An Analysis of the non-verbal social behavior of a child in a kindergarten setting, by Jerome R. Freund

Material Information

An Analysis of the non-verbal social behavior of a child in a kindergarten setting, by Jerome R. Freund
Freund, Jerome Robert, 1940- ( Dissertant )
Wolking, William D. ( Thesis advisor )
Avila, Donald L. ( Reviewer )
Pennypacker, Henry S. ( Reviewer )
McGee, Richard K. ( Reviewer )
Ramey, Madelaine ( Reviewer )
Place of Publication:
Gainesville, Fla.
University of Florida
Publication Date:
Copyright Date:
Physical Description:
viii, 84 leaves. : illus. ; 28 cm.


Subjects / Keywords:
Bowls ( jstor )
Cultural customs ( jstor )
Experimental psychology ( jstor )
Observational research ( jstor )
Recordings ( jstor )
Research methods ( jstor )
Social events ( jstor )
Social interaction ( jstor )
Social structures ( jstor )
Social systems ( jstor )
Child development ( lcsh )
Dissertations, Academic -- Psychology -- UF ( lcsh )
Kindergarten ( lcsh )
Nonverbal communication ( lcsh )
Psychology thesis Ph. D ( lcsh )
Social interaction ( lcsh )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


A descriptive and experimental study was made of the ways in which one kindergarten child touched his classmates and was touched by them. The descriptive portion of the study attempted to determine whether the touching responses demonstrated a structure both within and across sets of responses. The experimental portion of the study attempted to determine what stability the social system would demonstrate when an experimental intervention was introduced. The social system of physical contact was studied by using a multiple baseline operant paradigm. Response rates of ten variables for the target child's responses and his classmates' responses respectively were collected during the baseline phase. During the phase of experimental manipulation one of the carpet child's response rates was increased via the use of a form of token economy. Multiple baselines were collected during this phase as well as during the final phase, the return to baseline. During this last phase no experimental manipulations were in force. Two statistical methods were used in the data analysis, factor analysis and correlational analysis. Factor structures were found within each set of responses. Forty significant correlations were found when comparing across phases the respective rates of the two sets of responses. Stability was found across phases in terms of the factor structure, the variable composition of factors and the cross-correlation of response variables. It was concluded that the social system of physical contact studied in the kindergarten setting exhibited a definite structure, both within and across sets of responses. This structure demonstrated three types of self-regulation; mechanisms when the system was experimentally disturbed. Additionally it was observed that the behavior repertoires of the target child and his classmates successively approximated each other across phases. The data suggested that the phase of experimental manipulation stimulated this mutual approximation.
Thesis -- University of Florida.
Bibliography: leaves 77-83.
Additional Physical Form:
Also available on World Wide Web
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
022657014 ( AlephBibNum )
13924691 ( OCLC )
ADA4874 ( NOTIS )


This item has the following downloads:

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EF954RQ01_TT2ILU INGEST_TIME 2017-07-14T22:03:58Z PACKAGE UF00097612_00001


ilysis of the Non-verbal Social Behavior of a Child in a arten Setting By Jerome H. Freund 3IL OP i iRTIA . IE OF ... OS PHILOSOPHY. UNI 1'-




To Elizabeth, Robbie and Natashya whose love and patience created the conditions for completion of this manuscript.


Acknow] edgments It is with great appreciation that the help and assistance of the following individuals are acknowledged: Dr. Don Avila, Dr. Richard McGee, Dr. Madelaine Raraey, Dr. Henry Fennypacker and Dr. William Wolking, members of my committee whose gentle corrections and kind advice were much appreciated during my labors; Dr. Warren and Sara Rice, whose technical advice and hours of labor re critical for the completion of the data analysis. irs was truly a labor of love, a gift of friendship the author will not soon forget; and finally Kathy Evans, who worked tirelessly and painstakingly to transform my rough copy into finished form. A special thank you goes to Miss Mary Bell, the kindergarten teacher who so ciously shared her class and her time with the experimenter. Final appreciation goes to "David" and his classmates, the subjects of the study, who accepted the experimenter into their midst with such grace and charm. iii


Table of Contents Acknowledgments ij^ List of Tables v Abstract vll Literature Review and Statement of Problem i Contribution of Social Psychology 2 Social Environment: The Operant Approach 3 Summary of Literature fleview 7 Strategy of the Study 8 Research Tactics 10 Summary of Research Strategy 13 Method ^ Selection of Behavior 14 Selection of Research Site 15 Recording Method X 6 Baseline ^g Cons equation 2 Return to Baseline 22 Methods of Data Analysis 22 Results 26 Experimenter Effects 26 Reliability 30 Formal Factor Structure 31 Factor Composition by Variables kj Cross-correlation Study 4-7 Discussion and Summary 52 Recommendations for Future Research 58 Appendices A. Category Definitions 63 B. Selection of Consequated Variable 65 C. Building the Kesponse Chain 67 D. Daily Rates of "David" and "Class" 74 List cf References 77 Biographical Sketch g^ iv


List of Tables 1 . Daily Rates of "Attending" Behavior Class Rates 27 2. Summary of the Reinforcement of "Attending" Behavior Table of Daily Rates 29 3. Reinforcement of "Attending" Eehavior Comparison of Means Differences 29 ^. Agreement Percentages of Two Observers 31 5. Orthogonal Factor Solution "David" Baseline 32 6. Orthogonal Factor Solution "Class" Baseline 33 7. Oblique Factor Solution "David" Baseline 3^ 8. Oblique Factor Solution "Class" Baseline 3^ 9. Target Child Responding on "Grasp" Variable Successive Observation Days 35 10. Orthogonal Factor Solution "David" Consequation 37 11. Oblique Factor Solution "David" Consequation 38 12. Orthogonal Factor Solution "Class" Consequation 39 13. Oblique Factor Solution "Class" Consequation 39 1*4. Orthogonal Factor Solution "David" Return to Baseline ^0 15. Oblique Factor Solution "David" Return to Baseline ^0 16. Orthogonal Factor Solution "Class" Return to Baseline *n


17. Oblique Factor Solution "Class" Return to Baseline 42 18. Summary of Factor Structure Patterns 42 19. Salient Wi thin-Factor Variable Groupings "David" Data 43 20. Salient Within-Factor Variable Groupings "Class" Data 45 21. Cross-correlation of "David" and "Class" Data 48 22. Correlation Stability 49 23. Factor Solution of David's Baseline First Approximation 65 24. Sating Hate Comparisons 67 25. Token Exchange Latencies 69 26. Average Probability of Token Exchange 70 vi


Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment Of th i ents for the Degree of Doctor of Philosophy "'315 OF THE NON-VERBAL SOCIAL BEHAVIOR OF A CHILD III A KINDERGARTE] JETTING By Jerome R. Freund June 1972 Chairman: William D. Wolking Major Department: Psychology A descriptive and experimental study was made of the ways in which one kindergarten child touched his classmates and was touched by them. The descriptive portion of the study attempted to determine whether the touching responses demonstrated a structure both within and across sets of responses. The experimental portion of the study attempted to determine what stability the social system would demonstrate when an experimental intervention was introduced. The social system of physical contact was studied by using a multiple baseline operant paradigm. Response rates of ten variables for the target child's responses and his classmates' responses respectively were collected during the baseline phase. During the phase of experimental manipulation one of bhe carpet child's response rates was increased via the use of a form of token economy. Multiple baselines were collected during this vii


phase as well as during the final phase, the return to baseline. During this last phase no experimental manipulations were in force. Two statistical methods were used in the data analysis, factor analysis and correlational analysis. Factor structures were found within each set of responses. Forty significant correlations were found when comparing across phases the respective rates of the two sets of responses. Stability was found across phases in terms of the factor structure, the variable composition of factors and the cross-correlation of response variables. It was concluded that the social system of physical contact studied in the kindergarten setting exhibited a definite structure, both within and across sets of responses. This structure demonstrated three types of self-regulatinp; mechanisms when the system was experimentally disturbed. Additionally it was observed that the behavior repertoires of the target child and his classmates successively approximated each other across phases. The data suggested that the phase of experimental manipulation stimulated this mutual approximation. viii


An Analysis of the Non-verbal Social Behavior of a Child in a Kindergarten Setting Literature Review and Statement of Problem The purpose of the present study was to describe the structure and some of the relationships of the social world of a kindergarten child and to demonstrate a nevi combination of methods for doing so. In this study an effort was made to define precisely and measure a certain set of social behaviors in or'"-": to describe one aspect of the complexities of a child's social interactions. The aspect studied was the various ways in ch one child touched and was touched by his classmates. It was hoped that: (a) a factor structure could be found bin each set of responses; (b) response classes or clusters of variables would be found within each factor; (c) that evidences of system homeostasis would be found after the social system was experimentally disturbed; and (d) evidences of the successive approximation of each set of responses to the other might be found, i.e., that an asinp; number of stable interaction patterns might occur across time, '.'his approach, though utilizing operant technology, follows in I ition of developntal studies made nearly forty years ago (Arrington, 1932; Murphy, 1937). Arrington looked at a variety of classroom behaviors in two-year-old children in an attempt to determine some developmental norms. Kurphy I.


