DIMENSIONS OF COMMUNICATION IN THREE- AND FOUR - YEAR OLD CHILDREN’S PEER INTERACTIONS . Dissertation for the Degree of PhQ ‘ D. MICHIGAN- STATE UNIVERSITY MARY LOU'ISE’PASS‘ANDREWS' 1975 ' III/BI III/III III/II a) a? (D m _ N uh I F- 293 00 I ‘ 2 f '2 N T ' . l 1:. T x ' f :15}, 5‘ l 0.1,. Jamar-.2“. This is to certify that the thesis entitled DIMENSIONS OF COMMUNICATION IN THREE-AND FOUR-YEAR-OLD CHILDREN'S PEER INTERACTIONS presented by MARY LOUISE PASS ANDREWS has been Inccépted towards fulfillment of the requirements for PhoDo degree in Family Ecology Wm %/ , a]?— Major professor Date NOV . 1011975 0-7639 egll:-, . “no 09 7.3.,» WW ' nR'ozjg M/ N, 93 . c—as ’xhfi\m mm_ 1+1}:st If“: .. ” "9 I D t- _,p K. H ‘ ‘3 '\ ABSTRACT DIMENSIONS OF COMMUNICATION IN THREE- AND FOUR-YEAR-OLD CHILDREN'S PEER INTERACTIONS By Mary Louise Pass Andrews This study was designed to describe communication among three- and four-year-old children in a naturalistic peer inter- action setting. An objective was to identify factors that are related to children's communicative competence and the differential incidence of communication in the play group. In so doing, the information transmission potential of young children's peer inter- actions was assessed. A small group social-interaction methodology was used. The sample consisted of 139 three- and four-year-old children enrolled in eight urban day care centers. The children were randomly assigned to participate in play groups. Ten minutes of the children's play was video-taped and subsequently rated using a time-sampling obser- vational procedure, the revised Observation of Socialization Behavior instrument (Boger, Cunningham, & Andrews, l973). At 20 second intervals the first play behavior was characterized by l4 behavioral categories. Mary Louise Pass Andrews Communication was determined by an analysis of the actions and contingent reactions of children in the group. If a child's behavior produced a reciprocal response from at least one other peer, communication was said to have occurred. Communication was viewed as an element of the ecosystem that linked individual sys- tems to each other and to the environment. A primary dimension of communication operationalized in this study was communicative competence, an attribute of the child system. It reflected the relative degree to which a child impacts on the social environment. Analysis of variance and multiple regression analyses were implemented to explore the relationships between children's communi- cative competence scores and other dimensions of social interaction behavior and selected demographic characteristics. Communicative competence was found to be positively related to the relative pro- portion of the time that children initiated to others and the qual- ity of their play involvement. Age and experience in group care were also significant predictors of communicative competence. The older and more experienced children exhibited higher communicative competence scores. Male children also had higher communicative competence scores than females. However, socioeconomic group member- ship, ethnicity, the number of children in the family or even the presence of siblings in the family were not related to differential communicative competence. An analysis of the overall incidence of communication in the play groups revealed that much of children's play behavior was com- municated to others. Communication occurred in 65% of all Mary Louise Pass Andrews time-sampled intervals. Even when behavior was not directed to other peers, it often communicated. The probability of communication increased when behaviors were accompanied by a verbalization, when initiations were directed to individual peers rather than the generalized group or physical environment, and when behaviors were elicited with negative rather than neutral or positive affect. It appeared that these young children noticed dissonant or anti-social events rather than neutral or pro-social events, more so than would be expected by chance. The generalizability of these findings to other social set- tings needs to be determined before implications can be drawn. However, a step has been taken, through this study, to objectively quantify an illusive element of children's social interactions-- communication. It is hoped this and other methodological approaches can be used to further an understanding of the ontogeny and role of communication in human development. DIMENSIONS OF COMMUNICATION IN THREE— AND FOUR-YEAR-OLD CHILDREN'S PEER INTERACTIONS By Mary Louise Pass Andrews A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Family Ecology 1975 ACKNOWLEDGMENTS A dissertation is but one of many products representing the pursuits of one's graduate program. But to me it is significant, in that it stands at the apex of a long and diverse path of pro- fessional development. Many people and events have contributed to shaping the course and perhaps the ultimate outcome of that expe- rience. Looking back to analyze it is like trying to unravel the complexities of the socialization process, a phenomenon I am only beginning to understand and appreciate. I have been fortunate in being able to pursue a breadth and depth of experiences and associations while studying at Michigan State University. It is to those friends, colleagues, and associates to whom I have made contact along the way that I offer my gratitude and blessings. Specifically, I wish to thank Robert Boger, my guidance com- mittee chairman and supervisor of my "apprenticeship." His encourage- ment, tolerance, and unfailing faith in my ability to cope and grow with challenging experiences have provided me with an invaluable foundation from which to move forward. Through him I have been able to develop some special interests and skills that have made my gradu- ate program unique. To Beatrice Paolucci, my unofficial advisor, I express appreciation for helping me feel a part of a larger professional ii community. Through contact with her I have grown in confidence and sc0pe. Her model for professional excellence has been an under- lying influence on my work. I thank Don Melcer for his continuing support and unwavering confidence in me and my ideas. In him I have had a colleague with whom to share my interests and insights, working together to crystal- ize ideas. In Jack Bain I found a warmth, Openness, and acceptance that facilitated my work in communications and encouraged me to pursue my interests as reflected in this research. I also wish to acknowledge the pervasive influence of JoLynn Cunningham throughout my entire program as well as in this research. She provided the encouragement and leadership that helped me define and establish a role as a graduate student and researcher. As a model, a critic, and a friend, she has helped shape the course of my graduate experience. Innumerable others have supported me during the course of my studies. I would expecially like to recognize Eileen Earhart and Margaret Bubolz, with whom I have been privileged to work. Some of my most memorable experiences and the most challeng- ing opportunities for critical discussion have been with fellow graduate students at the Institute. I thank Ann Wilson, Terry Boyd, Sandy Evers, and especially Verda Scheifley for their friendship and support. Others, who contributed to a team effort to implement the research of which this study was a part were Paul Muhs, Tito Reyes and Nancy Hand. Finally, the support of Alice Lucas and Mary Voth of the clerical staff have made my work easier. Last, but most important, I express my sincere appreciation to my family: to Harry and Willie, who patiently tolerated my absence from family activities during the past few years and espe- cially during these hectic months of writing; and to my parents, Alice and Wendell Pass, for their early and continuing encouragement to pursue the difficult. Their faith and support have been strengths to me throughout my educational career. iv TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES . Chapter I. INTRODUCTION Statement of the Problem Overview of the Study. Elements of the Multiperson Interaction to Be . Studied. . Conceptualization of Communication Objectives . Assumptions . Definitions . II. RELATED LITERATURE Theoretical Perspective: The Child as an Open System. Socialization Communication Review of Research on Children's Peer Communica-. tion . . Consequences of Peer Communication Review of Methodological Approaches to Studying . Children's Peer Interaction . Time Sampling . Event Sampling Specimen Description Time and Event Sampling . Review of Systematic Differences in Children's Peer Interactions . . . Sex . . Socioeconomic Status Race or Ethnicity Age . . Experiences with Siblings and Peers Page viii -J 00030101 «D00 ll Chapter III. METHODOLOGY Research Design Hypotheses Sampling Procedures Sample Selection . Description of the Sample Instrumentation Content Measures . . Data- -Gathering Procedures Reliability and Validity Reliability . Validity Operationalization of Variables Communication Operationalizatidn of Demographic Characteris-o tics and Background Variables Operationalization of Social Interaction Variables . Operationalization of Modes of Communication Data Reduction and Analysis . . . . Data Reduction Procedures . Analyses Concerning Attributes of the Child Analyses Concerning Attributes of the Communi- cation IV. RESULTS OF DATA ANALYSES Relationships Between Communicative Competence and Background Characteristics of Children . Relationships Between Communicative Competence and Other Social Interaction Variables Differential Incidence of Communication Summary of the Results . . V. DISCUSSION . The Relationships Between Background Character— istics and Communicative Competence Relationships Between Social Interaction Varia— . bles and Communicative Competence . Description of the Incidence of Communication Summary . . . . . . . . . vi 93 102 107 Ill I14 ll4 I17 120 l23 Chapter Page VI. CONCLUSIONS, IMPLICATIONS, AND SUGGESTIONS FOR FUTURE RESEARCH . . . . . . l25 Conclusions . . 125 Children' 5 Play Group Communicative Competence 125 Methodological Contributions . . . . . . l28 Limitations of the Study . . l30 Implications and Suggestions for Future Research l35 APPENDIX . . . . . . . . . . . . . . . . . I38 BIBLIOGRAPHY . . . . . . . . . . . . . . . . 157 vii Table TO. ll. 12. 13. LIST OF TABLES Characteristics of the Sample . Results of Tests for Internal Consistency of Behav- ioral Categories: Revised Observation of Social- ization Behavior Instrument . . . Results of Kolmogorov-Smirnov Test of Goodness of Fit on the Distribution of the Communicative Compe- tence Variable . . . . . . . . . . Results of Levine's Test of Equality of Dispersion . Results of Multiple Regression Analysis Predicting Communicative Competence From Selected Background Characteristics . . . . . . . . Pearson Product-Moment Correlation Coefficients for Background Characteristics With Communicative Competence . . . . . . . . Results of a One-Way Analysis of Variance Test for Age Effects on Communicative Competence Scores Means and Standard Deviations of Communicative Competence Scores Based on Age Partial Correlation Coefficients . Results of the Two-Way Analysis of Variance Test for Sex and SES Effects on Communicative Competence Scores Means and Standard Deviations for Communicative Competence Scores Based on SES Group Membership . Means and Standard Deviations for Communicative Competence Scores Based on Sex . Kruskal-Wallis One-Way Analysis of Variance Test for Ethnic Group Differences in Communicative Competence Scores . . . . . . . viii Page 55 7O 82 87 94 95 96 96 97 99 99 100 TOO Table Page 14. Means and Standard Deviations of Communicative Competence Scores Based on Ethnicity . . . . . lOl 15. Results of Analysis of Variance Tests for Age Dif- ferences in Sex, SES, and Ethnic Groups . . . . 102 I6. Pearson Product Moment Correlation Coefficients Representing the Interrelationships Among the Social Interaction Variables Derived From the 058 . . . 103 I7. Partial Correlation Coefficients Holding Activity Level Constant . . . . . . . . . . . 104 18. Results of Multiple Regression Analysis Predicting Communicative Competence From Social Interaction Variables . . . . . . . . . . . . . . l05 l9. Results of Multiple Regression Analysis Predicting Communicative Competence From Both Background Characteristics and Social Interaction Variables . l06 20. Chi Square Test of Relationship Between Communica- tion and Verbalizations . . . . . . . . . . l08 21. Chi Square Test of Relationship Between Communica- tion and Behavioral Affect . . . . . . . . . l09 22. Probabilities of Communication Based on Behavioral Affect, Object of Interaction, and Physical Contact llO 23. Chi Square Test of Relationship Between Communica- tion and the Object of the Initiation . . . . . lll A-l. Minimum Suggested Rater Reliability Indices for Observation of Socialization Behavior . . . . . 154 ix LIST OF FIGURES Elements of the multiperson interaction Model of Communication Intentional and unintentional communication Schemata for data analyses . Diagram of mobile unit Interaction as observed in the original 058 Interaction and impact as observed in the revised OSB . . . . . Data gathering procedure Example of a contingency table An example of one observational frame Assignment of points for 058 rater reliability Observation of socialization behavior rating Protocol . . . . . Page 18 49 57 61 62 67 91 146 155 156 CHAPTER I INTRODUCTION The child, as a social being, learns to behave in a manner that shows consideration of how one's behavior is interpreted by others. This "social" awareness or ability to understand (I) that one is responsible for and can regulate one's behavior, and (2) that one's behavior makes an impact on others who initiate reciprocal acts is a basic association learned early in a child's life. How the child then expresses this understanding in developing increas- ingly more complex and meaningful patterns of interacting with the social environment is a subject of interest to those studying the socialization process. Studies of early parent-child interaction have substantially increased understanding of the inputs into this process from the family system. The pervasive nature of the substance and manner in which parents communicate to their children seems to mediate the child's cognitive, social, and emotional development (Hess & Shipman, l965; Cunningham, 1972). But the child is also an active partici- pant in this process. Earliest and expanding communication patterns influence the child's relationships with parents and the course of subsequent experiences with the social environment (Lewis & Rosen- blum, 1974). As one expands the sphere of one's interactions during child- hood, new encounters provide necessary feedback about the utility of one's behavior in eliciting desirable responses from others. Thus the child builds a systematic perspective toward social exchange that is used in meeting new experiences, being generalized from one experience to another so that appropriate communication patterns can be differentially selected from a repertoire of possible lines of action. The level of the child's communication skill mediates the success of experiences in social interaction and thus colors social development. Basically, the study of the ontogeny of communication skill has only recently attracted research interest. Parallel areas of inquiry into the social and language activity of children date back to the early l930's. But each discipline studying these phenomena investigated different components of this more complex process; hence integration and comparability of data are lacking. One of the major difficulties hindering research in commu- nication has been the lack of adequate instrumentation to capture the holistic dimensions of what may be communicative behavior. Ethologists have begun to explore this problem by reducing complex behavior to basic units (Brazelton, Koslowski, & Main, l974; Bran- nigan & Humphries, l972; Smith & Connolly, l972). Observational techniques have been developed to measure individual behavior and even dyadic interaction, but methods of describing communication among more than two individuals are not readily available in the literature. Thus, the present study explores a specific methodological approach to assessing the impact of children's behavior on others in a play setting. Descriptive and comparative information derived from this study can make a contribution to understanding the social communication skill of young children and therefore the potential for information transmission among peers. Such data are necessary to identify the potentially complex social forces that influence children's social behavior and development. Statement of the Problem The primary purpose of this study is to describe communica- tion among three- and four-year-old children in a naturalistic peer interaction setting. Two aspects of children's communication are investigated: (1) children's communicative competence and (2) modes of communicative exchange. In the first instance, a communicative competence score is derived, which denotes the extent to which a child's behavior is received as communication by his peers. This variable represents the degree to which a child impacts on his environment and thus actively shapes his social experiences. The relationships between this variable and other social interaction variables and selected demographic characteristics of the children are explored. A second focus of this study is on describing a cross- sectional representation of the communication flow that results from children's spontaneous play behavior. Modes of exchange are investi- gated to determine resultant differential incidences of communication. Such an analysis of the information flow contributes to under- standing children's attending and responding patterns during peer interaction. Overview of the Study The data for this study are part of the data collected for a larger study that investigated the relative effects of short-term supplemental parent and classroom programs on the self-concept and social involvement of children in day care settings (Boger & Andrews, l975). The observational data collected from the larger project before the implementation of treatment conditions are used in this study. A small group social interaction methodology is used. This methodology was developed by Boger and Cunningham (T969) to provide the setting in which to observe children's socialization behavior during spontaneous free play. Videotapes of children's play behavior are subsequently rated, using a time- and event-sampling observa- tional procedure, The Observation of Socialization Behavior (058). Categories of both verbal and nonverbal behavior as well as the affective and play context of such behaviors are observed. This original observational procedure was adapted for the present study. The observational unit was redefined so that all categories of behavior refer to the same behavioral phenomenon. Also, a procedure was added to observe the consequence of actions on all members of the group, whether or not such individuals were the object of the interaction. By effecting these changes, the observational procedure facilitates an analysis of the dimensions of communication of inter- est in this study. Elements of the Multiperson Inter- action to Be Studied Three elements of the social milieu can be investigated during multiperson interaction (Golembiewski, l962): (l) the characteristics, attributes, or behaviors of individual subsystems; (2) the nature and direction of the communication; and (3) the character and setting of the system itself. These elements are illustrated in Figure l. Figure 1. Elements of the multiperson interaction. Dimensions of two of these elements are operationalized in this study. l. The attributes and characteristics of the children are defined by: Ethnicity, sex, socioeconomic status, age, experience with siblings and peers, and the following social interaction variables: social behavior, responsivity, initiative, activity level, and communicative compe- tence. 