. McFall (1982), Trower (1982) and Dodge (1986) each presented process approaches, including series of both overt and covert steps that the individual will perform and are necessary for creating competent or skillful behaviors. The theoretical paradigm thus shifted from a purely skill-centered approach to the "process" involved in creating skillful behavior.

    Using a variation of McFall's (1982) and Trower's (1982) model of social skills, Park and Gaylord-Ross (1989) were able to teach social problem-solving social skills to adolescents with moderate mental retardation in a supported employment setting. Instead of focusing on molecular behavioral components of social interactions, Park and Gaylord-Ross demonstrated that the three participants were able to engage in social problem-solving. The authors showed that social skills training need not (or perhaps, should not) focus solely on overt behaviors, but could concentrate on the development of complex social behaviors through the mediation of covert events (Meichenbaum & Goodman, 1971; Moore, 1995).

    Gumpel (1994) expanded McFall's (1982) and Trower's (1982) process approach into a six stage model of social competence. The innovation inherent in this model lies primarily in its focus on self-regulatory mechanisms (Bandura, 1986). It consists of five iterative and recursive skills: decoding, decision, performance, self-monitoring and environment-observational skills whose components have been validated using both theoretical (Gumpel & Wilson, 1996) and applied (Gumpel, 1989; Gumpel & Frank, 1999) research methodologies. The model differentiates between two different causes of poor social skills: (a) skill deficits which are a result of a lack of acquisition of the appropriate target skills, and (b) performance deficits which result from inappropriate use of skills already in the individual's repertoire. Whereas, the skills deficit model would be suggested by the complete absence of the target skill in the individual's behavioral repertoire, performance deficits describe situations where the appropriate skill is being inappropriately or unstably used. This may result from insufficient previous reinforcement or the lack of appropriate self-regulatory mechanisms to ensure the skill's correct performance.

    This distinction between "not having the skill at all" and "having it, but not using it" has clear theoretical and applied implications. For the first instance to be correct it would mean that at no point has the individual performed the appropriate behaviors. In other words, the target behavior does not exist in the individual's skill repertoire and the individual can be assumed to possess an acquisition or learning deficit (Gresham & Elliott, 1989). Social learning theory (Bandura, 1977), however, clearly supports the second instance, namely that the behavior is almost always present in the individual's behavioral repertoire. To meet this assumption, the individual must have exhibited the behavior or have learned the behavior in the past. Clearly, this is an easier criterion to meet and has been discussed by Gresham and Elliot (1987) who differentiated between overt, performance, self-control and, self-control performance skills. We feel that this breakdown is unnecessary and that a simpler and more parsimonious dichotomous conceptualization is sufficient: skills versus performance deficits.

    If appropriate social skills are absent from the individual's skill repertoire, then in order to promote more socially skillful behavior, the treatment paradigm will involve the direct instruction of prosocial skills using either a molecular or a process approach. If, on the other hand, these skills are in the individual's skill repertoire but are not being adequately or appropriately activated, then social skills training should focus more on activation strategies. One such activation methodology for increasing the frequency of skills is the general mechanism of self-management strategies (DuPaul & Eckert, 1994) through its reliance on self-mediated stimuli to promote generalization and maintenance of behavioral gains (Stokes & Baer, 1977).

    Several terms fall under the general rubric of self-management: self-monitoring, self-evaluation, self-instruction and self-correction. All terms are based on the use of self-regulatory mechanisms to facilitate the completion of behavioral chains (Bandura, 1986) and are grounded in ethical, legal and moral concerns that often the client has only limited control over his own behavior change process (Kadzin, 1975; Shapiro, 1981). A common form of these self-management procedures is to have a client problem-solve (self-instruct) and subsequently compare (self-evaluate) these observed behaviors to a specified criterion level of performance, finally resulting in self-correction (Meichenbaum & Goodman, 1971). Even though self-monitoring was originally employed as a method of data collection, a wide body of research has pointed to problems regarding the accuracy of client-collected data (McFall, 1977). Problems with data accuracy, coupled with the obvious and documented treatment gains associated with self-monitoring procedures, led researchers to explore the reactive effects of these procedures (Kazdin, 1974; Nelson & Hayes, 1981). They found that even though data collected by the client was often inaccurate, data collection procedures were sufficient to serve as an impetus for behavioral change.

