This item is only available as the following downloads:
1 PATTERNS OF SOCIAL INTERACTION DURING GROUP CONTINGENCIES By JILL M. HARPER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 2012 Jill M. Harper
3 ACKNOWLEDGMENTS Many thanks are due to those who have helped me to complete this dissertation project. I would like to thank my family and friends for their continued support throughout the entire process. I would also like to thank my labmates, in particular Tara Fahmie and Kathryn Horton, for their contributions throughout the course of this project. I thank the members of my committee, Dr. Steven Smith, Dr. Donald Stehouwer, and Dr. Timothy Vollmer. And finally, I would like to thank Dr. Brian Iwata for all of his inval uable guidance, instruction, and support throughout my graduate career.
4 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 3 LIST OF TABLES ................................ ................................ ................................ ........................... 6 LIST OF FIGURES ................................ ................................ ................................ ......................... 7 ABSTRACT ................................ ................................ ................................ ................................ ..... 8 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................... 9 Definition and Classification of Group Contingencies ................................ ............................. 9 Applications of Group Contingencies ................................ ................................ .................... 14 The Relative Efficacy of Group Contingencies ................................ ................................ ...... 17 Advantages and Drawbacks of Group Contingencies ................................ ............................ 19 Non programmed Social Contingencies ................................ ................................ ................. 21 2 GENERAL METHOD ................................ ................................ ................................ ............ 24 Subjects and Setting ................................ ................................ ................................ ................ 24 Preferen ce Assessments ................................ ................................ ................................ .......... 24 Response Measurement and Reliability ................................ ................................ .................. 25 General Procedure and Experimental Design ................................ ................................ ......... 28 3 EXPERIMENT 1: REINFORCEMENT OF SKILL ACQUISITION ................................ ... 31 Response Measurement and Reliability ................................ ................................ .................. 31 Procedure ................................ ................................ ................................ ................................ 31 Results ................................ ................................ ................................ ................................ ..... 33 4 EXPERIMENT 2: REINFORCEMENT FOR BEHAVIOR REDUCTION .......................... 40 Response Measurement and Reliability ................................ ................................ .................. 40 Procedure ................................ ................................ ................................ ................................ 40 Results ................................ ................................ ................................ ................................ ..... 42 5 EXPERIMENT 3: PUNISHMENT FOR BEHAVIOR REDUCTION ................................ .. 49 Response Measurement, Reliability, and Procedure ................................ .............................. 49 Results ................................ ................................ ................................ ................................ ..... 50 6 DISCUSSION ................................ ................................ ................................ ......................... 56 LIST OF REFERENC ES ................................ ................................ ................................ ............... 63
5 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ......... 68
6 LIST OF TABLES Table page 2 1 Subject characteristics. ................................ ................................ ................................ ..... 29 2 2 Results of edible and activity preference assessments ................................ ..................... 29 2 3 Social i nteraction c ategories and e xamples ................................ ................................ ..... 30 A 1 Scripted instructions across conditions and experiments ................................ ................. 62
7 LIST OF FIGURES Figure page 3 1 Frequency of sorting for Group A (Experiment 1) ................................ ............................ 37 3 2 Rate of positive and negative interactions across conditions for Group A (Experiment 1) ................................ ................................ ................................ ................... 38 3 3 Proportion of interactions per social category across conditions for Group A (Experiment 1) ................................ ................................ ................................ ................... 39 4 1 Percentage of intervals of out of seat behavior for Group B (Experiment 2) ................... 46 4 2 Rate of positive and negative interactions across conditions for Group B (Experiment 2) ................................ ................................ ................................ ................... 47 4 3 Proportion of interactions per social category across conditions for Group B (Experiment 2) ................................ ................................ ................................ ................... 48 5 1 Percentage of intervals of o ut of seat behavior across conditions for Group B (Experiment 3) ................................ ................................ ................................ ................... 53 5 2 Rate of positive and negative social interactions across conditions for Group B (Experiment 3) ................................ ................................ ................................ ................... 54 5 3 Proportion of interactions per social category across conditions for Group B (Experi ment 3) ................................ ................................ ................................ ................... 55
8 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PATTERNS OF SOCIAL INTERACTION DURING GROUP CONTINGENCIES By Jill M. Harper August 2012 Chair: Brian Iwata Major: Psychology Group contingencies are distinct from individual contingencies in that the performance of one individual influences the delivery of reinforcement to another. Thus, group contingencies may occasion other, non programmed contingencies of a cooperative or competitive nature although this unique feature of group contingencies has rarely been examined. The purpose of this study was to examine the relative effects of competitive and cooperative group contingencies relative to those of individual contingenc ies, on performance and social interaction s In addition, we examined both behavior acquisition (sorting) and behavior reduction (out of seat behavior) because it seemed likely that reinforcement ( Experiments 1 & 2) and punishment (Experiment 3) might occasion d ifferent t ypes of interactions. Result s showed that all contingencies improved performance from baseline across the three experiments, but that the relative efficacy of group contingencies was mixed when applied to behavior acquisition and reduction using reinforcement and punishment. Some dif ferences in the general pattern as well as specific types of social interactions across cooperative and competitive contingencies were noted. The general results highlight the correlation between changes in performance and interactions during different gro up contingencies.
9 CHAPTER 1 INTRODUCTION Definition and Classification of Group Contingencies Group contingencies refer to contingencies of reinforcement (or punishment) in which common consequence s are determined by the behavior of an individual member of the group, a subset of the group, or the behavior of the entire group (Cooper, Heron, & Heward, 2007; Tankersley, 1995 ) Group contingencies have been classified in several different ways, but are most often desc ribed in terms of the structural arrangement or whose behavior determines the consequences and to whom these consequences are delivered (Litow & Pumroy, 1975; Neumann, 1977). Litow and Pumroy proposed a classification system based on potential variations in response requirement and reinforcement which yielded three general ca tegories of group contingencies: independent, dependent, and interdependent contingencies. Although their classificati on system remains prominent in the literature ( Tingstrom, Sterling Turner, & Wilczynski, 2006 ) it has been expanded to include additional variants in both response requirement and reinforcement allocation. An independent group contingency is a common contingency applied to the individual behavior of all members of a group simultaneously Hopkins, Schutte, and Garton (1971) applied an independent group contingency to the handwriting performance of first and second grade students. Each student was allowed to leave his or her seat and enter the playroom contingent on transcribing a passage from the blackboard Thus, a common reinforcer, entrance into the play area, was applied to the individual behavior (assignment completion) of all group members simultaneously. The contingency resulted in an increase in both the r ate and accuracy of letter formation across both classrooms.
10 The dependent group contingency as described by Litow and Pumroy (1975) involves consequences delivered to all members of a group based on the behavior a single or subset of individuals. The i ndividual whose behavior determines reinforcement for the group is sometimes Coleman (1970) provided a nice demonstration of the modification of classroom behavior using a dependent contingency. The on task and disruptive behavio r of four elementary students across different classrooms was targeted A timer signaled the end of each 10 s interval, and the target student earned a piece of candy if ( s ) he was on task for that interval All earned candy was then divided among the class at the end of each session. In this example of a dependent contingency, reinforcement for the entire class was determined by the behavior a single individual who was also the lowest performer in the class. The individual whose behavior determines reinfor cement for the group dur ing dependent contingencies also has included a single, unidentified group member the best performer, and the worst performer (Litow & Pumroy, 1975; Buckholdt & Wodarski, 1978 ) Drabman, S pitalnik, and Spitalnik ( 1974 ) compared levels of disruptive behavior for the most and least disruptive students in a classroom under independent and several dependent contingencies. Points were earned for the absence of disruptive behavior, which later were exchanged for minutes of f ree time. Disruptive behavior of all target students (most and least disruptive students) decreased when the amount of free time for the entire group was based on the performance of the least disruptive, most disruptive, or a randomly selected group member Phillips, Phillips, Wolf and Fixen (1973) described a nother variation of the dependent group contingency, known as the dependent manager contingency T he manager delivered points to or removed points from his peers for the ir completion of assigned work T he manager also earned or lost points based on the number of chores completed by other group members.