2. The following dimensions of the communication flow are operationalized: channel of exchange--verbal and nonverbal affective connotation--positive, neutral, negative directionality--individual peer, group, environment 3. Characteristics of the group and setting are controlled to the greatest degree possible. The setting or environment, the composition of the group to represent one low- and one mid-SES male and female, and the nature of the experience in terms of its novelty value are similar across groups. Although the course and content of the interactions differ from group to group, it is assumed that individuals have an equal opportunity to interact and express their own style of behavior. As these group members are chosen at random on an ad hoc basis, no group expectations nor proscriptions of roles should be operative (Larson & Hill. 1958)- Conceptualization of Communication To identify the communication that occurs as children inter- act in a small group, a descriptive communication model developed by Nan Lin (1973) is operationalized. This conceptual model is based on a functional view of human communication. With this model, communication can be viewed as occurring in a number of phases. These phases overlap, but each has distinctive characteristics that contribute to the structure and process of communication. These phases can be identified as: (l) encounter, (2) exchange, (3) influ- ence, and (4) adaptation/control. The model is illustrated in Figure 2. encounter] —--> [exchangej —->[ influence I \ Ladaptation/controlj/ t Figure 2. Model of Communication (Lin, 1973). During the first phase, encounter, the initial linkage between the information and the receiver is established through a specific medium. In the second phase, exchange, a flow of shared meaning occurs. The third phase, influence, then verifies the communication; it is defined as the discrepancy between a person's behavior or attitudes before the encounter or exchange and his behavior or attitudes after the encounter and exchange. Influence therefore represents the psychological or behavioral impact of the information flow. Finally, the effectiveness of communication over time depends on the control-adaptation process, whereby the fidelity of information flow is maintained through feedback and adapted transmission. The influence phase of Lin's model is operationalized in this study to identify when and if communication occurs. Any behavioral interaction is considered communication if a behavior elicited by one peer influences a receiver who attends or reacts in some manner as a result of the original behavior. Thus the communi- cation is inferred from the analysis of actions and contingent reactions of others in the play group. Because of the nature of time-sampling observations, only the linear flow of information from encounter to influence is observed. Thus an interaction is identified, not a communication transaction, which would involve the adaptation/control phase. Objectives The primary objectives of the study are: 1. To determine if the following demographic factors pre- dict or differentiate degrees of communicative competence: Sex Socioeconomic status Age Ethnicity Number of siblings Number of months since entering center Number of months in group care 2. To determine the relationship between communicative competence and the following social interaction variables: Level of social behavior Activity level Responsivity Initiative 3. To identify the use of the following differential modes of communicative exchange and the resultant differences in the incidence of communication: channel of exchange--verba1 vs. nonverbal affective connotation of physical behavior--negative vs. positive or neutral directionality--initiations directed to individual peers vs. the group or the materials and environment Assumptions 1. The ability to effect an impact on one's social environ- ment is a communication skill that develops through the interaction of expanding cognitive structures and experiences with the social environment. Therefore, a communicative competence score that measures this skill will vary across a sample of three- and four- year-old children. 2. Spontaneous play behavior among like-aged peers offers an opportunity to assess an individual's skill in communication. 3. In a play situation in which the grouping of children is random, the setting is equally novel for all participants and the opportunity is provided for everyone to participate as he wishes; the resulting play behavior observed can be attributed to internal factors within the child. Definitions Communication: Information flow, in the form of the behavior .of an individual, that influences a receiver who attends to the initiator and/or reacts in some manner as a result of the original behavior. No intent to communicate is necessarily inferred. Interaction: An initiation-response sequence between two or more individuals. 10 Initiation: An overt muscular or verbal activity of an individual, which involves a change from the immediately preceding activity. Resgonse: Change in muscular activity of an individual as a result of stimuli that are manifested by another individual. Transaction: Reciprocating patterns of interactions; one's response contributing to another's initiation, which is responded . to by the original participant in a continuous fashion. Communicative Competence: The relative ability of an individual to impact upon or initiate communication with the social environment. "Bit of Interaction": A stretch of play behavior that includes an initiation, response, or ongoing action (terminated by another initiation or response) directed to the self or the environ- nent. CHAPTER II RELATED LITERATURE Theoretical Perspective: The Child as an Open System The theoretical perspective upon which this study is founded is an integration of a variety of theories of human development and conceptual frameworks related to human activity. It is primarily an application of a human ecological (ecosystems) approach (Auserwald, 1968; Hook & Paolucci, 1970), drawing heavily from general systems theory (Buckley, 1967; Laszlo, 1972; Bertalanffy, 1968), communi- cation theories (Miller, 1965; Mead, 1934; Theyer, 1968), and devel- Opmental interaction theories of human development (Piaget, 1950; Brunner, 1965). The application of an ecosystem perspective to human development, although relatively new and still lacking unity, offers exciting possibilities for investigating the complex inter- actions of the organism with the environment. The reader is directed to Gardner (1971) and Sims, Paolucci, and Morris (1972) for further information about this approach. At a very basic level, human organisms can be viewed as systems, sets of interdependent components in dynamic interaction that metabolize matter-energy and information (Thayer, 1968). Infor- mation can be defined as matter-energy in a patterned flow that conveys meaning. The intake of food, water, air, and sensory 11 12 stimulation from the environment is processed within the system to maintain life. The output from the body is human activity, along with waste products. Although the metabolism of food or fuel to maintain the physical machinery of the body may be easier to concep- tualize, information exchange is just as critical to sustaining life. The exchange of information with the environment is a baSic activity of all living systems. The internal processing of stimulus inputs from the environment organizes human behavior and relates the organism to the environment (Thayer, 1968). As human systems usually exist in larger suprasystems such as the child in a family, the exchange of information across system boundaries links individual systems to one another and provides the building blocks for estab— lishing human relationships. A characteristic of open systems, as contrasted to closed mechanical systems, is the ability to exchange information with the environment. All living systems are somewhat open, but the degree of openness varies considerably. The environment can be viewed as all other factors outside of the system that impinge on the system (Chin, 1969). The environments of the child system are other human and socio-cultural systems, as well as physical and material aspects of the setting. Information exchange with the environment not only helps to maintain the system but also allows it to change or evolve to new organizational levels. Buckley (1967) referred to these processes as morphostatis and morphogenesis, respectively. The implication in terms of the child is that active information 13 exchange with the environment produces changes within the information- processing structures of the organism itself. Thus development occurs, and the organism moves to higher levels of organizational structure and functioning. Thayer (1968) noted this complex process: The way the child perceives, processes, and gives meaning to life experiences involves information flow. Man creates the reality and environment he knows and at the same time, proba- bly, the mind by which he knows it and therefore himself (p. 138). A feature that distinguishes human systems from other living systems is the information—processing capability that guides crea- tive action. The child is not only an open system but also a self- governing system (Gardner, 1971). The dynamic control processes that regulate behavior require that the effect of the system's activ- ity enter the system to provide information relevant to future activ- ity. These are often referred to as cybernetic control processes (Bertalanffy, 1968). Feedback is information about system output that reenters the system to help the system regulate and adapt to the newly created environment. Communication or interaction across system boundaries provides the vehicle by which informational out- put not only impacts on the environment and changes the system because of its release, but also is fed back or recycled to the system as environmental input. This array of information enters the system's internal processes. It is within the control processes that input is integrated with dynamically stored material (memory) and choices are made directing human behavior. All transactions of the organism with the environment provide input that helps the child organize his world into predictable patterns and relationships 14 necessary for what is often termed learning. Gardner (1971) noted: The child actively filters his perceptual world. He actively integrates the stored materials with whatever he pulls in from that world. And he actively decides on the basis of this integration the course of action to follow from among the alternatives available to him (p. 64). At least two dimensions of the system determine its capacity for communication: the level of development of internal information- processing structures and the openness of the system. The internal structures limit the kinds of perceptions, discriminations, and integrations possible. The openness of the system determines the amount and type of information exchanged across system boundaries. Human systems usually exist in a variety of larger systems, groupings of significant people and institutions; these are multi- person systems. The individual system at this level is a component of the larger system. As parts are interdependent and rule governed in a system's framework, the child and the larger system are inti- mately linked. The raw data with which the individual organizes himself and his concept of reality are products of interactions with the various multiperson systems in which he exists. The inputs to the individual system are the outputs of the multiperson system and conversely . . . (Thayer, 1968, p. 140). These intimate, interdependent transactions are the components of the socialization process. Socialization Socialization can be defined as the adaptation of the organ- ism to a social environment. Both the organism and the environment shape the process. 15 A child is born with a biological structure that allows behavior to be patterned and adapted to the surroundings. Growth and adaptation take place in a social world. A mother is much more than a caretaker; she can and does struc- ture the infant's environment. She and the other people that surround the infant relate their actions to his and so provide him with a means of building connections with their world and an entry to their social culture. They also act as mediators for him of the wider social order and it is through them that the child begins to learn about his place in the world (Richards, 1974, p. 7). The human infant in entirely dependent on his social envir- onment for survival. Communication initiated by the infant activ— ates caregiving and reciprocal need—satisfying interactions of infants with caregivers. The infant's cry, the parent's contingent physical contact, and the infant's quiescence is an example of earliest signaling systems reciprocally interacting. This inter- action lays the foundation for forming a series of interpersonal relationships that insure the emotional interdependence of the participants (Bowlby, 1969). Such relationships further motivate individuals to be sensitive to each other's signals, and, in the case of children, to imitate behaviors of significant others. This is a primary mode of social learning (Bandura, 1967; Weinstein, 1969). These complex processes (often referred to as attachment), common to all human groups, serve survival and adaptation functions and insure the child's gradual integration into the social complex of the larger suprasystem. Vickers (1967) aptly described the socialization phenomenon: "In-so-far as I can be regarded as human, it is because I was claimed at birth as a member of a communication network, which programmed me for participation in itself." 16 As patterns of communication evolve and relationships based on expectations and need satisfaction develop with primary care- givers, the child uses these experiences to relate to an increasé ingly wider array of social contacts. The child actively explores his world, selectively interacting with other systems and uniquely assimilating information from the environment. Internal representa- tions of functional behavior guide future interactions with the social environment. Two forces, therefore, converge: the individual's develop- mental strivings and the social system's organizational and normative influences that are reflected in the attitudes, beliefs, and skills of the child that comprise the child's social development at any one point in time. Communication Communication is the mechanism that links living systems to one another and to their environments. Through the exchange of signs and symbolic activity, meaning is shared, uncertainty is reduced, and understanding and action are facilitated. Depending on how one defines the system of interest, one can think of communica- tion as the process of sense receptors feeding impulses to the nervous system, or at a macro level, international treaties binding nations of people together. Communication serves an organizing function among interdependent parts, coordinating activity so that the system can maintain itself (Thayer, 1968). In the case of the human system, the flow of information within the system and between 17 the organism and the environment helps to move the system to greater elaboration of structure for higher-order functioning. This role of communication is critical in the elaboration of the human system during development. Various definitions and conceptualizations exist relative to human communication. Basically, all approaches include elements of participants (source and receiver) linked by a message or a symbolic exchange of information that reduces uncertainty (Berlo, 1960; Cherry, 1961). Communication can further be classified by intent. Intentional conmunication is that which participants initi- ate to establish some shared meaning or influence over each other. Unintentional communication refers to stimulus output that is received as communication, with no conscious intent to be communi- cated. The diagram in Figure 3 illustrates how both intentional and unintentional communication can occur simultaneously within the same event. Most definitions of communication refer to intentional commu- nication. This is reflected in Miller and Steinberg's statment of the purpose of communication: "The basic function of all communica- tion is to control the environment so as to realize certain physical, economic, or social rewards from it" (1975, p. 71). This perspective is consistent with the active nature of the human system seeking to interact with the environment through communication. However, ethologists and psychologists studying human inter- actions have suggested that much of human behavior is communicated and often without conscious, purposeful intent (Blurton Jones, 1972; D? unintentional m response original behavior ‘1 r~799 l 33¢.z' (2% . . ref/ezrew- -‘/" 1" I I 1 {Igfi§i, : 4:2 la Figure 3. Intentional and unintentional communication. Richards, 1974; Bandura, 1969). The informal nature by which children "pick up“ adult mannerisms and speech patterns is an example of this phenomenon. Watzlawick, Beavins, and Jackson (1967) suggested that all behavior communicates. One cannot not communicate. In such a con- ceptualization the delimiting concept of intent is not necessary; all behavior has the potential to communicate. What is actually communicated depends on the attributes of the individual participants and the nature of the message. The sender and receiver bring to the encounter characteristics and experiences that influence their inter- pretation of the meaning of messages, both intentionally and unin- tentionally sent by others. 19 Nan Lin's (1973) descriptive model of communication can apply to both intentional and unintentional communication. As described in Chapter 1, Lin suggested that communication be viewed as consist- ing of four phases: encounter, exchange, influence, and adaptation/ control. Within any one communicative act, influence can result even if intent to communicate is not present. However, over time, or in any communication "transaction," intent would probably exist as a motivating force to proceed in the communication process. With this model, the measurement of influence, in either an inter- action or a transaction, would reflect the effectiveness of the communication act. One factor contributing to effective transmission is communi- cation skill. Habermas (1970) defined communicative competence as the mastery of skills that establish intersubjectivity of meaning. The communication act is then seen as the vehicle whereby skillful communicators establish shared meaning. In this regard, both recep- tive and expressive functions are important. The neonate has well-developed receptive abilities to assimi- late stimuli from the environment, but limited infbrmation-processing and expressive capabilities. Thus the infant's communicative ability is limited, but is definitely present (Brazelton et al., 1974; Richards, 1974). Studies of earliest mother-infant interaction have provided evidence that children as young as 2 to 3 weeks intentionally modify their behavior based on feedback from the environment (Brazelton et al., 1974). Experimental manipulations of maternal behavior 20 have produced expectancy behaviors and modifications in children's responses to dissonant events‘during the first five months of life. These studies verify the child's active participation in communica- tion transactions with the social environment during early infancy. And yet the literature on the ontogeny of communication skill of children barely recognizes the ability of children to respond to and initiate appropriate signals in light of environmental feedback. Review of Research on Children's Peer Communication The research literature available on the young child's development of communication skill is fragmented. This may be a result of the failure of many researchers to distinguish between the linguistic and communicative ability of the developing child (Krauss & Glucksberg, 1970). Much attention has been focused on the development of language in children, especially as a result of the work of Piaget (1926), Vygotsky (1962), Bernstein (1968), and others who have emphasized the critical role of language in cogni- tive development. The development of communicative effectiveness, however, is even more comprehensive than the development of lin- guistic skills. Ryan (1974) noted: The so-called "one-word stage" cannot be regarded as simply the mere accumulation of a vocabulary, in preparation for its incorporation in later sentences. Rather, during this time, a child is learning a lot about how to talk, how to do many different things with words . . . and to partici- pate in mutual dialogues with others. Further, the use and understanding of standard words develops at a time when the ability to communicate non-verbally is well established, in the sense of being able to influence the behavior of others, and of indulging in reciprocal interchanges of various kinds p. 186 . 