2. al^o looked at a variety of discrete behaviors performed by young children in order to relate personality factors to behavior. Though verbal behavior was noted in these studies, its content did not receive particular attention. C ontribution of Socia l Psychology nee Moreno's invention of sociometry (Koreno, 1956) bhe literature of social psychology has been replete with studies analyzing social interactions in natural, laboratory, and "game" settings (Bales, 1950; Bavelas, 1950 ; Beauchamp, 1970; Bormann, 1969; Carson, 1969; Collins and Guetzkow, 196^; Deutsch, 1953; Gamson, 196^; Isaacs, 1965; Kaufmann, 1967; Luce and Half fa, 1957; 1, 19?1). In addition to describing the formal characteristics of groups, much work has been done in an effort to describe the feelings and attitudes generated in social interactions (Abrahamson, 1966; Bales, 1969; Carkhuff and Berenson, 1967; Rogers, 1970) and how these are related to the more formal group characteristics (Borraanh, 1969; Liberman , 1970). A real difficulty with both the formal and feelin level approaches to social interaction analysis is that thci actual behavior of people in groups is not measured, "sign" (Weiss, 1968), or other behavior suppos-


3. edly representing the behavior of interest, is utilized. Thus with respect to rating scales, choice patterns , etc. , the reader is not at all assured that the choices and attitudes reported by r^roup members actually obtain during the ongoing social interactions. As Skinner (1969) notes, behavior used to describe or predict other behavior is an uncontrollable variable. In the types of analyses 3 by .-ales (1950, 1969), Rogers (1970), Liberman (1970), Ilishler and Waxier (1968) and others of this genre, a great deal of precision is sacrificed in order to rain comprehensiveness. The problem remains of determining whether the behavior measured really does represent the attitude or feeling it is said to represent. A great deal of scoring is left to the judgement of the raters. Thus it is possible that the results of such studies measure in fact a good deal of the raters 1 attitudes (Rosenthal, 1966). Social Environ vent : The Ope rant Approach Social psychologists have in general confined their group studies to groups of adolescents and adults (Hare, 1962; Bradford, Gibb and Benne, 196*0. Very little effort hai, been expended in the analysis of the social interactions and transactions of young children. It is with the experimental analysis of behavior that studies of this sort as well as theory bui] be found


<*. (Azrin and Llndsley, 1956; Baer and Wolf, 1966, 1970; Cohen, 1962; Iiop.De, 1970; Homme, C'deBaca, Cottingham and Homme, 1963; Lovitt, Kunzelmann, ::olen and Hulten, 1968; eisworth, Deno and Jenkins, 1969; Patterson, 1966; Patterson and Anderson, 1964; Patterson and Gullion, 196S; Patterson and Reid, 1970; Staats and Staats, 1963). The approach represented by these researchers continues the tradition of direct recording of behaviors of interest, noted in the early studies of Arrington and Murphy. It adds to this approach the precision and objectivity of a well-developed operant technology. The operant approach to the analysis of behavior seems to be well adapted to expand its efforts from the investigation of organisms singly to their investigation in groups. One of the foundations of this approach assumes the functional relationship between the organism and its environment. This is stated frequently as the relationships which obtain among the antecedent events, the response itself and the consequences to the responding organism of its response (Skinner, 1969). This formulation is sufficiently broad to include social stimuli and responses (Lindsley, 1963). Thus social events may be considered as eliciting, reinforcing, discriminative, facilitative or suppressive stimuli; responses may be social and schedules of reinforc immed


5. by the social milieu of an organism rather than by automatic equipment (Lindsley, 1963). ihereas in the classical "Skinner-box" operant work stimulation was one way, i.e., towards the responding organism, research meaningfully investigate two-way or n-way stimulation in the analysis of social interactions (Lindsley, 1963). In the social setting, behavior becomes interdependent (Patterson, i960 ; Weingarten and liechner, 1966). This interdependence becomes a new independent variable (Burgess and Bushell, 1967) in the study of behavior. Patterson (1966) notes that interdependence is that condition where the social milieu exerts pressure for behavior change in specified directions. Individual behavior which is "valued" by the group, i.e., which is rewarding to it, is awarded positively reinforcing consequences. Behavior which is not desired by the group is not reinforced. Thus, the behavior of individual members of a group is shaped to some set of group "norms" (-Burgess and Bushell, 1967; Patterson and Anderson, 1964; Thibaut and Kelly, 1959; Watzlawick, Beavin and Jackson, 1967). Interdependence becomes a patterned interaction or patterned set of social relationships , which is a social system ( Jushell, 1967 ) . A social structure is the specific form these inter-


6. ttiona assume (Burgess and Bushell, 196?). One of the major thrusts of the present study was to discover the nature and form of the pattern of social relationships occurring in a kindergarten classroom. Interdependence rives rise to a corollary variable, i.e., homeostasis (Burgess and Bushell, 196?). This is the way a social system adapts to change in its environment by the acceleration or deceleration of certain interaction patterns which limit change in other of its patterns. During the system's adaptation to new circumstances it is in a state of disequilibrium. iving attained a new level of adaptation the interaction patterns stabilize and reach equilibrium. There are three essential features of a social situation (Sidowski, 1957): 1. two or more subjects' responses produce rewards or punishments for each other; 2. the principal source of reward or punishment for any subject is the response made by the oLher subject; 3. the responses controlling reward and/or punishment are subject to learn inr. The dyad or two-person group is the smallest social sh exhibits these features and is the


7. simplest with which to begin a functional analysis of social interaction. Many investigators have adopted the tactic of studying dyadic interaction (Carson, 1969; Kastorf, i960; Homans, 1961; Lindsley, 1963; Patterson and iieid, 1970; Sidowski, 1957; Thibaut and Kelly, 1959). The present study focused upon the dyad as the primary social unit of study. It also assumed as axiomatic the view that social interaction is the emission of behavior by two individuals in each other's presence (Thibaut and Kelly, 1959), whose behavior may function as discriminative stimuli, social responses or reinforcers for participants (Watzlawick, Beavin and Jackson, 1967). Summary of Literature Review The preceding sections have presented an overview of current concepts in the study of human social interaction. Early work examining the behavior of children in groups directly examined the behavior of interest. From this, inferences were drawn concerning child development and personality traits, .later work in the study of groups turned to more indirect analysis in order to build inferences with reference to formal gi^oup structuring, communication patterns, leadership emergence, cohesiveness , affections, etc. I.ethodolo^ical problems with respect to validity, reliability and experimenter


8. contamination of effects became important. A further problem was the relative neglect of the study of group behavior of young children. It was found that very good, precise work with children's groups has been done by those who espouse the experimental analysis of behavior. This suggested a return to the direct investigation of group behaviors ranging from groups of yc children to adults. Though experimenter effects are still a problem to contend with, this approach greatly reduces problems of validity and reliability as compared to previous approaches. Strategy of the Study Bpite the increasing interest in human interactions in operant research, a survey of the literature seemed to demonstrate a paucity of work in describing the whole complex of interrelationships of individuals, especially children, in a social setting. The basic thrust of the present work was to study an entire small social system in order to measure and describe as accurately as possible its nature. Descriptive research , Peterson and Ault, 1969; lelltiz, Jahoda, Deutsch and Cool:, 1959) pts t urately portray characteristics of an individual, group or situation by determining the frequency of occurrence of the events of interest or the )f association of events


9. with other events. This type of research is frequently employed in the field. Another type of study employed in both lab and field is the experimental study (Bijou, Peterson and Ault, 1969; Kerlinger, 196*0 which describes the functional relationships between events. The present study employed the ideas of both descriptive and experimental research. Phase I was limited to the description of the relationships which were already existent in the social system studied. Phases II and III were experimental in that Phase II involved the manipulation of one variable of interest while monitoring the others and Phase III removed the experimental manipulation of . ase II. The assumption was that change in one part of the response system would result in corresponding changes in other parts of the system functionally related to the part manipulated. Experimental manipulation then was the test of the nature and strength of the functional relationships of the variables of interest, Another way to conceptualize the study was in the form of the single subject, A-5-A design familiar in operant work (Baer, Jolf and Sisley; 1968),