11 This contingency highlights one in which reinforcement is delivered to a single member of the group based on the behavior of the entire group. I n an interdependent group contingency, consequences are delivered to the entire group based on behavior of the entire group. Allen and Iwata (1980) applied such a contingency for participation in a physical education class E veryone in the class was requir ed to complete a series of exercises before any class member could participate in a game. Exercise participation increased and the time required to complete the exercises decreased during the group contingency. H eering and Wilder (2006) used an interdepen dent group contingency to increase on task behavior in two general education classrooms. The c students in a given row were on t ask during random observations. Observations alternated across different rows throughout the session and the entire clas s received reinforcement if on task behavior was observed across rows for 75% or more of observa tions. This is an example of an interdependent group contingency in which the behavior of the entire group, although at different periods o f time throughout the session, determined reinforcement for the entire group. As the above examples illustrate, many variations of group contingencies can be derived from the structural components, or whose behavior determines reinforcement for whom. The key feature of group contingencies is that the behaviors of multiple individuals interact. Therefore, group contingencies, by definition, include a social component. The social aspect of group contingencies is likely the result of changes in the relative d istribution of reinforcement among group members ; t hat is, whether an increase in the probability of reinforcement for one member of the group increases or decreases the probability of reinforcement for other members. Several authors have proposed classification scheme s that emphasize these social, or functional,
12 characteristics of group contingencies (Buckholdt & Wodarski, 1978; Michaels, 1977; Slavin, 1977) Buckholdt and Wodarski (1978) described three general categories of contingencies based o n the relative effects of performance across individual group members : individualistic, competitive, and cooperative contingencies. The individualistic contingency prescribes consequences based on individual behavior ; thus performance by one member does not influence the probability of reward for another T he performance of one group member affects the chance of reward for other members during both the competitiv e and cooperative contingenc i es, although for different reasons. In the competitive contingenc y, the availability of rewards is limited such that the success (or failure) of one individual in obtaining reinforcement decreases (or increases) the probability of reinforcement for others. By contrast the interdependent nature of the cooperative conti ngency results from consequences resulting from performance by the group as a single unit rather th an the behavior of individuals in the group Michaels (1977) and Slavin (1977) extended previous classification systems to include both the structure of the contingency (individual versus group) as well as the criterion for reinforcement (standard versus relative) although the authors used somewhat different terminology in their descriptions The interactions between these two compon ents resulted in four general contingencies including individual (independent) contingencies group cooperative contingencies in dividual competitive contingencies and group competitive contingencies I ndividual contingencies are no different than the ind ependent contingency described by Li t ow and Pumroy (1975) and involve the simultaneous application of a contingency to the behavio r of more than one individual such that the probability of reinforcement for one person does not affect the probability of rei nforcement for another
13 C ompetitive contingencies involve the distribution of reinforcement based on the relative performance of group membe rs. Gresham and Gresham (1982) used a competitive contingency during which each student in a classroom had the opportunity to earn reinforcement if five or fewer instances of disruption were observed Howe ver, only the individual(s) who emitted the fewest number of disruptions actually received reinforcement. For example, if one subject engaged in three instances o f disruption whereas another student engaged in only two instances of disruption, only the second subject would receive reinforcement. When competitive contingencies are applied at the group level, the performance of one group is evaluated relative to th e performance of other groups. Therefore, as the probability of reinforcement increases for one group, the probability of reinforcement decreases for other groups. By contrast p e rformance under group cooperative contingencies usually is compared to some s tandard criterion and t he probability of reinforcement for all members of the group increase s as the probability of reinforcement for any member of the group increases The Good Behavior Game (GBG) first described by Barrish, Saunders, and Wolf e (1969) may be categorized as either a group competitive or group cooperative contingency, depending on whether the classroom is divided into teams or performs as a single unit, respectively. In the competitive arrangement instances of problem behavior by any tea m member result in a tally mark for the whole team, and the team(s) with the fewest tally marks at the end of the session receives reinforcement. In the cooperative arrangement, usually implemented with a s single group (classroom), instances of problem beh avior by any group member would result in a tally mark for the entire class, and reinforcement for all class members would be contingent on receiving fewer than a specified number of tally marks as a group.
14 Slavin (1977) noted the important point about gro up reward structures (contingencies) and their effects on behavior Although group contingencies can be applied in a competitive or cooperative arrangement, the occurrence of competition or cooperation is no t guaranteed. C ompetitive and cooperative contingencies refer simply to descriptions of the allocation of consequences f or specified response s C ompetition /cooperation on the other hand, refer to s ets of specific responses that may or may not occur during group contingencies In fact, Slavin dec whose occurrence needs to be documented ( Applications of Group C ontingencies Group contingencies produce changes in behavior via programmed consequences, and in th is sense are no different tha n individual contingencies. Therefore, g roup contingencies lend themselves to the application of a range of independent variables based on both positive and negative reinforcement as well as punishment. I n addition group contingencies have been applied across a variety of dependent variables including skill acquisition and behavior reduction, as well as across a variety of settings and populations. Most applications of g roup contingencies to increase behavior have focused on the use of reinforcement b ased procedures to improve academic performance Lloyd, Eberhardt, and Drake ( 1996 ) described an example of a common contingency. The experimenters increased performance on Spanish vocabulary quizzes using an independent contingency coupled with group stud y. Students were given the opportunity to study for the quiz in a small group format, but reinforcement was contingent on individual quiz performance. A small tangible and social praise were delivered to each student who received a score of 90% or above on the quiz. Williamson, Williamson, Watkins, and Hughes (1992) compared the effects of independent and dependent contingencies on performance during an arithmetic estimation task (guessing the
15 number of items in a jar). The independent contingency was compe titive within each class such that a single winner received reinforcement for the correct estimation. During the dependent contingency, a correct guess by any student resulted in reinforcement for the entire class, and more than one class could win each we ek. Many examples of the application of interdependent group contingencies to skill acquisition can be found in the literature. A common interdependent contingency applied to behavior acquisition is the cooperative version of the GBG, which involves reinf orcement based on the performance of a single team as compared to some standard criterion. Lutzker and White Blackburn, (1979) used a variation of the GBG during which either positive reinforcement (candy) or negative reinforcement (early dismissal from w ork) was contingent on an increase in the number of boards sorted by work teams in a rehabilitative work program. Another common variation of the interdependent contingency requires all group members to engage in the target response simultaneously. For ex ample, Packard (1970) increased the attentive behavior of elementary students using an interdependent contingency. A light signaled the inattentive behavior of any student in the class, and reinforcement (token) was contingent on the attentive behavior of all students in the class for a specified amount of time. Other experimenters have applied positive reinforcement during interdependent contingencies to increase play skills (Kohler, Strain, Hoyson, & Davis, 1995), homework completion ( Olympia, Sheridan, J enson, & Andrews, 1994), and oral hygiene ( Swain, Allard, & Holborn, 1982), as well as negative reinforcement to increase prosocial peer behavior ( Skinner, Cashwell, & Skinner, 2000 ) and vocational performance (McNally, Norusis, Gentz, & McConathy, 1983). The application of group contingencies using both reinforcement and punishment based interventions for behavior reduction is also well documented. Several variations of differential
16 reinforcement have been applied to behavior reduction across different typ es of group contingences including differential reinforcement of alternative behavior (DRA), differential reinforcement of low rate behavior (DRL) and differential reinforcement of other behavior (DRO) Herman n DeMontes, Dominguez, Montes, and Hopkins (1 973) decreased the occurrence of tardy arrivals to work with an independent DRA contingency Tardy arrivals decreased when workers earned monetary vouchers for timely arrivals (the alternative behavior) that were paid out at the end of each workweek. Reid, Schuh Wear, and Brannon (1978) also targeted employee behavior through differential reinforcement. The authors decreased staff absenteeism through an interdependent contingency that involved preferred days off (weekends) for low rates of absenteeism (DRL) Differential reinforcement of other behavior involve s the delivery of reinforcement contingent on the absence of the target response, or the delivery of reinforcement contingent on other behavior. For example, Putnam, Handler, Ra mirez Platt, and Luisell i (2003) used an independent DRO contingency under which students earned lottery tickets contingent on the absence of inappropriate bus riding behavior. Group DRO contingencies have also been applied to stealing (Switzer, Deal, & Bailey 1977) and disruptiv e classroom behavior (Drabman et al., 1974), as well as interventions for drug addicts (e.g., Kirby, Kerwin, Carpenedo, Rosenwasser, & Gardner, 2008; Liebson, Cohen, & Failace, 1972). The most common punishment procedure applied to behavior reduction during group contingencies is a response cost procedure which involves the contingent removal of reinforcement following the occurrence of the target response. Within the context of group contingencies, response cost procedures often consist of the removal of conditioned reinforcers such as tokens or points. Witt and Elliot (1982) reduced inappropriate classroom behavior with
17 an independent response cost contingency. Students began each ses sion with a fixed number of paper slips on their desk but lost one slip of paper for each occurrence of inappropriate classroom behavior. At the end of the session, remaining slips, if any, were entered into a lottery for the opportunity to earn various re inforcers. Group respons e costs also have been implement under interdependent contingencies that have targeted inappropriate vocalizations within the classroom (Salend & Lamb, 1986), disruptive behavior (Axelrod, 1973 ; Fabiano et al., 2008) and cash short ages of store clerks (Marholin & Gray, 1976 ). The Relative Efficacy of Group Contingencies Although the general effectiveness of group contingenc ies has been extensively demonstrated, comparisons across varia tions are necessary to establish their relative efficacy (Litow & Pumroy, 1976; Theodore, Bray, & Kehle, 2004). Two way comparative studies have almost exclusively compared the effectiveness of independent and interdependent contingencies and have reported mixed results Several studies have reported equal effectiveness between contingencies (Axelrod, 1973; Frankosky & Sulzer Azaroff, 1978; Shores, Apolloni, & Norman, 1976) whereas other studies have reported superior effects under the interdependent contingency ( Lloyd et al., 1996; Long & W illiams, 1973 ; Pigott Fantuzzo, & Clement, 1985 ). Mixed results have also been reported in three way comparison studies (Alric Bray, Kehle, Chafouleas, & Theodore 2007, Gresham & Gresham, 1982; Speltz Shimamura, & McReynolds 1 982; Theodore et al., 20 04; Sha piro & Goldberg, 1986). Alric et al. ( 2007 ) conducted a comparison of the effects of independent, interdependent, and dependent group contingencies on the reading fluency of fourth grade elementary students. Similar effects were observed across the three contingencies: reading fluency improved when reinforcement was delivered contingent on individual performance (independent), the performance of a randomly selected student (dependent), or the average performance of the group (interdependent). Thus,
18 the r esults of the Alric et al. study replicated the outcomes of previous studies on the relative efficacy of group contingencies in increasing academic performance (Shapiro & Goldberg, 1986; Speltz et al., 1982). G resham and Gresham (1982 ) published one of the first three way comparative studies on group contingencies applied to behavior reduction The authors compared the efficacy of independent, interdependent, and dependent group contingencies in reducing disruptive behavior in the classroom Both the interdependent and dependent contingencies were effective in decreasing disruption, with slightly greater reductions observed under the interdependent contingency. However, the results of this study may be limited by the unique, competitive nature of the i ndependent contingency. The independent contingency is typically set up such that each member of the group who meets some standard criteri on receives reinforcement. Gresham and Gresham used a competitive independent contingency during which only the best p erformers to comparative studies that employed the more typical independent contingency. More recently, Theodore et al. (2004) compared the efficacy the three typ es of group contingencies using a DRL procedure to decrease the disruptive behavior of three students with emotional disorders. During the independent contingency, any student who engaged in five or fewer instances of disruption received reinforcement. The behavior of an unidentified student determined reinforcement for the entire class during the dependent contingency. Reinforcement during the interdependent contingency required on all students to meet the DRL criterion of five or fewer instances of disrup tion. Although all three contingencies produced noticeable reductions in disruption for all subjects, the dependent contingency produced superior effects for two of the subjects and the independent contingency produced the greatest effect for the third su bject.
