My Major Area Paper, Vanderbilt University (1994)
©1992, 2013 by Dallas Denny
Source: Denny, Dallas. (1994). Generalization and maintenance in social skills training of adult persons with mental retardation: A review of the literature. Major Area Paper, Department of Special Education, George Peabody College of Vanderbilt University.
Generalization and Maintenance in Social Skills Training
Of Adults With Mental Retardation
A Review of the Literature
Major Area Paper
By Dallas Denny
Department of Special Education
George Peabody College of Vanderbilt University
Abstract
Abstract
I discuss the importance of generalization and maintenance in social skills training. After defining the terms used in the paper, I examine a special issue of the Journal of Applied Behavior Analysis (JABA) on the state of the art in social skills training. I then examine and discuss seven reviews which include studies of social skills training in adults with mental retardation, examining them for discussion of generalization and maintenance. Following this, I review fourteen papers in which social skills were taught to adults with mental retardation, concentrating on the programming and measurement of generalization and maintenance. I conclude with a discussion of what my findings disclose about the state of the art in social skills training.
Mark Twain once remarked that everyone complains about the weather, but that nobody ever does anything about it. Something akin to this has happened in applied behavior analysis, a field in which the importance of measuring the generalization and maintenance of behavior change was explicitly pointed out more than 25 years ago by Baer, Wolf, & Risley (1968). Stokes & Baer, in a widely-cited 1977 paper titled, “An implicit technology of generalization,” called for their peers to take steps to systematically measure behavior change across time, setting conditions, behaviors, and trainers, and described ways in which it could be done.
While Stokes & Baer (1977) certainly made the behavior analytic community aware of the importance of measuring and programming generalization and maintenance, the extent to which their call to action has been heeded is generally considered to be somewhat akin to the situation with the weather: that is, maybe some people have done or are doing something about it, but not that many, and not that much (Fox & McEvoy, in preparation).
As Baer, Stokes, Holman, Fowler, & Rowbury (1981) pointed out, it was not that the “outline of a technology” of generalization and maintenance did not exist in applied behavior analysis, but that it was “largely implicit in current techniques, rarely recognized as a technology or as a target for technological research, and … largely uncodified” (pp. 39-40). Even today, when generalization is found, we are often unable to say how or why it has occurred (Stokes & Osnes, 1989).
In this paper I will do several things: a) define generalization, maintenance, social skills training, and the characteristics of the subject population in which I am interested; b) look at the extent to which generalization and maintenance have been programmed and measured in social skills training in general with a variety of special populations (by examining a special issue of the Journal of Applied Behavior Analysis); c) examine existing reviews of social skills training in persons with mental retardation in order to determine the findings of the reviewers regarding generalization and maintenance; and d) examine papers not covered by the reviews in order to look at the extent to which generalization and maintenance have been programmed and measured in studies of social skills training of adults with mental retardation. In the discussion section, I will a) compare research with adults with mental retardation with research with subjects who exhibit other disabilities; and b) determine the extent to which a technology of programming and measuring generalization and maintenance has been demonstrated in social skills training for adults with mental retardation.
Defining Generalization
As early as 1938, in his work with nonhuman animals, Skinner addressed the issue of stimulus generalization (transfer of stimulus control to objects without a history of reinforcing the organism) and maintenance (exhibition of operant response after the discontinuation of the artificial reinforcement schedule). However, Skinner’s definition of generalization was a strictly functional one which may not lend itself to the practical and ethical considerations of applied settings (Fox & McEvoy, in preparation). Later workers have found his definitions and those of other early workers (e.g. Keller & Schoenfeld, 1950) too restrictive (cf. Drabman, et al., 1979; Allen, et al., 1991).
In 1977, Stokes & Baer expanded the definition of generalization to include exhibition of a target behavior under nontraining conditions across settings, behaviors, subjects, and/or time. In their framework, there were several different types of generalization. Maintenance, in Stokes & Baer’s system, is considered to be but one type of generalization.
Stokes & Baer’s definition is a topographical rather than a functional one, and hence does little to specify functional relations among variables in the treatment setting (Stokes & Osnes, 1988). In a later paper (1989), Stokes & Osnes refined Stokes & Baer’s intervention tactics, but did not further develop the classification system.
In 1979, Drabman, et al. introduced the generalization map as a method of classifying generalization effects. Their system, which is perhaps the ultimate in topographical models, gives sixteen different classifications of generalized treatment effects, as many as are possible with Stokes & Baer’s (1977) four domains (time, setting, behavior, and subject). In this system, also, maintenance is seen as just one of several types of generalization.
I am unaware of a clearer or more succinct description of the subtleties of measuring generalization in applied settings than that of Fox & McEvoy (in preparation). They begin by describing a basic study in which measuring generalization is a relatively straightforward process. Their example is that of a pigeon in an operant conditioning chamber.
For the pigeon, pecking at an electrically-lit button connected to an electronic counter, it is easy to structure an experiment so no reinforcement is available under conditions which share stimulus characteristics with the reinforcing condition. Under the strict control allowed by the operant conditioning chamber, it is possible to assess the functional relationships between dependent and independent variables. Fox & McEvoy give the example of the discriminative stimulus being a particular level of illumination of the button. A schedule of reinforcement previously determined by the experimenter is in effect so long as the button is lit; when the button is not lit, extinction conditions prevail. After a certain time in the chamber, the pigeon will learn to peck only, or at least more frequently, when the key is lit than when it is not. [Macintosh, in his 1974 text on animal learning, gives a similar example, using wavelength (color) rather than intensity as the discriminative stimulus].
Fox & McEvoy then postulate a range of conditions of illumination, in which the button is sometimes more intensely lit, and sometimes less intensely lit than previously. These new conditions, although sharing stimulus characteristics with the condition which results in reinforcement, are not reinforced. Rates of key-pecking will typically be intermediate, with levels of illumination close to the originally reinforced level eliciting more key pecks than more dissimilar levels.
Generalization, in this instance, is measured under conditions of non-reinforcement. The strictly controlled setting allows us to say with some certainty that the behavior has generalized [in a Skinnerian (functional) sense] to different intensities of illumination. It is possible to prepare a graph, with number of key pecks plotted on the Y-axis and level of illumination plotted on the X-axis, showing a generalization gradient (Macintosh, 1974). From this gradient, we can predict the pigeon’s behavior when it is presented with a key illuminated at a previously untested level of illumination.
Fox & McEvoy contrast the pigeon experiment with an applied study, in which a child with a disability is trained to exhibit desired social behaviors. The child is prompted and praised by the experimenter to share a toy with a particular classmate (peer).
In this study, as with the pigeon, generalization is measured; in this case, the experimenter looks to see whether the child shares the toy with other (i.e., untrained) classmates, and whether the subject shares toys in other settings. However, the strict functional measurement which works for the pigeon is not possible here, for the variables are more complex and under less complete control. Unlike a electrically operated key, which can accept input, but which does not respond, the selected classmate as well as the classmates used to measure generalization interact with the subject child, responding to his social bids, and the child himself may behave differently because of the change of behavior of the classmates. Not only may other aspects of the subject’s behavior have been changed by the intervention, but the responses of the classmates may have been affected, and the ensuing social interactions can become quite complex and impossibly difficult to measure. This change in interactions may or may not occur in other settings, with other trainers, with other classmates, or across time. Kazdin (1975) would say the social behavior of the subject has come under control of the natural community of reinforcement.
Fox & McEvoy point out that in this situation, we cannot be sure generalization, in the strict sense that Skinner meant it (having happened in the absence of reinforcement) has occurred.
… we have not accomplished generalization in the traditional, functional sense; rather, we have taken advantage of naturally occurring contingencies… by teaching a response, sharing, likely to be reinforced by the “test stimuli” (the non-training peers) themselves, even though the teacher does not prompt or praise subject interactions with them.
—Fox & McEvoy, manuscript in preparation
We know we have seen a change across setting, time, subject, or behaviors, but we are not sure whether the change has occurred because the behavior has generalized to conditions which contain stimulus elements similar to those which elicit it, or because of changes in the nature of the social interactions with peers. Most likely, it is some combination of these.
An example can be taken from Fox, et al. (1984), who used a sequential multiple peer exemplar training procedure to train an autistic girl to make social interactions. During the intervention phase, a teacher would use a graduated guidance procedure to prompt the girl to initiate social contact with a nondisabled peer. Generalization was measured daily in a playsetting three hours after training. All peers were present during the generalization sessions, but peers were sequentially introduced throughout the intervention phase. After training with three peers, social initiations and time spent in interactions with a fourth nondisabled peer increased “spontaneously.”
We could say the behavior had generalized across peers. However, the authors point out that virtually all of the subject’s initiations to peers with autism were unsuccessful (i.e. were not responded to), and consequently she very rarely made social initiations to them. Obviously, the fact that the nondisabled peers responded to her social bids played a part in maintaining the subject’s behavior. To this extent, it was impossible for the authors to demonstrate generalization in a strictly functional sense. Despite this, however, the findings are highly significant, and of practical use in classroom settings.
What an applied researcher needs to be able to demonstrate is that under prevailing social conditions, intervention will result in lasting behavioral change (i.e., will continue after treatment is terminated), and that desired behavior change will occur spontaneously (i.e. without programming) across settings, subjects, and behaviors other than those initially programmed.
Strict control of variables in social skills training is not possible— and will probably never be possible. Behavioral coding schemes, even with the aid of computers (cf Denny & Fox, 1989), are quite complex, and must be limited so acceptable levels of inter-observer agreement can be maintained. With present technologies, the many interactions taking place in the classroom cannot be accurately measured; barring unforseen breakthroughs in behavioral observation technologies, researchers must necessarily limit their measurements.
It would of course be possible to put the subjects in a strictly controlled environment— something much like the operant conditioning chamber used for the pigeon, but there are two problems with this. First, human interactions do not ordinarily take place in such artificial settings. Findings would have little social validity. We would learn little which would help us under the complex social conditions in which human beings operate. Second, and more importantly, ethical considerations allow us but limited latitude to control peoples’ lives. Unlike the pigeon, which rests in a wire cage while not in the operant chamber, human beings live and interact with other human beings during nearly all of their waking hours. It would be impossible and undesirable to limit the social interactions of subjects while they were not in the experimental setting.
It is for these reasons that topographic definitions of generalization have arisen. While the functional picture is clouded and is likely to remain so, topographic schemes can give us much practical information about the real effects our interventions are having in the everyday settings in which our subjects operate. They can provide strategies for teachers and parents which can lead to improved social interactions with their disabled students and children, enhancing their quality of life.
There are reasons to consider both topographic and functional definitions of generalization. A functional analysis gives us more information in a strict behavior analytic sense, while a topographic analysis, though less “scientific,” gives us information which can help us as clinicians. In the literature review which follows, I will look at generalization in both the topographic and functional senses, calling the former (functional definition) generalization, and the latter (topographic definition) generality, a term proposed by Fox & McEvoy (in preparation).
Defining Maintenance
In the functional sense, maintenance is viewed as generalization across time. We can suddenly stop reinforcing our pigeon for pecking the illuminated key, and continue to count key pecks. Or, with extinction conditions prevailing, we can place the pigeon after an extended lapse of time back in the operant conditioning chamber and illuminate the key and compare the number of key pecks with the number which occurred during training.
Maintenance, as conceptualized in social skills training, is rather like the last example. Researchers, after a greater or lesser period of time, look to see whether the target behavior still occurs. However, here the picture becomes muddy, for it is often difficult to determine whether extinction conditions are or have been in effect. For the same reasons it is difficult to control variables in such a way that measurement of generalization is possible, uncontrolled variables make measurement of maintenance difficult. Moreover, researchers have not been consistent, either in their terminology or in their view as to what maintenance actually is. Maintenance has been called, among other things, persistence of behavior change, resistance to extinction, follow-up, and generalization across time (Fox & McEvoy, in preparation). Fox & McEvoy point out that some workers test for maintenance under conditions of reinforcement, or under partial conditions of reinforcement. Others test under conditions of extinction. Fox & McEvoy note that sometimes a broad definition will allow for findings that maintenance has occurred, while a more restricted definition would not.