21 Thus communication skill precedes the use of language and continues to develop simultaneously with linguistic skills throughout the child's early years and even throughout the life span. Another force that has limited the scope of communication research has been reliance on operationalizing the cognitive role- taking skills underlying intentional communication. Communication studies during childhood have focused primarily on verbal informa- tion exchanges wherein role taking is required as a prerequisite for communication. This perspective follows from Piaget's (1926) writings. Piaget (1926) propounded a view that the young preschool child is egocentric. He suggested that young children lack the ability to perceive a point of view other than their own. It has been thought that this egocentrism limits interpersonal communica- tion, as speakers fail to provide the necessary information the listeners need in order to effect a sharing of meaning (Flavel, 1968). Piaget (1926) wrote: Before the age of 7 or 8 . . . understanding between children occurs only in so far as there is contact between two identical mental schemes already existing in each child . . . . In all other cases the explainer talks to empty air. He has not, like the adult, the art of seeking and finding in the other's mind some basis on which to build a-new (p. 133). Vygotsky (1962) further characterized young children's verbal exchanges as representing "inner speech," which is the abbreviated, condensed verbal coding used to communicate internally or mentally. Young children often fail to differentiate this form 22 of speech from public speech, which needs to convey meaning to others. Thus young children's communications are inappropriate in the social context, reflecting an egocentric perspective that only the speaker understands. Based on these theoretical perspectives, investigators have examined communication skill in terms of children's ability to take the perspective of another and to use this information in creating messages to inform another person about a specific topic (Flavel, 1968; Glucksberg & Krauss, 1967; Shantz, 1969). An example of this type of study is the experimental research of Krauss and Glucksberg (1970) concerning the adequacy of communica- tion between two like-aged children. As reported in that study, a two-person communication task was developed. The children were separated by a barrier and given similar sets of blocks with novel graphic designs. One child's blocks were stacked in a specific order, which he was requested to communicate to the other child so that that child could stack his blocks similarly. Training trials were given using blocks with familiar objects to check on the children's ability to comprehend the task, so that defective performance in the experimental task could be attributed to communication rather than understanding the rules of the game. Only the older children, 52 to 63 months, were able to perform the training task. Of the seven pairs who were given the experimental task, none could complete the task. To determine whether the difficulty lay in formulating the messages or in the listening skills of the children, a second task 23 was introduced with an experimenter playing the role of speaker. Eight of the twelve children at this age (46 to 63 months) were able to perform satisfactorily as listeners. Similar studies with experimental tasks have been conducted to see if the children could modify their directions based on the listener's feedback (Glucksburg & Krauss, 1967). Children were scored for modifying messages as a result of feedback requesting more information. Younger children failed to modify their original messages and often repeated the initial message. Skill in changing one's message to fit the needs of the listener increased consider- ably at the third and fifth grades and among adults. Although the results of these and similar studies have been offered as evidence in support of the notion of egocentrism, it is questionable whether it can be concluded that children fail to take another's perspective into account in such communication. It is possible that children performed poorly because of the complex nature of the cognitive and linguistic skills required in the tasks. To describe distinctive features of objects requires considerable classification and vocabulary skill. The level of difficulty of the cognitive processing required rather than the child's role-taking skill may be the factor limiting successful communication. Observational studies of children's language suggest that four-year-old children can take the perspective of another in com- munication. Shatz and Gilman (1973) reported significant differences in the message content observed in a series of studies comparing 24 the communication of four-year-olds when speaking to adults, peers, and two-year-olds. In the first study, 16 four-year-olds were asked to tell an adult and then a two-year-old about a toy. The specific topic was given to control somewhat for message content. Later a similar request was made without specifying the topic. Under both conditions content analysis of the conversations revealed significant differences in the length and complexity of sentence utterances when children spoke to the two-year-olds compared to when they spoke to adults. Another study compared eight four—year-olds' speech in spontaneous conversations with their mothers and then with peers. In this case, with respect to utterance length, the use of various constructions, and attentional utterances, peers were treated like adults. These results suggest that in verbal communication four- year-olds do make allowances for the characteristics of the listener. In both structured and spontaneous situations, children adjusted their speech to younger listeners, whether or not the speaker had a sibling. The investigators reported that children seemed to be able to respond to cues from the listeners as well as to perceive differing comprehension abilities of various listeners. Since the speech directed to both like-aged peers and adults was the same, it may be that children communicate with peers at their most developed level of ability. Although the studies described above stressed the content of messages, observations of the social consequences of children's 25 verbal communication also support the notion that children orient their communication to listeners and do not merely talk "to empty air," as Piaget (1926) noted. Mueller (1972) observed 24 pairs of children 3 1/2 to 5 years of age in a videotaped 20-minute free play session. Children were matched by age (within eight months of each other) and sex. None of the children were acquainted with each other. Descriptions of the frequency and social consequence of the verbal behavior of the children were reported. Mueller found that 94% of all utterances showed intent to communicate; i.e., the speaker looked at the lis- tener, talked about things of interest to the listener, or used attention-getting techniques. Further, 85% of the utterances received replies or at least attracted the listener's attention. These results suggest that children do not just egocentrically speak in monologues, but do communicate with each other. Although three- year-olds talked less than the older children and boys talked more than girls, there were no differences across age and sex groups concerning the number of communications that received attention or replies. The probability that an utterance would be communicated was .89 if the speaker attracted the listener's attention and only .78 if the content was of mutual interest. In total, only 5% of the utterances were so unclear as to be incoherent. The social skills of looking at a listener and getting his attention increased the chances that one's messages would be received. 26 Garvey and Hogan (1973) further investigated dyadic play behaviors of children 3 1/2 to 5 years of age to determine the degree to which utterances were used to facilitate social interaction. These utterances were termed social speech, defined as speech that is strictly adapted to the behavior of the partner. Eighteen white middle-class dyads were formed, and 15 minutes of their spontaneous free-play behavior was recorded on videotape. The transcripts of their verbal behavior were subsequently analyzed to determine the frequency, length, context, and consequence of utterance units (stretches of speech separated by pauses greater than one second or interrupted by another person's speech). Agreements by three independent judges determined whether or not a mutual engagement occurred. The consequences of utterances were coded in one of five categories: no consequence, unrelated speech, attending behavior, appropriate nonspeech behavior, and appropriate speech. The mean rate of utterance units was one every 4.6 seconds. Younger children had fewer utterances and the number of words per utterance was less than with older children. Dyads were considered to be mutually engaged on an average of 66% of the time. These engagements included games and activities with objects as well as verbal interactions. The mean percentage of utterances that entered into communicative exchange was 59% (s.d.=13.2%). Older children communicated more often than younger children, and did so through more lengthy exchanges. But even the youngest dyads were capable of conversations that included more than three components (initiation- response-initiation). 27 These results suggest that during the majority of their play- time children participate in mutual social engagements using speech as a way to achieve and maintain contact with each other. Although instances of private speech (Vygotsky, 1962) and monologuing (Piaget, 1926) existed, even the youngest children were able to initiate verbal messages in a manner that involved the other person and con- tributed to the social nature of the play activities. Hence even 3 l/2-year-olds communicated successfully with peers. Two similarities of all of these studies are their reliance on verbal exchange as the mode of communication and their focus on dyadic interaction. Observations of social interactions in a natural setting, such as the nursery school classroom, provide further support that in a social context young children communicate intent as well as information that elicits responses from others. Honig, Caldwell, and Tannenbaum (1970) reported on the use of continuous recordings to describe the social interactions of children in a natural classroom setting. They focused on information- processing behaviors--i.e., categories such as: confirms, shows to, converses, writes or draws, reads to, corrects or disconfirms, inquiries, informs or teaches, informs about culture, and role plays with. Both verbal and nonverbal modes of communication were included. The frequency and distribution of these behaviors were examined as a function of age. The sample consisted of eight children in each of four age groups: one-, two-, three-, and four-year-olds. Significant differ- ences were observed among age groups on total number of 28 information-processing behaviors emitted. The total number of information-processing behaviors increased with age, but the per— centage emitted by children to adults declined with age, as did the percentage that were self-directed. Communication to other children increased with age, especially from two years onward. The authors reported that showing or demonstrating behavior was a primary mode of communication for the preverbal and minimally verbal child, but was replaced with verbal exchanges in older child- ren. They noted a positive relationship between age and the fre- quency with which behavior was accompanied by verbalizations. Also, with age, children's information-processing behaviors were increas- ingly directed to other children rather than to adults. These results support the notion that even three-year-olds elicit an exchange of information with the peer social environment. Reuter and Yunik (1972) reported that 30 to 40% of a nursery school child's free-choice time is spent in social inter- action with children or adults, 20 to 30% with peers. Hence the peer social environment of young children from two years onward becomes an important arena for using and developing communication skills. Communication with a shared meaning does occur among peers of this age, and verbal communication is increasingly used to ini- tiate and maintain social contact. Consequences of Peer Communication In terms of both children's cognitive development (Piaget, 1926) and the learning of socially acceptable behaviors (Kohlberg, 29 1969; Clausen, 1968), peer interactions during the preschool years are critical influences on children's development. Piaget (1926) suggested that during peer interactions the young child receives information dissonant from his own, thus creating a conflict whereby he is forced to accommodate another point of view or another state of reality. Proponents of social learning theories also recognize the importance of peer interactions as incidental learning expe- riences (Bandura, 1969). William Hartup (1970) provided a comprehensive review of the literature concerning the research on peer interactions in childhood. He noted: "It is both challenging and difficult to isolate the variance in children's socialization that derives from contact with peers" (Hartup, 1970, p. 361). However, he did conclude that the type, length, and onset of experiences with peers do con- tribute significantly to the social development of the human child. The present study focuses on the degree to which peers are both responsive to and elicit responses from each other during play. As muchlyfyoung children's learning is incidental, the nature of the child's social contacts has the potential to influence the child's developing repertoire of social behaviors. Hartup (1970) concurred, stating that "much of children's learning is the outcome of social interaction, interaction that may not have been initiated with the intent of evoking behavior change" (p. 405). Much of the social learning literature examines the effects on children's behavior of exposure to various models and reinforcers. 3O Bandura (1969) discussed the theoretical foundations and research support for this type of learning. In general, contingent reinforce- ment by both adults and peers can affect the type of social behavior displayed. Since even attending behaviors of peers may be reinforc- ing, much of a child's social behavior may be shaped during peer interactions. Experimental manipulations of peer reinforcements have been found to affect behavior. Levison (1971) paired socially active and socially isolated children during a dramatic play session. As the isolated children became recipients of positive reinforcement in the form of positive peer interaction, their classroom behavior also changed. They showed an increase in both verbal interaction and cooperative play. In this instance, children's own limitations in rewarding peers also influenced the frequency with which they were recipients of positive social exchange. To generalize to communica- tion behavior, it may be that children who are responsive to peer initiations in turn produce more behaviors that are responded to by peers. Another mode of learning social behavior is through imita- tions of models. Both altruistic behaviors (Hartup & Coates, 1967) and aggressive behaviors (Hicks, 1965) have been noted to increase in children's play after exposure to models. Mowrer (1950) suggested that rewarded models would be imitated to a greater extent than nonrewarded models. However, Hartup and Coates (1967) found that this depended on the child's own experiences with peer reinforce- ment. Those children who received little reinforcement from their 31 peers imitated the nonrewarded model. Other discrepancies in the listerature exist, suggesting that there are no easily describable relationships by which to explain imitation behavior. The fact is that children do imitate both adult and peer models. The ability to assess what behaviors are attracting child- ren's attention in everyday environments is critical to begin to understand the complex incidental learning that occurs during social development. One of the focuses of the present study is on describing the social aspects of communication, including both verbal and nonverbal behavior. How do individuals impact on one another? Considering all behavior as having information value in a social exchange is only one perspective in the field of communication. Yet it is a most appropriate framework for unraveling the complexities of social interaction and social influence in human groups. Review of Methodological Approaches to Studying Children's Peer Interaction To study how one child interacts with another, most investi- gators have chosen to use a systematic observational procedure. Elements of the environment and the behaviors of the participants are usually operationalized and observed. Although verbal inter- actions can be analyzed from audio recordings, any definition of behavior that includes nonverbal actions requires a live or mechani- cally recorded observation. A number of observational techniques have been employed to investigate children's social interactions through the years. 32 Although these methodologies nay not identify communication as a behavior of interest, they all refer to communication through a form of interaction behavior. Earliest observational studies of peer interactions identi- fied rather molar behavioral events (Berne, 1930). They did, how- ever, serve a useful function in noting children's increasing attempts to initiate peer contact and response with age (Maudry & Nukula, 1939; Bridges, 1933). Buhler (1933) used running accounts of behavior to describe six-month-old infants' reactions to the presence of other infants. Olson and Cunningham (1934) described in detail the use of a time-sampling framework as a methodological approach to describe normative social behavior. Ruth Arrington ' (1943) further elaborated on this and other approaches, and noted the difficulties of comparing research when observational units and behavioral categories vary. Her review of the research to that date concerning childrens' social behavior could have been written in 1963, as few changes in methodology and scope were forthcoming during the interim. Boyde and DeVault (1966) reviewed the variety of observational techniques employed during the 1960's and the corresponding breadth of subject matter studied. An element of sophistication and rigor appeared to surround observational method- ologies by this time, suggesting that these techniques had "come of age," so to speak. However, a radical improvement in the quality and complexity of observational data resulted when video-recording equipment began 33 to be used extensively to record behavior. The capacity to play back and reanalyze behavioral sequences aided immensely in studying complex interpersonal interactions. Paulson (1972) compared the quality of ratings derived from videotape and live observations. Basically, the only limiting factor with video-recordings is the placement of the camera to achieve maximum coverage. Similar prob- lems exist, however, when observers need to remain out of sight of the objects of the observation. Wright (1960) described a variety of observational techniques for use with young children. He listed diary descriptions, specimen descriptions, time sampling, event sampling, field unit analysis, and trait rating. Some of these techniques are more apprOpriate for research purposes than are others. Those that include predetermined behavioral categories to be observed provide more structure and objectivity to the observational data than mere running accounts or retrospective descriptions of behavior. The trait rating technique has advantages in describing children's general social tendencies, but lacks precision when studying actual behavior or behavioral interaction. The most appropriate techniques for observing overt behavior are time- and event-sampling techniques (Smith & Connolly, 1972). Instruments exemplifying each of these techniques and combinations of them are discussed in the following sections. An additional tech- nique that is described is a specimen-description technique combined with event sampling. The procedure was used in a study discussed 34 earlier concerning children's information-processing behavior in the classroom (Honig et al., 1970). Time Sampling Goodenough (1928) described time sampling as the observation of the behavior of an individual or a group of individuals for definite short periods of time and the recording of the occurrence or non-occurrence of certain Specified and objectively defined forms of behavior during each of these periods (p. 230). The information derived from this procedure is usually the proportion of time intervals in which the behavior occurs relative to total time. Over repeated observationalperiods, descriptions represent normative behavioral tendencies of individuals. Of importance to this tech- nique is the length of the observational time unit relative to the definitions of behavioral categories. A balance must be achieved so that a behavioral event can be adequately observed in one time frame. Time sampling is best suited to studying behaviors that occur frequently. Smith and Connolly (1972), in a study of children's peer interactions, described the use of a time-sampling procedure to obtain an overall picture of the behavior of preschool children with differing age, sex, and background characteristics. Based on an ethological approach, molecular behaviors are considered. In this procedure, over 60 categories of facial expressions, vocali- zations, bodily postures, and motor patterns are observed. The observational unit is a lO-second interval rated continuously for five minutes. 