    The purpose of this research was to investigate whether the non-activation of appropriate and previously acquired social behavior can account for some social skills deficits. We used a social skills training package, which focuses on metacognitive activation skills (i.e., self-monitoring and group monitoring), and examined whether this treatment would increase the frequencies of prosocial game-playing behavior. In other words, we hypothesized that re-focusing the social skills training paradigm from one based on skill deficits to one based on performance deficits (through self-management procedures) would cause an increase in socially skillful behaviors. We hypothesized that if our model is correct, participants would exhibit low and unstable baseline levels of social behaviors (i. e., the target behaviors are present in their behavioral repertoire). We created two treatment conditions: self-recording and group recording. The first condition would examine the reactive effects of self-recording. We hypothesized that this procedure would substantially increase the performance of social skills. The second treatment condition was designed to phase out external controls (i. e., reinforcement for individual performance levels). We hypothesized that treatment gains would be improved or maintained in this condition. We operationally defined an increase in socially skillful behaviors as the increased frequency of positive social interactions and concomitant reductions negative/aggressive behaviors.

Dependent and Independent Variables
    Data were collected on two dependent measures: (a) "Positive Interaction" (PI): any instance where the child was engaged in appropriate game-playing behavior (e.g., following the rules and appropriate turn-taking), not including teasing, raising of voice, cursing or physical contact with a peer or his or her property (e.g., school bag or game tokens), and (b) "Negative Interaction" (NI): an interaction which involved teasing, raising of voice, cursing or physical contact with a peer or his or her property (e.g., school bag or game tokens). The independent variable was defined as the separate game playing sessions in the after-school club.

Experimental Design
    To evaluate the effectiveness of the different components of the social skills training program on the participants' social skills, an alternating treatments design (ABAC) was used. The two different treatment conditions were dependent on the type of behavioral monitoring employed. The first treatment condition involved the children's' self-monitoring of their own behaviors; the second condition involved external monitoring of those same behaviors. Generalization data were collected in three ways: (a) probe sessions scattered across the two experimental treatment conditions and collected in non-treatment settings, (b) generalization data across settings as different types of board games were introduced, and (c) maintenance data across time.
Baseline. During the baseline phase, the five participants were physically separated from the rest of the after-school program with all play encounters conducted in a separate room. Children were told to play a board game together for 30 min. Board games were different for reach gaming-session and were randomly chosen from a game-pool of known and age appropriate games (e.g., the Hebrew versions of "Chutes and Ladders(tm)," "Candy Land(tm)," "Sorry(tm)" and "Trouble(tm)"). No feedback or instructions were given to participants during either baseline or reversal conditions.

Procedure. During training sessions, the experimenter introduced the daily board game with instructions that it should be played for 30 minutes. Games for each game-period were randomly chosen from the game-pool. Before beginning play, five roles were assigned to participants (Johnson & Johnson, 1985; Olympia, Sheridan, Jenson, & Andrews, 1994): (a) the Chair was responsible for teaching the game and rules to the other group members, (b) the Observer was responsible for monitoring group members' behavior (which were explained to him or her by the experimenter), (c) the First Recorder was responsible for recording appropriate group behaviors (PI) provided to him or her by the experimenter, (d) the Second Recorder was responsible for recording the game's outcomes, and (d) the Substitute was responsible for assisting the experimenter in dispensing reinforcements as well as serving as a substitute game-player for any absent group member.

    All games took place in the same room as during baseline. Three posters were hung on the wall: (a) "Today's Rules" included the specific rules for that day's game, (b) "Today's Jobs" introduced and defined each of the participant roles and responsibilities, and (c) "Today's Behaviors," which was divided into two sections: the first section showed the cumulative scoring of PI points and the second section kept a cumulative tally of the game results. PI behaviors were described to the children as "cooperative and continuous play, around the game-board with appropriate turn-taking until the game was completed."