As I am concerned primarily with generalization in this paper, I will attempt to circumvent the definitional quagmire by limiting my discussion to maintenance as a type of generalization in the purely functional sense. That is, I will define maintenance as a treatment effect which persists over time in the absence of experimental intervention. Social factors which may have served to maintain the behavior will be identified whenever possible.
Defining Social Skills Training
Social skills training is a broad topic. Social skills are required for every human interaction, from purchasing an item in a mini-mart or fast food restaurant to getting along with one’s peers at a work site. But what constitutes social skills training? Researchers can focus on the skills associated with functional tasks such as purchasing coffee, concentrating upon only those social skills necessary to successfully carry out the task being programmed (cf. Storey, Bates, & Hanson, 1984), or they can have as their primary focus the interactions themselves. Some special educators (e.g. Brown et al., 1983, Wilcox & Bellamy, 1982) have strongly advocated the former type of training, noting that it makes an immediate and observable difference in the individual’s life. Nonetheless, many of these types of studies only incidentally train social skills as they are necessary as part of the chain of behavior required to complete the functional task, and, while the researchers sometimes measure generalization of the task itself, they have not typically attempted to see if the social skills acquired generalize to other types of tasks. For this reason, I will concentrate on studies which have as their stated goal the training of social skills; however, I will discuss one example of a study which incidentally trained social skills, and compare it with a study which had as its purpose the training of social skills.
Storey, Bates, & Hanson (1984) investigated the ability of six adults with mental retardation to acquire and generalize skills necessary to purchase coffee in a restaurant. These researchers used a 46-step task analysis in which seven of the steps involved social interaction with the server (e.g., communicating order, communicating thank you, waiting for order). The other 39 steps involved the physical process of entering the restaurant, selecting a seat, preparing and drinking the coffee, disposing of the empty cup, and leaving the restaurant. The authors measured generalization in two other restaurants and at a fast-food establishment, and at follow-up in the original restaurant two to five months after the end of training. The training was successful for all six subjects and generalized for each of the subjects to the other settings (i.e., the subjects were able to successfully order and consume coffee). Training effects were maintained at follow-up two to five months later.
Storey, Bates, & Hanson (1984) did not attempt to measure the social skills they trained outside the context of purchasing coffee. There is some question, as noted by the authors themselves, as to the social validity of this study, as five of the six subjects were institutionalized, and the opportunity to purchase coffee was unlikely to present itself in their daily lives. The authors would transport the subjects to a section of town where they ordinarily did not and could not go in order to purchase coffee.
Matson & Andrasik (1982) is an example of an attempt to directly teach social skills. These researchers taught leisure-time social-interaction skills to adults with mental retardation. They noted that although early researchers had trained social skills in adults with mental retardation, generalization and maintenance had not been adequately addressed or assessed. They did cite one study (Matson & Earnhart, 1981), in which researchers found generalization of social skills was most likely to occur when at least part of the training was conducted in the natural environment. They noted that constraints on staff time limited work in natural settings, and speculated that training subjects to be more independent in the natural environment might increase the likelihood that staff would actually train them in natural settings.
In their first experiment, Matson & Andrasik (1982) attempted to see whether a self-reinforcement procedure was sufficient to reinforce appropriate verbal interactions. They used a multiple baseline design across subjects (Baer et al., 1968) with two experimental conditions, baseline and self-reinforcement. Subjects were two adults with moderate mental retardation. Behaviors measured included appropriate and inappropriate verbal behaviors. Subjects self-reinforced by giving themselves tokens whenever they thought their social behavior was appropriate. Tokens could be exchanged for edibles.
Data indicated a moderate increase in appropriate verbal interactions and corresponding decrease in inappropriate verbal interactions for one subject. The authors concluded self-reinforcement alone might not be sufficient to change verbal behavior, but might function as one component in a treatment package.
In a second experiment, Matson & Andrasik (1982) used a treatment package that added social skills training and self-management to self-reinforcement. They used a multiple baseline design across subjects; subjects were three adult persons with mental retardation. Following a baseline period, there was an attention condition in which the subjects were shown role-playing scenes. The subjects were reinforced for attending training and following the instructions and requests of the experimenter.
After the attention phase, a treatment package was implemented. Social skills training (which included instructions, information feedback, social reinforcement, and modeling, as in Matson, et al., 1980) and self-monitoring (in which subjects were taught to monitor their target responses) were added to self-reinforcement. All three subjects showed significant increases in appropriate verbal interactions and corresponding decreases in inappropriate verbal interactions when the treatment package was implemented. A third experiment examined the relative effectiveness of social skills training versus a combination of social skills training and self-management procedures.
The difference of the approaches of Matson and Andrasik and Storey, et al. (1984) is clear. Matson and Andrasik had as their goal the training of social skills, while Story, et al. trained functional skills which incidentally included social skill components. I have considered for purposes of this paper only those studies which have as their primary goal the training of social skills, and excluded those which incidentally teach social skills; that is, the study by Storey, et al (1984) would be excluded, but Matson and Andrasik (1982) would be included.
But what exactly are social skills? I have considered social skills to be any skill which involves direct social interaction with another human being. Training an individual to shake hands would, then, be considered social skills training by this definition, whereas training an individual to open a package would not.
Other Definitions
Because I am interested in studies of social skills training in adults with mental retardation, it is important that I define what an adult is, and what mental retardation is. After an initial examination of the literature, I decided to set the lower age limit for an adult at age 21. This is culturally appropriate for American society, but more importantly, it works empirically better than age 18 or any other age to differentiate studies with primarily adolescent subjects from those whose with adult subjects.
So I can differentiate studies with adolescent subjects from studies whose subjects are children, I have considered adolescence to begin at age 12.
I have considered any subject specifically identified as mentally retarded or mentally handicapped, or with measured IQs below 70, to be mentally retarded. I have excluded subjects identified as mentally handicapped as the result of mental illnesses (psychosis, etc.) from this definition, although I included subjects who were identified as both mentally ill and mentally handicapped.
The State of the Art in Social Skills Training
The operant training of social skills is an area in which generalization and maintenance are of particular importance. This is because skills are typically trained in one or at most a few settings. They are expected or hoped to transfer to new settings in which the original trainers are not present, and are hoped to endure after the experimenters have withdrawn, with the ever elusive “natural community of reinforcement” (Kazdin, 1975) maintaining the shaped behavior across both time and place. Resources are scarce, and the need for lasting change is great.
What is the state of the art in social skills training? What populations are being trained, and to what extent are generalization and maintenance being programmed, or even measured? Are topographic or functional definitions used? Do our interventions produce generalization or generality?
One approach to determining the state of the art of social skills training would be to examine a recent issue of a professional, peer-reviewed periodical which publishes reports of “cutting-edge” research. The Journal of Applied Behavior Analysis is a professional publication which is the central journal of the field of behavior analysis. The Summer 1992 issue of JABA (Volume 25, Number 2), contains a series of thirteen articles on improving social competence. The editors define social competence as “the effective and appropriate use of social behavior in interactions with an individual or individuals” (Odom & McConnell, 1992). Twelve of the articles are experimental analyses, and the thirteenth is a review of generalization and maintenance of social skills training studies of preschool children (Chandler, Lubeck, & Fowler, 1992). Certainly, taking a look at the articles the editors selected could shed some light on the state of social skills training, the subject populations most frequently studied, and the extent to which generalization and maintenance are programmed and measured.
Chandler, Lubeck, & Fowler (1992) surveyed twenty-two behavioral and nonbehavioral journals from 1976-1990, identifying articles that “a) provided training or consequences for behaviors during intervention; b) were successful at producing behavior change during intervention; c) included preschool children as subjects; and d) focused on peer interaction.” They identified 73 papers meeting these criteria. Fifty-one of these studies also measured generalization of social behavior, and were selected for analysis. The authors did a descriptive review, comparing the characteristics of the studies most successful in producing generalization and maintenance with characteristics of those studies which were least successful. They found a number of factors differentially related to success; for instance, those most likely to produce generalization addressed functional target behaviors, specified a criterion of fluency, used indiscriminable contingencies; used mediation techniques; used prompting, positive reinforcement, and feedback as behavior change strategies; assessed more than one dimension of generalization, used withdrawal designs; used multiple comparative measures; used baseline and normative data as comparative measures; and had a longer intervention. The authors noted that while it would be tempting to conclude that use of the practices they found to be more successful would lead to generalization, the effectiveness of those practices must be experimentally verified. They concluded it is time to specifically address questions of generalization in social studies with children. That is, generalization should be a focus of research rather than one of several variables which are measured.
Table 1: Social Skills Studies in JAMA, 1992, V. 25, No. 2
The other JABA articles are intervention studies, most of which use children with disabilities as subjects (eight of twelve; see Table 1). Ten of the twelve used children or adults with disabilities, their peers, or some combination of peers and persons with disabilities as subjects. Two studies took a more indirect route to measuring social skills, intervening with supervisors of preschools (Hundert & Hopkins, 1992) or asking community judges to rate the social skills of actors (Quinn, et al., 1992).
The studies used a variety of specific intervention tactics: setting events (Chandler, Fowler, & Lubeck, 1992); teaching sociodramatic scripts (Goldstein & Cisar, 1992); increasing social interactions (Hundert & Hopkins, 1992; Kamps, et al., 1992; Kern-Dunlap, et al., 1992); peer-mediation (Goldstein, et al., 1992; Stewart, Van Houten, & Van Houten 1992); teacher prompts (Odom, et al., 1992); a social network intervention (Haring & Breen, 1992); self management techniques (Koegel, et al., 1992); video-assisted training (Morgan & Salzberg, 1992); and judging of social skills training packages (Quinn, et al., 1992).
The twelve studies show a clear predominance of subjects with disabilities, and especially of children. Only two studies (Stewart, Van Houton, & Van Houton, 1992; and Morgan & Salzberg, 1992), used adults with disabilities as subjects, and only Morgan & Salzberg (1992) used adults with mental retardation.
To what extent did the twelve articles addresses issues of generalization and maintenance? Although Odom & McConnell (1992), in their introduction, stated that “the research agenda for the 1990s and beyond will be to establish a technology of generalization and maintenance for social competence interventions,” only one of the twelve (Stewart, Van Houten, & Van Houten, 1992) used the word generalization in the title; none used the word maintenance. Only Stewart, Van Houten, & Van Houten (1992) used either word as a descriptor in the abstract, but several studies measured or at least discussed generalization or maintenance.
Goldstein & Cisar (1992), using a multiple probe design, taught three sociodramatic scripts to six nondisabled children and three children with autistic characteristics; the children were grouped in triads consisting of one disabled child and two nondisabled children. Social interactions of the subjects were trained (via the scripts) after a baseline period during which the children were told how they were expected to behave, but were not given specific instructions. Following script training, there was a posttraining period which was almost identical to the baseline period. There was then a generalization condition, which was like the posttraining condition, but with children regrouped. Generalization, then, was measured across both settings and across peers. The authors found generalization in both posttraining and generalization conditions.
Stewart, Van Houton, & Van Houton (1992) were the only researchers in the 12 JABA studies who specifically set out to increase generalized social interactions. They trained mentally disabled adults to serve as peer therapists for other, more severely withdrawn adults. In a treatment phase called “peer therapy for generalization,” they prompted interactions with individuals other than the peer therapists. Follow-up data were collected at 1 month and 4 months after the study had concluded.