10. Re search Tactics A major procedural question concerned how actually to study the tremendous complexity of n-way interactions occurring rapidly in a viable social system. It seemed clear that any social interaction is really the existence of several dyadic interactions concurrently, the memberships of which are continually changing. If "public appearances," such as speeches, panel discussions, etc., are eliminated, the routine face-to-face interactions in which persons engage mainly involve dyadic interactions. These interactions may be longor short-lived and any one person may have many or few persons with whom he interacts. In view of considerations such as these, one of the major assumptions of the research was that the interactions of a group could be broken down into dyadic relationships. When the social group contains twenty-one persons, as it did in the present study, two hundred and ten possible combinations of dyads may occur. This raised the next issue, the matter of whom to observe. A second assumption of the study was that "n" persons in a group do not constitute one social entity in any real sense, but that they constitute "n" social entities interlocked in various ways. V/ith a class of twenty-one persons there would therefore be twenty-one different social worlds, each having a different person as its center and each interlocking in slightly different ways with the


11. others (Proshansky, Ittelson and Sivlin, 1970). The present study limits itself to the description of the single individual's social relationships within the specific context of the classroom. The next issue involved the behaviors themselves, 'arieties of touching behavior were chosen because they were easily observable and because they formed a part of each chili's repertoire of social responses to other children. The next major issue concerned the research strategy. An assumption was that if a relationship was found between an aspect of the target child's behavior and the behavior of those he touched, then this relationship would show some stability if the system was experimentally chanred. Another possibility was that if the relationships were disrupted during an experimental change, then the relationships would reappear once the experimental manipulation was removed. It was felt that the ,.-A pattern, or baseline-consequation-return to baseline -.em, was the stratery of choice for testing these assumptions . In order to further elucidate functional relationships, the method of multiple baselines (Bijou,


12. Peterson and Ault, 1969; Baer, Wolf and riisley, 1968; Sidman, I960) was employed. The assumption was that the kinds of touchingbehavior within each child were not independent of each other but could be regrouped into larger response classes. Keeping multiple baselines would demonstrate whether this assumption were true or not. Since response classes were assumed rather than known, factor analysis was the statistical method of choice with which to analyze the data (3urt, 1966; noon, 1970). The keeping of multiple baselines throughout all phases made such an analysis possible. An additional assumption was that either the factor structure of responses would remain stable across the three phases of the study or that, if the consequation phase disrupted the structure, the baseline structure would reappear during the return to baseline. A third possibility was that if a new factor structure appeared during consequation it would endure throughout the return to baseline. The implied assumption is that the social system studied would have self-regulatory mechanisms stabilizing the system despite interference. The ..-A plan of attack coupled with multiple baseline recording was believed to be adequate for the description of homeostasis in the system, if it appeared.


13. Sunrary of research Strategy The concept of the study, in summary, was a description of one child's social environment in a kinderrten classroom. The purpose of this work was: (a) to discover the formal pattern of responses within each set of responses; (b) to discover response classes stable over at least two phases of the study; (c) to discover stable relationships in the interaction patterns of the target child and his peers; and (d) to diccover the nature of the homeostatic mechanisms operative within the social system studied. The specific social behaviors studied were the various kinds of touching behaviors made by and to this child to and from other children.


3 'chol ion o ' . v i or Before c t of the study proper, a series of extended observatic re made in a kindergarten class of Sidney Lanier Elementary School. A preliminary list of categories of touching was drawn up based upon these observations. The original list contained sixteen separate categories of touchinr. During the baseline phase of the present study, four categories were drop because there were no instances of behavior to fit into these categories . Two more categories were omitted 'he data ] r sis. Uo categories re added after the sta.rt of the study. The ten final categories with their respective final code numbers were: (1) "Brush off"; ( it"; (3) "Push"; {h) "Grab"; (5) "Grasp"; (6) "Pat"; (7) "Hand Hold"; (8) "Hug"; (9) "Arm Around Body"; and (10) "Vouch." Definitions of each category may be found in Appendix A. Ideas for the categories were obtained from prolonged acquaintance with kinder .viors c-.nd. an anticipatory ai' j (Jellti , 3eutsch and Coo . 59) of all responses. .


15. Further, it was attempted to make the category list exhaustive and mutually exclusive (Selltiz, Jahoda, Deutsch and Cook, 1959; Kerlinger, 1964) so that all possible touching behaviors of interest could be recorded unambiguously. Selection of .'ie search Site The research was carried out in a kindergarten classroom of Prairie View Elementary School in Gainesville, Florida. This proved to be a good choice because the room design facilitated data collection, the classroom teacher was cooperative, and the children were experimentally naive. Children were chosen as subjects because of the paucity of research in the social dynamics of children's p-roups . The decision was made to examine a children's group from the point of view of a single child. Two criteria were employed in the selection of such a target child. 1. 3 child's rate of touching liad to be sufficiently high for the collection of adequate rates. ?. ie child's touching repertoire had to be suffi-


16. ciently varied so that the number of recorded baselines in the multiple baseline procedure could be maximized. David .:. was the child chosen for study. Recording Method he decision was made to continuously record the behaviors of interest. This event sampling technique :>r, 196^) v/as chosen because it was believed that a continuous record v/as necessary in order to discover what relations existed in the target child's social world. This method also facilitated the computation of behavior rates, the raw data of operant research (Honig, 1966). The observation period was about 100 minutes in length during the baseline phase, and reduced to 60 minutes during the following two phases. Thus the study concerned itself with the analysis of a child's social system during periods of free play. Dataware fathered along three dimensions: 1. who David touched and who touched David; 2. what type of touching behavior was exchanged; 3. who initiated the interchange and whether there was a response. Analysis of the third dii excluded from the


17. study because of difficulty in precisely recording the "initiation" and "response" categories. Data were recorded upon a single sheet composed of a series of two-by-two tables. Columns were labeled "Initiation" and 'Mesponse" respectively while rows were labeled "David" and one of the other children's names respectively. Since there were about twenty other children in the room with David, there was the possibility of having twenty separate two-by-two tables, one for each of the possible dyads in which David participated. Cell entries were composed of the number codes of each of the touching categories, a fresh sheet was used for each new day of observation. Dataware taken for about three days per week. Total observation days for baseline, consequation and return-tobaseline phases were 20, 21 and 21 respectively. Standard procedure for the typical operant paradigm suggests (Sidman, I960) that rates should attain stability before a phase change is introduced. Since it was not known in advance whether touching rates in the interaction situation would attain rigorous stability, it was decided to arbitrarily divide the observation times into three roughly equivalent -nts. This was done to insure the completion of the three-phase study before the termination of the school year. A sample two-by-two table is inclu/J .


. Por purposes of illustration, this would represent the data collection format for the David-Allan dyad of day fit


19. was allowed between observer and children during the observation period proper. David was kept totally unaware of the fact that he was the target child during this phase. From the observer's position, almost all of the activity of the room could be kept under surveillance. Cons equation Two issues were involved in the planning of the consequation phase: (1) what behavior to consequate; and (2) the method of consequation itself. One of the purposes of the study was the attempt to uncover factors or response classes. Insofar as behaviors belonging to any one response class are highly correlated with each other, the increase in any one of them would produce an increase in the rates of the oth^r response class members. If this were to be demonstrated, the possibility of spurious correlations among behaviors would be reduced. Consequently, it was believed that the behavior to be consequated should be one highly correlated to a response class. refore the baseline data of the behavior where David was touching others were factor analyzed and rotated to x criterion (Guertin and 3ailey, 1970) in order to discover in some prelimiwhat th« factor structi • ht t . ariable


20. number sever, subsequently renumbered to five, the •asp" variable, was chosen for consequation. Appendix resents the factor matrix and rationale for selection of this variable. 3 methods of reinforcement delivery were considered, i.e., schedules of 'intermittent reinforcement and a system of secondary reinforcement alon^ the lines of a token economy (Ayllon and Azrin, 196 r .'; Birnbrauer, Wolf, Kidder and Tague, 1965; Clark, Lachowicz and Wolf, 196."; Wolf, Giles and Hall, 1968). The second approach was adopted because it would keep the danger of satiation to a minimum, would be a stable source of reinforcing consequences over prolonged periods of time and would intrude as little as possible upon the ongoing social interactions of David and his classmates. The consequation phase was preceded by approximately one month of training for David. The reinforcement schedule itself was a rather complex chain of events which proceeded as follows: 1. \ihen a "grasp" occurred it was immediately followed by the sound of a muted bell. 2. A token was placed in a bowl by the observer immediately after the bell sounded.