19 In summary, the overall effectiveness of group contingencies has been repeatedly demonstrated; however the relative effectiveness of specific variations of group contingencies is less clear. The mixed results obtained in comparison studies may be t he result of a number of variables. First, procedural differences across studies might account for discrepant results obtained (e.g., a dependent group contingency applied in one study may not be comparable to a similar contingency used in another) Michaels (1977) warned against cross study comparisons due to the vast number of ways group contingencies differ across applications Second, different results across comparisons may be related to the way in which dependent variables are measured and summa rized. Several authors have noted differences in relative effects across individuals within a single study. Similarly, conclusions across studies may differ based on whose behavior is taken into account and at on what level the analysis is conducted. Final ly i t could be the case that unident ified variables are responsible such t hat discrepant results across similar direct contingencies may be the due to other non programmed events. For example, d iscrepant results could be attributed to variability in soci al interactions across comparison studies. As Slavin (1977) pointed out although competition and cooperation are not guaranteed to occur within group contingencies, they are possible outcomes. Given the number of variables that may account for the effects of group contingencies, a closer examination of the behavioral mechanisms operating during group contingencies is warranted. Advantages and Drawbacks of Group Contingencies Sever al advantages of g roup contingencies over individual contingences have been discussed in previous research. The most obvious benefit is simplicity; group contingencies are easier to implement given that the same contingency and reinforcers are applied to the behavior of multiple individuals simultaneously ( Alexander, Corbett, & Sm igel, 1976; Axelrod, 1973; Gresham & Gresham, 1982; Long & Williams, 1973; Tingstr o m et al., 2006 ). The other potential
20 advantage of group contingencies is the use of peers as behavior change agents (Alric et al., 2007; Crouch, Gresham, & Wrigh t, 1985; Lit ow & Pumroy, 1975). C onsequences for the group may occasion as prompts or reminders as well as consequences such as praise or disapproval. Several studies have included anecdotal reports of such social behavior within group contingencies. For example, Gresham and Gresham (1982) reported that social interactions during an interdependent group contingency included both peer praise for appropriate behavior and reprimands for inappropriate behavior. Similarly, Packard (1970) reported that peer pressure in t he form of reminders to pay attention and praise in the form of congratulations was observed during an interdependent group contingency on classroom attending behavior. One criticism of group contingencies is that s ome group members may lose reinforcement based on the behavior of others (Gresham & Gresha m, 19 82; Marholin & Gray, 1976). This criticism has been discussed in terms of skill deficit in which case some members of the group may not have the skills required to meet the contingency (Litow & Pumroy 1975), as well as lack of motiv ation, in which case some group members may be reinforced by sabotage of group reward ( Crouch et al., 1985; Kelshaw Levering, Sterling Turner, Henry, & Skinner, 2000). Although the social interaction among group members is considered one of the major advantages of group contingencies, particular interactions, such as negative peer pressure are also the source of criticism (Harris & Sherman, 1973; Salend & Lamb, 1986; Shores et al., 1976 ; Tingst ro m et al., 2006) The occurre nce of negative peer pressure has been reported during group contingencies that applied both reinforcement and punishment procedures. Shores et al. (1976) noted the occurrence of verbal threats during the initial implementation of an interdependent DRA con tingency to increase on task behavior. Similarly, Axelrod (1973) reported that peer pressure in the form of verba l threats was observed during an interdependent response cost
21 contingency to reduced classroom disruption, whereas no instances of verbal threats were observed during the independent contingency. Non programmed Social Contingencies A s sugg ested by several authors, non programmed social contingencies may play a critical role in the effectiveness of group contingencies (Cosden & Harring, 199 2; Litow & Pumroy, 1975; Tingstr o m, et al., 2006). However, few studies to date have systematically examined non programmed social interactions within group contingencies. Shores, et al. (1976) measured the frequency of task related and non task related v erbalizations among a group of five boys during independent and interdependent group contingencies for on task behavior. An increase in task related comments which includ ed statements regarding schoolwork, the token economy, or other contingencies such as back up reinforcers was observed during the interdependent contingency but not the independent contingency. No change in non task related comments, which included general conversation or unintelligible speech, was observed across conditions. The authors further noted that the type of task related comments changed throughout the group contingency : M ore negative comments such as threats were initially observed, whereas comments in the form of prompting and social praise were observed during later sessions. Frankosky and Sulzer Azaroff (1978) examined positive and negative social interactions among groups of men with developmental disabilities during independent and interdependent contingencies for block sorting. Data were collected during both the task peri od itself, and a subsequent snack period to examine potential carryover effects of changes in social interaction observed during the group contingency. The authors reported decreases in negative interactions and increases in positive interactions during bo th task sessions and subsequent snack periods during the interdependent contingency as compared to the independent contingency. One
22 limitation of the study was that general discussion among group members was scored as positive interaction; therefore, it is unclear if the increases in positive interactions observed during the group contingency was in the form of task related interactions, such as encouragement or prompting, or if they were simply an increase in unrelated conversation. Furthermore, all groups maximized reinforcement during both the independent and interdependent group contingencies, so the effect of increases in positive social interactions on task performance remains unclear. Speltz et al. (1982) extended the findings of Frankosky and Sulzer Azaroff (1978) by examining social interactions under variations of group contingencies for the completion of math worksheets. Social interactions were categorized as positive (encouragement, assistance), negative, or neutral (general conversation). Low le vels of negative interactions among peers were observed across all variations of group contingencies, and an increase is positive interactions was observed during the dependent (hero) contingency. A more detailed analysis showed that the increase in positi ve interactions observed during the dependent contingency was largely directed at the target student. Although results of this study indicated that the most frequent interactions across all contingencies were neutral, different types of contingencies may p romote different types of interactions. Together, these studies provide preliminary data to suggest that the frequency type of social interactions may differ across variations of group contingencies. However, data from additional subjects would be required to support this finding because previous studies that examined social interactions across group contingencies all involved cooperative contingencies It may be the case that social interactions differ along o ther dimensions during group contingency procedures. For example, no study has analyzed social interactions across group contingencies targeting increases in responding as compared to contingencies targeting decreases
23 in responding. It is possible that the proportion of negative to positive interactions would be different across such applications. More specifically, difference s in social interactions may exist within the same type of group contingencies across contingencies aimed at behavior augmentation or behavior reduction. In addition, previous research has been limited to examining social interactions under cooperative contingencies and it seems likely that interactions may differ under cooperative and competitive contingencies. In summary, although a great deal of research has been conducted on the effects of group contingencies, conclusions about relative effectiveness have been mixed. In addition, it was often unclear how behavior was changed because the unique feature of group contingencies, non programmed social interactions has rarely been examined. Thus, the purpose of this study was to examine both performance and social interaction across competitive and cooperative contingencies applied to behavior acquisition and reduction using similar methodology. Experiment 1 examined patterns of interaction under different contingencies of reinforcement for behavior acquisition. Experiments 2 and 3 focused on interactions under conti ngencies of reinforcement (Experiment 2) and punishment ( Experiment 3 ) for behavior reduction.