The authors found three types of generalization: disabled persons showed increased social interactions with persons other than the peer therapist; peer therapists showed increased social interactions with persons other than the target peer; and treatment effects were maintained after treatment was concluded.
Kamps, et al. (1992) taught social skills to children with autism. Specific skills trained included initiating and maintaining interactions, making conversation, giving and accepting compliments, turn taking, sharing, helping others, asking for help, and including others in activities. Following training, there was a feedback condition, during which children in free play were shown a printed form which marked their social interaction with paste-on stars. Follow-up one month after the feedback condition, again in free play, showed that both frequency of social interactions and time engaged in social interactions had maintained at the levels found in the feedback condition. Feedback on social interaction was provided to the entire group twice during each free play period.
Odom et al. (1992) taught nondisabled peers to engage in social behavior with children with disabilities, then systematically withdrew verbal prompts and visual feedback. Social interactions continued at intervention levels as prompts and feedback were faded, and throughout a maintenance phase. As in the baseline period, the teacher did not provide feedback during the maintenance phase, except as was necessary to keep order.
Haring & Breen (1992) developed a peer network in which nondisabled children would meet with children with disabilities in a support group-like setting, with an adult facilitator present. After group goals for the disabled students had been attained, the peer network was partially dismantled (i.e., frequency of support groups was reduced, data collection by peer participants was terminated, and teaching of specific skills was discontinued); this was the maintenance phase. Frequency of interaction and appropriate responding continued at intervention levels throughout the two-month maintenance period for one student with disabilities; scheduling difficulties prevented a maintenance phase for the second student.
Morgan & Salzberg (1992) studied the effects of video-assisted training on employment-related social skills in adults with severe mental retardation. In their first study, subjects were trained in a 5 x 3 m room and probed in various work settings. Training was in several phases: video discrimination training; video discrimination training plus self-model; self-model plus familiar supervisor; and self-model, familiar supervisor, and behavioral rehearsal. The researchers noted when generalization to the probe condition occurred. Follow-up 60 days after conclusion of the study showed a decrease in correct responses at follow-up for two of three subjects.
Two studies, then, measured one or more types of generalization (Goldstein & Cisar, 1992; Stewart, van Houton, & van Houton, 1992), and four had a follow-up or maintenance phase of some sort (Haring & Breen, 1992; Kamps, et al., 1992; Morgan & Salzberg, 1992; and Odom, et al., 1992). The other six studies did not discuss maintenance or generalization.
Not surprisingly, Stewart, Van Houton, & Van Houton (1992), who had specifically set out to produce generalized responding, were most successful at producing generalization; the other studies were more or less successful.
If these JABA studies are a microcosm of the state-of-the-art of social skills training, then it is clear we have not yet reached a level of sophistication in which all of our studies program or even measure generalization and maintenance, or in which maintenance and generalization are always successfully produced when measured. If, however, Stewart, Van Houton, & Van Houton (1992) is representative of studies which specifically seek to produce generalization, then perhaps we have reached the stage at which we can successfully program, measure, and (usually) produce generalization.
Reviews of Social Skills Training
Another way of determining the state of the art of social skills training is to examine reviews of research papers. I was able to locate eight such reviews of social skills training, of which seven contained at least one study with subjects identified as adults with mental retardation. They were Davies & Rogers, 1985; Harchik, Sherman, & Sheldon, 1992; Hollin & Trower, 1988; Jackson, King, & Heller, 1981; Robertson, Richardson, & Youngson, 1984; Singh & Winton, 1983; and Storey, 1987; see Table 2.
It should be noted that reviewers often do not specify all relevant details of the articles they cite. It was sometimes difficult to determine from the text the characteristics of the subjects, the methods used, and whether generalization and maintenance were measured. Similarly, it was not always explicit whether studies were single-subject in design, but in most cases it was possible to make a good guess. I indicate in the text and in the various tables when subjects were group studies; when I do not specifically mention the nature of a study, the reader can assume that it is a single-subject study.
I attempted to identify citations which were clearly identified as having adults with mental retardation as all or part of the subject population. For those articles, I searched the text for mention of generalization and maintenance, and indicated when they were measured, and, if possible, whether generalization and maintenance were found.
For sake of comparison, I included studies with children and adolescents with mental retardation in the tables for the various reviews.
I have used the authors’ judgements of who is and is not an adult. It should be noted the various authors identified subjects as adults, adolescents, or children, without defining their terms (although they often gave their ages). It is likely their definitions differed from mine.
Table 2: Summary of Review of Studies Training Social Skills with Adults
Hollin & Trower (1988)
Hollin & Trower (1988; see Table 3) did a global overview of training social skills in subjects with a variety of disabilities ranging from transsexualism to schizophrenia. Their section on mental disability cited several review papers. Although they did not review individual papers, they included discussion of several representative individual studies with adults with mental retardation (Andrasik & Matson, 1985; Matson & DiLorenzo, 1986; Robertson, et al; Matson & Earnhart, 1981; Matson & Andrasik, 1982; and Reese, Sherman, & Sheldon, 1984). Three of these specifically addressed generalization issues. Matson & Earnhart (1981) found a combination of two procedures (self-monitoring and prompting of target behavior in vivo) resulted in successful training and generalization of verbal behavior. A follow-up by Matson & Andrasik (1982) found a combination of self-monitoring and self-reinforcement was more likely to lead to generalization than social skills training alone. Reese, Sherman, & Sheldon (1984), who trained social skills in a group home setting, found self-recording was critical for maintenance of learned skills.
Of the remaining reviews, Harchik, Sherman, & Sheldon (1992) reviewed the use of self-management procedures in people with developmental disabilities. Wehman, Abramson, & Norman, 1977 examined transfer of training in behavior modification programs (no studies reviewed used adult subjects). Five papers, Davies & Rogers, 1985; Jackson, King, & Heller, 1981; Singh & Winton; 1983; Storey, 1987; and Robertson, Richardson, & Youngson, 1984) specifically reviewed social skills training in persons (of all ages) with mental retardation.
Table 3: Social Skills Studies Cited by Hollin & Trower
Davies & Rogers (1985)
Davies & Rogers (1985) reviewed social skills interventions in persons with mental retardation (Table 4). The authors divided the interventions into three categories: Motoric-Physical Social Skills (that is, physical behaviors such as hand waving or smiling); Social Skills and Behaviors (more complex behaviors which required understanding and awareness of social situations); and Social-Cognitive Skills and Behaviors (in which emphasis was placed on the cognitive components of social skills).
Davies & Rogers identified nine studies which trained motoric-physical skills, using a variety of strategies (Table 4). None of them were identified in the text as having adult subjects, although two studies did utilize adults (Lowther & Martin, 1980; and Edmonson & Han, 1983). Of these two, only Lowther & Martin (1989) addressed generalization.
In the second category, Social Skills and Behaviors, Davies & Rogers identified 14 studies. Nine studies were identified as having adult subjects and four as having children; the status of the remaining study (Zisfein & Rosen, 1974) is unclear from the text; however, Robertson, Richardson, & Youngson (1984) identified the subjects of this study as adults in their review (Table 4). In the ten studies (including Zisfein & Rosen, 1974) using adult subjects, nine addressed generalization (two in a limited way), and one did not. Davies & Rogers (1985) considered the fact that assessment of generalization was done in an analogue manner in some studies a weakness, and noted that generalization was only rarely measured in the natural environment; nor was the social impact of acquiring new skills generally evaluated.
In the category Social-Cognitive Skills and Behaviors, Davies & Rogers identified eight studies (Table 4). Four used adult subjects, two adolescents, and one children. It was not possible to tell from Davies’ and Rogers’ text or the title the subject population of the remaining study (Davies, 1982), which was a doctoral dissertation and so not readily available. All of the studies using adult subjects addressed generalization, and generalization occurred in all five of the studies.
Table 4: Social Skills Cited by Davies & Rogers
Davies & Rogers (1985) concluded the majority of studies in their review failed to examine or demonstrate generalization, noting that strategies for improving generalization to natural environments are lacking and need to be identified.
Although Davies & Rogers appreciated the importance of maintenance, they unfortunately did not mention which if any of the studies they reviewed provided maintenance or follow-up.
Harchik, Sherman, & Sheldon (1992)
Harchik, Sherman, & Sheldon (1992) reviewed the use of self-management procedures in persons with developmental disabilities. Six of the 54 studies they identified trained social skills. Four of these were comprised partially or totally of adult subjects (Table 5).
Table 5: Social Skills Studies Cited by Harchik et al.
It was not possible to tell from Harchik, Sherman & Sheldon’s text which, if any of the studies training social skills in adult subjects addressed generalization and maintenance, but the authors did mention maintenance was measured in only 28 of the 54 studies, and generalization in only 13 of the 54— this despite the fact that improving generalization and maintenance were given by the authors of the various articles as primary reasons for teaching self-management procedures. Harchik et al. further note that although generalization and maintenance were sometimes assessed, they were rarely programmed for or analyzed.
Jackson, King, & Heller (1981)
Jackson, King, & Heller (1981) identified 11 studies in which social skills were taught to persons with mental retardation (Table 6). Subjects of six of the studies were adults. Of these six studies, four tested for generalization, (Bates, 1980; Bornstein, et al., 1980; Perry & Cerreto, 1977; and Rycharik & Bornstein, 1979), but only two did so in the natural environment (Bates, 1980; and Perry & Cerreto, 1979). Four of the 11 studies provided follow-up ranging from 3 weeks to 6 months, but only one of the adult studies addressed the issue of maintenance (Bornstein, 1980).
Jackson, et al. concluded “researchers have not taken deliberate steps to enhance maintenance and generalization of social skills.”
Table 6: Social Studies Cited by Jackson et al.
Robertson, Richardson, & Youngson (1984)
In a review published in the British Journal of Clinical Psychology, Robertson, Richardson, & Youngson (1984) located 20 single-subject studies in which social skills were taught to persons with mental handicaps (Table 7). Nine of the 20 used adult subjects with mental retardation. Of these, six measured generalization (four in natural environments), and four included follow-up at periods ranging from four weeks to six months.
Robertson et al. (1984) identified 16 group studies, of which 10 had a subject population of mentally handicapped adults. Some of the subjects may not have met the criterion for mental retardation; in several studies the mean IQ was above 60, and in one study, only 66% of the subjects were identified as mentally handicapped. Generalization was measured in six of the ten studies, and was found to have occurred in five of the six. Maintenance was measured in five of the ten, and occurred in four of five cases.
Robertson et al. found the single-subject studies more effective in demonstrating a treatment effect, and noted that those studies with follow-up demonstrated maintenance, but noted that more work is needed in demonstrating the social relevance and generalizability of behavior change.
Table 7: Social Skills Studies Cited by Robertson et al
Singh & Winton (1983)
Singh & Winton (1983) examined existing studies in which institutionalized children, adolescents, and adults with severe and profound mental retardation were taught social play (Table 8). They concluded that individuals with severe and profound mental retardation can be taught social skills and “can sometimes even assist the development of these skills in their peers.” They noted that although the studies they examined in their survey were for the most part methodologically sound in terms of their basic design, maintenance and follow-up were often lacking, and that although some studies measured generalization effects, only a few actively trained for generalization.
Singh & Winton located 22 studies, all published in the 15 years preceding their report (1967-1982) (Table 8). (Although they claim 23, I could only find 22 in the text). Only five had as subjects adults with mental retardation (Wehman, et al., 1976; Wehman, 1977; Samaras & Ball, 1975; Nietupsky & Svoboda, 1982; and Dy, et al., 1981. Just two of the 22 (Cone, et al., 1978; and Morris & Dolker, 1974, neither with adult subjects) included data on maintenance, and only 14 of the 22 studies tested for or trained generalization (it is not possible to tell from the text of Singh & Winton’s review which of the 22 assessed generalization). Among the five studies with adult subjects, Samaras & Ball (1975) and Nietupsky & Svoboda (1982) provided anecdotal evidence of generalization, but only Dy, et al., (1981) actively measured generalization. None of the studies with adult subjects programmed or measured maintenance.