21. 3. Approximately one second of time out followed the placement of the token in the bowl while the rewarded response was noted on the data collection sheet. if-. This sequence of steps 1-3 above was repeated five times. 5. Upon the fifth repetition, a buzzer was sounded immediately after the token was placed in the bowl. 6. At the sound of the buzzer, David came over to the bowl while the fifth response was noted upon the data collection sheet. 7. David exchanged his five tokens for either one of several varieties of food or for pennies. The exchange ratio for food was 1:1, while the ratio of tokens to pennies was 5«1. rood had to be eaten on the spot and pennies had to be pocketed. 8. A ten-second limited hold was in effect after the sound of the buzzer. Uncollected tokens were removed thereafter. 9. At this point the whole schedule started over a.^ain, As with all response-chain building, the chain was shaped in reverse (Ferster and Perrott, 1968). The various stages in the buildin" of this chain may be found in Appendix C.


22. to "-v.-oline The return-to-baseline phase was bep;un immediately after 21 days of consequation had been recorded. All reinforcement of behavior was abandoned and all the paraphernalia associated with the consequation phase was removed. The observer limited himself to recording the ten touching variables in the manner done duri: the baseline phase. After 21 days of observation, the study was brought to a close. ;:et'iods of br-:ta Analysis e study addressed itself to: (a) the investigation of the interrelationships among the persons of a small social system; and (b) the effect upon the structure of these relationships when the system is experimentally turbed. Two statistical tools were used in the attempt to answer these questions: factor analysis and crosc-eorrelational analysis . The factor study attempted to determine the stence of response classes intrinsic to the target child's behavioral output and also to t'n ivior he :-ived at the hands of his classmates. 3 correlational study att 1 to describe the relatlonshi] Lch existed between the hibited and the behavior received by the targel


23. Factor analysis is robust with respect to assumptions concerning normalcy of score distributions. Factor analysis makes no assumptions about normality (Cattell, 1966b) though the resulting simple structure is clearer if scores are normally distributed. Also, if non-linearity is a possibility in the data, the usual factor analytic procedures produce very good first approximations to simple structure (Cattell, 19o6b; )igman, 1966). types of data were factor analyzed. One type was labeled "David" while the other was labeled "Class." 3 "David" dataware in three parts, corresponding to the three phases of the experimental study, asv/ere the "Class" data. Both "David" and "Class" data were represented by a three-dimensional matrix in which behavior variables were columns, observation days were rows and phases of the study were the depth. Cell entries of the "David" and "Class" data represented the daily rates at which the target child emitted and received respectively the behavior of interest. ich phase was factored separate!;/ for both "David" and "Class" data types. The correlation matrices employed behavior variabl "relatives" and observation days as " attell, 1966a). Matrix


24. entries were uninodal (Cattell, 1966a) throughout. The associated factor analytic technique was the I -technique (Cattell, 1966a). Ihe factor loadir represented the correlations of variables v;ith the ins ions upon which a single "individual" varied (Kelly and Lingoes, 1962). Cattell (1966c) denoted so dimensions as "state tsions," unique to the ividual. i-'or purposes of the study, the resulting ensions were the unique response classes of behavior Lon and reception vis-a-vis the target child. Of various definitions of factors set forth by Cattell (1966b), the one most closely related fco the concerns of this study considers a factor as an emergent ongoing process freneratinp a set of relationships to variables. Thus the response classes sought were conceived of not as static entities but as relatively identifiable processes accounting for the covariance of groupings of behavior variables. objectives of the study were to demonstrate that response classes, i.e., a factor structure, could be found. What happened to the resulting structures dur he manipulation phase and the return to baseline the focus of interest. potheses were made concern!] structures Ld change, except that


25. some structure would be apparent in phases II and III . It was the expectation that the factor structure of Phase -Iwould bear a rood deal of resemblance to the structures of either Phases I . or II rather than bei ; a new structure. This expectation then either looked for evidence of reversibility during the return to baseline or looked for the maintenance of the ;ered level of functioning produced by the consequation or second phase. The decision was made to use both orthogonal and oblique solutions in an effort to find the simplest structure. Cattell (1966b) argues for oblique rotations asperting that factors usually tend to be correlated to some degree in samples. Secondly, the use of oblique rotations permits higherorder factor solutions or the construction of factor hierarchies (Schmid and Leiman, 1957). Since it was not known in advance whether the present study would or would not need to utilize higher order factor analyses to achieve its goals, oblique rotations were employed along with the more conventional ort: il rotations.


Results "ects Hecently an increasing amount of attention has been piven to the ways in which researchers influence the course of their investigations ( Bandura and '.'alters, 1963; Byrne, 1969; Rosenthal, 1966; Sara:on, i960). One of the consequences of this phenomenon is the nonreplicability of experimental results across experimenters (Sarason, 1966). 3y and lar^e, interest has concerned the effects of experimenter expectancies particularly in social interaction situations (Sarason, 1966; iiosenthal, 1966). One solution v/hich has been proposed (Rosenthal, 1966; Sarason, 1966) involves the removal of the experimenter from the situation and his replacement by completely automated equipment. In many studies, Including the present one, this procedure was not economically possible. An alternative (Rosenthal, 1966) involves strict curtailment of any information cues from experimenter to subjects. This was procedure adopted in the present study. At no time bhe study proper were the children, David or the cla^-t€ "'•'» informed of the true nature of the invesition. sd the investigation of social 26.


27. interactions in a free responding situation, it was important to gain some measure of the constraints put upon the social system by the mere presence of the observer. The assumption was made that the amount of attention received by the observer from the children was a function of the disturbance his presence caused in the classroom. The attending behavior considered relevant and recordable was "looking at the observer." A "look" was recorded whenever the gaze of a child paused for a moment or longer upon the person of the observer. Data were collected concurrently with the collection of the baseline data and later during the return to baseline. It was estimated that most of the attending behavior was recorded. Data were recorded over nine weeks. A table of results follows. Table 1 Daily Sates of "Attending" Behavior Class Hates ?k


28. The regression coefficient was found for the above data. This coefficient was used in a regression equation to locate the two endpoints of a line which best fit the data points. Linearity of data was assumed. This computed line was plotted upon six-cycle logarithmic paper. The average deceleration of the line was found to be 0.83 movements per minute per week. Assuming that an acceleration of 1.00 movements per minute per week indicates no change, the 0.83 index signified a moderate but definite downward trend in the data. It was assumed that the attraction of the children toward the observer was a function of the relative number of rewards and/or punishments associated with him (Byrne, 1969). Consequently the observer rigorously avoided any vocal or non-vocal cue which might acknowledge his awarenets of the children's presence. In order to test this assumption, a brief experiment was conducted during the return-to-baseline of the major study. There were five parts to this experiment: 1. collection of a short baseline; 2. following all children's "looks" with a smile from the observer;


29. 3. an extinction phase; 4. following all childrens' "looks" by both a smile and a greeting fron the observer; 5. an extinction phase. The data are summarized in the following table. Table 2 nary of the neinforcenent of "Attending" Behavior


30. The results suggested that "attending" behavior increases were a function of the social reinforcements dispensed by the observer. Reliability useful procedure is one suggested by Arrington (1932), Bijou, Peterson and Ault (1969). In this procedure the reliability is the percentage of agreement of tv;o or more judges with respect to whether the behavior or behaviors did in fact occur. In the present study the reliability measurement procedure was as follows. Following the suggestion of jijou, Peterson and Ault (1969), one behavior was scored at a tine by two observers. One observer scored only the target behavior selected for that day. The other observer scored his multiple baselines as he had always done. Behavior was recorded in consecutive five-minute segments. Each instance of the target behavior was recorded on the record sheet with a tally mark. Only those behaviors which seemed to occur most frequently were utilized in the reliability checks. .'lie assumption was that if a high decree of reliability wera apparent with them, this reliability would be generally applicable across all ten record . Lr\ all, five checks were made, including four of the ten variables. Ihese


31. four were: (a) f grasp":; (b)"hit n ; (c) "pus:-"; and (d)"touch B The indices in the following table represent the number of time segments in which the frequency counts of both observers matched, divided by the total number of time segments. ble 4A reenent Percentages of Two Observers


32. solutions, using the Varimax criterion, for "David" and "Class" baseline data folio-.:. Table 5 Orthogonal Factor Solution ivid" ] Lne


33. programmed cut-off point was at the absolute 0.30 level of loading. Factor loadings in all tables are rounded to the second decimal olace. Table 6 Orthogonal Factor Solution "Class" Baseline r ar.


3^. Table 7 Oblique Factor Solution "Uavid" Baseline /ar.


35. the variable which was chosen for consequation, variable five, occupied a central place, beinrr distributed among three of the four factors. This included its negative relationship with factor four. The "Class" baseline data also showed a circumplical pattern similar to that of the vid" data. Since primary and reference factor intercorrelations seemed to be quite low, second-order factor solutions were not sought for the "David" and "Class" data. During the consequation phase of the study, responding on variable five, the "grasping" variable, was consequated according to procedures previously outlined. That the acceleration actually occurred may be seen from the followLng table. Table 9 Target Child Responding on "Grasp" Variable iccessive Observation Days




37. changed from a complex circumplical, wheel-like, network to a simple chaining of factors. ictors four and three were chained through variable three, factors three and one were chained through variable nine, and factors one and two were chained through variable seven. It was slightly more complex in the orthogonal solution, but the basic interrelationship of factors remained the same, able 10 Ort al Factor Solution " David ,! C onsequat i on r, -


38. ble 11 Oblique Factor Solution 1 iv id" Consequation Var.