    Prior to each group meeting, the daily group Chair met with the second author to review the rules for that day's game and to hang the appropriate rule poster in the gaming area. At the beginning of the group meeting, the Chair explained the game's rules to all participants and checked for understanding. The Chair then gave each participant a badge to wear denoting his or her role for the day. The group then commenced playing their board game, with data collection and monitoring of play behaviors according to the treatment condition, as outlined below, with the Observer writing a short written narrative of the game. At the end of the allotted gaming period (approximately 30 minutes) the Observer provided behavioral feedback to the group members. The First Recorder received observational data from the researcher and summarized the data on the data poster and the Second Recorder recorded the final outcome of the game by writing the name of the game's winner.

Data Collection and Reliability. All data were collected during gaming sessions by two separate observers, using a momentary time sampling (20s) system. The gaming period lasted for approximately 30 minutes and included 90 possible observations. A round-robin approach was used to observe each game-player; in other words, each player was observed once every fourth 20s period. Each observer was equipped with a small "Walkman" type tape recorder, which was hidden under a sweater and used one small earpiece. Upon presentation of the auditory stimulus, the observer marked whether the target child was engaged in any of the two dependent variables (PI or NI behaviors). Observers were trained to code behaviors until they reached a level of inter-observer agreement for occurrences and non-occurrences greater than 90%.

    Inter-observer agreement was assessed for each child at least once for each experimental condition for a total of 40% of all possible observations. Occurrence and non-occurrence agreements were calculated on an interval-by-interval basis by dividing the agreements by the total number of agreements and disagreements and multiplying by 100. Inter-observer agreement for the occurrence of PI behaviors was 93% (range: 91%-96%) and for NI behaviors was 87% (range: 83%-91%); and for non-occurrence of PI behaviors was 95% (92%-98%) and for NI behaviors was 95% (range 83%-100%).

    Small edible prizes were distributed to each participant at the beginning and end of each group meeting to promote attendance, were non-contingent on gaming performance, and were faded out without prior warning at the end of the first treatment condition. Reinforcers were chosen through individualized reinforcer sampling procedures. Each child was presented with a tray of assorted potential reinforcers. These were sampled from each of the five categories outlined by Northrop et al. (1997): (a) edible items, (b) tangible items, (c) activities, (d) teacher attention, or (e) peer attention.

First Treatment Condition: Self-Recording. The first treatment condition took place over eight consecutive group meetings. During this treatment condition, each player recorded his or her own PI behaviors. Appropriate and personalized levels of PI behaviors were arrived at through calculating 20% more than the mean number of PI behaviors for the first baseline condition. For instance, if during the first treatment condition Avi exhibited 20 PI behaviors, his personal criterion level would be 24 (20+20%). Each child met with the second author following the end of each gaming period and verbally answered three questions: "Did I play the game, follow the rules and participate fully?" Each group member would then record those points on a self-recording sheet which was then summarized on the "Today's Behaviors" poster.

    Second Treatment Condition: Group Recording. The second treatment phase consisted of eight consecutive sessions and examined if skillful performance could be maintained without external control. This treatment phase was identical to the previous treatment conditions (i.e., each child continued to record his or her own PI behaviors), except that a group criterion level was determined based on the number of PI behaviors during the previous treatment condition (using the same procedure of increasing the criterion level by 20%). At the conclusion of each game period, group member's PI points were summed into a total PI score which was then recorded on the "Today's Behavior" poster. During this treatment condition group members were no longer scored individually.

Generalization. To promote generalization across gaming situations, game times were staggered during the after-school program. Additionally, games were changed to allow for training of multiple exemplars (Stokes & Baer, 1977). All possible games in the club were categorized as belonging to one of the four categories: (a) competitive, (b) physical, (c) cognitive, and (d) language oriented and were randomly chosen from each of these four groups. During generalization periods no treatment interventions took place. Tangible reinforcers provided to children for participating in the group were faded towards the end of the first treatment phase, and during the second and third treatment phases they were replaced by social and naturally maintaining reinforcers (Stokes & Baer, 1977). These reinforcers were not provided by the adult, but rather through the naturally maintaining satisfaction of game-playing.