Table 8: Social Studies Cited by Singh & Winton
Storey (1987)
Storey (1987) reviewed social skills training of persons with severe and profound handicaps in community settings (Table 9). He excluded studies which taught social skills in school or segregated sites without assessment of generalization, but included studies which taught primarily non-social skills, but with social skills components (as in interacting with a clerk when making a purchase). He found nine studies which met his criteria. All measured generalization across settings. Maintenance was assessed at periods ranging from 15 days to five months. Only three of the nine studies did not address maintenance. Unfortunately, only one of the studies (Storey, et al., 1984) used an adult subject; this study did not address maintenance.
Like the authors of the other reviews, Storey noted it is important that researchers focus on issues of generalization and maintenance in the future.
Table 9: Social Skills Studies Cited by Storey
Discussion of the Reviews
The seven reviews yielded a total of 102 studies of social skills studies with persons with mental retardation, dating from 1968 to 1988. Of these, 54, or 53% used adult subjects. Thirty-seven of the 54, or 69% were identified by the various reviewers as having addressed generalization, either anecdotally or in a more formal fashion, and 10 of the 54, or only 18% were considered by the authors to have addressed maintenance. It should be noted that two of the reviews (Davies & Rogers (1985); and Harchik, Sherman, & Sheldon, 1992) did not address the issue of maintenance.
Obviously, some of these reviews utilized the same studies. Of the 54 studies with adult studies identified in the seven reviews, 38, or 70% were distinct (Table 10). Of these thirty-eight distinct studies, 28, or 74% were identified as having addressed generalization, but only 7 (18%) were identified as having addressed maintenance.
When one compares the percentage of studies (42%) addressing generalization in adults in the reviews to that found by Chandler, Lubeck, & Fowler (1992) in their survey of articles from 1976 to 1990, in which 51 of 73 or 70% of studies of social skills training in children measured generalization, results are surprisingly similar.
From this summary of seven reviews, it would seem that despite a widespread understanding that it is important to program and measure maintenance and generalization, more than one-quarter of studies do not program or measure generalization, and most do not program or measure maintenance. Moreover, some reviewers do not discuss issues of maintenance in their papers.
Current Review of the Literature
I have reviewed the literature, searching for studies with adult subjects with mental retardation which train social skills. I located papers by searches of computer databases, by examining the reference sections of papers thus located, and by spot-searching several journals. Additionally, Dr. Jim Fox provided me with his own materials which listed a number of papers which addressed issues of maintenance and generalization in both theoretical and empirical manners.
Table 10: Master Table of Social Skills Studies with Adults Subjects
Method
I was interested in locating papers reported on the attempts of researchers to train social skills to adults with mental retardation, and papers which reviewed such studies. I searched several computer databases, including PsychLit and ERIC, with a start date of 1970 and an end date of 1993, using the keywords “social skills training” and “social skills & mental retardation.” At least 500 “hits” were saved to disk; I then sorted the file into three categories: 1) articles which were clearly not appropriate according to my definition of the subject population (i.e. social skills training of World War II veterans or of chronically shy persons); 2) articles which were clearly appropriate (i.e. “Teaching appropriate social behaviors to a young man with moderate mental retardation in a supported competitive employment setting; Wheeler, Bates, Marshall, & Miller, 1988); and 3) articles in which either the subject population or type of training were unclear from the title and/or abstract. I located additional articles by examining the reference lists of at least 50 papers located in the Eric/PsychLit searches, and spot-checked back issues of several journals, including the Journal of Applied Behavior Analysis, Mental Retardation, and Journal of the Association for Persons with Severe Handicaps. I then obtained copies of all articles from Category 1 not cited in the seven review papers (Davies & Rogers, 1985; Harchik, Sherman, & Sheldon, 1992; Hollin & Trower, 1988; Jackson, King, & Heller, 1981; Robertson, Richardson, & Youngson, 1984; Singh & Winton, 1983; and Storey, 1987) by visiting libraries at Emory University, Georgia State University, and Vanderbilt University, and by use of the Interlibrary Loan Department at Georgia State University.
I located eight reviews of social skills studies, of which seven included papers with adults with mental retardation as subjects; these reviews are discussed in the previous section of this paper. I removed papers reviewed in any of these seven papers from Category 1. I additionally removed any papers in which it was clear from discussion in other papers, or from examination of the paper itself, that the subject population did not consist predominantly of adults with mental retardation, and/or in which social skills were not trained.
I obtained the Category 2 articles I thought most likely to fit Category 1 criteria in the same way I had obtained Category 1 articles. Those which met the criteria of having adult subjects and having trained social skills were moved to Category 1, and those which did not fit were removed to Category 3. Additionally, studies which sounded from their title as if they fit in Category 1, but did not, were moved to Category 3.
Final Selection
I identified fourteen articles which had not been reviewed elsewhere, and which met the criteria of having exclusively or predominantly adult subjects with mental retardation, and in which social skills were trained. The final cut for Category 1 is listed in Table 11.
The review which follows differs from the reviews of Davies & Rogers, 1985; Harchik, Sherman, & Sheldon, 1992; Hollin & Trower, 1988; Jackson, King, & Heller, 1981; Robertson, Richardson, & Youngson, 1984; Singh & Winton, 1983; and Storey, 1987 in that my primary interest is determining whether and how generalization and maintenance were programmed and measured in these studies of social skills training. Generalization was addressed in all seven reviews, and maintenance in five, but the primary focus of the reviews was on the social skills training itself, and not maintenance and generalization issues.
In the review which follows, I will first describe the subjects, experimental design and the most important procedures used in the fourteen studies, and briefly discuss the results and conclusions of the authors, especially with regard to maintenance and generalization. I will then ask the following questions about each article: 1) Did the authors attempt to measure generalization and maintenance? 2) How? 3) What were their findings? In the Discussion, I will address 1) What the review tells us about the effectiveness of operant training of social skills in persons with mental retardation; and 2) How the papers reviewed compare to the issues raised by Baer, Stokes, Holman, Fowler, & Rowbury (1981) McEvoy & Fox (in preparation), and others about what should be happening with regard to maintenance and generalization.
Table 11: Articles Selected for Review
The Review
Agran, Salzberg, & Stowitschek (1987)
Agran, Salzberg, & Stowitschek (1987) investigated the effectiveness of a training program for social skills on five moderately to severely mentally retarded adult employees of a sheltered workshop in Utah. Training took place at the workshop, but in a room separate from the work setting.
The authors selected as dependent variables (target behaviors) initiations with a supervisor when the subject ran out of work; and initiations with the supervisor when the subject needed assistance. Supervisors were asked to rate the social abilities of the subjects on an ongoing basis.
The experimental design was a multiple baseline across subjects and responses. Following a baseline period in which the subject received no feedback or instruction, there was a period of social skills training, followed by a period in which the training was partially withdrawn, and then a maintenance period. A fifth period was introduced for one subject who did not show target behaviors in the generalization setting (the actual work setting). Maintenance data were collected for 13 weeks.
Agran et al. (1987) found that for four of the five subjects there were “rapid and dramatic” gains in both training and generalization settings in the first target behavior during the training period. A fifth subject, the last to remain in the baseline period, showed spontaneous improvement in both settings for the first target behavior. This spontaneous change did not occur for the second target behavior.
Chadsey-Rusch, Karlan, Riva, & Rusch (1984)
Chadsey-Rusch, et al. (1984) attempted to train conversational skills to three men with mental retardation in a competitive employment setting. Subjects were selected because of poor conversational skills and lack of interaction with their co-workers. Subjects were trained to ask questions, and other conversational behaviors (topic initiations and topic continuations) were measured. The experimental design was a multiple-baseline across two subjects, who were given prompted training after a baseline. An ABABC design was used for the third subject; this was done to examine the effects of prompted and unprompted training. Prompted training differed from unprompted training in that subjects were periodically prompted to ask a question during a 5-minute social skills training period.
For the third subject, a trainer followed a structured conversational format during a baseline period, giving no feedback or instructions to the subjects. After an unprompted training phase, there was a second baseline, which was followed by a second unprompted training phase and then a prompted training phase. The remaining two subjects received prompted training following a baseline period.
Results indicated that prompted training was effective in increasing the frequency of conversational questions without negatively affecting the conversational flow. No attempt was made to program or measure generalization or maintenance, nor did the authors even mention generalization or maintenance.
Fantuzzo, Wray, Hall, Goins, & Azar (1986)
Fantuzzo, et al. (1986) systematically replicated studies by Foxx and his colleagues (Foxx, McMorrow, & Mennemier, 1984; Foxx, McMorrow, & Schloss, 1983) to teach a modification of the Sorry board game to three single mothers with mental retardation who had been identified as child maltreaters. Parenting skills and social skills were trained, with generalization probes and follow-up at two weeks. Rather than presenting data for each subject, the authors collapsed the data for the three subjects into one figure and two tables.
During a baseline phase, a facilitator read situation cards to subjects, who, after responding, moved a marker the number of places indicated on the card. The facilitator modeled and elaborated appropriate responses.
In the Parent Skills: Treatment Series I phase, a deck of “best answer” parenting skills cards was introduced. These cards indicated the correct responses to the parent situation cards used in the baseline. After their response, the facilitator gave verbal feedback to subjects, using the second set of cards as a guide. When subjects answered incorrectly, or after 10 seconds, they were not allowed to move their pieces. Correct and elaborated responses resulted in a gold star being awarded and movement of the marker on the gameboard. Correct but non-elaborated responses resulted in a gold star and a limited (2 spaces) move of the marker. Additionally, individual and group performance criteria were set above the baseline level.
The Social Skills: Treatment Series II phase, was the same as the Parent Skills: Treatment Series I phase, with the exception that a deck of “best answer” social-skills cards was introduced.
A pregeneralization and four post-generalization tests were conducted two weeks before the baseline and one month after the termination of treatment. During home visits, an individual not otherwise involved in the experiment read the situation cards to the subjects and asked them what they would do in each situation. No feedback or reinforcement was given.
Two weeks after postgeneralization tests, there was a follow-up game session at the experimental site; the session was identical to baseline sessions.
The post-generalization and follow-up tests indicated treatment gains were maintained in the absence of treatment for 2.5 months following training.
Foxx, McMorrow, Bittle, & Ness (1986)
Foxx, et al. (1986) set out to measure the generalization of social skills taught to six elderly, female, mentally retarded residents of a community-based facility. They were especially interested in investigating the finding by Foxx, et al. (1984) that generalization of social skills to the natural environment might not occur until some time after the conclusion of training. They were also interested in determining how the presence of untrained peers would affect the social interaction of subjects who had been training in a generalization setting.
After a preliminary interview, the subjects were divided into matched pairs and placed in two groups according to their response to training situations. Subjects in both the training and contrast groups were presented with some of the situations from the Stacking the Deck program (Foxx & McMorrow, 1983); no feedback was given.
The authors measured the social interactions of the subjects in two generalization (natural) settings. In one setting, untrained peers were also present. After a baseline period, the training group was given the Gender Social Skills program from Foxx, et al. (1983), while the contrast group played the Sorry game from which the Stacking the Deck game was modified. Data were collected on six social behaviors used by Foxx, McMorrow, & Schloss (1983). The responses of subjects to 48 training situations were also scored.
For the training group, the percentage of correct responses to game situations increased during training. The groups did not show a difference in the level of appropriate social interactions during training, but in a follow-up, two of the three subjects in the training group showed an increase in appropriate interactions. An appropriate pattern of interaction seemed to increase as follow-up continued. The contrast group did not shows these types of changes. Neither group showed the use of trained skills with untrained peers.