35 The children were observed in a nursery classroom during two times of the day for 12 separate five-minute records. Examples of the behavioral categories are: facial expressions--smile, pucker; vocalization--talk, laugh, scream; motor activity--climb, run, point; antagonistic behaviors--threat, hit, flight; and social interactions--wa1ks to, helps, and cooperates with. In this study, behavioral frequencies relative to total time were factor analyzed to describe molar patterns of social behavior. Three principle components were derived, reflecting social maturity, play with or without materials, and an age-related construct. Other analysis strategies can also be used with this procedure. This particular example of a time-sampling technique uses rather discrete behavioral events as the objects of observa- tion. Others have used broad composite categories of behavior to assess children's play participation. Parten (1932) used the technique with one-minute observation intervals. Her six categories of play are frequently used in child development texts. Event Sampling Event sampling is a slight modification of time sampling. A time period is required as a reference point, but the frequency or sequence of behavioral events is examined. The Appalachia Education Laboratory developed a categorical rating system as a continuous event-sampling observational procedure (Pena, 1971; Pena and Miller, 1971). Occurrences of specific cate- gories of behavior were recorded. Categories concerned such 36 behaviors as vocalizations, peer interactions, attention to tasks, and affective acts. A pair sequence analysis procedure was developed to note the sequence as well as the quantity of activity. Event samples usually allow greater breadth in defining types of behaviors that are possible to rate. As each category needs to be noted only when it occurs, observers can be trained to look for a variety of possible events. Also, an advantage is the ability to note the sequence and context of behavioral events as well as their frequency. The event-sampling procedure is most advan- tageous for observing behaviors that occur infrequently or sporadi- cally. Specimen Description Specimen descriptions are usually written records of short behavioral episodes. Their objectivity and precision usually depend on the skill and effort expended by the observer. Although specimen description is less appealing for research purposes than other methods, Caldwell (1969) described the application of this technique in a research instrument. It was developed by Caldwell, Honig, and Wynn and is called APPROACH--A Procedure for Patterning Responses of Adults and Children. A trained observer walks through the classroom or place in which a child is playing and whispers a running account of the child's behavior into a portable audio tape-recorder. Later these recordings are analyzed using a system that describes "behavioral clauses." These units are running records that contain a verb. 37 Each clause is further divided into subject, predicate, object, and action qualifiers. The subject is usually the central figure. Predicates can take on a variety of categories such as environmental action--ignors, attends; information processing--converses, writes; negative reinforcement--resists, rejects; and body activit -- increases or accelerates. Nine different predicate categories are available, plus behaviors within each grouping. The object of the behavior can be other children, adults, or the environment. Qualifi- ers note the intensity, quality, or motivational basis for the behavior. Last, the setting is nated in terms of the child's play activity, geographic location, or social milieu. Since behavioral clauses vary in length and sequence, a wide variety of variables can be derived. Caldwell (1969) suggested that the main advantage of this fbrmat is that the setting of the behavior, in terms of time and space, is observed in as much detail as is the actual behavior. Thus an ecological perspective can be derived. Time and Event Samplhng The last observational methodology to be discussed in this section is a combination time- and event-sampling procedure devel— oped by Boger and Cunningham (1971) to describe the socialization behavior of children in peer interaction. It is called the Observa- tion of Socialization Behavior (OSB). A basic time-sampling frame of ZO-second intervals was established so that a systematic profile of children's typical 38 behavior could be derived. However, in addition, behavioral cate- gories were designed with subcategories to describe a variety of possible behavioral events. Such categories are mutually exclusive and exhaustive. Rating scales are also incorporated to note inten- sities of activity and affect. Therefore, both quantitative and qualitative data can be derived. Ten separate scales are included, eight based on observed behavior and two based on situational inference. The scales are: social behavior, involvement, verbalization, physical behavior, play context, peer interaction, group interaction, adult interaction, inferred motivation, and emotionality. Each of the scales focuses on aspects of children's play involvement with the environment. The social behavior scale was developed by Mildren Parten (1932) to measure children's social participation in play. Cate- gories are ordinal, and are labeled as follows: unoccupied behavior, solitary play, onlooker behavior, parallel play, associative play, and cooperative play. The involvement factor refers to the nature and intensity of children's activity. Categories are initiation; response with three qualifiers, accept, reject, or ingore; and ongoing activity. Each category of involvement is also rated according to intensity-passivity, moderation, and intensity. The object of one's involvement is coded as either a specific individual, a pair of individuals, the group, or an adult. Dimensions of children's verbal and nonverbal communication behavior are also included in the rating instrument. The verbaliza- tion codes are those developed by Robert Bales (1951) in the Bales 39 Interaction Process Analysis instrument. Twelve categories of ver- balizations are available, including task-oriented and socio- emotional affectual verbalizations. Fantasy verbalizations are also noted. Nonverbal communication was recognized through a physical behavior category, which refers to the social-affective nature of the physical activity. The setting of the interaction is coded as large or small muscle, dramatic play, cognitive activity, or routine. The two inferred behavioral states are emotionality--the degree of happi- ness or sadness displayed--and inferred motivation. The latter category contains 14 behavioral states that could be identified if inferred from behavioral cues. Examples of these states are empathy, independence, imitation, and attention seeking. The function of the OSB instrument is to describe character- istics of children as derived from both proportional data and aver- age ratings. Contextual dimensions of the interaction can also be studied. The directionality of behavior, the child's play context, his verbal and nonverbal communication, and his inferred motivational states can be determined. This diversity of data provides an opportunity to investigate a number of research questions through a variety of analysis strategies. The OSB offers more standardization of behavioral descrip- tions than does the specimen record, because the time-sampling framework does not allow children's rate of activity to influence the results. The OSB offers the same versatility in describing 40 interactions in terms ofthe social and play conteXt, but is limited in describing the environmental context of behavior. Likewise, it offers similar scape in noting qualitative dimensions of behavior as the specimen description and more than the time- or event-sampling methodologies. Last, the objectivity and precision of observing overt behavior directly are retained, with the same investment in time and observer effort as with the specimen description. Thus, in many ways, the time- and event-sampling procedure has advantages over other techniques. Its main disadvantage is that the sequence of behavior cannot be determined because of the cross- sectional nature of the data-collection procedure. This is a limita- tion of all time-sampling procedures. In addition, however, a limi- tation of this and other methodologies discussed is that they only focus on one subject of interest at a time. Although the object of the subject's behavior is defined, the reciprocal actions of others in the environment in response to the subject's behavior are not determined. Interaction is viewed from only one participant's point of view at a time; hence the total impact of an individual's behavior on the social environment is not observed. This specific dimension of behavior becomes critical when analyzing children's communication. Since any behavior has the potential to influence any receiver in proximity, a more holistic view of social exchange is required. The revised version of OSB (Boger, Cunningham, & Andrews, 1973) used in the present study includes this dimension of group 41 response. However, in operationalizing this procedure, children are observed while playing in small groups. Implementing this additional procedure in a natural classroom has not been explored. All of the observational procedures discussed in this section can be used in the natural classroom setting or in any smaller social grouping. The classroom offers the advantage of observing naturally occurring behavior in an everyday setting. However, the controlled small group methodology has advantages, too. Small Group Social-Interaction Methodology.--Small group social-interaction studies are predicated on an assumption that the elements of behavior exhibited by individuals in small groups would be similar if observed in other social units. Thus patterns of behavior or tendencies to behave in Specific ways are being derived as enduring personality characteristics (Hare, 1962). If such an assumption can be accepted, the standardization of environmental conditions offered by a controlled setting has a substantial impact on the quality of data received and the potential for deriving relational propositions. Although new confounding effects may be introduced, as critics have suggested (Golembiewski, 1962) (i.e., individuals' reactions to novelty, or a decreased field of choice of playmates), these effects can be systematically manipu- lated and studied once the environment is standardized. Naturally occurring variation in even the physical environment is difficult to assess and control in natural settings (Arrington, 1943). In class- room settings an indirect influence on social behavior is the nature 42 of the cognitive and normative set associated with teacher and class expectations (Emmerich, 1973). No such normative control is opera- tive in a nonacademic setting. Therefore such a setting hypotheti- cally provides an opportunity for children to elicit spontaneous behavior. Hence, small group controlled settings may offer advantages over natural settings when one wishes to derive comparable data across children, groups, or time. Small groups are also more amenable to video recordings, because all of the children can be observed simultaneously by a single observer. Review of Systematic Differences in Children's Peer Interactions An objective of the present study is to determine if communi- cative competence differs among children from various demographic groups. The primary independent variables are sex, socioeconomic class, ethnicity, age, and past experience with siblings or peers in group settings. Extensive research exists on age, socioeconomic status (SES), and experiential influences on children's behavior and development (Clausen, 1966; Hess, 1970; Hartup, 1970). However, in screening this literature for propositions concerning effects on children's social behavior or even more specifically, communication behavior, one derives few definitive links. The discussion that follows is based on observational research of social interaction or communica- tion behavior. 43 Sex In the Appalachia Preschool Study, Pena (1971) reported significant sex differences in the social behavior exhibited in a structured play setting. Boys talked more, initiated to others more, and stayed at tasks longer than girls. Jersild and Markey (1935) reported that boys of the preschool age engaged in more con- flicts during free play than did girls. They also reported this relationship for lower-SES children over middle-SES peers. Smith and Connolly (1972) similarly noted that boys engage in more rough- and-tumble play than do girls. In a longitudinal study of children's socialization behavior, Boger and Cunningham (1972) noted a sex-by-social-class interaction (P < .05) on the amount of time children were engaged in adult interaction in a controlled play setting. In this study, the original OSB instrument was used to describe children's social interactions. More child-initiated adult interaction occurred for lower-class subjects than for their middle-class peers, and this behavior was more pronounced for females than for males. In general, then, one would expect males to have higher communicative competence scores than females. Socioeconomic Status The literature on social class differences is more explicit in regard to cognitive behaviors than it is for social behavior. However, a great deal of research has focused on children's language development relative to social class. This interest has been based 44 on Whorf (1956) and Bernstein's (1964) view that different linguistic codes prevail in the communication among different social groups and influence the way individuals perceive, process, and transmit infor- mation. The child's own linguistic and cognitive development is influenced by these pervasive styles of communication. Hess and Shipman (1965), Brophy (1970), and Cunningham (1972) found support for Bernstein's proposition in noting different patterns of interaction between lower-class parent-child dyads and middle-class dyads. Generalizing from these studies, one would expect that lower-class children communicate less precisely than do middle-class children. This hypothesis was supported by Krauss and Rotter (1968) and Baldwin, McFarlane, and Garvey (1971) in experimental communica- tion studies with older children, and by Bearison and Cassel (1975) with six-year-olds. However, these communication studies reflect cognitive skills, and whether this relationship would be supported with measures of social communication is presently unknown. In respect to different modes of exchange, Miller and Swan- son (1960) suggested that lower-class individuals express themselves more effectively through nonverbal means, whereas middle-class persons express themselves best verbally, through the manipulation of ideas. Schachter and others (1970) reported that low-SE5 and black children exhibited more frustration talk, for example the expression "Stop!" White and middle—SES children were able to verba- lize more detail and were thus more explicit in their conversations. 45 As the Operationalization of communication in this study includes both verbal and nonverbal behavior, SES effects are diffi- cult to predict. Race or Ethnicity Race and social class variables are often confounded in social interaction research. A longitudinal study conducted by Boger and Cunningham (1972), however, did include race as an inde- pendent variable. In a completely crossed and balanced design, race, sex, and SES were the three independent variables. In their study, during classroom observations, black children exhibited significantly higher average levels of social behavior than did their white peers. Because the social behavior variable reflects quality of interaction, communication scores may also vary across ethnic groups. Ase A number of studies have demonstrated that during the pre- school years children play with peers, in lieu of playing with objects or adults, increasingly with age (Blurton Jones, 1972; Honig et al., 1970; Adams, 1967). Blurton Jones (1972) reported that age differences in some social behaviors are more persistent than experience with peers or siblings. Among younger children, those who had siblings showed more social interaction behavior, but by the time they were four years old, children with or without siblings behaved similarly. 46 Experiences with Siblings and Peers Blurton Jones (1972) further reported a slight but nonsig- nificant tendency for children without siblings to exhibit less aggressive behavior. Although siblings can have a significant effect on children's sex-role attributes (Clausen, 1966), little evidence is available to describe the impact of siblings on communication or social inter- action variables. A study of fifth and sixth graders' answers to questions about how they influence their brothers or sisters led Sutton-Smith and Rosenberg (1970) to suggest that sex and birth- order effects present rather stable influences on children's social interactions. Males were reported to use more physical techniques, as compared to females' use of symbolic or psychological techniques to receive compliance from siblings. Sutton-Smith and Rosenberg concluded that children who are larger or older usually dominate social interactions. Ralph and her associates (1968) found that the number of child-child negative interactions at age five was highest in children with one year of nursery experience, lower in children with two years of nursery experience, and lowest in children with three years of nursery experience. In the study by Smith and Connolly (1972), length of nursery school experience was more highly correlated with the following variables than was age: sociability in play, rough- and-tumble play, laughing, and smiling. Thus, experiences with other children can impact on children's patterns of social inter- action. 47 In a discussion of the contributions of parents and peers to children's development of interpersonal skill, Weinstein (1969) noted a basic "division of labor" between parents and peers in the early socialization process. He wrote: Basic capacities like empathy and personality orienta- tions conducive to effective interpersonal control are seen as coming primarily from parents. But the refinement of these orientations and their impact on the acquisition and utilization of lines of action seem to be more heavily molded by peer interaction (p. 773). Since the primary variable of interest in this study is a measure of overt social influence, communicative competence may be less sensitive to demographic characteristics of families than to experiences with peers. However, as measures of both of these factors are included, this study provides an opportunity to explore these potential relationships. CHAPTER III METHODOLOGY In the preceding chapters, an overview of the research project was provided and the study was given perSpective through an identification of a theoretical orientation and a review of relevant research. The task in this chapter is to describe the research design and hypotheses, sampling procedures, and resulting sample, and to examine in detail the observational instrumentation, the Operationalization of specific variables, and the data reduction and analysis strategies implemented to test the research hypotheses. Research Design This is a descriptive, correlational study. Therefore the data contained herein reflect children's behavior and demographic status at one point in time. The focus of inquiry in the study is on communication, a measure of system openness. The communication flow among the children was inferred from observations of individuals' actions and the subsequent responses of others in the play group. The degree to which each child impacted on the social environment was considered a dimension of openness within the individual child system and was termed communicative competence. 48 49 The relative incidence of communication across time-sampled intervals represents a measure of openness in children's play groups and was analyzed to describe normative patterns of social communica- tion among three- and four-year-old children. The relationships that were investigated in the study are illustrated in Figure 4. I A[Background Characteristiil I C Peer A Behavior->- Peer B / \ / <—-—Behavior / / I Commuzficafive competence Attributes of Social Interaction Behavior Individual Child System \ \ \ \ \ I Communieation I Communication within the Group | I I | I I 1 Figure 4. Schemata for data analysesa The following variables were operationalized: A. Background Characteristics Sex Social Economic Status aLetters designate categories of hypotheses. 50 Ethnicity Age Number of siblings Number of months since entering center Number of months in group care 8. Attributes of Social Interaction Level of Social Behavior Activity Level Responsivity Initiative C. Modes of Communicative Exchange Channel of Exchange Verbal Nonverbal Affective Connotation of Physical Behavior Negative Neutral or positive Directionality of Initiations Individual peer Group Materials or environment Hypotheses The research hypotheses explored in this study were organized around the following three basic research questions: A. What are the relationships between children's background characteristics and communicative competence scores derived from observations of children's small group peer interaction? H1. The number of children in the family is a predictor of communicative competence. H2. The length of time in group care and the length of time enrolled in the present day care center are predictors of communicative competence. H3. The age of the child is a predictor of communicative competence. H4. Children from middle-SES homes have higher communi- cative competence scores than children from low- SES homes. 