Figure 1 depicts the frequencies of the three dependent variables for all five children. PI behaviors for all five participants were at low levels during baseline conditions. The introduction of the first treatment condition had a strong stabilizing influence on the prosocial behaviors for all five participants and caused a continued increase in desirable behaviors, and reduced to almost a zero-level the presence of NI behaviors. This trend is most pronounced for Avi, Boaz and Danny. However, even for Gila and Haddas, this trend is strong despite slight regressions towards the middle of the treatment phase.

    The removal of treatment during the reversal phase had a weak effect on all five participants. For all of the children a slight downward trend is noticeable during this second baseline phase. Within the time and ethical constraints of this study, it was not possible to allow this second baseline phase to continue. This downward trend, however, was reversed quickly with the introduction of the second treatment condition of Group Monitoring. The use of group monitoring for PI behaviors was able to maintain high levels of the target behaviors for all five participants. Treatment gains clearly generalized in the first generalization phase (across situations) as well as maintained themselves during the maintenance phase of six weeks. Additionally, generalization data taken from the behavioral probes during both treatment conditions show changes similar to those in the treatment conditions.

    Overall, the results show: (a) it is possible to train social skills using an alternative to the traditional social skills training paradigm of direct practice, verbal training, modeling, role-play and reinforcement of target skills by focusing on behavioral monitoring procedures, (b) self-recording along with personalized criterion levels was effective in promoting social skills, (c) group recording was able to maintain these behavior changes, and (d) these behavioral changes generalized across game-tasks and probe situations and maintained themselves at six week follow-up.

    We assessed (a) whether socially inappropriate behaviors can be viewed as resulting from an absence of those skills in the performer's behavioral repertoire or are a result of the inappropriate (or lack of) use of those skills, and (b) whether training in behavioral monitoring skills would facilitate the use of prosocial skills already in the individual's behavioral repertoire. The results presented show that it is possible to conduct successful social skills training without relying on the traditional deficit based package of direct practice, verbal training, modeling, role-play and reinforcement of target skills, but rather it is possible to use methods designed to require increased self-regulation. We were able to show that collecting data on the frequencies of one's appropriate social interactions can have a strong mediating influence on the performance of those same skills.

    The social skills training package presented consisted of two different types of self-regulatory training: (a) self-recording, and (b) group monitoring. Both components were presented along with non-contingent reinforcement as well as public displays of data collected. Self-recording caused a pronounced increase in appropriate behaviors with concurrent decreases in inappropriate ones. However, a limitation of this study is the presentation of the social skills training package as consisting of these two components (i.e., self-recording and the public display of behaviors); further research should investigate the efficacy solely of self-recording. Additionally, due to the fact that the study examined children's behavior in a group situation, it was not possible to control for order effects. Future research should control for this variable by expanding the number of groups and by counter-balancing treatments across groups.

    The children used in this study were all enrolled in special education frameworks for children with emotional or behavioral disorders and exhibited primarily externalizing behaviors. Future research in this area should formally evaluate these behavior disorders using an instrument such as the Child Behavior Check List (Achenbach, 1988) or the Social Skills Rating System (Elliott, Gresham, Freeman, & McCloskey, 1988), and should control for both internalizing and externalizing disorders. Such between group comparisons would increase generalizability of the findings.

    We hypothesized that once the participants are trained to attend to their own behavior, these metacognitive skills would quickly maintain and generalize; hence, we anticipated a weak return to baseline effect. The return to baseline was, indeed, only moderately effective. It may be that a reversal design is not appropriate with a powerful component such as self-recording, a design which does not necessitate a reversal stage may be more appropriate (i.e., a multiple baseline design). Additionally, a multiple baseline across children would also allow for data collection in a series of different non-game related situations. The exploratory nature of this research examined only limited generalization and maintenance effects. Further research should look at generalization in unstructured and free play behaviors with group members as well as other children, and should examine generalization in non-play realms (e.g., in-class behaviors).

    Data referring to generalization across settings, people and other social skills would greatly increase our understanding of the efficacy of self-recording procedures in overcoming the generalization problems endemic in social skills training. Additionally, social validity data should be collected from both teachers and children. To further examine the efficacy of the activation model, a more sophisticated between-group experimental design is needed. Such a design would empirically compare the efficacy of the two treatment models and would further support the hypothesis that social skills deficits may not necessarily be due to skill deficits (in need of training), but rather may be caused by the incorrect use of those skills. Further research, which more fully examines this issue of skills versus performance is needed to further elucidate this important theoretical issue.