The authors felt their finding of delayed generalization of social skills might explain why other studies had not found generalization of trained skills; furthermore, their finding that generalization does not occur in situations with untrained peers might further explain the failure of studies in which follow-up occurred in settings like day rooms, in which there are untrained peers.
Foxx, McMorrow, & Mennemeier (1984)
In an attempt to determine why social skills research had failed to reliably demonstrate generalization, Foxx, McMorrow, & Mennemeier (1984) used a multiple baseline across groups design. Six high-functioning adults with mild to moderate mental retardation were matched into two groups by predetermined criteria; I was unable to find in the text any way in which these groups were different in any way or any way in which their experimental treatment differed.
Target behaviors were in six areas used by Foxx, McMorrow, & Schloss (1983). The subjects were videotaped while working on a structured task, and later evaluated for appropriate social interactions; the productivity of each group was also measured.
A facilitator played the modified Sorry game with the subjects in one of the groups. During baseline, players could move their pieces whether or not they made a correct response. No feedback was given.
During a training phase, players were allowed to move their game pieces only when they responded correctly. The facilitator provided feedback, and subjects self-monitored their performance and later graphed it. When each subject reached a predetermined criterion, he or she received a reward.
In a third phase, generalization across trainers was measured. The facilitator was replaced by the workshop supervisor of the subjects. There were additional generalization tests. During a pre-training test and a post-training test, the subjects acted out situations with peers who were not involved in the experiment. Also, the social interaction of the subjects was measured from a videotape. Finally, there was follow-up once per week for one month.
Correct responses in the game situation showed an increase over baseline in the training phase in all six target areas. Scores continued to improve when the facilitator was replaced by the workshop supervisor. Both groups showed a higher rate of correct game responses in the post-training generalization test than in the pre-training test. Group productivity was variable, as was appropriate social interactions between residents. Group One’s level of appropriate interactions was lower during the third (supervisor) phase than during the training (facilitator) phase; Group two showed little change. Four of the six subjects showed their highest level of appropriate interactions during the follow-up phase, although the authors note that their responding was variable.
Foxx, McMorrow, & Mennemeier (1984) concluded their intervention resulted in an increase in appropriate social responses to relevant vocational situations. They noted that a collateral increase in the number of words per response suggests that the responses of the subjects had become more complex.
The real purpose of this experiment, however, was to determine whether the skills they trained generalized across settings. Although the results of the post-training test were significantly higher than the pre-test, variability within and between subjects resulted in a judgement of the levels of appropriate social behavior during the videotaped work session “equivocal;” that is, generalization did not occur in all of the subjects. The authors noted that some subjects appeared to have increased their social behaviors, while others did not. The fact that the subjects were grouped resulted in an inability to clearly demonstrate generalization. The authors concluded that it did seem clear that the social behavior of the subjects were interdependent. They noted that subjects with long histories of inappropriate social behavior can confound results of studies like theirs and suggested that including peers during generalization sessions might be a productive strategy.
Garris & Hazinski (1988)
Garris & Hazinski (1988) noted that a shortcoming of single-subject studies was little or no generalization or maintenance of the target behavior. They used a group study to measure the generalization of social skills training on 42 adults with mental retardation. Target behaviors were conversational questions, self-disclosing statements, positive conversational feedback, eye contact, and appropriate affect. Group I received social skills training, Group II reviewed social skills training and self-monitored videotape feedback, and Group III was a nontreatment control group. Subjects were randomly assigned to the three groups.
Training for Group I consisted of instructions, modeling, role play, and feedback. Subjects were trained to ask conversational questions, make self-disclosing statements, and to give positive feedback in conversation. They were also trained to make appropriate eye contact and show appropriate affect. Group II received this training, and were allowed to view a videotape of their role-lay performance. Subjects were instructed to tell the rest of the group whether specific behaviors had occurred. They were then given feedback by the instructor and their peers.
Generalization was measured by rating the interaction of subjects in an conversation with an unfamiliar college student who was blind to the nature of the study. Generalization sessions were videotaped and rated by independent observers.
All subjects in Groups I and II reached predetermined levels of social interaction during training sessions. However, these findings did not generalize to the conversations with college students. The authors concluded that social skills training and social skills training augmented with self-monitored videotape feedback were insufficient to cause generalization to novel persons. They were surprised by this, for they had made “a major effort” to program for generalization, “using multiple exemplars for a variety of settings, encouraging several conversational exchanges, and using a variable training format which allowed the subjects to initiate role-play responses related to their interpersonal situations.” More generally, they concluded that social skills training “may not be an effective training procedure to teach mentally retarded persons normative social functioning in natural settings.” Certainly, other experimenters in this review reached different conclusions.
Jackson & Martin (1983)
Jackson & Martin (1983) were interested in determining whether a 30-year-old female with diagnosed chronic schizophrenia and mild-moderate mental retardation could benefit from social skills training. The study was conducted in a small room. The experimental design was a multiple baseline across behaviors. Following a baseline period in which the subject was rated on her response to role-play situations, social skills training was introduced. In this phase, scenes consisting of an audiotaped narrative of a situation and a prompt by a male or female role model were presented to the subject. Following the subject’s response, she was given feedback and appropriate and inappropriate responses were modeled for her. Follow-ups were done three, five, and six months following the completion of the training phase. Conditions during follow-up were the same as for the baseline.
Results showed an increase in eye contact, facial gestures, verbal content, voice volume, and overall skill in the social skills training phase. Treatment gains were maintained at all follow-ups. There was an attempt to measure generalization; generalization and novelty scenes supplemented the training scenes. Generalization scenes were presented during the social skills training phase, presumably without the prompts provided for training scenes (this is unclear from the text). Novelty scenes were used as assessment tools, and were presented only at the 13th and 14th training sessions and at follow-up. The authors noted performance increments on both generalization and novelty scenes.
Lalli, Pinter-Lalli, Mace, & Murphy (1991)
Lalli (et al.) replicated and extended the study of Foxx, McMorrow, Bittle, & Ness (1986). Subjects were residents of a group home. Procedures differed from Foxx, et al. (1986) in several ways: a) the specific skill deficits of the subjects were identified before training; b) all residents in the generalization environment were trained; c) the training focused on subject-subject dyads; and d) the subjects were required to participate in a role playing exercise. Subjects were divided into groups. The experimental design was a multiple baseline across groups. After baseline, there was a training phase during which generalization probes were taken, and then a follow-up phase which lasted for six weeks (Group I) and three weeks (Group II).
Results were consisted with those of Foxx et al. (1986); social interactions increased, and were more frequent in the presence of trained peers. However, Lalli, et al. (1991) found an immediate improvement in social interactions rather than the delay found by Foxx et al. They suggest that this may have occurred because they trained all individuals in the residence. They further note that by selecting target behaviors which were relevant to and supported by the natural environment, generalization may be facilitated.
Misra (1992)
Misra (1992) used a multiple-baseline design to study the effects of self-monitoring on generalization and maintenance of social skills in three adults with mild mental retardation. There was a baseline period, after which the subjects received individualized social skills training. After a second baseline, the subjects were taught to monitor their performance in a training and in a generalization setting. Self-monitoring was then faded and a maintenance phase begun.
The social skills training resulted in an increase in appropriate social skills in the training setting, but they did not generalize consistently until self-monitoring was introduced. The author concluded that self-monitoring can be of important in the generalization of social skills to natural settings.
Mueser, Valenti-Hein, & Yarnold (1987)
Mueser, Valenti-Hein, & Yarnold (1987) were interested in improving the heterosexual competence of adults with mental retardation. Their subject population initially consisted of 41 men and women between the ages of 18 and 51, but 15 subjects were discontinued before the conclusion of the study. Subjects were randomly divided into three groups (but balanced for gender; this was, after all, a study of heterosexual competence): traditional problem solving, flexible problem solving, and relaxation training.
Before the initiation of treatment, subjects engaged in a series of role plays; role play also occurred after the last session and one month following the end of the experiment. Five of the six role plays cast opposite sex characters, and concerned introducing oneself, asking for a date and being refused, asking for a date and being accepted, responding appropriately to a compliment, and mutually deciding where to go on a date. The sixth role play cast same-sex actors, one of which pestered the other about details of a date. This sixth role play was not included in ratings.
A variety of measurement strategies were used. A 20-item test on knowledge of dating skills was given subjects before and after training. Also, ratings by the partner of each subject were performed before treatment and at follow-up. Additionally, subjects in all groups were taught to self-monitor and record their social interactions. Finally, the social interactions of subjects were measured during a refreshment break held during each session.
The three treatment groups were given similar problem situations. Subjects in the flexible problem-solving group were encouraged to examine a variety of solutions to the problems, which were then discussed, modeled by the trainer, and rehearsed with other subjects (Spivack & Shure, 1984). Subjects in the traditional problem-solving group were also encouraged to examine alternative solutions, which were modeled by the trainer, discussed, ranked, and rehearsed by the subjects. The primary difference between the two groups was that the flexible problem-solving group placed more responsibility on the subjects. The relaxation training group was given relaxation training during the initial phase of each session. Following this, subjects in the group were given problem situations and asked to discuss what in the situations made them nervous. The situations were then modeled by the trainers, and subjects acted out the situations.
The authors found subjects in all three groups showed significant improvement in their dating skills. However, this improvement did not persist at follow-up for subjects in the relaxation group.
Although Mueser et al. (1987) did a follow-up, they made no attempt to program or measure generalization.
Realon, Favell, Stirewalt, & Phillips (1986)
In two experiments, Realon, et al. (1986) taught individuals with mental retardation to provide multiply handicapped persons (who were the clients of the subjects) with interactions and materials during leisure
In the first experiment, the subject was a 24-year-old woman who was diagnosed as severely retarded. Data were collected on verbal interaction (talking with other residents; physical interaction (holding hands, patting, or rubbing residents on the back for more than three seconds); providing leisure materials; wheelchair rides (pushing a resident in his or her wheelchair); and providing music. During baseline, one of the authors gave the subject verbal prompts to use the target behaviors to interact with residents. There was then an intervention phase, in which one 15-minute period of her two-hour work assignment was used to prompt and reinforce her interactions with her clients. The subject was directed to engage in the target behaviors and asked to repeat her work duties. She was given praise for every 30 seconds of unprompted interactions with her clients, and then verbally prompted and reinforced for returning to work. The instructor modeled appropriate behavior when the subject did not work. When she reached a pre-determined performance standard, descriptive social praise and prompts were gradually reduced to one prompt per training session, and then training sessions were discontinued and replaced with brief probes, which were done every fifteen minutes or so throughout her two-hour work period.
Following intervention, there was a return to baseline. There was then a maintenance phase, in which probes were conducted about every 20 minutes during a work period in her home living unit. She received a token and descriptive social praise if she was working when a probe was conducted; otherwise, she received a fine and an explanation of why she received it. Finally, follow-up sessions were done at one, four, six, and twenty-four months. During follow-up, unit staff used the reward/fine system previously used during the maintenance period.
Results showed an increased level of interaction during the training phase over baseline, which was maintained (but at a slightly lower rate) during the second baseline period. Social interaction slowly increased during the maintenance phase. The follow-up indicated that treatment gains were maintained two years after termination of the study.
The results suggest it is possible to increase and maintain with minimal supervision the social interactions of an individual with severe mental retardation working with multiply handicapped individuals.
In a second experiment, three women diagnosed as severely retarded, and one man diagnosed in the mild-moderate range of retardation served as subjects. All four subjects had additional handicaps including severe scoliosis, spastic paresis, deafness, and all had histories of problem behavior. The setting was similar to that in Experiment 1.
The experimental design was a multiple baseline across subjects. Following a baseline identical to that in Experiment 1, intervention was sequentially begun for the subjects. When a predetermined performance standard was reached, reinforcement and prompts were faded. There was a single follow-up done 12 months after training for two of the subjects; the remaining two subjects did not receive follow-up because they had become involved in other vocational and residential programs.