24 CHAPTER 2 GENERAL METHOD Subjects and Setting Two groups, consisting of three individuals diagnosed with intellectual disabilities, participated in Experiment 1, 2, and/or 3. Table 2 and diagnosis as well as group assignment. Each subject was observed prio r to the study to confirm that he/she met the language criterion of a minimum vocal repertoire of two word All subjects excee ded this minimum repertoire and communicated using complete sentences In addition, a ll subjects were able to follow verbal instructions. Members of each group were classmates at the beginning of the experiment, and therefore, shared a history of social interactions. Formal observations of group interactions were not conducted prior to the start of t he experiment. Sessions were conducted at a special education school in either a classroom that contain ed typical materials (toys, books, desks, chairs, etc.) or a therapy room equipped with tables, chairs, and materials necessary for each session. All se ssions were 7 min in length and were divided into a 5 min task period followed by a 2 min break period. Two to 4 sessions were conducted per day, typically 3 to 5 days per week, based on subject and experimenter availability. Sessions were conducted only o n days that all group members were available. Preference Assessments Two assessments were conducted prior to the start of each experiment. A paired stimulus (PS) preference assessment (Fisher et al., 1992) was conducted with each subject to identify prefe rred edibles to be used as reinforcers in subsequent sessions. The subject first sampled each of nine edible items Subsequently, the therapist presented the items in pairs, and the subject was instructed to pick the preferred one. An approach response was recorded if the subject touched or
25 pointed to one of the items. The subject was allowed to consume the approached item, and the therapist removed the non approached item. If the subject did not emit an approach response on a given trial, the therapist rem oved both items and represented the trial. If still no approach response was emitted, the items were removed and the next trial was presented. Each edible was paired with every other item, and the position of items was counterbalanced across trials. Prefe rence rankings were calculated following the completion of the assessment based on the percentage of trials on which each item was approached. Only edible items that all group members selected on 75% or more of trials were selected for use as reinforcers. A multiple stimulus without replacement (MSWO) preference assessment (DeLeon & Iwata, 1996) was conducted with each subject to identify high preferred (HP) leisure items to be used in subsequent conditions. The subject was exposed to eight activity items prior to the assessment. The therapist then presented an array of all items to the subject, who was instructed to choose the preferred one. The subject was allowed access to the selected item for 30s, at which time the therapist removed the item from the a rray and instructed the subject to choose another item from the remaining array. When all items had been selected or when no item was selected across two presentations, a new trial began with the presentation of all items. Following the completion of two t rials, a preference hierarchy was calculated by dividing the number of times each item was selected by the number of times each item was presented in the array. HP activity items were defined as items that were ranked 1 st to 3 rd in the preference hierarchy Response Measurement and Reliability During the preference assessments, trained observers collected data on the subjects selections on each trial a mean of 76.3% (range, 33.3% to 10 0%) of trials Reliability scores were calculated by comparing recorded
26 selection of the same item. The number of trials with agreement s was divided by the total number of tria ls (agreements plus disagreements) and multiplied by 100. Mean agreement scores across subjects were 99.7% (range, 97.2% to 100%) for edible PS assessments, and 98.6% (range, 93.8% to 100%) for activity MSWO assessments. All baseline and treatment sessions were videotaped using a RCA handheld video recorder and were scored later for social interactions during group activities. Frequency data were collected across 13 categories of vocal and/or non vocal social interactions (see Table 2.3 for examples ) and were categorized as either positive or negative in tone. The categories of social interactions were derived from th ose used in previous research (Frankosky & Sulzer Azaroff, 1978; Speltz et al., 1982) as well as from pilot observations. Three types of interactions were summarized as Facilitate performance : Offer help was defined as providing information or providing materials related to the task to another group member, Explain was defined as any statement about the current contingency or session, and Demonstrate/Prompt was defined as any instruction or demonstration of action directed at another group member. Perform others work was scored if one group member completed the work assigned to another group member. Praise/Encouragement includ ed both vocal and non vocal interactions that expressed approval of the current behavior of another group member. Threaten/Criticize/Compl a i nt included statements about aversive results for another group negative comments to another memb er of the group about his/her appearance or behavior, and statements expressing disappointment or annoyance of the current situation. Seek Help was divided into two subcategories: Ask for Information was defined asking a question about the task or contingency that required a verbal response from another group member and Ask for Assistance was defined as asking for help that required a physical response
27 from another group member. Both Self Instructio n defined as statements about the current contingency not directed at another group member and Tact of Task, defined as statements about as subcategories under Nonspecific task related comment. Sabotage included misleading or untrue statements /information about the contingency, prompt s for other group members to engage in non task related behavior, and attempts to Non task related comments included any statement or question unrelated to the task materials, task behavior, or reinforcers. Other Problem Behavior was defined as aggression (hitting, kicking, throwing objects), self injurious behavior, or destruction to prope rty. Social interaction data were summarized for each subject and each group as rate of positive and negative interactions per session component (task and break). The specific types of positive and negative interactions were summarized as proportion of tot al interaction s per condition, which was calculated by summing the total number of interactions per condition and dividing the number of positive or negative interaction s within a given category by the total number of interactions within that condition. A n independent observer collected data during 30.5% (range, 28.4% to 37%) of all sessions across groups and experiments. Each videotaped session was time stamped which allowed two observers to independently watch and score the session at different times. R eliability was determined for social interactions by dividing each session into consecutive 10 s intervals, dividing the smaller number of responses by the larger number of responses for each interval, averaging the fractions across all intervals, and mult iplying by 100. Mean agreement scores were then calculated across categories for all groups and experiments. Mean agreement was 99.3% (range, 70.8% to 100%). Lower agreement scores usually were the product of
28 disagreements in tone (positive versus negative) across observers while both observers agreed on the category of the interaction. General Procedure and Experimental Design All sessions were 7 min in length and were divided in to continuous 10 s intervals. The first 5 min of each session consis ted of a task period, and the last 2 min of each session consisted of a re inforcement, or break period. An array of HP activity items was present across all task periods but not during the break period that followed each task period. Break periods were co nducted following every task period, including task periods in which the subjects did not meet the criteria for reinforcement. The inclusion of the break period allowed for the examination of social interactions following the delivery (or non delivery) of consequences for performance. Prior to each session, the therapist read a scripted instruction to the group ( see Table A 1 ), prompted all group members to engage in the task, and reviewed the contingency for the behavior of each group member. Following the task period, the therapist instructed all subjects to of edibles earned and/or lost, and delivered reinforcement to the corresponding group members. The follow ing conditions were evaluated in an ABAC ADAX reversal design in which (A) was Baseline, ( B ) the Independent Contingency, (C) the Competitive Contingency, (D) the Cooperative Contingency, and (X) a replication of the most effective contingency, if one was i dentified. Different colored materials and t shirts worn by the subjects served as discriminative stimuli (S D ) and correspond ed to the contingencies in place across conditions as follows: Baseline: All subjects were given white t shirts to wear and white m aterials. Independent and Competitive: Each subject was given different colored t shirt to wear and individual materials that corresponded to each t shirt color. Cooperative: All subjects were given orange t shirts and orange materials.
29 Table 2 1. Subject characteristics. Subject Age Group Diagnosis Experiment Cal 20 A Trainable mental handicap, speech/language impaired 1 Tess 19 A Intellectual Disability, language impaired 1 Diane 21 A Intellectual Disability 1 Glen 19 B Orthopedically impaired 2,3 Scott 19 B Intellectual Disability 2,3 Walt 18 B Intellectual Disability, speech/language impaired 2,3 Table 2 2. Results of edible and activity preference assessments Group (Experiment) HP edible item HP activit ies A (Exp 1) Smarties Paper and markers, Light Bright, beads with string, blocks Perfection, and piano B (Exp 2) Smarties Paper and markers, blocks piano, guitar, cars, and sports magazine B (Exp 3) Cheese Balls Paper and markers, blocks piano, guitar, Light Bright, Perfection game and clay
30 Table 2 3. Social interaction categories and e xamples Category Example Vocal Non Vocal Facilitate Performance Offer help Sharing materials Explain ou get snacks for sorting the Demonstrate / Prompt Model response Pointing to materials Perform others work Sorting into someone Praise / Encouragement Thumbs up Pat on the back Threaten / Criticize / Complaint Shake head in disappointment Seek Help Ask for information Ask for assistance Signing help, holding items up Nonspecific task related comment Self instruction Tact of task Non task related comments Stealing materials work Other Problem Behavior Vocal stereotypy Aggression, disruption
31 CHAPTER 3 EXPERIMENT 1: REINFO RCEMENT OF SKILL ACQ UISITION The purpose of Experiment 1 was to examine the types of performance and social interactions generated by group contingencies applied to behavior acquisition. The effects of the contingencies were evaluated during a categorical sorting task that required th e subjects to sort picture cards onto corresponding category cards. Group A ( Diane, Cal, and Tess ) participated in Experiment 1. Response Measurement and Reliability Data were collected on the frequency of correct and incorrect sorting for each subject a nd were summarized as the frequency of responses per subject and as a group mean per session. Correct and incorrect sorting were scored as product measures after the completion of the 5 min task period. Correct sorting was defined as a single picture on th e corresponding category board. Incorrect sorting was defined as a picture on a category board that did not correspond to the category of the picture card A n independent observer collected data during 51.5% (range, 42% to 60%) of all sessions across subj ects. Reliability for the number of pictures sorted correctly and incorrectly was determined by dividing the smaller number of responses scored by the larger number of response scored and multiplying by 100. Mean agreement for cards sorted correctly and incorrectly across subjects was 99.4% (range, 91.4% to 100%) and 99.2% (range, 75% to 100%), r espectively. Procedure During each session, the subjects were seated at a table with task materials. Sorting .5 ing to each of the five categories: play, work, hospital, holiday, and nature. Three sets of materials consisting of 50 sorting cards (10