    Research and practice in the fields of learning disabilities and emotional and behavioral disorders has recently focused on the use of metacognitive strategies to promote academic achievement (Coleman, Wheeler, & Webber, 1993; Tralli, Colombo, Deshler, & Schumaker, 1996; Warner, Schumaker, Alley, & Deshler, 1989). It appears that the metacognitive skill of self-regulation can also be applied to social behaviors. This research demonstrates the relevance of metacognitive activation strategies for socially skillful performance. The use of such executive metacognitive strategies has gained momentum over the last decade (Wong, 1991); and it appears that this field is also relevant to the social skills domain. Indeed, these findings appear to support the relationship between learning disabilities and social competence problems (Gresham & Elliott, 1989). Further research should continue to examine this critical issue.

        Achenbach, T. M. (1988). Child behavior checklist. Burlington, VT: Department of Psychiatry, University of Vermont.

        Bandura, A. (1986). Social foundations of thought and action: A social cognitive approach. Englewood Cliffs, NJ: Prentice Hall.

        Bryan, T. (1997). Assessing the personal and social status of students with learning disabilities. Learning Disabilities Research & Practice, 12, 63-76.

        Chadsey-Rusch, J. (1992). Toward defining and measuring social skills in employment settings. American Journal on Mental Retardation, 96, 405-418.

        Coie, J. D. (1990). Toward a theory of peer rejection. In S. R. Asher & J. D. Coie (Eds.), Peer rejection in childhood (pp. 365-401).
Cambridge, England: Cambridge University Press.

        Coleman, M., Wheeler, L., & Webber, J. (1993). Research on interpersonal problem-solving training: A review. RASE: Remedial and Special Education, 14, 25-37.

        Dodge, K. (1986). A social information processing model of social competence in children. In M. Perlmutter (Ed.), Cognitive perspective on children's social and behavioral development: Minnesota symposium on child psychology (Vol. 18, pp. 77-125). Hillsdale, NJ: Erlbaum.

        DuPaul, G. J., & Eckert, T. L. (1994). The effects of social skills curricula: Now you see them, now you don't. School Psychology Quarterly, 9, 113-132.

        Elliott, S. N., Gresham, F. M., Freeman, T., & McCloskey, G. (1988). Teacher and observer ratings of children's social skills: Validation of the Social Skills Rating Scales. Journal of Psychoeducational Assessment, 6, 152-161.

        Foxx, R., McMorrow, M., Bittle, R., & Ness, J. (1986). An analysis of social skills generalization in two natural settings. Journal of Applied Behavior Analysis, 19, 299-305.

        Foxx, R. M., & McMorrow, M. J. (1985). Teaching social skills to mentally retarded adults: Follow-up results from three studies. The Behavior Therapist, 8, 77-78.

        Foxx, R. M., McMorrow, M. J., Bittle, R. G., & Fenlon, S. J. (1985). Teaching social skills to psychiatric inpatients. Behaviour Research and Therapy, 23, 531-537.

        Gresham, F. M., & Elliot, S. N. (1987). The relationship between adaptive behavior and social skills: Issues in definition and assessment. Journal of Special Education, 21, 167-181.

        Gresham, F. M., & Elliott, S. N. (1989). Social skills deficits as a primary learning disability. Journal of Learning Disabilities, 22, 120-124.

        Gumpel, T. (1989, April). Social skills training with mentally retarded adults: A reformulation of a construct. Paper presented at the Council of Exceptional Children, Toronto, Canada.

        Gumpel, T. (1994). Social competence and social skills training for persons with mental retardation: An expansion of a behavioral paradigm. Education and Training in Mental Retardation and Developmental Disabilities, 29, 194-201.

        Gumpel, T., & Frank, R. (1999). An expansion of the peer-tutoring paradigm: Cross-age peer-tutoring of social skills among socially rejected boys. Journal of Applied Behavior Analysis, 32, 115-118.