Three of the four subjects showed low levels of social interaction during the baseline period; the social interactions of the four subject were greater, although highly variable. Following intervention, subjects showed increased levels of social interaction which continued throughout the duration of the study. Results support the conclusions of the authors in Experiment 1 that it is possible to increase and maintain with minimal supervision the social interactions of an individual with severe mental retardation working with multiply handicapped individuals.
It is interesting the authors used in Experiment 1 the phase maintenance for what might more accurately be considered to be a test of generalization, since it tested whether the target behavior occurred in an alternate setting.
Stowitschek, McConaughy, Peatross, Salzberg, & Lignugaris/Kraft (1988)
Stowitschek et al. (1988) were interested in looking at work-related social skills in an in vivo (non-institutional) setting. They chose as a setting a work-activity center in which clients did piece-work for reimbursement. Their subjects were ten adults with mental
Four social behaviors were chosen as target behaviors: appropriately saying thank you; appropriately saying please; not interrupting others; and appropriately saying excuse me. Social situations to measure the target behavior were set up by the experimenters.
A group design was used, with clients serving as their own controls. Two of the four target behaviors were randomly selected as “trained” and two as “not trained.” Training consisting of a review of the rules and daily reminders for the four target behavior was given (Phase I); this was followed by a second training phase during which subjects were asked to repeat the reminders (Phase II). There were a pre-test and four post-tests to measure the target behaviors.
The authors found that reminders of rules for interaction by a supervisor increased the appropriate social interactions of the subjects (Phase I training). The effect was more pronounced when the subjects repeated the rules (Phase II training).
Stowitschek et al. (1988) did not program or measure generalization or maintenance.
Van Hasselt, Hersen, Egan, McKelvey, & Sisson (1989)
Van Hasselt, et al. (1989) used as subjects two deaf-blind young adult males with intellectual abilities estimated to be within the severe range of mental retardation. Using a withdrawal design within a multiple baseline design, these researchers attempted to increase the amount of time which the subjects were on-task (manipulating a leisure-type object) and the amount of time in which the subjects engaged in social interaction. After baseline, there was an experimental phase in which each subject was reinforced for being on-task, followed by a second baseline, which was in turn followed by reintroduction of reinforcement. During this last phase, in multiple baseline fashion, social interactions began to be reinforced for each subject. Results shows higher levels of on-task behavior during the first treatment phase, followed by a return to baseline levels (or below) during the second baseline. On-task behavior once again increased during the second training phase. When social interactions began to be reinforced, the level of social interactions increased for both
The authors were quite self-congratulatory in that theirs were the first data showing that the social behavior of multiply handicapped deaf-blind individuals could be improved. They mention, however, that their study should be considered preliminary in that they did not program or measure maintenance and generalization.
It should have noted that during the study, both subjects, who were 21 years old, were in the process of moving from their residential institution to a community residence and vocational placement. Their involvement in the study was prematurely terminated when they left the institution.
Wildman, Wildman, & Kelly (1986)
Wildman, et al. (1986) attempted to assess the effectiveness and validity of conversational-skills training with adults with mental retardation in an outpatient setting. Their seven mild and moderately retarded subjects were randomly paired into conversational dyads, with the constraint that the same subjects would not be successively paired. The odd-numbered subject was attached to one of the existing dyads.
During a baseline period and after the training sessions which followed, the dyads were isolated and told to “get to know each other better.” Audiotapes were made of their conversations. In the training phase, subjects were given thirty minutes of conversational-skills training twice per week as part of a group. The design was a multiple baseline across the three target behaviors (appropriate question asking, complimenting the partner, appropriate self-disclosure).
The audiotapes of the subjects’ conversations were rated by college undergraduates. Generalization data were collected by making new dyads of subjects and nonretarded volunteers at the end of each training session and at follow-up and audiotaping their conversations. Social validity was measured by asking volunteers unfamiliar with persons with developmental disabilities questions about the social desirability of the subjects. There was follow-up at one, three, and six months.
The authors concluded that the conversational-skills training package they used can increase the frequency of appropriate conversational behaviors of the subject population. They noted that the improvements generalized to nonretarded unfamiliar peers, and were maintained over the six-month follow-up period.
Analysis
An easy way to determine whether the authors were interested in generalization and maintenance was to look at the titles. Four of the fourteen mentioned generalization. The abstracts of six of the remaining ten mentioned generalization. Only Realon et al. (1986) programmed generalization without mentioning it in the abstract or title, and only Stowitschek, et al. (1988) mentioned it in the abstract without programming it (they noted in the text that staff limitations made generalization training unfeasible).
In ten of the fourteen studies reviewed, the authors programmed and measured generalization; this rate of 71% is strikingly similar to the 70% rate (51 of 73 studies) found by Chandler, Lubeck, & Fowler (1992) in their survey of social skills training in children from 1976 to 1990; and to the overall percentage of studies (74%) I found for the pooled studies form the seven reviews. Eight of the ten, or 80%, successfully produced generalization.
All ten of the studies which addressed maintenance showed it was successfully produced. The number of studies addressing maintenance (10 of 14, or 71%) is significantly higher than the 7 of 38 (18%) of the pooled studies for the seven reviews which were identified as having addressed maintenance. It should be remembered that two of the reviewers (Davies & Rogers, 1985; and Harchik, Sherman, & Sheldon, 1992) did not address maintenance, making this number no doubt an underestimate.
The two reviews which did not address maintenance had between them a total of eleven studies which were not reviewed by one or more of the other five reviewers. If these are removed from consideration, then 7 of 27, or 26% of studies addressed maintenance, as compared to ten of fourteen, or 80% of the studies I reviewed.
The authors of the papers I reviewed, then, would seem to have addressed issues of generalization at a rate about the same as reported in previous reviews, and maintenance at a rate much higher. Most realized the importance of both generalization and maintenance, and built probes into their experimental design.
What types of generalization did the authors program and measure, and what were their findings? Of the ten studies in which generalization was programmed and measured, three measured generalization across settings, two generalization across three generalization across both settings and persons, and two generalization across conditions. All tests for generalization were positive, with the exception of Garris & Hazinski (1988), who measured generalization across persons, and Foxx, et al. (1984), who were unable to clearly demonstrate generalization across settings and persons.
Agran, et al. (1987) measured generalization across settings; they were interested in seeing if the social skills they had trained in a special setting generalized to the workplace. Target behaviors were directly measured in the generalization setting. Maintenance was measured by collecting data in the generalization setting under natural conditions of reinforcement (i.e., treatment was not in effect). Findings were positive for both generalization and maintenance.
Fantuzzo, et al. (1986) measured generalization across settings by administering pre- and post-tests in which a confederate would go to the home of the subjects and determine their responses to the card-based situations used in the training setting. There was a follow-up two weeks after the last of the postgeneralization tests; the subjects returned to the training setting and were presented with the situations used in training, under baseline conditions. Findings were positive for both generalization and maintenance.
Foxx, et al. (1986) measured the generalization of the social interactions of their subjects across both settings and persons. There were two generalization settings; in one, there were untrained peers. There was follow-up weekly for 15 weeks; this occurred because the authors were specifically interested in whether there would be delayed effects of their training. Findings were positive for both generalization and maintenance.
Foxx, et al. (1984) also measured generalization across both settings and persons. Generalization settings were a simulated workshop and the actual workshop. The authors found that the game skills they had trained had generalized, and that skill improvement continued when a facilitator was replaced by a workshop supervisor, but judged that changes in their workshop productivity and social interactions were equivocal.
There was follow-up weekly for one month in the workshop (one of the generalization settings), in which data were collected. Treatment effects were maintained at follow-up.
Garris and Hazinski (1988) measured generalization across persons by replacing one-half of a dyad with a novel partner, but generalization was not shown to have occurred. Maintenance was not measured.
Jackson & Martin (1983) measured generalization by presenting their subjects with alternative training scenes (generalization across conditions). They determined that generalization had occurred. They did follow-up under baseline (non-reinforcement) conditions at three, five, and six months, demonstrating that their training effects persisted.
Lalli, et al. (1991) measured generalization across conditions by conducting generalization probes under baseline (non-reinforcement) conditions. They demonstrated that generalization had occurred. Training effects were maintained at follow-up probes conducted for up to six weeks.
Misra (1992) found generalization of training effects occurred across settings and persons, and training effects persisted during a two-week follow-up.
Mueser, et al. (1987) did not program or measure generalization, but found treatment effects in their experimental groups persisted at follow-up at one month.
Realon, et al. (1986) used the term maintenance for what might more accurately be considered to be a test of generalization across settings. They found that generalization did occur. They followed-up after 12 months for two of the four subjects in their second experiment, finding that treatment gains were maintained.
Wildman, et al. (1986) measured generalization across persons by replacing one-half of a pair of dyads with nonretarded stooges. Generalization was shown to have occurred. A six-month follow-up revealed that treatment effects had persisted.
Discussion
My review indicated most experimenters who address issues of social skills training in adults with mental retardation are aware of the importance of demonstrating the durability and usefulness of training programs by programming and measuring generalization and maintenance. Among them, the authors addressed all four (time, setting, behavior, and subject of the types of generalization pointed out by Stokes & Baer’s (1977). Furthermore, the authors who addressed the issues were for the most part successful in demonstrating generalization, and completely successful in demonstrating maintenance.
Does this mean that we have finally made explicit what Baer, Stokes, Holman, Fowler, & Rowbury (1981) called “largely implicit in current techniques, rarely recognized as a technology or as a target for technological research, and … largely uncodified” (pp. 39-40) ? Perhaps. Certainly, in the papers I have reviewed, the majority of the authors made a conscientious attempt to build in procedures to address generalization and maintenance. Certainly, we have the “outline of a technology” (Baer, Stokes, Holman, Fowler, & Rowbury, 1981)—but do we have more than an outline?
I would argue we are perhaps filling in that outline, but that we have a long way to go before we can consistently demonstrate generalization and maintenance, much less claim to have understand how and why it occurs (Stokes & Osnes, 1989). There are a variety of reasons for this, not the least of which is that in social skills the systems of reinforcement in which generalization and maintenance are addressed are complex and difficult to measure. It is often unclear whether we are measuring Skinnerian generalization or generality (Fox & McEvoy, unpublished). Other problems include the lack of a standardized terminology for maintenance and generalization, and the confusion between generalization and generality.
It certainly would help if we could all learn to use the same terms when talking about generalization and maintenance, and to learn to consistently differentiate between generalization and generality, for only then will we be able to begin to tease apart the complex systems of reinforcement operating during generalization probes in social skills training.
Perhaps the term generality is too close in spelling and punctuation to the word generalization. Perhaps a new term should be introduced which would help researchers distinguish between two very different things: behavior change which occurs in new situations under because of reinforcement systems under the control of the experimenter (generalization) and behavior change which occurs because of the complex interplay of social behaviors which is out of the experimenter’s control.
Certainly, the gauntlet thrown out by Baer, Wolf, & Risley (1968) is being addressed, however incompletely and unsuccessfully. Certainly, we are still talking about the weather, but just as certainly, we are doing our best to change it.
References
Agran, M., Salzberg, C.L., & Stowitschek, J.J. (1987). An analysis of the effects of a social skills training program using self-instructions on the acquisition and generalization of two social behaviors in a work setting. Journal of the Association for Persons with Severe Handicaps, 12(2), 131-139.
Allen, J.S., Jr., Tarnowski, K.J., Simonian, S.J., Elliott, D., & Drabman, R.S. (1991). The generalization map revisited: Assessment of generalized treatment effects in child and adolescent behavior therapy. Behavior Therapy, 22, 393-405.