32 cards per category) and five category boards were rotated throughout baseline, independent, and competitive contingenc ies and all sets of materials were combined during the cooperative contingency. Individual sets of materials were identical except that the picture cards for each category differed such that when the sets were combined there would be no duplicate pictures with a given category. An array of HP activity items was placed on an adjacent table. Subjects had the opportunity to earn up to 10 pieces of a preferred edible per session for correct sorting. Baseline. Following the 5 min task period, the therapist cou nted the number of cards sorted correctly and incorrectly. No edible reinforcement was delivered for correct sorting during baseline. All subjects sat at an empty break table for 2 min following each task period. Independent. Each subject earned one edibl e for every 5 pictures sorted correctly (fixed category cards and placed one edible on a plate for every 5 pictures sorted correctly. The therapist stated the number of edibles earned by each subject, and subjects were allowed to consume their earned edibles during the subsequent 2 min break period. Competitive. The individual in each group who sorted the most pictures correctly earned one edible for every five pictures sorted correctly. In the case of a tie, such that two or more subjects correctly sorted the same number of pictures, the number of earned edibles was divided among the on, stated how many pictures each individual correctly sorted, stated the individual(s) who sorted the most pictures correctly, and delivered the edibles to that or those individual(s). Cooperative. The group was given a group bin with 150 picture cards ( 30 per category) and one set of category cards. Each subject earned one edible for every 15 pictures sorted correctly by the group (FR15). For example, if the group correctly sorted 120 pictures correctly,
33 24 edibles were delivered to the group, eight piec es to each group member. The therapist placed time 5 pictu res were placed in front of all plates. The therapist stated and delivered the number of Results Figure 3 1 shows Group A e Cal, and Tess) performance on the sorting task. The top panel shows the mean frequency of correct and incorrect sorting for the group across the different contingencies, and the bottom panel shows individual performance by the three group member s. All three contingencies resulted in increased levels of correct sorting at the group level. However, somewhat different patterns of responding across the contingencies were observed at the individual level. Group A sorted during only the first session of the initial baseline condition. A pattern characteristic of acquisition was observed at both the individual and group level during the independent contingency. That is, the frequency correctly sorted cards increased, whereas the frequency of incorrectly sorted cards decreased (although only slightly) across this condition. Sorting was highly variable during the second baseline condition. The competitive contingency resulted in an immediate increase in mean correct sorting to a level slightly higher than that observed during the independent contingency and mean correct sorting continued to increase slightly across this condition. Similarly, rates of incorrect sorting, low to begin with, de creased gradually during the competitive contingency condition. Non e of the subjects in Group A sorted during the third baseline condition. Correct sorting increased during the cooperative contingency, although not as high initially as was observed during the competitive contingency. However, by
34 the fourth session of this condition, responding increased further to levels similar to those observed during the competitive contingency. Near zero levels of incorrect sorting were observed throughout this condition. In s ummary, all three contingencies independen t, competitive, an d cooperative baseline levels. However, the competitive and cooperative contingencies resulted in somewhat superior effects as compared to the effects of the independent contingency. The bottom panel of Figure 3 1 shows the performance data for individual group members. Of the three group members, Tess engaged in the highest frequency of sorting during the independent contingency. Diane and Cal engaged in similar levels of correct sorting during the of subjects sorted at higher frequencies during the competitive contingency relative to the ndition, whereas Cal and during the cooperative contingency as compared to individual performance during both the independent and competitive contingencies : s correct sorting decreased as compared to her performance during the competitive contingency. Thus, although the competitive and cooperative contingencies resulted in similar patterns of responding at the group level, the same was true for only one of the three (Diane) individual perfor mance. Figure 3 2 shows the rate of positive and negative interactions across the task and break periods for Group A. All types of interactions increased across both the task and break periods during the initial baseline condition. An initial decrease in all interactions occurred with the
35 implementation of the independent contingency. Positive interactions during the task period then increased before again decreasing to moderate levels. The group engaged in moderate to high, but variable rates of positive interactions during the break period of the independent contingency. Negative interactions remained low during the task period but increased during the break period of the independ ent condition During the second baseline condition, positive interactions d uring the task and break periods were similar to those observed in the initial baseline. Negative interactions were variable during the task period, and initially high rates of negative interactions occurred during the break period followed by a decreasing trend to moderate levels. Positive interactions increased during the task period of the competitive contingency, whereas positive interactions during the break period occurred at moderate levels. Near zero levels of negative interactions occurred during the task period of the competitive contingency, whereas nega tive interactions remained low during the break period. All interactions increased during both the task and break periods of the third baseline condition; positive interactions increased to modera te rates, and negative interactions remained relatively low. During the cooperative contingency, the group engaged in moderate rates of positive interactions across the task and break periods. Negative interactions throughout the cooperative contingency we re low but variable Figure 3 3 shows the proportion of positive and negative interactions that occurred across nine different social categories for each condition. Positive, non task related comments accounted for the largest proportion of interactions a cross the three baseline conditions followed by other problem behavior in the form of minor aggression (e.g., throwing shirts, slapping) and positive threats, criticisms, and complaints in the form of teasing. When the independent contingency was in effe ct, positive and negative threats, criticisms, and complaints accounted for nearly half of all interactions among the group members. As compared to baseline, there was
36 noticeable decrease in non task related comments and other problem behavior during the i ndependent contingency and a slight increase in non specific task related comments in the form of labeling the categories while sorting. The group engaged in the largest proportion of non task related comments and the smallest proportion of problem behavi or during the competitive contingency. The most significant change in the distribution of social interactions was observed during the cooperative contingency. The largest proportion of task related facilitative interactions was observed under the cooperative contingency. More specifically, the group engaged in labeling of the categories requesting help, and verbal and gestural prompting.
37 Figure 3 1. Frequency of sorting for Group A (Experiment 1). Mean frequency of correct (closed) and incorrect (open) sorting by the group is shown in the top panel. The frequency of correct (solid) and incorrect (dashed) sorting by individual group members is shown in the bottom panel.
38 Figure 3 2. Rate of positive and negative interactions ac ross conditions for Group A (Experiment 1).
39 Figure 3 3. Proportion of interaction s per social category across conditions for Group A (Experiment 1)
40 CHAPTER 4 EXPERIMENT 2: REINFO RCEMENT FOR BEHAVIOR REDUCTION The purpose of Experiment 2 was to examine the types of performance and social interaction s generated by group contingencies in which reinforcement was used to reduce inappropriate behavior. The effects of the contingencies wer e evaluated during a DRO procedure applied to out of seat behavior. Group B (Walt, Scott and Glen ) participated in Experiment 2. Response Measurement and Reliability Data were collected on the frequency of out of seat and in seat behavior of each subject and were summarized as percentage of intervals with out of seat behavior per subject and as a group mean per session. Out of seat was defined as no part of the subjec with In ontact A n independent observer collected data during 40.6% (range, 37% to 45.8%) of all sessions across subjects. Reliability was calculated by dividing the number of intervals in which both observers agreed on the occurrence or non occurrence of the target behavior by the total number of intervals (agreements plus disagreements), and multiplying by 100. Agreement scores for out of seat and in seat behaviors were 96.2% (range, 80% to 100%) and 96.7% (range, 80% to 100%), respecti vely. Procedure therapist assigned each subject a specific seat within the task area (left, right, or center), and this position was held constant across conditions. HP a ctivity items were not available to subjects in the designated task area; however, an array of HP items was placed in an adjacent area 5 feet
41 from the task area. Subjects had the opportunity to earn up to 10 pieces of a preferred edible per session for not engaging in out of seat behavior. Baseline. terms of the number of intervals in which he was out of his seat, but delivered no edible reinforcement for the absence of out of seat behavior. All subjects sat with empty plates during the subsequent 2 min break period during baseline. Independent. Each subject earned one edible for every three, 10 s intervals during which out of seat behavior was not observed (FR 3 intervals). Th e therapist counted the number of intervals for each individual that did not contain out of seat behavior, divided this number by three, and delivered the corresponding number of edibles. The therapist stated the number of edibles earned by each subject, a nd subjects were allowed to consume earned edibles during the subsequent 2 min break period. Competitive. The subject who did not get out of seat for the largest number of intervals earned one edible for every 3, 10 s intervals for which out of seat behavi or is not observed (FR3 intervals). The therapist counted the number of intervals for each individual that did not contain out of seat behavior, divided this number by three, and delivered the corresponding pieces of edible to the individual who accumulate d the largest number of in seat intervals. In the case of a tie, such that two or more subjects were not out of seat for the same number of intervals, the amount of earned edibles was divided among all subjects who met criterion for reinforcement. Coopera tive. All group members earned one edible for every three, 10 s intervals during which out of seat behavior was not observed across all group members. That is, all members were required not be out of their seats during an entire 10 s interval for that int erval to be counted towards reinforcement. If any one of the three subjects was out of his seat during any