51 H5. Males exhibit higher communicative competence scores than females. . H6. There are no differences in anglo and black child- ren's communicative competence. ' B. What are the relationships between characteristics of children's social interaction behavior and communicative competence? H7. Children's social behavior and communicative compe- tence scores are positively related. H8. Children's initiative scores are more highly related to communicative competence scores than to respon- sivity scores. H9. Responsivity scores and communicative competence scores are positively related among children with similar activity patterns. C. What is the effect of the mode of communicative exchange on the incidence of communication? H10. More communication is elicited in intervals with verbalizations than in intervals without verbaliza- tions. H11. More communication is elicited in intervals with negative interactions than in intervals with positive or neutral interactions. H12. More communication is elicited in intervals with initiations directed to individual peers than in intervals with initiations directed to the group or the materials and environment. SamplinggProcedures The present study is part of a larger research endeavor sponsored by the Office of Child Development and implemented through the Institute for Family and Child Study, Michigan State University (contract number OCD-CB-4851- The larger study, entitled "Early Social Development: Parent and Child Programs," was an evaluation 52 of the relative effectiveness of short-term parent and classroom programs on day care children's self-concept and social involvement with peers (Boger & Andrews, 1975). As a quasi-experimental design, eight day care centers were sampled and randomly assigned to one of four treatment conditions. Data were collected before and after the program implementation period during the fall of 1973 and spring of 1974. Part of the first wave of data collected for the larger study was used in this study. Sample Selection Initial information concerning potential cooperating centers was secured through the State Department of Social Services Day Care Licensing Division and area 4-C coordinators. After screening lists of potential centers against basic criteria, staff contacts and visits ensued. The original criteria for center eligibility included the following: 1. Distance from MSU--maximum 70 miles 2. Listing with the licensing division of the State Depart- ment of Social Services 3. Offering a full-day program 4. Comparable philosophy, program, and staff qualifications 5. No simultaneous participation in other research or program obligations 6. Heterogeneous enrollment of children to meet the follow- ing needs: a. age range--3 1/2 to 5 years b. enrolled for four half-days per week 53 c. minimum of eight children in each of four groups (excluding kindergarteners): low— and mid-SE5 males and femalesl d. racial balance across SES groups or all one race To secure an adequate number of children within the age and enrollment range, medium— to large-sized centers were approached. All centers considered met the first five criteria. The distribu- tion of children across sex, SES, and ethnic groups was the most difficult sampling criterion to satisfy. The centers selected offered the best balance in enrollment of those centers available and willing to participate. The eight centers selected were located in four large cities in lower Michigan. Their enrollments ranged from 70 to 166 children per center. The characteristics of the children enrolled in these eight centers varied somewhat. All children within the age range of 3 1/2 through 5 years (as of January 1, 1974) who were enrolled for at least four days per week at each center were considered eligible to be included in the sample. Parents were notified of the research project and their cooperation in permitting their children to be tested was solicited. Parents were also asked to sign a permission form and complete a general information sheet that requested facts about the child's family and the child's past experiences in group care.2 The information needed to determine socioeconomic group membership was secured from the parents through this general 1SES membership was initially determined by eligibility for Social Service assistance. 2Copies of these forms are presented in the Appendix. 54 information sheet. The criteria for delineating social economic group membership were adopted from the short form of the McQuire and White (1955) research tool, The Measurement of Social Status. Weighted scales composed of the social status components for occupa- tion, source of income, and education were evaluated for the princi- pal wage earner of the family. In cases in which both parents were fully employed, the father's index score was used. If there were extreme variances between maternal and paternal SES index scores, a subjective evaluation employing the median or the mother's index score was selected as the characteristic for the child's SES value. When determining SES groups, low or middle, an index score value of 51 (:3) was considered the critical cut-off point between middle- and low-SES groups. Scores from 48 to 54 were considered flexible and could be placed in either group, based on other idio- syncratic information. All of the children who satisfied the basic age and enroll- nent criteria and whose parents consented to their participation were subsequently tested using a battery of four instruments. As only the observational data from the controlled play setting would be used in the present study, the total sample of children was further screened to select a subset of only those children who appeared in play groups in which all four children were of the same ethnic identity and whose ages ranged between 36 to 60 months. As a result, the present sample consisted of 139 children enrolled in eight different day care centers in lower Michigan. 55 Description of the Sample The distribution of children by sex, social economic status, ethnicity, and age is described in Table 1. Table 1 Characteristics of the Samplea Characteristic N % SEX Male 68 49 Female 71 51 Socioeconomic Status Low 72 52 Middle '67 48 Ethnicity Black 33 24. Anglo 106 76 Age Three-year-olds 32 23 Four-year-olds 107 77 aN = 139 The sample was rather evenly divided by sex and SES. How- ever, larger numbers of anglo children than black children were included in the sample. Also, about one-fourth of the sample were three-year-olds, whereas three—fourths were four-year-olds. The average age of the children in the sample was 53.14 months (s.d.= 5.26 months). 56 Most of the children were enrolled full time in the day care programs. The average enrollment was 9.5 half days per week (s.d.= 1.45). A wide variation existed in the length of time these children had been attending the day care centers. On the average, children had been enrolled fbr 7.77 months prior to September 1, 1973. How- ever, the standard deviation was 7.24 months, suggesting wide varia- tion in enrollment patterns. Variation was also evident in the children's previous experience in group care (Mean=18.00 months; s.d.=15.6 months). The average number of children in these families was 2.01 (s.d.=l.46). Of the total sample, 65 or 47% of the children were the only child in the family, whereas 74 or 53% of the children had siblings. The number of siblings per family ranged from 1 to 9. Approximately 63% of the children lived in single-parent families and 37% lived in two-parent families; a larger proportion of low-SES than middle-SES children lived in single-parent families. The ethnic distribution by family status was approximately equal. The children comprising the sample could be considered typical of most urban day care populations. About 90% of the mothers of these children worked while the child was at the center; 67% of the mothers were employed full time. Instrumentation A small group social-interaction methodology was operation- alized in this study developed by Boger and Cunningham (1969). The method consisted of videotaped observations of peer interaction 57 in a naturalistic setting. The situation for gathering the obser- vational data was a ten-minute unstructured play session in which four children (one male and one famale from each socioeconomic group --low or middle) were randomly selected and asked to play in a mobile playroom. The only play materials were cardboard cartons. This playroom was an interior space of a travel trailer, approximately 11 feet by 8 feet. It was completely carpeted, lighted, and heated similar to any indoor space. A portable wooden expanding gate extended across the room at the point marking the limit of the lower visual field of the camera. A Space behind the expanding gate was provided for the examiner to sit outside of the children's interaction range. A diagram of the mobile unit is shown in Figure 5. PLAY AREA ‘- ll' . / ‘e / (9 (38 A ‘ on e-way observation window I I\ L MOBILE UNIT VTR ' Camera E Figure 5. Diagram of mobile unit. 58 The observation in the mobile playroom provided a sample of children's behavior in an open-field standardized setting. The situation was designed to allow an opportunity for the children to manifest preferred modes of behavior or behavioral "styles." The children were not directed in their play, and the materials (boxes) presented no inherent play mode. "There is no overt indication of behavior expectations, and there is no attempt to guide, limit, or structure behavior" (Boger & Cunningham, 1971). All of the children were brought into the mobile classroom before data collection to become familiar with the setting and equipment. Then, upon entering the room for the play session, the children were read a brief statement. It explained that they could play in any way they wished so long as they did not hurt each other. They were also reminded to remain behind the expandable gate. The adult observer was present but outside of their play range. The observer remained in the room working on papers so that it did not appear that the children were being watched! The ten-minute play session was videotaped for subsequent rating and coding using the revised version of the Observation of Socialization Behavior instru- ment (Boger, Cunningham, & Andrews, 1974). The observational rating procedure was an adaptation of the original Observation of Socialization Behavior instrument (Boger & Cunningham, 1969). Many of the procedures and measurement scales of the original instrument were preserved. However, the method of sampling behaviors and the perspective toward "interaction“ were 59 changed considerably. These changes are discussed in the following section, along with a description of the instrument. The revised Observation of Socialization Behavior (OSB) instrument is an observational technique using a combination time- and event-sampling procedure.3 At 20-Second intervals, a mechanical beep is superimposed on the audio portion of the tape. Raters record the first play behavior at each ZO-second mark, thus securing a time sampling of behaviors across the ten-minute play session for each child. The observation interval chosen for this study was 20 seconds. This time span was selected as sufficient to record a meaningful sequence of behavior in a manageable and recordable manner. The video medium, however, was necessary to encode the total complexity of the behavioral interaction as proscribed by the rating procedure. Three and up to four playbacks were usually required to complete the rating process. 99113393 Based on an ethological approach, more global styles of behavior are produced from the analysis of more molecular behavioral units. At each 20-second mark, various behavioral dimensions of the play involvement of each child are recorded. Fourteen behavioral dimensions were chosen as mutually exclusive, objectively describ- able categories of behavior: 3Only the time sampled data are used in this study. See the description of the instrument in the Appendix for a discussion of the time and event sampling procedure. 60 Interaction (responses, ongoing play, initiations) Object of interaction Level of involvement Peer impact Verbalization Verbal fantasy Voice tone Physical behavior Physical tone 10. Social behavior 11. Autonomy -—-——-— 12. Leadership 13. Social competency l4. Emotionality gogowmm-wa—a Socio-emotional Scales The unit of observation is a "bit" of play behavior or sequence of interaction. This "bit" is defined as a stretch of play behavior that includes an involvement code of either response, ongoing activity, or initiation that is terminated by another response or initiation. All 14 categories of behavior are rated in reference to this single stretch of behavior. Because of this behavioral contingency, patterns or styles of interaction can be analyzed. This contingency dimension is a unique feature of this adapted version of the OSB. The original version of the instrument sampled each behavioral category separately whenever it first occurred in each ZO-second interval. Thus, an advantage of this adapted version is that tallies of behavioral categories can be compared to each other as well as to total time. Thus more versatil— ity is available in terms of analysis strategies. Measures Interaction and Involvement.--The form, sequence, intensity, and object of the play involvement are recorded. Since an inter- action sequence is of interest, response and initiation categories 61 were established. Responses include acceptance of another's initia- tion (A), rejection of another's initiation (R), no acknowledgment or awareness of another's initiation (N), ongoing interaction (0), and behavioral transition or eminent initiation (X). Following a response, an initiation may or may not occur. Initiation (I) is defined as an introduction of self or change in activity. Each of these two major categories is rated on degree or intensity of involvement. Three levels range from intense to passive. The object of the involvement is also recorded as group (undifferen- tiated), adult, individual or pair of individuals, materials, or environment. impagt.--Another major change in the revised OSB is that a new level of interaction is observed. Responses, initiations, and consequences can all be derived from each "bit" of interaction. The system of observation in the original OSB is illustrated in Figure 6. 03 _s_ s 02 2\02 _S_ responds to 01 _S_ initiates to 01 & at a moderate level: 02 at a moderate level. Figure 6. Interaction as observed in the original OSB. 62 In the revised OSB, the following is observed (see Figure 7); 2 C’1 7- 1 1 / 3 / 035 § 03 §\1\ 03 s 02 O2 02 _S_ responds to 01 at § initiates to 01 8: § impacts on 01 and 02 a moderate level: 02 at a moderate moderately and 03 passively. level: Figure 7. Interaction and impact as observed in the revised OSB. The impact codes were developed specifically to note the impact of a subject's behavior on others in the environment. Based on communication theory, it reflects a measure of environmental control or the influence stage of communication (Lin, 1973). In operationalizing communication, the consequences of the subject's involvement are recorded as reflected in the immediate behavior of the other three peers. Three response categories are available: acceptance, rejection, or no acknowledgment. Three levels of intensity of response are also rated if a response is noted: intense, moderate or average, and passive. This behavioral dimension of the interaction measures the environmental impact of children's behavior and is useful in determining differential communication patterns and the behavioral context of various comuni cation acts. 63 Verbalization.--The time-sampling procedure allows for a measure of the frequency of occurrence of various types of verbali- zations, and the behavioral context permits analysis of the rela- tionship between verbalizations and other behaviors. The Bales Interaction Process Analysis (1951) provides the basis for coding verbalizations. Twelve categories plus mumbling (unintelligible) are included. These categories are mutually exclusive and exhaus- tive; a complete verbal interaction is considered the unit. A more affective dimension of voice tone is also rated. It is a three- point scale--positive, negative, or neutral. The voice tone refers to the delivery, not the content, of the verbalization. In addition, each verbalization is rated in terms of fantasy or nonfantasy. The incorporation of the voice tone and fantasy codes is a change in the revised OSB. Physical Behavior.--The physical behavior codes were also changed in the revised OSB. As much of the young child's behavior is nonverbal in nature, a physical behavior rating is included. One aspect, physical contact, is rated in respect to the object of the interaction. When both materials and people are objects of interaction, the human interaction is considered first. Contact then refers to physical touching of another peer directly or indirectly through the medium of the play materials (boxes). When no human interaction is involved, the contact may be with materials. Another aspect of physical behavior is its positive or negative quality. As with voice tone, a physical behavior tone is 64 rated in reference to the affective nature and social aCceptability of the behavior. Hitting, pushing, and kicking are considered negative qualities. Tapping, patting, and caressing are considered positive qualities. Neutral behaviors refer to nonaffective activi- ties such as building or running. Social Behavior.--The ordinal scale developed by Parten (1932) was adopted as a measure of the child's social behavior. The categories include unoccupied play, solitary play, onlooker behavior, parallel play, associative play, and cooperative play in order of increasing sociability and maturity. The criteria for the various categories include spatial proximity to other children, similarity of materials being used, degree of reciprocity of interaction, and goal-directedness of the play. The social behavior dimension pro- vides a measure of quality of social interaction as well as an over- all measure of social maturity. Socio-emotional DimensionS.--The general tone of the child's social and emotional behavior is also rated, but admittedly is based on more subjective judgments on the part of the raters. Specific behavioral cues help define the dimensions, and a five-point scale based on the observability of the behaviors helps to objectify the rating procedure. The 14 inferred motivational states of the original OSB were modified and included in the present version as three new socio- emotional scales. The emotionality scale remains from the original instrument. 65 Thus four dimensions of the child's inferred socio-emotional state are rated: autonomy, social leadership, social competency, and emotionality. As defined, these dimensions are mutually exclu- sive. A five-point bipolar ordinal scale is used to rate them. The extreme positions, both positive (5) and negative (1), are designated for overt behaviors representative of the dimension. The central position (3) is a neutral, nonobservable indicator. The two inter- mediate positions (4) and (2) represent covert behavioral cues or mild overt behavioral indications of the dimension. These ratings provide an indication of the general social and emotional nature of the behavioral interactions and are rated contingent upon the other categories of behavior encoded. Oata-Gathering_Procedures All of the play sessions were organized and supervised by trained members of the research staff of the Institute for Family and Child Study, who were graduate students from the Department of Family and Child Sciences, Michigan State University. Their activ- ities were coordinated by the present investigator. The responsibility of this field staff was to select the groups of four children, accompany them from the classroom to the mobile playroom located on the schooi property, and implement the standardized procedures described earlier. A media technician operated the video-recording equipment during the play sessions. Undergraduate students were trained to rate the videotapes using the OSB rating procedure. These students were pursuing 66 degrees in the social sciences and had had previous experience work- ing with young children. The training period lasted approximately two to four weeks. Discussions, practice sessions, and simultaneous ratings by more than one rater, accompanied by discussions, were the format of the training experience. As more than one rater was involved in this data-processing procedure, an inter-rater agree- ment of 85% on total recordable positions was required. Once raters established reliability they proceeded to rate the tapes indepen4 dently. An additional problem with observational ratings, however, is "instrument decay"--a gradual drift away from consensus. To counteract this phenomenon, periodic group discussions and inter- observer checks were conducted. The videotaped play session and subsequent rating procedure are illustrated in Figure 8. Reliability and Validity Reliability The measure of reliability derived for this rating procedure is inter-rater reliability—~an indication of how consistently behav— iors are identified by more than one person. To maintain high inter-rater agreement, behavioral units must be recognizable and objectively defined, therefore, also reflecting the validity of the categories of behavior. All raters needed to establish a minimum level of inter- rater agreement of 85% on total recordable positions with the present investigator. The actual percentages of agreement achieved 67 Figure 8. Data gathering procedure. ranged from 86% to 98% between two raters over a ten-minute sequence of play activity. Validity The validity of observational measures derived by time samp- ling can be assessed by examining three factors: the naturalness of the behavior observed, the accuracy with which it was recorded, and the adequacy with which it was sampled (Arrington, 1943). Each of these factors contributes to understanding the type and quality of the data derived and thus the limitations of the data in making generalizations. 