Baer, D.M., Stokes, T.F., Holman, J., Fowler, S.A., & Rowbury, T.G. (1981). Uses of self-control techniques in programming generalization. In S.W. Bijou & R. Ruiz (Eds.). Behavior modification: Contributions to education. Hillsdale, NJ: Erlbaum Press.
Baer, D.M., Wolf, M.M., & Risley, T.R. (1968). Some current dimensions of applied behavioral analysis. Journal of Applied Behavior Analysis, 1, 91-97.
Bates, P. (1980). The effectiveness of interpersonal skills training on the social skills acquisition of moderately and mildly retarded adults. Journal of Applied Behavior Analysis, 13, 237-248.
Bornstein, P.H., Back, P.J., McFall, M.E., Friman, P.C., & Lyons, P.D. (1980). Application of a social skills training program in the modification of interpersonal deficits among retarded adults: A clinical replication. Journal of Applied Behavior Analysis, 13, 171-176.
Breen, C., Haring, T., Pitts-Conway, V., & Gaylord-Ross, R. (1985). The training and generalization of social interaction during breaktime at two job sites in the natural environment. Journal of the Association for Persons with Severe Handicaps, 10(1), 41-50.
Bregman, S. (1984). Assertiveness training for mentally retarded adults. Mental Retardation, 22, 12-16.
Chadsey-Rusch, J.C., Karlan, G.R., Riva, M.T, & Rusch, F.R. (1984). Teaching mentally retarded adults conversational skills in employment settings. Mental Retardation, 22, 218-225.
Chandler, L.K., Fowler, S.A., & Lubeck, R.C. (1992). An analysis of the effects of multiple setting events on the social behavior of preschool children with special needs. Journal of Applied Behavior Analysis, 25(2), 249-264.
Chandler, L.K., Lubeck, R.C., & Fowler, S.A. (1992). Generalization and maintenance of preschool children’s social skills: A critical review and analysis. Journal of Applied Behavior Analysis, 25(2), 415-428.
Cone, J.D., Anderson, J.A., Harris, F.C., Goff, D.K., & Fox, S.R. (1978). Developing and maintaining social interaction in profoundly retarded young males. Journal of Abnormal Child Psychology, 6, 351-360.
Davies, R.R., & Rogers, E.S. (1985). Social skills training with persons who are mentally retarded. Mental Retardation, 23(4), 186-196.
Denny, D., & Fox, J. (1989). Collecting and analyzing continuous behavioral data with the TRS‑80 Model 102 portable computer. Journal of Special Education Technology, 9(4), 183‑189.
Drabman, R.S., Hammer, D., & Rosenbaum, M.S. (1979). Assessing the generalization in behavior modification with children: The generalization map. Behavioral Assessment, 1, 203-219.
Dy, E.B., Strain, P.S., Fullerton, A., & Stowitschek, J. (1981). Training institutionalized, elderly mentally retarded persons as intervention agents for socially isolate peers. Analysis and Intervention in Developmental Disabilities, 1, 199-215.
Edmonson, B., & Han, S.S. (1983). Effects of socialization games on proximity and prosocial behavior of aggressive mentally retarded institutionalized women. American Journal of Mental Deficiency, 71, 1017-1024.
Elias-Burger, S., Sigelman, C., Danley, W., & Burger, D. (1981). Teaching interview skills to mentally retarded persons. American Journal of Mental Deficiency, 85(6), 655-657.
Fantuzzo, J.W., Wray, L., Hall, R., Goins, C., & Azar, S. (1986) Parent and social-skills training for mentally retarded mothers identified as child maltreaters. American Journal of Mental Deficiency, 91(2), 135-140.
Fox, J.J., Gunter P., Brady, M., Bambara, L.M., Spiegel-McGill, P., & Shores, R. (1984). Using multiple peer exemplars to develop generalized social responding of an autistic girl. Monograph in Behavioral Disorders: Severe Behavior Disorders of Children and Youth, 7, 17-26.
Fox, J.J., & McEvoy, M.A. (In preparation). Assessing and enhancing generalization and social validity of social skills interventions with children and adolescents. Center for Childhood Learning and Development, East Tennessee State University, Johnson City, TN.
Foxx, R.M., & McMorrow, M.J. (1983). Stacking the deck: A social skills game for retarded adults. Champaign, IL: Research Press.
Foxx, R.M., McMorrow, M.J., Bittle, R.G., & 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., & Mennemier, J. (1984). Teaching social/vocational skills to retarded adults with a modified table game: An analysis of generalization. Journal of Applied Behavior Analysis, 17, 343-352.
Foxx, R.M., McMorrow, M.J., & Schloss, C.N. (1983). Stacking the deck: Teaching social skills to retarded adults with a modified table game. Journal of Applied Behavior Analysis, 16(2), 157-170.
Garris, R.P., & Hazinski, L. (1988). The effect of social skills training procedures on the acquisition of appropriate interpersonal skills for mentally retarded adults. Journal of Psychopathology and Behavioral Assessment, 10(3), 225-240.
Gentile, C., & Jenkins, J.O. (1980). Assertive training with mildly mentally retarded persons. Mental Retardation, 18(6), 315-317.
Gibson, F.W., Laurence, P.S., & Nelson, R.O. (1976). Comparison of three training procedures for teaching social responses to developmentally disabled adults. American Journal of Mental Deficiency, 81(4), 379-387.
Goldstein, H., & Cisar, C.L. (1992). Promoting interaction during sociodramatic play: Teaching scripts to typical preschoolers and classmates with disabilities. Journal of Applied Behavior Analysis, 25(2), 265-280.
Goldstein, H., Kaczmarek, L., Pennington, R., & Shafer, K. (1992). Teaching social skills to students with autism to increase peer interactions in an integrated first-grade classroom. Journal of Applied Behavior Analysis, 25(2), 289-306.
Grinnell, R.M., & Lieberman, A. (1977). Teaching the mentally retarded job interviewing skills. Journal of Counseling Psychology, 24, 332-337.
Hall, C., Sheldon-Wildgen, J., & Sherman, J.A. (1980). Teaching job interview skills to retarded clients. Journal of Applied Behavior Analysis, 13, 433-442.
Harchik, A.E., Sherman, J.A., & & Sheldon, J.B. (1992). The use of self-management procedures by people with developmental disabilities: A brief review. Research in Developmental Disabilities, 13, 211-227.
Haring, T.G., & Breen, C.G. (1992). A peer-mediated social network intervention to enhance the social integration of persons with moderate and severe disabilities. Journal of Applied Behavior Analysis, 25(2), 319-334.
Holley, S. (1980). A social skills group with mentally retarded subjects. British Journal of Clinical and Social Psychology, 19, 279-285.
Hollin, C.R., & Trower, P. (1988). Development and applications of social skills training: A review and critique. In M. Hersen, R.M. Eisler, & P.M. Miller, (Eds.), Progress in behavior modification, Vol. 22, pp. 165-214. Newbury Park, CA: Sage Publications, Inc.
Hundert, J., & Hopkins, B. (1992). Training supervisors in a collaborative team approach to promote peer interaction of children with disabilities in integrated preschools. Journal of Applied Behavior Analysis, 25(2), 385-400.
Jackson, H.J., King, N.J., & Heller, V.R. (1981). Social skill assessment and training for persons with mental retardation: A review of research. Australia and New Zealand Journal of Developmental Disabilities, 7, 113-123.
Jackson, H.J., & Martin, R. (1983). Remedying social skills deficits in a chronic schizophrenic-retarded person. Australia and New Zealand Journal of Developmental Disabilities, 9(2), 55-63.
Kamps, D.M., Leonard, B.R., Vernon, S., Dugan, E.P., Delquadri, J.C., Gershon, B., Wade, L., & Folk, L. (1992). Teaching social skills to students with autism to increase peer interactions in an integrated first-grade classroom. Journal of Applied Behavior Analysis, 25(2), 281-288.
Kazdin, A.E. (1975). Behavior modification in applied settings. Homewood, IL: Dorsey Press.
Kazdin, A.E., & Polster, R. (1973). Intermittent token reinforcement and response maintenance in extinction. Behavior Therapy, 4, 386-391.
Keller, F.S., & Schoenfeld, W.N. (1950). Principles of psychology. New York: Appleton-Century Crofts.
Kern-Dunlap, L., Dunlap, G., Clarke, S., Childs, K., White, R.L., & Stewart, M.P. (1992). Effects of a videotape feedback package on the peer interaction of children with serious behavioral and emotional challenges. Journal of Applied Behavior Analysis, 25(2), 355-364.
Koegel, L.K., Koegel, R.L., Hurley, C., & Frea, W.D. (1992). Improving social skills and disruptive behavior in children with autism through self-management. Journal of Applied Behavior Analysis, 25(2), 341-354.
LaGreca, A., Stone, W., & Bell, C. (1983). Facilitating the vocational-interpersonal skills of mentally retarded adults. American Journal of Mental Deficiency, 88, 270-278.
Lalli, J.S., Pinter-Lalli, E., Mace, F.C., & Murphy, D.M. (1991). Training interactional behaviors of adults with developmental disabilities: A systematic replication and extension. Journal of Applied Behavior Analysis, 24(1), 167-174.
Lee, D.Y. (1977). Evaluation of a group counseling program designed to enhance social adjustment of mentally retarded adults. Journal of Counseling Psychology, 24, 318-323.
Lowther, R.M., & Martin, G.L. (1980). Multiple settings and trainers for programming generalization of a social response of severely retarded persons. Behavior Research of Severe Developmental Disabilities, 1, 131-145.
MacIntosh, N.J. (1974). The psychology of animal learning. New York: Academic Press.
Matson, J.L., & Adkins, J. (1980). A self-instruction social skills training program for mentally retarded persons. Mental Retardation, 18, 245-248.
Matson, J.L., & Andrasik, F. (1982). Training leisure-time social-interaction skills to mentally retarded adults. American Journal of Mental Deficiency, 86(5), 533-542.
Matson, J.L., & DiLorenzo, T.M. (1986). Social skills training and mental handicap and organic impairment. In C.R. Hollin & P. Trower (Eds.), Handbook of social skills training: Vol. 2: Clinical applications and new directions. Oxford: Pergamon.
Matson J.L., & Earnhart, T. (1981). Programming treatment effects to the natural environment: A procedure for training institutionalized retarded adults. Behavior Modification, 5, 27-38.
Matson, J.L., Esveldt-Dawson, K., Andrasik, F., Ollendick, T.H., Petti, I., & Hersen, M. (1980). Direct, observational and generalization effects of social skills training with emotionally disturbed children. Behavior Therapy, 11, 522-531.
Matson, J.L., & Senatore, V. (1981). A comparison of traditional psychotherapy and social skills training for improving interpersonal functioning of mentally retarded adults. Behavior Therapy, 12, 419-427.
Matson, J.L., & Zeiss, R. (1979). The buddy system: A method for generalized reduction of inappropriate interpersonal behaviour of retarded psychiatric patients. British Journal of Social and Clinical Psychology, 18(4), 401-405.
McDonnell, J.J., Horner, R.H., & Williams, J.A. (1984). Comparison of three strategies for teaching generalized grocery purchasing to high school students with severe handicaps. Journal of the Association for Persons with Severe Handicaps, 9, 123-134.
McLure, R.F. (1968). Reinforcement of verbal social behavior in moderately retarded children. Psychological Reports, 23, 371-376.
Meredith, R.L., Saxon, S., Doleys, D.M., & Kyzer, B. (1980). Social skills training with mildly retarded young adults. Journal of Clinical Psychology, 36(4), 1000-1009.
Misra, A. (1992). Generalization of social skills through self‑monitoring by adults with mild mental retardation. Exceptional Children, 58(6), 495‑507.
Morgan, R.L., & Salzberg, C.L. (1992). Effects of video-assisted training on employment-related social skills of adults with severe mental retardation. Journal of Applied Behavior Analysis, 25(2), 365-383.