42 part of an interval, that interval was not counted towards reinforcement. The therapist counted the number of intervals that did not contain any out of seat behavior, divided this number by group earned 5 edibles, so you each get 5 edi Results Figure 4 Scott and Glen ) performance data. The top panel shows the mean percentage of interval s of out of seat behavior for the group across the different contingencies, and the bottom panel shows the percentage o f interval of out of seat behavior for individual group members. The cooperative contingency resulted in the lowest percentage of out of seat behavior at both the group and individual level. The group was out of their seats between 60% and 10 0% of interv als during the initial baseline condition. Out of seat behavior remained high for the first two sessions of the independent contingency before dropping to low to moderate levels. Out of seat behavior increased to near 100% across the last three sessions of the second baseline condition. Low levels of out of seat behavior were observed initially during the competitive contingency. The of seat behavior then became highly variable throughout this condition with several sessions of near zero out of seat behavior followed by spikes of high levels of out of seat behavior. Out of seat behavior increased and continued to be highly variable across the following baseline condition. Out of seat behavior occurred at moderate levels for the first three sessio ns of the cooperative contingency; however, near zero levels were then observed across 18 consecutive sessions. We then returned to baseline before replicating the effects of the of seat behavior immediately increased to moderate to high levels. When we reintroduced the cooperative
43 of seat behavior immediate decreased to near zero levels and remained low for the next 22 sessions. The bottom panel of figure 4 1 shows performance at the individual level. Walt and Glen of seat behavior was somewhat variable during the initial baseline condition, whereas Scott of seat behavior was high throughout this condition. Individual performance was simi lar to the group mean during the independent contingency; however, Walt and Scott were out of seat more often than Glen during this condition. All three subjects showed an increase in out of seat behavior during the second baseline condition; Glen the most immediate. All of seat behavior were observed for the majority of sessions, but high levels of out of seat behavior also were observed periodically. S cott and Glen engaged in moderate levels of out of of seat behavior immediately dropped to near zero levels. After the third session, all group members engaged in low levels of out of seat behavior with Scott periodically getting out of his seat toward the end of the condition. During the final baseline condition, Scott and Glen of seat of seat be havior increased of seat behavior immediately decreased to near zero levels during the return to the cooperative contingency and remained low for the remainder of the condition, with the exception of one session during with Scott was out of his seat for approximately 30% of intervals. Figure 4 2 shows the rate of positive (top panel) and negative (bottom panel) interactions across the task and break periods of Experiment 2 for Group B. Similar patterns in the rate of positive and negative interactions were observed across the first three baseline conditions and the
44 independent and competitive contingencies. During these conditions, the group engaged in moderate rates of positive interactions and low to moderat e rates of negative interactions with more variability in rate of interaction during the break period as compared to the task period. Increasing trends in positive interactions were observed across the cooperative contingency whereas negative interactions decreased to consistently low levels. Positive interactions returned to previous levels during the final baseline condition. Negative interactions increased slightly during the final baseline condition but were more stable than previous baseline interacti ons. During the replication of the cooperative contingency positive interactions again increased. Low rates of negative interactions were initially observed, followed by slightly higher rates at the midpoint of this final condition. However, rates of negat ive interaction once again decreased to consistently low levels by the conclusion of the experiment. Figure 4 3 shows the proportion of positive and negative interactions that occurred across the nine different social categories during each condition. Pos itive non task related comments accounted for the largest proportion of all baseline interactions, followed by positive and negative threats, criticisms, and complaints. The most frequent topography of the latter category was threats of tattling for steal ing items and engaging in varied inappropriate behavior s. The proportion of non task related comments was lowest during the independent contingency, and an increase in both positive and negative threats, criticism, and complaints occurred during this condition relative to baseline An interesting finding was that the largest proportion of interactions scored under facilitate performance and seek help were observed during the independent contingency. These interactions often involved asking about or exp laining the contingency (earning for not getting out of our seats) as well as the definition of in seat behavior Durin g the competitive contingency, comments related to the task
45 and threats, criticisms, and complaints were observ ed at levels similar to those observed during independent contingency, whereas interactions scored as facilitate performance and seek help occurred at levels similar to those observed during baseline. The group engaged in largest proportion of non task rel ated comments and the smallest proportion of threats, criticisms, and complaints during the cooperative contingency. Non task related comments often related to daily and weekly activities such as basketball games and cooking class.
46 Figure 4 1. Percentage of intervals of out of seat behavior for Group B (Experiment 2). The mean percentage of intervals of out of seat behavior for the group is shown in the top panel. The percentage of intervals of out of seat behavior for individual group members i s shown in the bottom panel
47 Figure 4 2. Rate of positive and negative interactions across conditions for Group B ( Experiment 2).
48 Figure 4 3. Proportion of interaction s per social category across conditions for Group B (Experiment 2)
49 CHAPTER 5 EXPERIMENT 3: PUNISH MENT FOR BEHAVIOR RE DUCTION The purpose of Experiment 3 was to examine the types of performance and social interaction generated by group contingencies in which a response cost procedure was used to reduce inappropr iate (out of seat) behavior. Group B (Walt, Scott and Glen ) participated in Experiment 3. Response Measurement, Reliability, and Procedure Data were collected on in seat and out of seat behavior as in Experiment 2. A n independent observer collected data during 33.3% ( range, 32.4% to 35.2%) of all sessions across subjects. Reliability was calculated as described in Experiment 2. Agreement scores across subjects for out of seat and in seat behavior were 96.8% ( range, 86.7% to 100%) and 93% ( range, 76.7% to 100%), respectively. The general procedure was identical to Experiment 2. However, subjects began each session with 10 pieces of preferred edibles on their plates, but lost one edible for each interval in which out of seat behavior was recorded (FR1). Th at is, the therapist removed one piece of edible for each interval during which out of seat behavior occurred. Baseline. This condition was identical to the baseline condition of Experiment 2. Independent. The therapist counted the number of intervals in which out of seat behavior was observed for each subject and removed the number of edibles lost by each subject from their respective plates. The therapist then stated the number of edibles lost by each subj ect, and subjects were allowed to consume any remaining edibles during the subsequent 2 min break period. Competitive Snacks were delivered only to the subject with the fewest number of intervals of out of seat behavior. The therapist counted the number o f intervals for each
50 individual that contain out of seat behavior and removed one piece of edible for each interval of out of seat behavior. The therapist delivered any remaining edibles to the individual with the fewest number of intervals with out of sea t behavior. In the case of a tie in which an equal number of intervals with out of seat behavior was observed for two or more subjects, half or a third of the amount of remaining edible Cooperative. The therapist pla the session as in previous conditions. The therapist counted the number of intervals during which out of seat behavior was observed for any subject and removed one piece of edible from each s of seat behavior was observed. The Results Figure 5 1 shows the performance for Group B (Walt, Scott and Glen ) when out of seat behavior resulted in loss of reinforcement during Experiment 3. The top panel shows the mean percent intervals of out of seat behavior for the group across the d ifferent contingencies, and the bottom panel shows the percent intervals of out of seat behavior for individual group members. All contingencies independent, competitive, and cooperative resulted in near zero levels of out of seat behavior. Group B en gaged in high levels of out of seat behavior across all four of the baseline conditions. This pattern of responding was somewhat delayed during the third baseline condition ; however, near 100% of intervals with out of seat behavior was observed across the last four sessions of this condition. Immediate decreases to near zero levels of out of seat behavior were observed during all contingencies. In fact during the independent contingency there was only one session in which out of seat behavior was observed and out of seat behavior
51 was not observed at any point during the cooperative contingency. Although out seat behavior occurred during several sessions of the competitive contingency, the mean percentage of intervals with out of seat behavior never exceede d ten percent during the fist exposure and remained near zero during the second exposure of this condition As the bottom panel of figure 5 1 shows, there was some individual variability in s highly variable during the of seat behavior was observed whereas Scott competitive conditions shows Walt and Scott were out of their seats more often than Glen who was only out of his seat during one session of the fist exposure to the competitive co ndition Figure 5 2 shows the rate of positive (top panel) and negative (bottom panel) interactions across the t ask and break periods of Experiment 3. Similar rate s of positive interaction during the task or break period were observed across all baseline and treatment condition s The group engaged in low to moderate rates of negative interactions during both the task and break periods across all conditions ; however, the level of variability differed somewhat across conditions. Variability in negative interactions decreased during the cooperative condition and remained low for the remainder of the experiment. Figu re 5 3 shows the proportion of positive and negative interactions that occurred across the nine different social categories for each condition. Non task related comments accounted for the largest proportion of interactions across all c onditions A larger p roportion of positive and negative threats, criticisms, and c omplaints occurred during the inde pendent and competitive contingencies relative to baseline and the cooperati ve contingency. Similar to Experiment 2, the largest proportion of non specific task related comments occurred during the independent
52 contingency. An interesting and somewhat expected finding was that the largest proportion of sabotage in the form of removing the seat of another group member or pushing other group members out of their s eat s occurred during the competitive contingency. A r elated finding was that group members often requested hel p and prompted each other to assist in instances of sabotage during the competitive condition
53 Figure 5 1. Percentage of intervals of out of seat behavi o r across conditions for Group B (Experiment 3). The m ean percentage of intervals of out of seat behavior for the group is shown in the top panel. The percentage of intervals of out of seat behavior for individual group member s is show n in the bottom panel.