        Gumpel, T., & Wilson, M. (1996). Application of a Rasch analysis to the examination of the perception of facial affect among persons with mental retardation. Research in Developmental Disabilities, 17, 161-171.

        Johnson, D. W., & Johnson, R. T. (1985). The internal dynamics of cooperative learning groups. In R. Slavin, S. Sharan, R. H. Zazarowitz, C. Webb, & R. Schmuck (Eds.), Learning to cooperate, cooperating to learn (pp. 103-124). New York: Plenum.

        Kadzin, A. E. (1975). Behavior modification in applied settings. Homewood, IL: Dorsey Press.

        Kazdin, A. E. (1974). Reactive self-monitoring: The effects of response desirability, goal setting, and feedback. Journal of Consulting and Clinical Psychology, 42, 704-716.

        McFall, R. (1977). Parameters of self-management. In R. B. Stuart (Ed.), Behavioral self-management (pp. 196-214). New York: Brunner/Mazel.

        McFall, R. (1982). A review and reformulation of the concept of social skills. Behavioral Assessment, 4, 1-33.

        Meichenbaum, D., & Goodman, J. (1971). Training impulsive children to talk to themselves: A means of developing self-control. Journal of Abnormal Psychology, 77, 115-126.

        Moore, J. (1995). Radical behaviorism and the subjective-objective distinction. The Behavior Analyst, 18, 33-49.

        Nelson, R. O., & Hayes, S. C. (1981). Theoretical explanations for reactivity in self-monitoring. Behavior Modification, 5, 3-14.

        Northup, J., Jones, K., Broussard, C., DiGovanni, G., Herring, M., Fusilier, I., & Hanchey, A. (1997). A preliminary analysis of interactive effects between common classroom contingencies and methylphenidate. Journal of Applied Behavior Analysis, 30, 121-125.

        Olympia, D. E., Sheridan, S. M., Jenson, W. R., & Andrews, D. (1994). Using student-managed interventions to increase homework completion and accuracy. Journal of Applied Behavior Analysis, 27, 85-99.

        Park, H. S., & Gaylord Ross, R. (1989). A problem-solving approach to social skills training in employment settings with mentally retarded youth. Journal of Applied Behavior Analysis, 22, 373-380.

        Shapiro, E. (1981). Self-control with the mentally retarded. In M. Hersen, R. Eisler, & P. Miller (Eds.), Progress in behavior modification (Vol. 2, ). New York: Academic Press.

        Stokes, T., & Baer, D. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10, 349-367.

        Storey, K., & Gaylord Ross, R. (1987). Increasing positive social interactions by handicapped individuals during a recreational activity using a multicomponent treatment package. Research in Developmental Disabilities, 8, 627-649.

        Storey, K., Rhodes, L., Sandow, D., Loewinger, H., & Petherbridge, R. (1991). Direct observation of social interactions in a supported employment setting. Education and Training in Mental Retardation, 26, 53-63.

        Tralli, R., Colombo, B., Deshler, D. D., & Schumaker, J. B. (1996). The Strategies Intervention Model: A model for supported inclusion at the secondary level. Remedial and Special Education, 17, 204-216.

        Trower, P. (1982). Toward a generative model of social skills: A critique and synthesis. In J. P. Curran & P. M. Monti (Eds.), Social skills training: A practical handbook for assessment and treatment (pp. 82-115). New York: Guilford.

        Trower, P., Bryant, B., & Argyle, M. (1978). Social skills and mental health. London: Methuen.
Warner, M. M., Schumaker, J. B., Alley, G. R., & Deshler, D. D. (1989). The role of executive control: An epidemiological study of school-identified learning disabled and low-achieving adolescents on a serial recall task. Learning Disabilities Research, 4, 107-118.

        Wong, B. Y. L. (1991). The relevance of metacognition to learning disabilities. In B. Y. L. Wong (Ed.), Learning about learning disabilities (pp. 231-258). San Diego: Academic Press, Inc.

        Wong, S. E., Morgan, C., Crowley, R., & Baker, J. N. (1996). Using a table game to teach social skills to adolescent psychiatric inpatients: Do the skills generalize? Child & Family Behavior Therapy, 18, 1-17.

Game-playing 20