Morris, R.J., & Dolker, M. (1974). Developing cooperative play in socially withdrawn retarded children. Mental Retardation, 12, 24-27.
Mueser, K.T., Valenti-Hein, D., & Yarnold, P.R. (1987). Dating skills groups for the developmentally disabled: Social skills and problem-solving versus relation training. Behavior Modification, 11(2), 200-228.
Nelson, R.O., Lipinski, D.P., & Black, J.L. (1976). The reactivity of adults retardates’ self-monitoring: A comparison with token reinforcement. Psychological Record, 26, 189-201.
Nietupski, J., & Svoboda, R. (1982). Teaching a cooperative leisure skill to severely handicapped adults. Education and Training of the Mentally Retarded, 17, 38-43.
Odom, S.L., Chandler, L.K., Ostrosky, M. McConnell, S.R., & Reaney, S. (1992). Fading teacher prompts from peer-initiating interventions for young children with disabilities. Journal of Applied Behavior Analysis, 25(2), 307-318.
Odom, S.L., & McConnell, S.R. (1992). Improving social competence: An applied behavior analysis perspective. Journal of Applied Behavior Analysis, 25(2), 239-244.
Perry, M.A., & Cerreto, M.C. (1977). Structured learning training of social skills for the retarded. Mental Retardation, 15(6), 31-33.
Peterson, G.A., Austin, G.J., & Lang, R.P. (1979). Use of teacher prompts to increase social behavior: generalization effects with severely and profoundly retarded adolescents. American Journal of Mental Deficiency, 84, 82-86.
Quinn, J.M., Sherman, J.A., Sheldon, J.B., Quinn, L.M., & Harchik, A.E. (1992). Social validation of component behaviors of following instructions, accepting criticism, and negotiating. Journal of Applied Behavior Analysis, 25(2), 401-413.
Realon, R.E., Favell, J.E., Stirewalt, S.C., & Phillips, J.F. (1986). Teaching severely handicapped persons to provide leisure activities to peers. Analysis and Intervention in Developmental Disorders, 6(3), 203-219.
Reese, R.M., Sherman, J.A., & Sheldon, J. (1984). Reducing agitated-disruptive behavior of mentally retarded residents of community group homes: The role of self-recording and peer-prompted self-recording. Analysis and Intervention in Developmental Disabilities, 4, 91-107.
Robertson, I. Richardson, A.M., & Youngson, S.C. (1984). Social skills training with mentally handicapped people: A review. British Journal of Clinical Psychology, 23(4), 241-264.
Rychtarik, G.G, & Bornstein, P.H. (1979). Training conversational skills in mentally retarded adults: A multiple baseline analysis. Mental Retardation, 17, 289-293.
Samaras, M.S., & Ball, T.S. (1975). Reinforcement of cooperation between profoundly retarded adults. American Journal of Mental Deficiency, 80, 63-71.
Schloss, P.J., & Wood, C.E. (1990). Effect of self-monitoring on maintenance and generalization of conversational skills of persons with mental retardation. Mental Retardation, 28(2), 105-113.
Senatore, V., Matson, J.L., & Kazdin, A.E. (1982). A comparison of behavioral methods to train social skills to mentally retarded adults. Behavior Therapy, 13, 313-324
Singh, N.N., & Millichamp, C.J. (1987). Independent and social play among profoundly mentally retarded adults: Training, maintenance, generalization, and long-term follow-up. Journal of Applied Behavior Analysis, 20(1), 23-24.
Singh, N.N., & Winton, A.S.W. (1983). Social skills training with institutionalized severely and profoundly mentally retarded persons. Applied Research in Mental Retardation, 4, 393-398.
Skinner, B.F. (1938). The behavior of organisms. New York: Appleton-Century.
Spivack, G., & Shure, M.B. (1974). Social adjustment of young children: A cognitive approach to solving real-life problems. San Francisco: Gauss-Bass.
Stacy, D., Dolys, D.M., & Malcolm, R. (1979). Effects of social skills training in a community-based program. American Journal of Mental Deficiency, 84, 152-158
Stewart, G., van Houten, R., & van Houten, J. (1992). Increasing generalized social interactions in psychotic and mentally retarded residents through peer-mediated therapy. Journal of Applied Behavior Analysis, 25(2), 335-340.
Stokes, T.B., & Baer, D.M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10, 349-367.
Stokes, T.F., & Osnes, P.G. (1988). The developing applied technology of generalization and maintenance. R.H. Horner, G. Dunlap, & R.L. Koegel (Eds.), Generalization and maintenance: Life-style changes in applied settings. Baltimore, MD: Brookes Publishing.
Stokes, T.F., & Osnes, P.G. (1989). An operant pursuit of generalization. Behavior Therapy, 20, 337-355.
Storey, K. (1987). Teaching social skills to persons with severe handicaps in community settings: A review. Australia and New Zealand Journal of Developmental Disabilities, 13(3), 143-150.
Storey, K., Bates, P., & Hanson, H.B. (1984). Acquisition and generalization of coffee purchase skills by adults with severe disabilities. Journal of the Association for Persons with Severe Handicaps, 9(3), 178-185.
Stowitschek, J.J., McConaughy, E.K., Peatross, D., Salzberg, C.L., & Lignugaris/Kraft, B. (1988). Effects of group incidental training on the use of social amenities by adults with mental retardation in work settings. Education and Training in Mental Retardation, 23(3), 202-212.
Turner, S.M., Hersen, B., & Bellack, A.S. (1978). Social skills training to teach prosocial behaviors in an organically impaired and retarded patient. Journal of Behavior Therapy and Experimental Psychiatry, 9, 253-258.
Van Hasselt, V.B., Hersen, M., Egan, B.S., McKelvey, L.J., et al. (1989). Increasing social interactions in deaf-blind severely handicapped young adults. Behavior Modification, 13(2), 257-272.
Wehman, P. (1977). Research on leisure time and the severely developmentally disabled. Rehabilitation Literature, 38, 98-105.
Wehman, P., Abramson, M., & Norman, C. (1977). Transfer of training in behavior modification programs: An evaluative review. The Journal of Special Education, 11, 221-230.
Wehman, P., Karan, P., & Rettie, S. (1976). Developing independent play in three severely retarded women. Psychological Reports, 39, 995-998.
Wheeler, J.J., Bates, P., Marshall, K.J., & Miller, S.R. (1988). Teaching appropriate social behaviors to a young man with moderate mental retardation in a supported competitive employment setting. Education and Training in Mental Retardation, 23(2), 105-116.
Wildman, B.G., Wildman, H.E., & Kelly, W.J. (1986). Group conversational-skills training and social validation with mentally retarded adults. Applied Research in Mental Retardation, 7(4), 443-458.
Zisfein, R., & Rosen, M. (1974). Effects of a personal adjustment training group counselling program. Mental Retardation, 12, 50-53.
Proposal for Major Area Paper
A Survey of the Methodology of Generalization and Maintenance
In Operant Training of Social Skills
Of Persons With Mental Retardation:
Proposal for Major Area Paper
By Dallas Denny
Department of Special Education
George Peabody College of Vanderbilt University
15 January, 1992
Abstract
It is generally agreed that it is important to address issues of generalization and maintenance in social skills research with persons with mental retardation. It is similarly agreed that not enough is being done. I propose to examine existing reports of operant social skills training of persons with mental retardation and determine the extent to which maintenance and generalization are measured. Relevant papers will be obtained by request from researchers in the area, by computer search, by manual inspection of journals, and by examination of the reference sections of papers which are obtained. Results will be presented in tabular form and discussed in the text. Overall findings will be examined in the light of what a number of workers have indicated should be happening. I will provide suggestions for improvement of measurement of maintenance and generalization in operant social skills training with persons with mental retardation.
Specific questions I will be asking about each paper are: 1) Did the authors attempt to measure generalization and generalization? 2) How? 3) What were their findings? The overall questions I will be asking are: 1) To what extent are issues of generalization and maintenance addressed in this literature? 2) What do existing measurements tell us about the effectiveness of operant training of social skills in persons with mental retardation? 3) How do existing studies compare to the consensus of what should be happening?
The question of generalization and maintenance in applied behavior analysis is considered by most researchers to be one of crucial importance. Although Skinner addressed the issues of generalization (transfer of stimulus control to objects without a history of reinforcing the organism) and maintenance (exhibition of operant response after the discontinuation of the artificial reinforcement schedule) as early as 1938 in his work with nonhuman animals, it was Stokes & Baer (1977) who clearly and concisely brought the importance of these phenomenon to the collective attention of the behavior analytic community. Today, the importance of maintenance and generalization is widely discussed, but a number of researchers have lamented the fact that, like the weather, many researchers talk about these phenomena, but few adequately address (i.e. do much more than talk about) them in their studies. “Despite this longstanding concern with generalization and increasing attention to its assessment, it is difficult to point to any startling advances in our ability to reliably produce either generalization or maintenance of social behavior…” (Fox & McEvoy, manuscript in preparation).
The operant training of social skills is an area in which generalization and maintenance are of particular importance. This is because skills are typically trained in one or at most a few settings, are expected or hoped to transfer to new settings, and are hoped to endure after the experimenters have withdrawn, with the ever elusive “natural community of reinforcement” maintaining the shaped behavior across both time and place. Resources are scarce, and the need for permanent change is great. I propose to examine existing studies of the operant training of social skills of persons with mental retardation, and to determine the extent to which the authors addressed and attempted to measure maintenance and generalization of the skills they trained. I will also examine what this measurement revealed about the effectiveness of the training methods used. I will use a bibliography provided to me by Dr. Jim Fox, and supplement it with a search of computer databases such as PsychLit and ERIC and by manual examination of the last three years of journals I judge (on the basis of having previously published such studies) to be most likely to contain relevant material. I will examine the bibliographies of the papers I obtain for additional references.
I will present findings in tabular form, with case-by-case discussion of relevant issues. Trends and significant findings will be discussed in detail. The overall findings, as well as the work of representative papers, will be compared to what Stokes & Baer (1977), Stokes & Osnes (1989), McEvoy & Fox (in preparation) and others have indicated should be happening in the field. Lastly, I will provide suggestions for improvement of measurement of maintenance and generalization.
Papers in which issues of maintenance and generalization are not addressed will not be discussed, except to point out, in representative cases, what might have been done, but wasn’t.
The specific questions I will be asking about each paper, then, are: 1) Did the authors attempt to measure generalization? 2) How? 3) What were their findings? 4) Did the authors attempt to measure generalization? 5) How? 6) What were their findings? The overall questions I hope to answer are: 1) To what extent are issues of generalization and maintenance addressed in this literature? 2) What do existing measurements tell us about the effectiveness of operant training of social skills in persons with mental retardation? 3) How do existing studies compare to the consensus of what should be happening?
I will begin work on this project as soon as Dr. Jim Fox has had a chance to review this proposal and gives me provisional go-ahead, and will begin work in earnest when I have approval of the full doctoral committee. Because I find this particular literature slow going and difficult to understand, I anticipate that I will have a draft of the paper within four months after approval by the committee, and a final version six months after approval by the committee.
It seems likely the questions generated by this review of the literature will provide questions which it will be appropriate to address in someone’s doctoral dissertation— hopefully, mine.
References
Fox, J.J., & McEvoy, M.A. (In preparation). Assessing and enhancing generalization and social validity of social skills interventions with children and adolescents. Center for Childhood Learning and Development, East Tennessee State University, Johnson City, TN.
Skinner, B.F. (1938). The behavior of organisms. New York: Appleton-Century.
Stokes, T.B., & Baer, D.M. (1977). An implicit technology of generalization. Journal of Applied Behavior Analysis, 10,349-367.
Stokes, T.F., & Osnes, P.G. (1989). An operant pursuit of generalization. Behavior Therapy, 20, 337-355.