68 Inter-rater reliability measures offer some basis on which to judge the accuracy of the recording process. Assessments of the content of behavioral events can also contribute to understanding the nature of the behavioral units, and thus the degree to which inference and observer bias may mediate the accuracy and objectivity of the observational data. The other two factors, however, refer to the representativeness of the observed behavior. In regard to the issue of the naturalness of behavior, the small group setting provides a sample of behavior that is considered representative of spontaneous social behavior. This issue was explored in an earlier discussion of the advantages and disadvantages of the small group setting versus the natural classroom. In the larger study, from which these data were derived, children's social behavior scores measured in the small group correlated moderately (r=.208) with social behavior scores derived from classroom obser- vations (Boger & Andrews, 1975). Thus the behavior observed in' the controlled setting would appear to be similar but not identical to naturally occurring behavior in the classroom. Results of this study, therefore, must be interpreted in light of the observational setting. The second issue of representativeness refers to the adequacy of the sample of behavior. One method of assessing adequacy is through tests of concurrent validity with other instruments (Gellert, 1955) or with the same instrument under differing observational time periods. Another approach is through analysis of the internal 69 consistency of the behaviors observed across time intervals within one observational period. Such inter-item analysis across time intervals was implemented with the data derived from this study. With tests of internal consistency, both the adequacy of the sample of behaviors and the intrinsic stability of the occur- rence of behavioral events influence the results. Therefore, only those behavioral categories that required a rating during each interval were analyzed, using an analysis of variance technique (Kerlinger, 1963). The reliability coefficient is derived from the following formula: r = V (true) = Vind, -Verror = 1 _ verror vind. Vind. ind. A B C The total variance in the analysis of variance model is divided into three terms: variance associated with items, individuals, and resid- ual. The variance term associated with individual differences less error of measurement or residual derived from the analysis of variance test is used in the reliability formula as a substitute for true variance (see B above). Thus, the reliability coefficient reflects the degree of stability expected in responses across observational intervals after errors of measurement are deleted. The results of these analyses for the observational data derived from this study are reported in Table 2. 70 Table 2 Results of Tests for Internal Consistency of Behavioral Categories: Revised Observation of Socialization Behavior Instrument Standard Error of Behavioral Category Reliability Coefficient Measurement Level of Response .80 .38 Level of Initiation .84 .39 Level of Impact on Peer A .85 .49 Level of Impact on Peer B .86 .48 Level of Impact on Peer C .86 .47 Social Behavior .92 .33 Physical Behavioral Tone .87 .17 In a study using a similar time-sampling observational procedure, Smith and Connolly (1972) reported that only behavioral categories displaying a consistency of .50 or better were included in data analyses. All of the categories of the OSB exhibited a consistency well above that level, suggesting that in fact the behavior displayed in the small group was rather constant throughout the length of the play session. Operationalization of Variables The task in this section is to describe the Operationaliza- tion of the variables used in subsequent analyses. Three lines of research inquiry were combined in this study. Two of the foci were concerned with a measure of the child's commu- nicative competence. The relationships between this variable and 71 the child's demographic group membership and background character- istics were explored. In addition, as this variable reflects a feature of the child's overt behavior in a small play group, the relationship between communicative competence and other social inter- action variables were determined. The third line of inquiry moved away from the individual child and focused on the interactions that occur when four children meet in a specific setting. The focus here was on determining dif- ferential incidences of communication elicited through various modes of communicative exchange. The variables used in each of these lines of inquiry are described in the following section. But first the operationalization of communication is discussed, since it was the primary focus of interest in all analyses. Communication "Interaction is made up of single actions manifested by individuals in propinquity or contact to one another" (Chapple, 1940). By analyzing the sequence, content, and consequence of these single actions, the information flow among individuals can be examined. Communication can be viewed as behavior that elicits a response in a potential receiver. In this study, any behavior was considered communication if it influenced a receiver who attended to the initiator and/or reacted in some manner as a result of the original behavior. Operationalized, this means if an impact code of acceptance or rejection was recorded for any one of the three peers during an interval, communication had occurred. 72 Operationalization of Demographic Characteristics and Background Variables The primary independent variables in this study are described below. The information from which these variables were derived was provided by the children's parents on a general information sheet. The relationships between these variables and children's communica- tive competence were determined. Sex,--Both male and female children were included in the sample. Sex differences in communicative competence were explored. Socioeconomic Group Membership.--The socioeconomic status of each child was determined by using the McQuire and White Socio- economic Status Index (1955). This procedure was described earlier under sampling procedures. Children were categorized as middle or low SES. A continuous score value was also available from the index used in regression and correlation analyses. The higher the score, the lower the child's status. Agg,--Children's age was available in the form of months. A standardized reference point of January 1, 1974, was used in determining the child's age in months. Ethnicity.--Although a variety of ethnic groups were repre- sented in the sample of children, the only distinction made for this study was between anglo and black. A very small minority of children was Indian and Chicano. Since the number of children in each of 73 these groups was insufficient to be included as a design factor, these children were eliminated from the sample. The term "anglo" was ascribed to any child whose parents were both caucasian and of European heritage. The term "black“ was ascribed when either or both parents were Negro. Number of Children in the Family.--On the general informa— tion sheet, parents noted the number of children in the family and their ages. This variable was a summary of the number of children in the family who were less than 18 years of age. If the child was an only child, the number of children in the family was one. Number of Months since Enterinngenter.--Two experience variables were available in this study. The first variable repre- sented the number of months the child had been enrolled in the particular day care center included in the study. A reference point of September 1, 1973, was used to standardize this figure. Thus the month in which the child entered was subtracted from the September date to determine this figure. This variable also repre- sented the potential degree of familiarity with the children present in the play setting. Number of Months in Group Care.--The second experience vari- able represented the number of months the child had been in any form of group care previous to and including the present day care attendance. Part-week and part-day attendance were included on an equal par with full-time attendance. 74 Operationalization of Social Interaction Variables The following variables were derived from the observation in the naturalistic small group play setting. Each variable corre- sponded to a characteristic style of behavior of a child. Communicative Competence.--Communicative competence was the primary dependent variable in this study and was derived from the impact codes of the OSB. The variable is a measure of children's relative ability to influence their environment. Its conceptual formula is: the average over all peers and all intervals of the occurrence of an impact (acceptance or rejection) times its level of impact. The potential range of this variable is zero to nine. Higher scores represent more intense and diverse influence over others. Communicative competence is an attribute of individual systems. It reflects "openness" to the extent that it represents the system's ability to influence the social environment. Responsivity.—-The relationship of responsivity to communi- cative competence was also determined. Responsivity also reflects "openness" in the system as it measures the relative degree to which the child responds to others in the environment. Its con- ceptual formula is the logit4 of the proportion of intervals with response versus the proportion of intervals with ongoing activity. Responsivity was derived from the involvement codes. 4To stabilize proportional data for use in parametric analytic models, the logit of the proportion is formed and this new figure is used in the analyses. 75 Activity Level.--The average intensity with which the child interacts is measured by activity level. It is the combined average intensity of initiations and responses. The relationship between this variable and communicative competence was determined. Initiative.--The initiative variable represents the degree to which the child purposefully places himself in interaction with the physical or social environment. In contrast to communicative competence, initiative measures the relative frequency with which the child initiates interaction, regardless of whether or not that behavior elicits a response from others. Its conceptual formula is the logit of the proportion of intervals with initiations versus the proportion of intervals without initiations. Social Behavior.--The last variable reflecting the child's characteristic behavior in social interaction was social behavior. This variable is the average rating over all intervals of the child's level of play involvement. This scale includes six levels of play, from unoccupied to c00perative play. The average score denotes the relative social maturity or sociability of the child. Its relationship to communicative competence was also determined. Operationalization of Modes of Communication Much of young children's play is rather random--explorations of the body, materials, and enviornment with or without accompany- ing intent to communicate with others. Thus, an objective of this 76 study was to determine what characteristics of children's play behavior in groups communicates more than others. Thus, the modes of communication within the small group were examined. Whenever an activity of the subject could be classified as communication, i.e., when some impact was noted, the characteristics of the interaction were examined. The various codes used in this analysis were: verbal, physical behavioral tone, object of initia- tions, object of responses, and contact. The following character- istics of the information flow were used to determine which aspects of children's interactions elicited greater communication. Channel of Exchange,--Each bit of interaction was cate- gorized as verbal or nonverbal based on the occurrence or non- occurrence of a verbalization. The relative differences in inci- dences of communication through these two channels were determined. Affective Connotation.-—Each bit of interaction was rated according to the affective connotation of the behavioral delivery. Whether a verbalization or a physical behavior elicited the communi- cation, the nonverbal physical behavioral tone of the interaction was noted. This affective code had three levels: positive, neutral, and negative. The degree to which children responded differentially in intervals with negative versus neutral or positive behavior was determined. Additionally, the object of either the initiation or response was noted to determine to whom the behavior was directed and whether or not physical contact occurred. 77 Directionality of Communication.--Whenever an interaction was classified as conmunication, the initiation code was examined to determine to whom the child directed his initiation. The possi- ble codes were: individual peer, pair of peers, the entire group, the materials, or the environment. These objects of the initiation were examined to determine if different incidences of communication occurred when initiations were directed to specific individuals versus when they were directed to the more generalized group, or the nonhuman setting of materials or environment. Data Reduction and Analysis Data Reduction Procedures All observational ratings and demographic information were encoded to numerical codes and recorded on computer coding forms. This coding was then quality checked by the investigator before punching the codes on computer cards. AS a second form of quality checking, the computer cards were subsequently verified mechanically. Two frequency count programs were created to direct the computer to perform the initial mathematical computations needed to form the individual variables. One of these programs, written in Fortran, focused on the entire 30-interval sequence for each subject. Through it the frequencies and average ratings necessary to compute the variables related to dimensions of the child's behavior were calculated. For those variables that were average ratings, the variable was formed during the first transformation; i.e., individual ratings 78 for each interval were summed across intervals and divided by the number of intervals. Variables in the form of proportions, however, required two or more transformations. This process consisted of: l. summing across intervals to determine the frequency of the occurrence of an event 2. forming a proportion, the numerator being the fre- quency of the occurrence of an event and the denomin— ator being either the frequency of nonoccurrence of that same event or the frequency of the occurrence of another event 3. calculating the logit of the proportion The second program focused on each interval separately, computing frequencies and cross-tabulations of the occurrence of two or more simultaneous events. It was with this SPSS program that each interval was checked to determine the incidence of communication and the conditions surrounding the occurrence of communication. The variables formed during these computations and transformations were subsequently punched onto a new data deck and also stored on magnetic tape. The computer program used in data analyses were those available through the 5.8 version of the Statistical Package for Social Sciences (Wie, Bent, & Hyll, 1970) and the adapted version of Finn's Multivariance program (Schmidt & Schiefley. 1972). All analyses were implemented on the Control Data Corporation 6500 com- puter at the Michigan State University Computer Laboratory. Both descriptive and inferential statistics were used in the analyses of these data. The descriptive statistics were pri- marily percentages and probabilities of the occurrence of behavioral 79 events. The inferential statistics included both parametric and nonparametric techniques used in hypothesis testing. All statistid cal tests were two-tailed. An alpha level of .05 was used in 22,. determining the probability of a type one error; i.e., the null hypothesis was rejected when in fact it was true. Analyses Concerning Attributes of the Child The two research questions concerning the attributes of the children and asked of the data in this study were: (1) What are the relationships between children's background characteristics and communicative competence scores derived from the small group play observation? and (2) What are the relationships between character- istics of children's social interaction behavior and communicative competence? The analysis strategies needed to test the hypotheses of interest relative to the above research questions concerned both associations and differences between groups. Thus, correlation, multiple regression, and analysis of variance models were required. However, either parametric or nonparametric statistical models could be chosen. The underlying assumptions of the parametric models were scrutinized first, as they offered more versatility in investi- gating complex relationships and would, therefore, be favored if the data were appropriate to satisfy their assumptions. Two basic assumptions of most parametric tests are: (1) normality--the samples have been drawn from populations that are 80 normally distributed, and (2) measures are continuous with equa1~ appearing intervals so that mathematical computations such as addi- tion, subtraction, and division can be implemented. In response to the second assumption, all of the dependent variables in this study could be considered continuous and of an interval scale. The social interaction variables were either means or logs of proportions, both based on intervals derived by a divisor. Many of the independent variables were also continuous and interval, although their metric varied. In some cases, the metric was months; in others, it was number of people, or in the case of SES value, a specially derived index. Thus, all of these variables had equal- appearing interval metrics and satisfied the measurement criteria of parametric tests. In regard to the assumption of normality, the primary depend- ent variable, communicative competence, approached normality but was slightly positively skewed in this sample. Characteristics of the Sampling Distribution of the Communi- cative Competence Variable.--The following is a description of the sampling distribution of the communicative competence variable derived from the 139 three- and four-year-old children included in the sample. Central Tendency_ Variability Symmetry Mean--l.747 Variance--.984 Skewness--.588 Median--l.625 Standard Deviation--.992 Kurtosis-- -.l4 Mode--l.367 Range--4.467 Minimum--.100 Maximum--4.567 81 The distribution of the variable as observed in these data approached a normal distribution, but was slightly flatter and more skewed than the normal curve. The conmunicative competence scores had a restricted range. Greater numbers of children exhibited low scores than would be expected with a normal distribution. This may be a result of the time-sampling framework, wherein communication is measured whethercn'not eliciting behavior occurs. If an event- sampling procedure were implemented with the same operationalization of communication, perhaps higher scores would be derived. However, in spite of the limited range, the observed distribution was more nearly normal than any other shape. With a normal distribution, the mean and median would be the same. A slight difference between the mean (1.747) and the median (1.625) in the observed distribution indicated a slight asymmetry. This distribution was slightly positively skewed, having more values to the right of the center of the curve than extending to the left. This is reflected in the measure of skewness (.588). A perfectly symmetric normal curve would have a skewness measure of zero. Kurtosis is a measure of the peakedness of the curve. In this sample, the kurtosis was -.l4. Thus, this curve was flatter than the normal curve, which has a kurtosis of three. A Kolmogorov-Smirnov Goodness of Fit test was implemented to test the degree of similarity between this distribution and the theoretical normal distribution. The results of this test are reported in Table 3. 82 Table 3 Results of Kolmogorov-Smirnov Test of Goodness of Fit on the Distribution of the Communicative Competence Variable N=139 Max. + Difference Max. - Difference K-S Statistic Probability .078 -.042 .913 .375 The null hypothesis is rejected if the K-S statistic is too large. In this case, the statistic was small and the decision was to accept the null hypothesis that this distribution came from the normal dis- tribution. Therefore, it can be assumed that the communicative compe- tence scores were noramlly distributed. An additional safeguard, however, is that most parametric tests are robust in regard to this assumption of normality. Thus, it is safe to assume that these data are apprOpriate for parametric tests. However, each statistical model will be reviewedindividually. Such a discussion follows: Measures of Association.