39. able 12 Orthogonal Factor Solution " lass" Consecuation •


40. simpler t Lng consequation, the single chain of factors becoming split into two 2-factor chains. Factors one and four were chained through variable three while factors two and three were chained through variable five. Table 14 Ortho.ronal Factor .'"Solution -:urn t< ] ine r.


to. le orthogonal factor structure of the "Glass" data follovjed the same pattern as the "David" orthogonal factor structure. During the return to baseline the "Class" factor structure returned to the complexity observed during the baseline period. The central feature of this complexity was the distribution o£ variable three over factors one, three and four. :his included the ative relationship variable three had with factor four. On the other hand, the oblique solution represented a return to a chain-circumplex model, modified from Phase I . able 16 Orthogonal Factor Solution "Class" tie turn to Baseline •


42, Table 17 Oblique Factor Solution "Class" Return to Baseline Var.



44. As Table 19 shows, three groupings were salient across all phase changes. The target child had three main types of responses which endured despite manipulations. The first response class, involving the "hit" and "push" variables, was tentatively labeled "physical roughness." The second grouping involved "hand holding" and "hug" variables. This group was labeled as "friendly clasping ." The third grouping involved the "grasp" and "touch" variables. This group was labeled "gentle touch" behavior. It was interesting to note that during the consequation of variable five, the "grasp" variable, it became negatively associated with variable ten, the "touch" variable. This quite correctly conformed to the general finding reported in Table 9 that during Phase II "grasping" behavior almost entirely dominated the target child's repertoire. This condition was reflected again in the grouping of variable five with variables seven and eight, "hand hold" and "hug" respectively during the baseline period and return-to-baseline period, dropping out during the consequation. In fact, variable five assumed a negative relationship with variable seven during consequation. "Grabbing" and "grasping," variables four and five, became associated during baseline, to reappear again during the return to baseline. These two variables were related


«*5. in that they both involved the taking hold of another person with one hand. However, their association may also have been a recording artifact since rigorous separation of the two behaviors was a difficult matter during the data collection itself. Finally it was interesting to note a new association of "pushing" and "patting," variables three and six, beginning during consequation and persisting into the return to baseline. This response class was labeled "flathanded striking ." It appeared that this style of behaving was added more or less permanently to the target child's repertoire during the consequation phase. In summary, three response classes were salient in the target child's repertoire, one of which lost a variable during the consequation but regained it in the return to baseline. A fourth response class was added to the child's repertoire during the period of manipulation. Table 20 Salient Within-Pactor Variable Groupings "Class" Data Phase




47. the one hand the relationship may have been through the response parameter of grasping which occurred in both behaviors. However, its first appearance during consequation led to an alternative hypothesis. During consequation the target child was doing a great deal of grasping. This behavior was usually performed upon the shoulder of another child. Thus the pattern of shrugging off the target child's "grasp" may have been begun, followed by an increase in peer "grasping" of the target child. This increase in peer "grasping" may have been a result of observing the target child's reward for such behavior (Bandura, 1966). In summary, three salient response classes were present in the peer responses to the target child's behavior. A fourth response class appeared during consequation. It is hypothesized that this response class was a direct result of the effects of the target child's consequated behavior upon his classmates. Cross -correlation Study The study thus far concerned itself with the discovery of stable relationships within the target child's response system and within the response system of the class to the target child. The experimenter further desired to discover what relationships, if any, obtained between the response system of the target child and that of the class


48. towards him. Were there any behaviors performed by the target child which were highly correlated with behaviors received by the target child from his classmates? To answer this question, the daily response rates of the ten variables in the "David" data were correlated across days with the same ten variables in the "Class" data. This was done for each of the three phases of the study. Table 2l below summarizes the results. For ease


49. Table 21, cont. Return to Baseline


50. increased over phases. Of a total of 40 significant correlations, about 22;5 occurred during baseline, 28# occurred during consequation and $0% occurred during the return to baseline. This growth in the number of significant correlations is strongly suggestive of the growth of the mutual accomodation of class members' responses to target child's responses, and vice versa (Thibaut and Kelly, 1959). The lambda index of predictive association (Hays, 1963) was computed in order to determine more fully the association between the number of significant correlations per variable and the phase of the study. The lambda value equals .20. This means that given the number of significant correlations for any variable per phase, the error in predicting the phase in which these correlations occur is reduced by about 2Q#. This is not a high index, but is supportive of the view that target child and class behavior repertoires successively approximated each other across the phases of the study. " Further evidence that behavior repertoires tended to approximate each other across phases came from an examination of the entire correlation matrix. Out of a total of 100 correlation coefficients, 23 of them, or ZQ%, demonstrated an increase when the return to baseline co-


51 efficients were compared with their counterparts in the baseline. To be counted, a correlation coefficient had to increase in departure from zero by .23 or more. When two variables correlate by .23 approximately 5% of their respective variances are accounted for by their covariance. Thus an increase of 5% or more arbitrarily qualified a correlation to be counted in the above tabulation. Thus, ZQ% of the correlations increased in strength by more than 5/' from the baseline to the return to baseline phases. This was not considered conclusive, but rather supportive evidence for the assertion that the behavior repertoires of the target child and his classmates tended to successively approximate each other throughout the duration of the study.


Discussion and Summary In broadest outline, this study has been concerned with the discovery of a patterned set of relationships (Burgess and Bushell, 196?) or social system. Secondly, it has teen concerned vjith what happens to the relationships when the social system is manipulated. In other words, the second interest has been in the demonstration of homeostasis (Burgess and Bushell, 1967) or the dynamics of a social system undergoing change. There have been two facets of the social system under study, i.e., the set of responses emitted by the target child David and the set of responses received by David from his classmates. The above elements may be grouped so that four types of analyses may be made. The following two-by-two table illustrates the concept. Intra-set Inter-set System Statics System Dynamics "System statics" refers to the structure of the system at any point. "System dynamics" refers to the nature of the changes occurring in the system structure over time. "Intra-set" analyses are those carried out within either the "David" or "Class" data sets. "Inter-set" analyses are those carried out between the "David" and "Class" 52.


53. data sets. In terms of the statistical procedures used, the prefixes "intra-" and "inter-" refer to factor analysis and cross-correlation respectively. Tables 5 through 8 and 10 through 17 represent the internal formal structure of the target child's responses and the responses received by him from his classmates. It was clear that a meaningful, if complex, structure was found within each set of responses, "David" or "Class," for each phase. The consideration of the interrelatedness of factors within each phase was not dependent upon the specific nature of the variables which overlapped two or more factors. Table 18 summarizes both the system statics and dynamics for both "David" and "Class ." In cross comparing the orthogonal solutions phase by phase, there is great similarity in each phase between the structure of "David's" responses and those of the "Class ." This similarity also holds true for the oblique solution except in the return-to-baseline phase. This matching of syctem structure across phases suggests that both "David" and the "Class" were mutually responsive to the shifts in each others' behavior repertoires. This is to say that insofar as the factor structure of the two response sets reflects a certain "style" of responding, the form of that style was similar for both "David" and "Class ."


54. The nature of the systems' dynamics differs depending upon the factor solution employed. The complexity of the orthogonal solutions in both the "David" and "Class" data suggests that the response dimensions are not independent but are correlated to some degree. This is entirely within Cattail's ( 1966b) expectations that the natural dimensions of human behavior will be correlated to some extent. In the oblique factor solutions, both the "David" and "Class" response systems demonstrate homeostasis in the face of disruption of the ongoing system. The nature of the self-correcting mechanisms is different in each response set. The "Class" set demonstrates the type of homeostasis which reestablishes the conditions prevalent before the system disruption occurred. The "David" set demonstrates a type of homeostatic mechanism which does not return to a previous state, but stabilizes and continues the new system conditions provoked by the disturbance to the system. In brief, the "Class" system follows a sort of A-B-A pattern, while the "David" system follows an A-B-B pattern. The consideration of the interrelatedness of factors ignored the question of whether or not certain variables in each set of responses clustered together stably. Tables 19 and 20 summarized the stability of variable clusters