54 Figure 5 2. Rate of positive and negative social interactions across conditions for Group B (Experiment 3).
55 Figure 5 3. Proportion of interaction s per social category across conditions for Group B (Experiment 3)
56 CHAPTER 6 DISCUSSION The general purpose of this study was to examine not only the types of performance produced by independent, competitive, and cooperative group contingencies but also the patterns of social interaction among group members We extended previous research by examining positive and negative interactions across specific types of social interactions including both facilitative and suppressive categories. The combination of the three experiments attempted to highlight differences in performance and social interaction across variations of group contingencies that applied reinforcement and punishment procedures to behavior acquisition and behavior reduction. Group performance during Experiment 1 was consistent with previous research that has demonstrated the superiority of group over individual (independent) contingencies in their application to behavior acquisition (Fran kos ky & Sulzer Azaroff, 1978; Speltz et al., 1982). At the group level, both the competitive and cooperative contin gencies produced higher frequencies of correct sorting as compared to sorting during the independent contingency. One possible explanation for the difference in performance across the independent and group contingencies might be a sequence effect. S ubjects time under any one type of reinforcement contingency. To examine this possibility, future research might include replications of each contingency within a single group. The current study extended previous research by including analyses of performanc e for each member of the group, which allowed the examination of changes in individual performance across conditions. The group members in Experiment 1 showed somewhat different patterns of performance a cross conditions : One subject (Diane) performed at similar levels across the competitive and cooperative contingencies, another subject (Tess) performed best under the
57 competitive contingency, and the third subject (Cal) performed best under the cooperativ e contingency. An interesting note is that the three subjects also showed different patterns of responding during the independent contingency, with Tess performing at the highest level, Diane performing at moderate levels, and Cal performing at the lowest level of the three subjects. An interesting idea for future research might be to determine whether responding during the independent contingency is predictive of patterns of responding during subsequent group contingencies. Several interesting patterns of social interactions were observed during Experiment 1. Although no difference in the rate of positive interactions was observed across contingencies, the distribution of positive interactions across topographies varied across contingencies. Diane and Cal e mitted non specific task related comments in the form describing task materials and seeking help by asking for information about the task materials across all contingencies. Both Diane and Cal requested help from Tess across contingencies; however, Tess wa s more likely to reciprocate whereas she reciprocated both group likely to prompt one another and to ask fo r and provide assistance with the sorting task during the cooperative contingency. Our findings are consisted with those of McCarty Griffin, Apollo ni, and Shores correlated with an increase in performance on math worksheets during an interdependent group contingency. In summary, the patterns of interaction among group members seem ed to correspon d to performance during the competitive contingency might have been facilitated, at least in part, by
58 information provided by Tess this contingency might have been hindered by the lack of responses to his requests for information. Similarly, the convergence of individual performances during the cooperative contingency co varied with relative contributions in term of offering and recei ving assistance in completing different components of the task. The results of Experiment 2 further support the superiority of group contingencies over individual contingencies. Although the competitive contingency produced superior results to the indepen dent contingency, out of seat behavior was consistently lower during the cooperative contingency relative to the competitive contingency. Therefore, the cooperative contingency was determined to be the most effective contingency in reducing out of seat beh avior via differential reinforcement. Furthermore, these results were consistent across the performance of each individual group member. One reason for the difference in the relative efficacy of the competitive contingency across Experiments 1 and 2 might be that the momentary probability of reinforcement for each group member was more salient during Experiment 2. That is, the time spent out of seat was more easily observed by the subjects relative to the frequency of correctly sorted cards. Patterns of res ponding at the individual level during Experiment 2 support this hypothesis: When a group member was out of his seat in a given session, he was generally out of his seat for the majority of that session. Changes in the rate of positive and negative intera ctions during Experiment 2 corresponded to group performance. That is, the highest rates of positive interactions and the lowest rates of negative interactions among group members were observed during the cooperative contingency replicating the findings r eported by Frank os ky and Sulzer Azaroff
59 (1978). In contrast to the patterns of interaction observed during Experiment 1, similar rates of positive and negative interactions were observed during task and subsequent break periods of Experiment 2. This differ ence might be explained by differences in task structures across the two studies: Engagement with the sorting task (Experiment 1) likely required attention to the task materials and therefore may have decreased the opportunity for interact ions among group members. On the other hand, opportunity of interaction due to close proximity. In fact, non task related comments accounted for the largest proportion of interactions across all conditi ons. Unlike the results of Experiment 1, increases in facilitative interactions were not observed during the cooperative contingency of Experiment 2. This difference might be explained by the behaviors targeted across the experiments: Skill acquisition in Experiment 1 vs. behavior reduction in Experiment 2. Individual group members required assistance to correctly sort during the cooperative contingency of Experiment 1, where as no assistance was required to gency of Experiment 2. The largest proportion of non task related comments and the smallest proportion of threats, criticisms, and complaints were observed during the cooperative contingency. It seems likely that these changes in the distribution of interactions were a direct pr oduct of the cooperative nature of contingency. Threats, criticisms, and complaints directed at other group members might produce social avoidance that include s out of seat behavior, which would in turn decrease the probability of reinforcement for all gro up members. Group performance during Experiment 3 extended the findings of Axelrod (1973), who reported that both independent and interdependent response cost procedures were effective in decreasing classroom disruptions. We found that all contingencies independent, competitive,
60 and cooperative were equally effective in reducing out of seat behavior. I ndividual performance in Experiment 3 was similar to that in Experiment 2 No difference in the rate of positive or negative interactions were found acro ss contingencies, and the levels of interaction were consistent across both the task and break periods. Similar to the results of Experiment 2, n on task related comments again accounted for the largest proportion of interactions across all conditions of Ex periment 3 W e reintroduced the competitive contingency at the end of the experiment to examine performance under this contingency following a history with the cooperative contingency. Although out of seat behavior was low ac ross both exposures to the com p et it ive contingency, performance was suppressed to a greater degree following the cooperative contingency. D ifferences in performance during the two exposures to the competitive contingency were correlated to changes in the interactions among group member s. During both exposures to the competitive contingency, Glen attempted to sabotage other group members by pushing them off of their seats and attempting to remove their seats In addition, Glen often solicit ed help from another group member to sabotage th e performance of the third member. Instances of these behaviors were more frequent during the initial exposure to the competitive contingency. Although Glen continued to engage in sabotage during the second exposure to the competitive contingency, the othe r group members were more likely to actively resist sabotage directed at them and to ignore Glen Together, the three experiments highlight correlation s between changes in performance and changes in interactions among group members during different group contingencies. Although it is the contingencies of reinforcement (or punishment) that directly influence performance, corresponding changes in social interactions may also be observed when applying
61 co ntingencies to group behavior. In the current study, d ifferences in cooperative and competitive interactions were evident across group contingencies applied to behavior acquisition and behavior reduction and to a lesser degree, reinforcement and punishmen t procedures applied to behavior reduction. Future research might attempt to isolate the effects of social interactions during group contingencies. One way to further examine the effects of social interactions on performance during group contingencies migh t involve the inclusion of a confederate group member to introduce different types of interactions across contingencies. Detailed analyses of both performance and social interactions are critical to the understanding of how these two behaviors interact acr oss different types of group contingencies. Although previous research has demonstrated changes in the overall rate of positive and negative interactions across different contingencies (Frankosky & Sulzer Azaroff, 1987; Shore s et al., 1976 ; Speltz et al. 1 982 ), no study to date has reported on the specific types of interactions observed across types of group contingencies. The inclusion of such information is helpful because it is possible for the l evel of interaction to remain consistent across contingenci es while the distribution in the type of interactions may change, as seen in Experiment 1. Although the current stud ies provide a preliminary analysis of the relation between performance and social interactions during independent, competitive, and cooperat ive group contingencies, future research should continue to examine the relation between these two variables within group contexts.