-- Correlation: The relationship between two variables can be determined by computing a correlation coefficient. Such a coeffi- cient represents the degree to which two variables vary together. When both variables are measured on a linear interval scale, the Pearson product-moment correlation coefficient can be computed. The conceptual formula for this correlation coefficient is rxy = §§%—u X Y 83 The correlation coefficient is the quotient when the covariance of x and y is divided by the square root of each individual variance (Glass and Stanley, 1970). Correlation coefficients can range in value from -1 to +1. The sign indicates the direction of the rela- tionship, either inverse (-) or direct (+), and the value indicates the strength of the relationship, from no relationship (0) to a perfect relationship (1). The two assumptions underlying the use and interpretation of this statistical tool are: l. The predicted relationship is linear. 2. The measurement scales are of equal-appearing intervals. Both of these assumptions can be satisfied based on the nature of the variables and an understanding of the pr0posed hypotheses in relation to their foundations in past research. However, a third factor influencing the interpretation of the correlation coeffi- cient is awareness of the possibility that a third variable related to the two under investigation is responsible for the observed rela- tionship. To explore these potential spurious relationships, a variety of partial correlations will be computed when appropriate. Regression: An extended use of the coefficient of corre- 1ation to answer research questions posed in the predictive mode is regression analysis. The regression line or equation is used to predict one variable based upon its relationship to another variable. A test of the accuracy of the prediction can also be made, giving the regression analysis an additional advantage over correlations 84 (Armore, 1966). The primary task in regression analysis is to derive a regression line that minimizes the deviations between observed and predicted values of the dependent variable for each specified value of the independent variable. A least square method is implemented. The basic equation for the regression line is: Ye = a + bX The expected value of Y (dependent variable) is the sum of a (the Y intercept) plus b times X (the beta weight or constant represent- ing the slope of the line times the observed value of the independ- ent variable). This equation can be extended to include multiple independent or dependent variables. In a multiple regression analysis the task is to produce a linear combination of independent variables that will correlate as highly as possible with the dependent variable. The prediction equation in multiple regression is: Ye = b1X1 + b2X2 + b3X3 . . . + a + R The beta coefficients are chosen to make R (residual or error) as small as possible. A variation of multiple regression is stepwise regression, a procedure that provides a means of choosing independent variables in an order that gives the best prediction possible with the smallest number of independent variables. The factor that can determine whether or not an independent variable is included in the F statistic 85 associated with the significance of beta, and, in some programs, the degree to which a new independent variable is a linear combination of existing variables. A limitation of this procedure, however, is that the results of the regression analyses are dependent on the specific ordering of the independent variables. Other orderings may produce slightly different results. In all of these procedures, the statistic that indicates how 2 closely the two or more variables are associated is R , the coeffi- cient of determination. Its conceptual formula is: R2 = SS linear regression SS total Thus, R2 is the pr0portion of the total variation in Y associated with X. This term is often converted into the percentage of the variation in Y associated with or explained by knowledge of X. Measures of the accuracy of the prediction are the standard error of estimate and the F statistic. In the former caSe, the standard error represents the extent to which, on the average, the observed values of Y are dispersed around the line of regression. The second measure refers to testing the null hypothesis with the proportion of the variation accounted for by the regression equation (F = MS regression) MS residual ’ this hypothesis, it can be accepted that the specified equation is When the F statistic is large enough to reject better than a chance predictor of Y given X. 86 The assumptions of the simple and extended regression analy- ses are: l. Normality. 2. Homoscedasticity--the variation in Y is constant for all changes in the value of X. 3. Linearity--the relationship between the independent and dependent variables is linear. These assumptions can be assumed to be satisfied because there is no evidence to suspect the contrary. Since no such procedures are available in nonparametric tests,the stepwise regression procedure was implemented in this study to determine which of the independent variables were the strongest predictors of communicative competence. As an extension of correlations and partial correlations, the regression analysis provided a means to differentiate among inde- pendent variables. ae Measures of Group Differences.--The analysis of variance model (ANOVA) is the most versatile in exploring group differences. Such a model allows for tests of interactions as well as main effects in a variety of factorial designs with extensions for multiple dependent variables, analysis of covariance, and repeated measures models. Therefore, if the assumptions of this model can be satis; fied, it is the best strategy to implement. The assumptions of the analysis of variance model are: 1. Measures are continuous with equal-appearing intervals. 2. Normality--the samples have been drawn from populations that are normally distributed. 87 3. Homogeneity of variance-~the variances within groups are statistically alike. 4. Independence--observations are independent of one another both within and between groups. The first two assumptions can be considered satisfied, as discussed earlier. In the case of the third assumption, Levine's test of equality of dispersion was implemented to test for the equality of variances across groups. With this test, the absolute amount of variance in each observation relative to the group mean is entered as the score value into the analysis of variance computa- tions. The nonrejection of the null hypothesis of no differences between groups was sought. The results of these tests are summarized in Table 4. Table 4 Results of Levine's Test of Equality of Dispersion . . Level of . . Factor F-Stat1st1c Probability Dec151on Sex: male, female 2.5420 .113 Do not reject SES: low, middle 1.0263 .313 Do not reject Ethnicity: ' black, anglo 6.0077 .016 Reject Note: Dependent Variable—-communicative competence Independent Variables--sex, SES, ethnicity 88 These results indicated that the variance in the two levels of the independent factors within sex and SES were similar. However, with ethnicity, the variance across levels was not similar. There- fore, the assumption of equal variance was not satisfied for ethnicity but was satisfied for sex and SES. The consequence of violating this assumption when cell Sizes are unequal is that it becomes more difficult to find significant differences between groups. The ANOVA is robust in regard to this assumption with equal cell sizes. However, since the cell sizes in this study were unequal, a nonparametric analysis of variance test, the Kruskal-Wallis test, was used to test for ethnic group differ- ences. A fourth assumption of independence of observations can be satisfied, based on the fact that children were randomly selected to participate in the play groups. Subjects were not matched or paired in any systematic manner. Although the children were ran— domly selected based on their demographic group identity, this procedure did not violate the independent sampling criteria of this assumption. A factor that is of concern relative to this assumption is that the data were collected when children of mixed group identity (sex and SES) were interacting together. Methodologically, the sampling procedures satisfied the assumption of independence but the data collection procedures presented some ambiguity. A primary assumption of this and other small group social- interaction methodologies is that the observed behavior of an 89 individual is independent of others' behavior and representative of the individual's behavior as it would be exhibited in any group setting. Underlying the statistical assumption of independence is the knowledge that the unit of analysis is the unit that was randomly sampled. In this study, that unit was the individual. If this assumption of independence is violated and a different unit of analysis is used or deemed more appropriate, the lack of independ- ence may effect an increase in the within-group variance. If this occurred, the F statistic would be reduced and it would become more difficult to find significant group differences. In this study, this assumption of independence was assumed to be satisfied and the analysis of variance model was implemented whenever its other assumptions were satisfied. Thus, to determine sex and SES group differences, ANOVA's were applied. To test for ethnic group differences, the nonparametric Kruskal-Wallis Test was implemented. Analysis of Variance: The simplest form of the analysis of variance test is a one-way ANOVA with two levels of the independent variable. Two sets of scores exist, one for each level of the independent variable. The analysis of variance model can be repre- sented as: X = u + a + e The observed score is a linear combination of the population mean + group differences + error. The question asked in implementing this 90 model is: Is the variance attributed to group membership (a) greater than fihe error or within-group variance (8)? If this proportion (F = figs) is large enough to be significant, the question is answered in the affirmative. The two populations can thus be considered different. However, if the within-group variance is large enough to cause the proportion to be smaller than can be expected by chance, the two sets of scores must be considered as representing only one population and thus no group differences exist. Extensions of this model can be implemented to note the interaction and main effects of more than one independent variable on one or more dependent variables. Kruskal-Wallis One-Way Analysis of Variance: The nonpara- metric ANOVA uses data in the form of ranks. All observations are ranked, and the sums of the ranks for each group are tested using a chi square statistic. When this statistic is so large as to be unlikely to occur by chance, the null hypothesis of no group dif- ferences is rejected. A correction for tied ranks is also avail- able. Anaiyses Concerning Attributes of thE'Communication The research question related to the communication that occurs when groups of four children play together was: What is the effect of the mode of communicative exchange on the incidence of communication? In testing the hypotheses related to this question, fre- quency data were used, i.e., the frequency of the occurrence of 91 communication. Each interval for each child was the unit of analy- sis. This amounted to 4,830 intervals in which descriptive infor- mation concerning the play behavior of children was observed and encoded. To prepare for the analyses of the frequency of the occur- rence of communication, each interval was checked to determine whether or not communication had occurred. If an impact code was recorded for any one of the peers, communication was considered to have occurred. Contingency tables were formed with one variable representing the occurrence or nonoccurrence of communication and the other variable representing another behavior elicited during the interval. For example, a table may consist of the frequency of communication by the frequency of verbalizations (see Figure 10). Verbalizations Communication Yes No Yes No Figure 10. Example of a contingency table. Analyses of Crossbreaks.--Once the crossbreaks are deter-' mined, a nonparametric chi square test can be implemented to determine whether the frequencies observed in the sample deviate significantly from some theoretical or expected frequency distribu- tion. The expected distribution in this case is the chance distribu- tion. 92 The conditions under which the chi square test of indepen- dence can be applied are (Isaac and Michael, 1971); 1. Data must be in the form of frequencies. 2. Individual events or measures are independent of other events. 3. Each event category must have a theoretical fre- quency of 3_5. 4. There must be a logical or empirical basis for the way the data are categorized. 5. Each observation occurs in one and only one cell of the contingency table. All of these conditions could be satisfied with the frequency data from this study. In regard to the independence condition, the behavioral categories of the OSB were mutually exclusive. Thus the occurrence of an event in one category was not contingent upon the occurrence of an event in another category. Each was a separate event although they all referred to the same behavioral phenomenon. If the null hypothesis is rejected when using a chi square test of independence, the behavioral events are not independent but vary in some systematic manner. A coefficient of contingency is also avail- able to represent the degree of association among variables. This value, although in the range of zero to one, is not as precise as the Pearson product moment correlation coefficient. When tables greater than 2 x 2 are formed, post hoc comparisons can also be implemented. CHAPTER IV RESULTS OF DATA ANALYSES The results of the data analyses are reported in three sections of this chapter, based on the three areas of inquiry identified in the study. Each research hypothesis is stated and discussed relative to its corresponding statistical test. In conclusion, a summary of the results is presented. Relationships Between Communicative Competence and Background Characteristics of Children The first three research hypotheses are discussed together, as they were simultaneously investigated using a multiple regression technique. As stated, they are: H]: The number of children in the family is a predictor of communicative competence. 2: The length of time in group care and the length of time enrolled at the present day care center are pre- dictors of communicative competence. H3: The age of the child is a predictor of communicative competence. These four variables plus the continuous variable, SES value, were entered as independent variables into a stepwise regression procedure to predict communicative competence. The results of this analysis are reported in Table 5. 93 94 Table 5 Results of Multiple Regression Analysis Predicting Communicative Competence From Selected Background Characteristics fir Multiple R = .3546 R2 = .1257 F-Statistic = 4.0259 P < .002 Degrees of Freedom: 5 and 134 Stepwise Regression Procedure .1 Age .004 .001 11.441 (.000) .074 Experience .001 .000 6.927 (.010) .043 Months Entered .001 .001 1.125 (.219) .007 Number of Children .003 .006 .139 (.710) .009 SES Value .000 .000 .339 (.626) .002 Age and experience were the only variables that were signifi— cant predictors of communicative competence. They accounted for 12% of the variance in communicative competence. Age contributed more to the prediction than experience in group care, but both were highly related to communicative competence. The correlation coefficients representing the relationship between these variables and communicative competence are reported in Table 6. The magnitude of these correlation coefficients and the results of the regression analysis suggest that there was no rela- tionship between communicative competence and the number of children in the family. Thus, H1 was not supported. 95 Table 6 Pearson Product-Moment Correlation Coefficients for Background Characteristics With Communicative Competence Background Variables Correlation Coefficient With Level of Communicative Competence Probability SES Value -.029 .370 Months Entered .184 .015 Number of Children in Family -.O75 .189 Experience .237 .005 Age .340 .001 The length of time in group care (experience) was positively related to communicative competence, as was months since the child entered the day care center. Both of these experience variables, however, were only slightly related to communicative competence as reflected in the magnitude of the correlations (r=.24 and .18, respectively). Only prior experience in group care was predictive of conlnunicative competence. Therefore, only part of H2 was sup- ported. H3 was definitely supported. Age contributed more to the regression equation than did any other variable and was moderately correlated with communicative competence (r=.34). To explore further the relationship between age and communi- cative competence, an analysis of variance test was implemented. The sample was divided into two groups: three- and four-year-olds. The results of this analysis are reported in Table 7. 96 Table 7 Results of a One-Way Analysis of Variance Test for Age Effects on Communicative Competence Scores Source of Sum of Degrees of Mean F- Level of Variation Squares Freedom . Squares Statistic Probability Between Groups 12.533 1 12.533 13.931 .0003 Within Groups 123.252 137 .899 TOTAL 135.785 The null hypothesis of no differences in communicative competence scores between three- and four-year-olds was rejected. Four-year-olds exhibited higher communicative competence scores than did three-year-olds. These means are presented in Table 8. Table 8 Means and Standard Deviations of Communicative Competence Scores Based on Age Group Mean Standard Deviation N Three-Year-Olds 1.198 .716 32 Four-Year-Olds 1.911 1.006 107 As both age and experience were found to be predictors of communicative competence, a series of partial correlations was computed to note the magnitude of the relationship of these 97 variables with communicative competence when the other was controlled. These results are reported in Table 9. Table 9 Partial Correlation Coefficients Variable Control Correlation With Level of. Var1able Commun1cat1ve Competence Probab111ty Age -- .340 .001 Age Experience .315 .001 Experience -- .237 .005 Experience Age .197 .021 Although age was more highly related to communicative competence than was experience (r=.34 and .24, respectively), the change in the magnitude of the relationship when the effects of the third variable were controlled remained about the same for each pair of variables. Thus, age and experience contributed about equally to the nature of each other's relationship with communicative competence. When all experience variableS--experience, months entered, and number of children in the family--were controlled, the relation- ship between age and communicative competence was r=.295 (p < .001). When age, months entered, and number of children in the family were controlled, the relationship between experience and communicative competence was r=.l74 (p < .043). Thus, the relationship between age and communicative competence was only slightly influenced by 98 simultaneous associations with experience factors. Experience, however, exhibited more variation when other factors were con- trolled. Hypotheses 4, 5, and 6 were investigated using tests of group differences. Since the independent variables, sex and SES, satisfied the assumptions of the parametric analysis of variance model, these two variables were investigated together using a two- way analysis ofvariance test. The third independent variable, ethnicity, did not satisfy these assumptions and, therefore, Hypothesis 6 was tested using a nonparametric analysis of variance model, the Kruskal-Wallis test. The results of testing Hypotheses 4 and 5 are discussed below. H4: Children from middle-SES homes have higher comnunica- tive competence scores than children from low-SES homes. ‘ HO: There are rm) differences in the communicative competence scores of low- and middle-SES peers. H5: Males exhibit higher communicative competence scores than females. HO: There are no differences in the communicative competence scores of males and females. As no interaction effects were noted, a clear test of main effects was possible. As noted in Table 10, the null hypothesis for sex effects was rejected but the null hypothesis for SES effects was not rejected. No differences between low- and middle-SES peers exist. Therefore research hypothesis 4 was not supported. The means and standard deviations for these two groups are presented in Table 11. 99 Table 10 Results of the Two-Way Analysis of Variance Test for Sex and SES Effects on Communicative Competence Scores Source of Sum of Degrees of Mean F- Level of Variation Squares Freedom Squares Statistic Probability Sex by SES Interaction .446 l .446 .446 .999 Sex Main Effect 7.063 1 7.063 7.433 .007 SES Main Effect .000 1 .000 .000 .999 Residual 128.268 135 .950 TOTAL 135.785 138 .984 Table 11 Means and Standard Deviations for Communicative Competence Scores Based on SES Group Membership Group Mean Standard Deviation N Low SES 1.739 1.027 72 Middle SES 1.755 .961 67 Sex differences, however, were noted. Thus, Research hypothesis 5 was supported. Males exhibited higher communicative competence scores than did females. llwemeans and standard devia- tions are presented in Table 12. 