55. across phases for "David" and "Class" respectively. In both data sets three types of homeostasis were observed. There were three clusters in both data sets which appeared in all three phases. These clusters of variables were insensitive to the disruption of the social system. These clusters of variables were the most stable of the response classes found. The second type of homeostasis followed the A-E-A pattern. Clusters present in the baseline phase were disrupted during the consequation but were reestablished during the return to baseline. The third type of homeostasis was of the A-B-B pattern. Mew clusters established during the strengthening of one behavior stabilized and became a feature ©f the return-to-baseline phase. This latter type of self-regulation reflects the third point in Sidowski's definition of a social system (1957), that responses controlling reward and/or punishment are subject to learning. This latter type of self-regulation also reflects Cattell's definition of learning: H . . . multidimensional change in response to experience in a multidimensional situation" ( 1966c, p.360). Two facts concerning the data of Table 22 are important. It was clear that in significant correlations which extended over two phases, there were none which extended over the baseline and consequation, extinguishing


56. in the return to baseline. Secondly the return to baseline phase represents the sum of all stable and significant correlations originating in either of the previous two phases. This is perhaps the strongest evidence for the assertion that the behavior of the target child and that of his classmates accomodated to each other (Thibaut and Kelly, 1959). There were also a great many transient correlations. These transient correlations composed •$, 6^ L /o and &$% of the total number of correlations in the three phases respectively. Given such a variety of correlations, it was indeed remarkable to discover stable relationships. Thus the great number of transient correlations argued for the strength of those correlations which remained salient. Another interesting fact may be gleaned from the examination of Table 2l . There were nine significant correlations in the baseline phase, eleven in the consequation phase and twenty in the return-to-baseline phase. The number of significant correlations in the last phase exactly equals the sum of the significant correlations In the previous two phases. Furthermore, there was a steady increase in the number of correlations significant beyond the .01 level of probability across the phases. In the first phase the number of such was five; eight were found in the second phase and ten in the third. Both


57. these facts further support the view that the behavior repertoires of the target child and those of his classmates successively approximated each other across phases in both number of relationships and strength. The effect of the consequation phase was to stimulate the approximation to each other of the target child and class behavior repertoires. Table 2l shows the inter-set system statics. It is apparent that there were patterns of interaction within each phase of the study, as indicated by the tabled correlations. Of greater interest is what happened to these interaction patterns across phases. The increase in both number and strength of the significant correlations suggests that both "David H and his classmates were each shaping their behavior repertoires in response to the other (Burgess and Bushell, 196?; Patterson, 1966; Patterson and Anderson, 196^4-; Thibaut and Kelly, 1959; Watzlawick, Beavin and Jackson, 1967). The evidence of the great number of cross-correlations unique to each phase suggests a good deal of mutual shifting of behavior repertory (Thibaut and Kelly, 1959) to meet the new social demands of each new phase. Of greatest interest are the correlations which persisted over two or three phases. These present direct


58. evidence of the workings of self-regulatory mechanisms in the behavioral interactions of "David" and his peers. These homeostatic mechanisms were of the same three types discussed above. "Touching," or variable number ten, was correlated with its counterpart in the other data set across all three phases. It was insensitive to system disruption and therefore may be considered as a stable characteristic of the social system which existed among David and his peers during the period of the study. The second and third types of self-regulatory systems followed the A-B-A and A-B-B patterns respectively. In sum, there was evidence of structure both within and across the response sets of "David" and "Class ." Additionally there was evidence of three types of selfregulating mechanisms operating across the three phases of the study in both the "David" and "Class" response sets. These mechanisms included those which remained insensitive to radical system changes, those which followed the A-B-A pattern and those which followed the A-B-B pattern. Recommendations for Future Research An implicit assumption of the study has been that the system statics and system dynamics of David's social milieu would be similar to that of the social milieu of other kindergarten children, both those with whom David


59. interacted and kindergarten children generally. An alternate possibility is that systematic differences in system statics and dynamics might occur due to racial, sexual, social class or other differences. Replication of this study is desirable therefore to test both of these assumptions. A longitudinal study of a single child's non-verbal social relationships would be desirable in order to determine the influence of developmental factors upon the structure and dynamics of the child's social relationships. A persistent problem in the present study was the inefficiency of the data collection system. Occasionally behaviors occurred with such rapidity that scoring by hand was very difficult. To minimize the potential for inaccurate or missing rates a semi-automated system incorporating observer operated multi-channel event recorders might be useful. Whole new areas of research may be investigated by certain modifications of the data collection format of the present study. The temporal patterning of each child's responses within each observation day might be ascertained by dividing up each observation period into smaller seg-


60. ments. Statistical tools for the analysis of such data have already been developed (Cattell, 1966c). Furthermore the patterns of sequential interactions "between children may be easily plotted by recording the emitted behaviors along a time line, with the aid perhaps of event recorders. Typical interaction patterns may more easily and accurately be spotted using this technique rather than the more indirect method of crosscorrelation used in the present study. Finally, the whole matter of direct observation and recording of social behavior in the field setting lends itself to the exploration of personality itself. The objective analysis of a person's social functioning has hardly been exploited. Many judgements are made about personality types on the basis of behavior only remotely, if at all, related to social behavior, the behavior of interest. Objective methods of data collection, such as those employed in the present study, offer an opportunity to directly investigate the behaviors upon which personality judgements are made. Statistical tools have recently been developed (Cattell, Coulter and Tsujioka, 1966) with which to develop more adequate criteria for personality classifications.


6i. The pioneer work of Hoon (1970) demonstrated the effectiveness of factor analysis in the investigation of social behavior by teaming it with the precise recording of objective behavior. The present study goes one step beyond Hoon by teaming an entire operant paradigm with factor analysis. The present study demonstrated the practicality of the operant approach for the investigation of group phenomena as well as showing the analytic power of factor analysis in thediscovery and monitoring of the organization of social behavior. Furthermore, factor analysis teamed with the precise recording of behavior is useful in the location of behaviors which occupy a central place in the hierarchy of an individual or group's behavior repertoire. Such a location of central behaviors may be of great importance in the selection of appropriate treatment variables for family or milieu therapy. Work in the experimental analysis of social systems is in its infancy. It is believed that such objective analyses, as in the present study, may broaden immeasurably our understanding of Man, the social animal.




Appendix A Category Definitions 1. Brush Off: one person removes the hand or arm of the other person by using his own hand or arm in a brushing or pushing motion. 2. Hit: one person strokes another with his open or closed hand. 3. Push: one person places his open hand or hands upon the body of another physically displacing the other's body. 4. Grab: one person wraps one of his hands around a portion of another person's body and attempts to stop the other from moving or attempts to displace physically the other person's body or portion thereof. 5. Grasp: one person wraps one of his hands around a portion of another person's body but in no way attempts to initiate or terminate movement in the other's body. 6. Pat: one person touches the body of another with one hand two or more times in rapid succession. Included here is rubbing or stroking of another. 7. Hand Hold: one person "grasps" or "grabs" another's hand. In this case alone is the "grasp" or "grab" rescored as "hand hold". 8. Hug: one person wraps both arms around the body of another but does not attempt to bring the other to the floor. 63.


64. 9. Arm Around Body: one person wraps one arm around the shoulders or trunk or hips of another. 10. Touch: one person places the flat of his hand, either front or back of hand, upon the body of another and does not attempt to initiate or terminate any bodily movement of the other.


Appendix B Selection of Consequated Variable Table 23 Factor Solution of David's Baseline First Approximation Var.


66. Thus it was the most representative of that factor, It was assumed therefore that the whole response class represented by this factor could be consequated by the consequation of variable seven.


Appendix C Building the Response Chain Magazine training on a free-feeding schedule was instituted. David fed himself either M & M's or raisins from the bowl which was later to become the token bowl. The purpose of this pairing was to condition the physiological changes preceeding eating to the stimulus of the bowl filled with objects. The objects in this case happened to be the primary reinforcers themselves. The rate of feeding on both candy and raisins was recorded in order to gain some preliminary information on response strength. Over extended periods of observation David maintained a rate of 4.3 responses per minute and 2.4 responses per minute for candy and raisins respectively. Not only was response strength high for the candy, it was also practically twice that for raisins. Data comparing rates of eating candy and raisins was kept for two more days. The results are summarized below: Table 24 Eating Rate Comparisons Stage


68. It was clear from the data that David could eat candy at a reasonably high rate and that he preferred it to raisins. For the remainder of the training period, M & K's were the primary reinforcers of choice. 2. In stage two, another neutral stimulus, a buzzer, was paired with eating. The purpose was to condition the physiological "anticipation" of eating to the buzzer. 3. Stage three required that the candy be eaten one item at a time at the bowl. *K In stage four when David was approaching the bowl regularly, a muted bell was sounded immediately followed by the placing of candy in the bowl. The buzzer continued to be paired with the eating response as in previous stages. The bell was established later as a discriminative stimulus for candy delivery. It also was positioned in David's response chain so as to function as a reward for approaching the bowl. 5. The basic procedure for stage five was the same as for stage four, except that a token was placed in the bowl rather than the candy. The buzzer-eating pairing was terminated. Additionally, a form of reinforcement menu (Homme, deBaca, Cottingham and Homme, 1968) was employed. With the exchange ratio set at 1:1, David was allowed to exchange his token for M & M's, raisins, salted peanuts or Fruit Loops. As the last