62 APPENDIX PRESESSION INSTRUCTIONS Table A 1. Scripted instructions across conditions and experiments Condition Experiment : Instruction Baseline 2: Independent Competitive 2: person with the fewest intervals of out of Cooperative 1:
63 LIST OF REFERENCES A lexander, R. N., Corbett, T. F., & Smigel, J. (1976). The effects of individual and group consequences on school attendance and curfew violations with predelinquent adolescents. Journal of Applied Behavior Analysis, 9 221 226. Allen, L., & Iwata, B. (198 0). Reinforcing exercise maintenance: Using existing high rate activities. Behavior Modification, 4 337 354. Alric, J.M., Bray, M.A., Kehle, T.J., Chafouleas, S.M., & Theodore, L.A. (2007) A comparison of independent, interdependent, and dependent group contingencies with randomized reinforcers to increase reading fluency. Canadian Journal of School Psychology 22 81 93. Axelrod, S. (1973). Comparison of individual and group contingencies in two special classrooms. Behavior Therapy, 4, 83 90. Barrish, H. H., Saunders M., & Wolfe, M.M. (1969). Good behavior game: Effects of i ndividual contingencies for group conseque nces on disruptive behavior in a classroom. Journal of Applied Behavior Analysis, 2 119 124. Buckholdt, D., & Wodarski, J. (1978). The effects of different reinforcement systems on cooperative behaviors exhibited by children in classroom contexts. Journa l of Research & Development in Education, 12 (1), 50 68. Coleman, R. (1970). "A conditioning technique applicable to elementary school classrooms." Journal of Applied Behavior Analysis 3 293 297. Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied behavior analysis (2nd ed.) Upper Saddle River, NJ: Pearson Prentice Hall. Cosden, M. & Harring, T. (1992). Cooperative learning in the classroom: Contingencies, group interactions, and students with special needs. Journal of Behavioral Educatio n, 2 (1), 53 71. Crouch, P. L., Gresham, F.M., & Wright, W.R. (1985). Interdependent and independent group contingencies with immediate and delayed reinforcement for controlling classroom behavior. Journal of School Psychology, 23 177 187. DeLeon, I. G. & B. A., Iwata (1996). Evaluation of a multiple stimulus presentation format for assessing reinforcer preferences. Journal of Applied Behavior Analysis, 29 519 533. Drabman, R. Spitalnik, & K. Spitalnik (1974). Sociometric and disruptive behavior as a function of four types o f token reinforcement programs. Journal of Applied Behavior Analysis 7 93 101
64 Fabiano, G.A., Pelham, W.E., Karmazin, K., Kreher, J., Panahon, C.J., & Carlson, C. (2008). A group contingency program to improve the behavio r of elementary school students in a cafeteria. Behavior Modification 32 121 132. Fisher, W., Piazza, C.C., Bowman, L.G., Hagopian, L.P., Owens, J.C., & Slevin, I. (1992). A comparison of two approaches for identifying reinforcers for persons with seve re and profound disabilities. Journal of Applied Behavior Analysis 25 491 498. Frankosky, R. & Sulzer Azaroff, B. (1978). Individual and group contingencies and collateral social behaviors. Behavior Therapy, 9 (3), 313 327. Gresham, F.M. & Gresham, G.N. (1982). Interdependent, dependent, and independent group Contingencies for controlling disruptive behavior. The Journal of Special Education 16, 101 110. Harris, W. W. & J. A. Sherman (1973). Use and analysis of the "good behavior game" t o reduce disruptive classroom behavior. Journal of Applied Behavior Analysis, 6 405 417. Heering, P.W. & Wilder, D.A. (2006). The use of dependent group contingencies to increase on task behavior in two general education classrooms Education and Trea tment of Children, 29 (3), 459 468. Herman, J. A., DeMontes, A. I., Dominguez, B. J., Montes, F., & Hopkins, B. L. (1973). Effects of bonuses for punctuality on the tardiness of industrial workers. Journal of Applied Behavior Analysis, 6 563 570. Hopkins, B. L., R. C. Schutte, Garton, K.L. (1971). The effects of access to a playroom on the rate and quality of printing and writing of f irst and second grade students. Journal of Applied Behavior Analysis 4 77 87. Kirby, C.K, Kerwin, M.E., Carpenedo, C.N., Rosenwasser, B.J, & Gardener, R.S. (2008). Interdependent group contingency management for cocaine dependent methadone maintenance patients, Journal of Applied Behavior Analysis 41 (4), 579 595. Kelshaw Leverin g, K., Sterling Turner, H., Henry, J.R., & Skinner, C.H. (2000). Randomized interdependent group contingencies: Group reinforcement with a twist Psychology in the Schools, 37 (5), 523 533. Kohler, F., Strain, P. S., Hoyson, M., Davis, L W.M. Donnia, & N. Rapp (1995). Using a group oriented contingency to increase social interactions between children with autism and their peers: A preliminary analysis of corollary supportive behaviors. Behavior Modification, 19 10 32. Liebson, I., Cohen, M., & Faillace L. A. (1972). Group fines: Technique for behavioral control in a token economy. Psychological Reports, 30 895 900.
65 Litow, L., & Pumroy, D. K. (1975). A brief review of classroom group oriented contingencies. Journal of Applied Behavior Analysis, 8 341 347. Lloyd, J. W., Eberhardt M.J., et al. (1996). Group versus individual reinforcement contingencies within the context of group study conditions. Journal of Applied Behavior Analysis, 29 189 200. Long, J.D. & Williams, R.L. (1973). The comparative effectiveness of group and individual contingent free time with inner city junior high school students. Journal of Applied Behavior Analysis 6, 465 474. Lutzker, J. R., & White Blackburn, G. (1979). The Good Prod uctivity Game: Increasing work performance in a rehabilitation setting. Journal of Applied Behavior Analysis, 12 488. Marholin, D. and D. Gray (1976). "Effects of group response cost procedures on cash shortages in a small business." Journal of Applied Behavior Analysis 9 25 30 McCarty, T., Griffin, S., Apolloni, T., & Shores, R. E. (1977). Increased peer teaching with group oriented contingencies for arithmetic performance in behavior disordered adolescents. Journal of Applied Behavior Analysis, 10 313. McNally, R., Norusis, P., Gentz, S., & McConathy, L. (1983). Use of self delivered reinforcement and group contingency management to increase productivity of severely retarded workers. Psychological Reports, 52 (2), 499 503. Michaels, J. W. (1977). Classroom reward structures and academic performance. Review of Educational Research, 47 87 98. Neumann, J. K. (1978). A classification scheme of social conditions of reinforcement within group operant systems. Journal of Applied Behavior Analysis, 11 435. Olympia, D., Sheridan, S., Jenson, W., & Andrews, D. (1994). Using student managed interventions to increase homework completion and accuracy. Journal of Applied Behavior Analysis, 27 (1), 85 99. Packard, R.G. (1970). The cont complex b ehavior. Journal of Applied Behavior Analysis 3 (1), 13 28. Phillips, E. L., Phillips, E.A, Wolf, M.W., & Fixen, D.L. (1973). Achievement Place: Development of the elected manager system. Journal of Applied Behavior Analysis, 6 541 561 Pigott, H. E., Fantuzzo, J. W., & Clement, P.W (1986). "The effects of reciprocal peer tutoring and group contingencies on the academic performance of elementary school children ." Journal of Applied Behav ior Analysis 19 93 98
66 Putnam, R.F., Handler, M.W., Ramirez Platt, C.M., & Luiselli, J.K. (2003). Improving student bus riding behavior through a whole school intervention. Journal of Applied Behavior Analysis, 36(4) 583 590. Reid, D., Schuh Wear, C., & Brannon, M. (1978). Use of a group contingency to decrease staff absenteeism in a state institution. Behavior Modification, 2 251 266. Reinhardt, D., Theodore, L.A., & Bray, M.A., & Kehle, T.J. (2009). Improving homework accuracy: Interdependent group contingencies and randomized reinforcers. Psychology in Schools 46(5), 471 488. Ritschl, C., Mongrella, J., & Presbie, R.J. (1972). Group time out from rock and roll music and out of seat behavior of handicapped children riding a school bus. Psychological Reports 31 967 973. Salend, S.J. & Lamb, E.A. (1986). Effectiveness of a group managed interdependent contingency system Learning Disability Quarterly 9 268 273. Shapiro, E., Albright, T., & Ager, C (1986). Group versus individual contingencies in modifying two disruptive adolescents' behavior. Professional School Psychology, 1 105 116. Shapiro, E. S. & Goldberg, R. (1986). A comparison of group contingencies for increasing spelling performance among sixth grade students. School Psychology Review, 15 546 557. Shores, R., Apolloni, T., & Norman, C. (1976). Changes in peer verbalizations accompanying individual and group contingencies to modify on task behavior. Pe rceptual and Motor Skills, 43 1155 1162. Skinner, C. H., Cashwell, T. H., & Skinner, A. L. (2000). Increasing tootling: The effects of a peer mediated group contingency program on students' reports of peers' prosocial behavior. Psychology in the Schools, 37 (3), 263 270. Slavin, R. E. (1977). Classroom reward structure: An analytical and practical review. Review of Educational Research, 47 633 650. Speltz, M.L., Shimamura, J.W., & McReynolds, W.T. (1982). Procedural variations in Journal of Applied Behavior Analysis, 15 533 544. Swain, J. J., Allard, G. B., & Holborn, S. W. (1982). The good toothbrusing game: A school based dental hygiene program for inc reasing the toothbrushing effectiveness of children. Journal of Applied Behavior Analysis, 15 171 176. Switzer, E.B., Deal, T.E., & Bailey, J.S. (1977). The reduction of stealing in second graders using a group contingency. Journal of Applied Behavior Analysis 10 267 272.
67 Tankersley, M. (1995). A group oriented contingency management program: A review of research on the Good Behavior Game. Preventing School Failure, 40 19 24. Theodore, L. A., Bray, M. A., & Kehle, T. J. (2004). A comparative study of group contingencies and randomized reinforcers to reduce disruptive classroom behavior. School Psychology Quarterly, 19 253 271. Tingstr o m, D.H., Sterling Turner, H.E., & Wilczynski, S.M. (2006). The good behavior game: 1969 2002. Behavior Modifica tion, 30 225 253. Williamson, D. A., Williamson, S. H., Watkins, P. C., & Hughes, H. H. (1992). Increasing cooperation among children using dependent group oriented reinforcement contingencies. Behavior Modification, 16 414 425. Witt, J. C., & Elliot, S. E. (1982). The response cost lottery: A time efficient and effective classroom intervention. Journal of School Psychology, 20 155 161.
68 BIOGRAPHICAL SKETCH Jill Harper first studied behavior analysis at the University of New Hampsh ire where she received her B.S. in Family Studies and B.A. in Psychology in 2003. She then entered the M asters of Science in Applied Behavior Analysis (MABA) program through Northeastern University under the supervision of Dr. Jason Bourret. While working toward her ma degree Jill held several clinical positions at the New England Center for Children (NECC), a private, educational facility that serves individuals with Autism and other developmental disabilities. J of Florida (UF) in the fall of 2007. Jill participated in a variety of research and clinical activities while at UF. She worked on an array of research projects that involved topics ranging from the assessment and treatment of severe problem behavior to s kill acquisition. While at UF, Jill provided behavioral services to individuals with various developmental diagnoses such as autism, mental retardation, and Prader Willi syndrome, across a residential treatment facility and a public, special education scho ol. Additionally, Jill engaged in several different teaching activities including serving as a teaching assistant for an introductory psychology course and a behavior analysis lab course, as well as instructing an introductory behavior analysis course. Jil l has recently accepted a position as Assistant Director of Clinical Services, Research, and Training at Melmark New England. Sh e hopes to also begin teaching courses in behavior analysis at an affiliate university.