69. entry of Table 24 shows, the rate of eating more than quadrupled over the previous day's rate, using this procedure. 6. In stage six the procedure of stage five was continued with one change. Every minute the bell was sounded, followed immediately by the delivery of a token to the bowl. This procedure was followed for two days. During this period, exchange latency frequencies and the average probability of exchange per opportunity were recorded. To record the former, each sounding of the bell started a timer. When David exchanged the token, the timer stopped and the time was recorded to the nearest five seconds. If David did not come to the bowl and exchange his token within one minute, the timer was reset to zero, latency was recorded as one minute and the sequence of bell and token delivery began again. Table 25 Token Exchange Latencies Days


70. by the total opportunities provided. The table of exchange probabilities is included below. Table 26 Average Probability of Token Exchange Training Stap;e


71. could later function as a conditioned reinforcer as well. It was believed that one reason for the low exchange probability was a lack of attentiveness to the bell sound. Therefore, in stage seven the procedure of stage six was continued, but David was not allowed to come to the bowl to receive and exchange tokens until he heard a buzzer. On the average, the observer allowed three to four tokens to accumulate before sounding the buzzer. An additional feature of this stage was that David was allowed only ten seconds in which to come to the bowl to receive his tokens after the sounding of the buzzer. After the passage of ten seconds, the tokens were removed from the bowl and were no longer available for exchange. The increase in exchange probability, Table 26, may be seen as a consequence of this procedure. 8. In stage eight, a temporal discrimination was added to the procedure of stage seven. Since the response probabilities for stage six indicated that David preferred to let roughly three tokens accumulate before exchanging them for a primary reinforcement, it was decided to consistently "call" David to the bowl, via the buzzer stimulus, after every third token was delivered. The two minute temporal interval was marked at every minute by the usual bell-token delivery sequence. When a buzzer call was missed,


72. the succeeding temporal interval was shortened to one and one-half minutes, marked by the bell-token delivery sequence every thirty seconds. Presumably the dramatic speeding up of the temporal interval consequent upon a missed buzzer call refocussed David's attention upon the sound of the bell. No experimentation was done to prove this assumption however. In any event, the result of the entire procedure was to further increase the exchange probability from .7^ to .89. The second day of this procedure saw an increase to .92 probability. At this point the data indicated that David was responding to the whole chain of events quite reliably. A discriminative stimulus, a muted bell sound, began the whole chain. To consequate the chosen behavior the bell sound had to be transformed to a conditioned reinforcer. To summarize the training up to this point: every minute a bell would sound, immediately followed by the delivery of a token into a bowl. Consequent upon the third delivery of a token a buzzer was sounded, after which David had ten seconds in which to retrieve his tokens and exchange them for one of a variety of edible primary reinforcements. 9. In stage nine it was announced to David that he could "make the bell work" for himself by grasping


73. others. The consequation phase began with reinforced practice in grasping by David. Some minor refinements were added to the procedure as follows : (a) The ratio of token deliveries to buzzer "calls" was adjusted to 5H» (b) An introduction of a brief time out after every token delivery so as to facilitate accurate recording of behavior. The consequation data included in the data analysis comprise only the data gathered after the token to buzzer ratio was stabilized at 5:1 •


Appendix D Daily Rates of "David" and "Class" Observation


75. Observation


76. Observation


List of References Abrahams on, M. Interpersonal accomodation. New York: Van uostrand, 1966. Arrington, R. E. Interre lations in the beha vior of youn°; children . New York: Bureau of Publications, Teachers College, Columbia University, 1932. Ayllon, T., & Azrin, N. H. Token economy : A motivational system for therapy and rehabili tation . New York: Appleton-Century-Crofts, 1968. Azrin, N. H., & Lindsloy, 0. R. The reinforcement of cooperation between children. Journal of Abnormal and Social Psychology , 1956, $2, 100-102. Baer, D., & Wolf, N. M. Natural communities of reinforcement. In R. Ulrich, T. Stachnik, & J. Mabry (Eds.) Control of human behavior. Vol. 2. Glenview, 111.: Scott, Foresman and Co., 1966. Baer, D., & Wolf, M. M. Recent examples of behavior modification in pre-school settings. In C. Neuringer, & J. Michael (Eds.) Behavi or modi fication in clinical psychology . New York: Appleton-CenturyCrofts, 1970. Baer, D., Wolf, M. M. f & Risley, T. R. Some current dimensions of applied behavior analysis. Journal of Applied Behavior Analysis . 1963, 1, 91-97. Bales, R. P. In teraction process analysis . Cambridge: Addi son-Wesley, 1950. Bales, R. P. Personality a nd interpersonal beha vior. New York: Holt, Rinehart, & Winston, 1969. Bandura, A. Behavior modification through modeling procedures. In L. Krasner, & L. P. Ullman (Eds.) Resea rch in behavior modification. New York: Holt, Rinehart, & Winston, Inc., 1966. Bandura , A . , & Walters , R . H . Social learning and personality deve lopment. New York: Holt, Rinehart, & Winston, T%3. Bavelaa, A. Communication patterns in task-oriented groups. J our nal of the Ac oustical Society of arlca . i960, 22, 725-730. 77.


78. Peauchanp, M, Element s of mathematical sociology . New York: Random House, 1970. Bijou, S. W., Peterson, R. F., & Ault, M. H. A method to integrate descriptive and experimental field studies at the level of data and empirical concepts. In R. L. Bursress, & D. Bushell, Jr. (Eds.) Behavioral sociology : The experime ntal analysis of social process. New York: Columbia University Press, 1969. Birnbrauer, J., Wolf, M., Kidder, J., & Tague , C. Classroom behavior of retarded pupils with token reinforcement. Journal of Experimental Child Psychology , 1965, 2, 119-135. Bormann, S. G. Discussion and grou p met hods : T heory and practic e, ftew York: Harper & Row, 1969. Bradford, L. P., Gibb, J. R., & Benne, K. D. T -group theory and labor atory method . Mew York: John Wiley & Sons, 196T7 Burgess, R. L., & Bushell, D., Jr. A behavioral view of some sociological concepts. In R. L. Burgess, & D. Bushell, Jr. (Eds.) Eehavioral so ciology : T_he_ experimental analysis of socia l process . New York: Columbia University Press, 1967 . Burt, C. Appropriate uses of factor analysis and analysis of variance. In R. Cattell (Ed.) Handbook of multivariate experimental psychology . Chicago: Rand Hci.'ally , 1966. Byrne, D. Attitudes and attraction. In L. Berkowitz (Ed. ) Advances in experimenta l socia l psychology . Vol. 4. New York: Academic Press, 1969. Carkhuff, R. R., & Berenson, B. G. Beyond counseli ng i therapy . New York: Holt, Rinehart, & Winston, 1967, Carson, R. Interpersonal concepts of p ersonality . Chicago: Aldine Publishing Co., I9t>9. Cattell, R. B. The data box: Its ordering of total resources in terms of possible relational systems. In R. B. Cattell (Ed.) Handbook of multivariate e xperime ntal psychology . Chicago: Rand McNally, 190b. (a) Cattell, R. B. The moaning and strategic use of factor analysis. In R. B, Cattell (Ed.) book of multivariate experi mental psychology . Chicago: Rand McNally, iy~bo~, Cb"T"


79. Cattell, R. B. Patterns of change: Measurement in relation to state-dimension, trait change, lability and process concepts. In H. B. Cattell (Ed.) Handbook of multivariate experimental psychology . Chicago: Hand McNally, 1966. (c) Cattell, R. B. The principles of experimental design and analysis in relation to theory building. In R. B. Cattell (Ed.) Handbook of multivariate experimental psychology . Chicago: Rand McNally, 1966. (d) Cattell, R. B., Coulter, M., & Tsujioka, B. The taxonometric recognition of types and functional emergents, In R. 3. Cattell (Ed.) Handbook of multivariate experimental psychology . Chicago: Rand McNally, 1966 Clark, M., Lachowicz, J., & Wolf, M. A pilot basic education program for school dropouts incorporating a token reinforcement system. Behavior Research and Therapy , 1968, 6, 183-188. Cohen, D. J. Justin and his peers: An experimental analysis of a child's social world. Child Devel opment , 1962, 22, 697-717. Collins, B. E., & Guetzkow, H. A social psychology of group processes for decision-making . New York: John Wiley & Sons, Inc., 196^. Deutsch, M. The effects of cooperation and competition upon group process. In D. Cartwright, & A. Zander (Eds.) Group d ynamics : Rosearch and theory , Evanston, 111.: How, Peterson, dc Co., 1953. Digman, J. Interaction and non-linearity in multivariate experiment. In R. B. Cattell (Ed.) Handbook of multivariate experimental psychology . Chicago: Hand McNally, 1966. Perster, C. B. , & Perrott, M. C. Behavior principles . New York: Appleton-Century-Crof ts , 1968. Fruchter, B. Manipulative and hypothesis testing factor analytic experimental designs. In R. B. Cattell (Ed.) Handbook of multivar iate experimental psychology . Chicago: Rand McNally, 1966. Gamson, W. A. Experimental studies of coalition formation. In L. Bcrkov;itz (Ed.) Advances in experi ntal social psychol ogy . Vol. l. New York: Academic Press, I9b^~.


80. Guertin, W. K., & 3ailey, J. D., Jr. Introduction to modern factor analysis . Ann Arbor, Mich. : Edwards Bros., 1970. Guilford, J. P. Fundamental statistics in psychology and education. (4th ed.) New York: McGraw-Hill, 196~5. Hare, A. D. Handbook of small group research . New York: The Free Press, 1962. Hastorf, A. H. The reinforcement of individual actions in a group situation. In L. Krasner, & L. P. Ullman (Eds.) Research in behavior modification . New York: Holt, Rinehart, & Winston, 1966. Hays, W. L. Statistics. New York: Holt, Rinehart, & Winston, l9b3. Homans, G. C. Social behavior : 1 ts elementary forms . New York: Harcourt, Brace, & World, 1961. Homme, L. Contingency contracting in the classroom . Albuquerque, N. H.: Westinghouse Learning Corp., 1970. Homme, L., C'deBaca, P., Cottingham, L., & Homme, A. What behavioral engineering is. Psychological Record . 1968, .18, 425/ +3 / +. Honig, W. K. Introduction. In W. K. Honig (Ed.) Operant behavior : Areas of research and appli cation. New York: Appleton-Century-Crofts, 1966. Hoon, P. W. Looking at behavioral environments: A preliminary investigation. Research Training Paper ; , University of Kansas Bureau of Child Research and School of Education, 1970. Isaacs, R. P. Differe n tial games : A mathe matical theory with applications to warfare and purs uit, control a nd optimization . New York: Wiley, 1965. Kaufmann, A. Graphs . dynamic programmin g and finite games. Translated by h. C. Sneyd. New York: Academic Press, 1967. Kelly, E. L., & Lingoes, J. C. Data processing in psychological research. In H. Borko (Ed.) Compu ter a pplications in the behavioral sciences . Englewood Cliffs, N. J.: Prentice-Hall, 1962.


81. Kerlinger, F. N. Foundations of behavioral research . New York: Holt, Rinehart, & Winston, 196^. Kirk, R. E. Experimental design : Procedures for the behavioral sc iences . Belmont, Calif.: Brooks/Cole Pub. Co., 1969. Liberman, B. Reinforcement of cohesiveness in group therapy. Behavior Therapy , 1970, 1_, 1*44-175. Lindsley, 0. R. Experimental analyses of social reinforcement. The American Journal of Orthopsychiatry . 1963, 21, 62^33. Lovitt, T. C, Kunzelmann, K. P., Nolen, P. A., & Hulten, W. J. The dimensions of classroom data. Journal of Learning Disabilities , 1968, 1.(12), 710-721 Luce, R. D., & Raiffa, H. Games and decisions . New York: Wiley, 1957. Ill shier, E., & V/axler, N. Interaction in families . New York: John Wiley, a Sons, 1968. Moreno, J. L. Sociometry and the science of man . New York: Beacon House, 1956. Murphy, L. B. Social behavior and child personality . New York: Columbia University Press, 1937. Neisworth, J. T. , Deno, S. L., & Jenkins, J. R. Student motivation and classroom management . Newark, Del.: Behavior Technics, Inc., 1969. Patterson, G. R. Responsiveness to social stimuli. In L. Krasner, & L. P. Ullman (Eds.) Research in behavior modification . New York: Holt, Rinehart, & Winston, 1966. Patterson, G. R., & Anderson, D. Peers as social reinforcers. Child Development , 196^, 21, 951-960. Patterson, G. R., & Gullion, M. E. Livlnp; with children . Champaign, 111.: Research Press ,' 19&8. Fatterson, G. R., & Reid, J. Reciprocity and coercion: Two facets of social systems. In C. Neuringer, & J. Michael (Eds.) Behavior modification in clinical p nycholo/yy . New York: Appleton-Century-Crof ts, 1970.


82. Proshansky, H. !•'., Ittelson, W. H., & Rivlin, L. G. Environmental psycholog y. New York: Holt, Rinehart, & Winston, 1970. Rogers, C. R. Carl Rogers on encounter groups . New York: Harper & Row, 1970. Rosenthal, R. Experimenter effects in behavior research . New York: Appleton-Century-Crofts, 19b6. Sarason, I. G. The human reinforcer. In L. Krasner, & L. P. Ullman (Eds.) Research in behavior modification . . ew York: Holt, Rinehart, & Winston, 1966. Schmid, J., & Leiman, J. M. The development of hierarchical factor solutions. Psychometrika , 1957, 22(1), 53-61. Selltiz, C., Jahoda, M., Deutsch, H., & Cook, S. Research methods in social relations . (Rev. ed.) New York: Holt, Rinehart, & Winston, 1959. Shaw, I' . E. Group dynamics : The psychol og y of small group behavior . New York: KcGraw-Hill, 1971. Sidman, K. Tactics of scientific research . New York: Basic Books, i960. Sidowski, J. P. Reward and punishment in a minimal social situation. Journal of Experimental Psychology , 1957, £M5), 318-32S": Skinner, B. F. Contingencies of reinforcement : a theoretical analysis . New York: Apple ton-CenT;uryCrofts, 1969. Staats, A. W., & Staats, C. K. Complex human behavior : A systematic extension of learnin g principles. »w York: Holt, Rinehart, & Winston, 1963. Thibaut, J. W., & Kelly, H. H. The social psychology of groups . New York: John Wiley & Sons, 1959. Thurstone, L. L. Experimental study of simple structure. Psychometr ika. 1940, ±, 153-168. Cited by B. Fruchter, I-anipulative and hypothesis testing factor analytic experimental designs. In R. r >. Cattell (Ed.) H andbook of multivariate experim ental psychology . Chicago: Hand ly, 1961T pT 337.


83. Watzlawick, P., Beavin, J. H., & Jackson, D. D. Pragmatics of human communication. Mew York: W. W. Norton, 1967. Weingarten, K., & Mechner, F. The contingency as an independent variable of social interaction. In T . Verhave ( Ed . ) The experimental analysis of behavior : Selected readings . New York: AppletonCentury-Crofts, i960 . Weiss, R. L. Operant conditioning techniques in psychological assessment. In P. HcReynolds (Ed.) Advances in psychological assessment . Vol. 1. Palo Alto, Calif.: Science and Behavior Books, 1968. Wolf, M., Giles, D., & Hall, R. Experiments with token reinforcement in a remedial classroom. Behavior Research and Therapy , 1968, 6, 5l-6*K


Biographical Sketch Jerome R. Freund was born May 6, 19^0 in India of missionary parents. He received his B. A. in philosophy from Westminster College in Fulton, Missouri, in June of 1962. He studied for a year thereafter in St. Andrews University, Scotland, on a Fulbright Scholarship. Mr. Freund completed his B. D. at Union Theological Seminary, New York, on June, 1968. Between his middler and ssnior year, Mr. Freund served as a Methodist missionary to the Congo and Malawi for three years. Mr. Freund was ordained into the Protestant ministry in February of 1969. In December, 1970, he received his M. S. in psychology from the University of Florida. He received his Ph. D. in clinical psychology in June, 1972, from the University of Florida. Mr. Freund is married to the former Elizabeth Lee Patterson of Norfolk, Virginia. The couple have two children, Robert Lee and Natashya Lee. Mr. Freund is currently employed as an intern in clinical psychology at the Guidance Center, Inc., Daytona Beach, Florida. 8*K


I certify that I have read this study and t In ny opinion it co is to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. William D. Wolking, urn Associate Professor/ol Special Education I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Donald L. Professor Education Avila of Foundations of I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. _niiypacl Henry S. Pehhypacker , Jr. Professor of Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Ridhard Associate Professor Clinical Psychology


I certify that I have read this study and that in opinion it; conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and qv as a dissertation for the decree of Doctor of Philosophy. °2f, »^,*.„; .&& elaine Ramey Assistant Professor of /Psycho] This dissertation was submitted to the Department of Psychology in the College of Arts and Sciences and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. Dean, Graduate School May, 1972


Vi SM27 17. 1.17. if f