100 Table 12 Means and Standard Deviations for Communicative Competence Scores Based on Sex Group Mean Standard Deviation N Male 1.977 1.054 68 Female 1.526 .881 71 Research Hypothesis 6 was presented in the null form, as no differences were expected. Since the ethnicity variable did not satisfy the ANOVA assumption of equal variances, the nonparametric Kruskal-Wallis test was implemented to test for the effects of ethnicity. H : There are no differences in anglo and black children's communicative competence scores. The results of this test are reported in Table 13. Table 13 Kruskal-Wallis One-Way Analysis of Variance Test for Ethnic Group Differences in Communicative Competence Scores Group Sum of Ranks Mean Rank N Chi Square Probability Black 2.557.5 77.5 33 1.501 .221 Anglo 7171.5 67.67 106 The null hypothesis was not rejected. No ethnic group differences were observed; thus Research Hypothesis 6 was supported. 101 Although no significant differences were noted between the communicative competence scores of anglo and black children, the mean rank for black children was higher than the mean rank for anglo children. Thus the actual means and standard deviations were computed and are presented in Table 14. Black children's communi- cative competence scores were higher but also exhibited more varia- tion than those of anglo children. Table 14 Means and Standard Deviations of Communicative Competence Scores Based on Ethnicity Group Mean Standard Deviation N Black 2.016 1.259 33 Anglo 1.663 .883 106 AS significant differences in communicative competence scores were observed based on the child's age, the subsamples within the sex, SES, and ethnic groups were tested to see if they varied significantly in their age distribution. One-way analysis of vari- ance tests were implemented with age as the dependent variable. A summary of these results is reported in Table 15. AS noted in Table 15, none of these groups varied signifi- cantly in their age distributions. Thus, age could not be considered an influence in interpreting the results of the tests of Hypotheses 4, 5, and 6. 102 Table 15 Results of Analysis of Variance Tests for Age Differences in Sex, SES, and Ethnic Groups Independent Variable F-Statistic Probability Sex 2.05737 .1541 SES 1.1125 .2934 Ethnicity .6482 .4222 Similar tests were implemented with expereince as the depend- ent variable. Again, no differences were observed in the past expe- riences in group care of the children in the various sex, SES, and ethnic groups. Relationships Between Communicative Competence and Other Social Interaction Variables Both Pearson product-moment correlation coefficients and regression analyses were computed to test the research hypotheses concerning the interrelationships among the social interaction variables derived from the observation in the small group play session. Hypotheses 7, 8, and 9 are discussed together. As stated, they are: H7: Children's social behavior and communicative competence scores are positively related. H8: Children's initiative scores are more highly related to communicative competence than to responsivity scores. H9: Responsivity scores and communicative competence scores are positively related among children with similar activity patterns. 103 To explore the basic degree of association among these five variables, simple Pearson product-moment correlation coefficients were derived. They are presented in Table 16. Table 16 Pearson Product Moment Correlation Coefficients Representing the Interrelationships Among the Social Interaction Variables Derived From the OSB Communica- . Social . . . . . Activity t1ve . Respons1v1ty In1t1at1ve Competence Behav1or Level Communica- tive Competence 1.00 .76*** .27*** .52*** -.56*** Social Behavior 1.00 .15* .38*** -.61*** “959°“‘ 1.00 .74*** .12 s1v1ty Initiative 1.00 -.22** Activity Level 1.00 *p < .05 **p < .01 ***p < .001 In response to H7, children's communicative competence scores were highly correlated with average level of social behavior (r=.76). Children who were more socially oriented and interacted with peers at more involved levelsof play had a greater impact on their social environment through communication. 104 Observation of the simple correlation coefficients also suggested that H8 could be supported. Children's initiative scores were more highly related to communicative competence than were their responsivity scores (r=.52 and .27, respectively). However, when activity level or the average intensity of interaction was controlled, the partial correlation coefficients suggested a different interpretation (see Table 17). The actual frequency with which children responded was more highly related to communicative competence than the frequency with which they initia- ted. It appeared that children's responsivity and initiation patterns were highly associated with each other and with communi- cative competence scores, therefore suggesting that children who respond to peers tend also to initiate interaction with them and thus affect communication. Table 17 Partial Correlation Coefficients Holding Activity Level Constant N = 136 Initiative Responsivity Communicative Competence .49* .58* *p < .001 H9, referring to the relationship between responsivity and communicative competence scores, was supported. Children's 105 responsivity patterns were highly related to communicative compe- tence when the effects of activity level were removed (r=.58). A multiple regression analysis with a stepwise procedure was also implemented to determine the relative contribution of the various social interaction variables in predicting communicative competence. Four independent variables were entered into the regression equation: social behavior, responsivity, initiative, and activity level. The overall equation accounted for 65% of the variation in communicative competence scores. The results of this analysis are reported in Table 18. Table 18 Results of Multiple Regression Analysis Predicting Communicative Competence From Social Interaction Variables Multiple R = .808 R2 = .652 F-Statistic = 66.128 p<.ooo Degrees of Freedom: 4 and 135 Stepwise Regression Procedure Standard Error F-Statistic Additive Variable Beta of Beta (Probability) R2 Social Behavior .886 .102 197.515 (.000) .58 Initiative .202 .068 23.884 (.000) .06 Activity Level -.607 .271 5.509 (.020) .01 Responsivity .012 .059 .043 (.837) .00 106 The results of this regression analysis suggested that social behavior and initiative scores were most predictive of communicative competence, accounting for 64% of the variance. Activity level, although a significant predictor, only accounted for another 1% of the variation. Given the knowledge of the child‘s social behavior and initiation patterns, responsivity patterns did not contribute to the predictive equation for communicative competence. Thus, even though responsivity scores may be more highly related to communicative competence when activity level is held constant, knowledge of the child's initiation patterns is more important when predicting cornnunicative competence scores. This information lends further support to H8. Table 19 Results of Multiple Regression Analysis Predicting Communicative Competence From Both Background Characteristics and Social Interaction Variables R2 = .6601 F-Statistic = 29.34 P<.OOO Degrees of Freedom: 9 and 130 Multiple R = .8124 Stepwise Regression Procedure Standard Error F-Statistic Additive Var1able Beta of Beta (Probability) R2 Social Behavior .863 .107 197.515 (.000) .58 Initiative .192 .070 23.883 (.000) .06 Activity Level -.599 .281 5.508 (.020) .01 Experience .003 .003 1.612 (.206) .00 Age .004 .009 .174 (.677) .00 Responsivity .026 .061 .215 (.644) .00 Months Entered .001 .007 .073 (.787) .00 Number of Children .026 .038 .143 (.706) .00 SES Value -.004 .004 .984 (.323) .00 107 Knowledge of both a child's background variables and social interaction variables provided a regression equation that accounted for 66% of the variation in communicative competence scores. The social interaction variables contributed more to the equation than did the background information. Social behavior and initiative were the most highly predictive variables, accounting for 64% of the variance. Given information about the child's social interaction behavior, age and experience did not aid in predicting communicative competence. Differential Incidence of Communication In the following section, the discussion focuses on the analyses undertaken to determine the differential incidence of communication through various modes of exchange. Chi square tests of independence were implemented, with the unit of analysis being the interval. More communication is elicited in intervals with verbalizations than in intervals without verbaliza- tions. H10: The statistical hypothesis presented to test this relation- ship was: H0: The occurrence of communication is independent of the occurrence of a verbalization. This null hypothesis was rejected. The results of the chi square test are reported in Table 20. More communication occurred in the verbal mode, whereas less communication than expected by chance occurred in the nonverbal mode. Thus, Research Hypothesis 10 was supported. 108 Table 20 Chi Square Test of Relationship Between Communication and Verbalizations x2 Value = 889.45 df = 1 P < .000 Verbalizations Communication Nonverbal Verbal No Communication 1357 163 observed 895 625 expected Communication 1221 1639 observed 1683 1177 expected Contingency Coefficient = .41 Communication occurred through the verbal mode 57% of the time and through the nonverbal mode 43% of the time. However, as also noted in Table 20, a number ofintervals in which verbalizations occurred were not accompanied by communication (N=163). This fre- quency represented 9% of the intervals in which verbalizations occurred or 4% of the total intervals. More communication is elicited in intervals with negative interactions than in intervals with positive or neutral interactions. H1]: H0: The occurrence of communication is independ- ent of the type of behavioral affect observed. The null hypothesis associated with Hypothesis 11 was Icejected. Results of this test are reported in Table 21. More communication occurred in intervals with negative behavioral affect; less communication than would be expected by 109 Table 21 Chi Square Test of Relationship Between Communication and Behavioral Affect x2 df = 1 P < .000 Behavioral Tone Communication Negative Neutral or Positive No Communication 50 1470 observed 284 1236 expected Communication 769 2091 observed 535 2325 expected Contingency Coefficient = .28 chance occurred in intervals with neutral or positive affect. Thus Research Hypothesis 11 was supported. To investigate further the context of behaviors eliciting communication, the following probabilities were computed. These probabilities describe the relationship between behavioral affect and the object of the interaction when physical contact occurs and when it does not occur. Children responded more to negative behavioral affect when the object of the interaction was a peer than if it was the materials aind environment. This relationship held both with and without [Jhysical contact being established. However, when the object of the interaction was the materials or environment children responded Intare to negative behavior only when contact was established. 110 Table 22 Probabilities of Communication Based on Behavioral Affect, Object of Interaction, and Physical Contact Behavioral Affect Negative Neutral or Positive Iglgggclions Contact No Contact Contact No Contact Peer .98 .95 .93 .63 Materials or Environment .84 .38 .37 .39 On the other hand, when direct physical contact was made with a peer, the probability that the behavior would be communicated was very high, whether it was negative (.98) or neutral/positive (.93). H12: More communication is elicited in intervals with initiations directed to individual peers than in intervals with initiations directed to the group or the materials and environment. Ho: The occurrence of communication is independent of the nature of the object of the initiation. The null hypothesis associated with Hypothesis 12 was also rejected. These results are reported in Table 23. More communication occurred when the object of the initiation was an individual peer; less communication than would be expected by chance occurred when the object of the initiation was the generalized group or materials and environment. Thus Research Hypothesis 12 was supported. 111 Table 23 Chi Square Test of Relationship Between Communication and the Object of the Initiation x2 = 116.36 df = 2 P < .001 Object of Initiation Communication . Mater1als or . . Environment Group Ind1v1dual No Communication 36 31 20 observed 10 16 61 expected Communication 121 212 929 observed 147 227 888 expected In general, children's play behavior was communicated to at least one other peer 65% of the time. For three-year-olds, communi- cation occurred 54% of the time and for four-year-olds, 68% of the time. The probability that communication would occur increased if a verbalization occurred (.91), or if an initiation was intentionally directed at another peer (.98). Summary of the Results The results of the data analyses can be summarized as follows: 1. Relationships between communicative competence and back- ground characteristics of children. a. The number of children in the family was not related to children's communicative competence scores. 112 The child's previous experience with peers, as measured by total length of time in group care, was positively related to and a significant predictor of communicative competence. The length of time the child had been enrolled in the present day care center, a measure of familiarity as well as experience with peers, was only mildly related to and not predictive of a communicative competence score. The child's age in months was positively related to communicative competence. Four-year-old child- ren had higher communicative competence scores than did three—year-olds. There was no relationship between a child's social economic status and his communicative competence. There was no relationship between a child's eth- nicity and his communicative competence. Males exhibited higher communicative competence scores than did females. Relationships between communicative competence and other social interaction variables. a. A measure of the child's average level of social participation, social behavior, was the best pre- dictor of communicative competence. In comparison to responsivity scores, initiative scores were more highly related to and accounted for more of the variation in communicative compe- tence. When a measure of the intensity of interaction was held constant, responsivity scores were related strongly and in a positive direction to communicative competence. Relationships between communication and mode of exchange. a. Three- and four-year-old children's play behavior communicated 65% of the time. The three- and four-year-old children in this sample communicated with fellow peers through the verbal mode slightly more frequently (57%) than 113 through the nonverbal mode (43%). The proba- bility of communication when a verbalization occurred was .91. The probability of communi- cation occurring for a nonverbal behavior was .47. Behaviors with a negative physical affect communicated more than behaviors with a neutral or positive affect. This relationship held whether the object of the interaction was the physical or the social environment, and whether or not contact was made with that object. More communication occurred when children directed their initiations to indiviual peers than when initiations were directed to the generalized group or the materials and environment. CHAPTER V DISCUSSION This study was designed to investigate the relationships between three- and four-year-old children's background character- istics and social interaction behavior and the degree to which they communicate to peers in a small group play setting. A small group social-interaction methodology and time-sampling observational rating procedure was utilized to identify and measure communication. A discussion of the results of the data analyses is organ- ized around the three thrusts of these analyses: (1) the relation- ships between background characteristics and communicative compe- tence, (2) the relationships between selected social interaction variables and communicative competence, and (3) the incidence of communication. The Relationships Between Background Characteristics and Communicative Competence A measure of the child's communication skill was developed as the average intensity times frequency of behavioral impact on peers. This variable was called communicative competence. Child- ren's background characteristics were examined in relation to dif- ferential distributions of communicative competence scores. 114 115 The results of these analyses suggested that age, experience in group care, and sex had the most differential influence on com- municative competence. Older children and children with greater experience in group settings had higher communicative competence scores. Males also had higher scores than females. No significant differences were evident based on number of children in the family, months since child entered the day care center, socioeconomic group membership, or ethnicity. Most studies have reported age-related increments in the amount of peer communication observed (Garvey & Hogan, 1973; Honig, et al., 1970; Smith & Connolly, 1972). Mueller (1972), Smith and Connolly (1972), and Blurton Jones (1972) also reported significant sex differences. Males exhibit more verbalizations than females and interact in more socially involved play, therefore establishing greater opportunity for communication. Hence, the result of the present study confirm the results reported in similar studies of children's social interactions. Although age and experience factors are often difficult to separate, in this study age appeared to have a greater influence on communicative competence when experience was held constant than did experience when age was held constant. Smith and Connolly (1972) reported the opposite conclusion, using a social participation index as the dependent variable. The nature of the measure of children's experiences in these two studies may have contributed to these differing results. 116 The results of no relationship between communicative compe- tence and socioeconomic group membership,number of children in the family, and ethnicity are difficult to discuss in relation to prior research, as these relationships have not been specifically investi- gated in the literature. A large body of research exists relative to language differences among socioeconomic classes (Ervin-Tripp, 1966). However, when communication is defined as both verbal and nonverbal behavior, as it is in this study, communication advantages that middle-SES children may have in the verbal mode may be compen- sated for by low-SES children's potential superiority in the non- verbal mode (Miller & Swanson, 1960). Thus SES differences would not be evidenced. Therefore, the finding of no SES group differ- ences was not surprising. However, the socialization literature would seem to suggest that the number of children in the family would be related to child- ren's skills in interpersonal relationships (Clausen, 1968). Thus, it was expected that the number of siblings in the family would be related to communicative competence. This was not the case. One problem could be that this sample was largely composed of children who were only children or the oldest child in the family. Less than one-sixth of the sample of children had older Siblings. Thus, a restricted range of values existed on this variable. When the sample was divided by only child vs. child with siblings, still no significant differences were noted in children's communicative competence scores. 117 Another factor perhaps confounding this relationship was the fact that all of these children were presently in day care settings and already had widespread contact with peers. Thus, even the child- ren without siblings had numerous opportunities to gain skill in social interaction through their peer contacts. Based on the fact that black children have been reported to have higher activity levels and social behavior scores (Boger & Cunningham, 1972), it was expected that black children would also exhibit higher communicative competence scores. However, in this study, although black children's communicative competence scores were higher than anglo children's, these differences were not sig- nificant. The relatively small number of black children included in this sample (N=33) compared to the total sample size (N=139), and the unequal cell variances observed between black and anglo children both contributed to decreasing the chances of noting significant differences. When the more powerful parametric ANOVA was implemented to test for ethnic group differences, the F-statistic approached significance at P < .07. Thus, a trend may be revealed in future investigations. Relationships Between Social Interaction Variables and Communicative Competence The focus of this aspect of the study had two purposes: (1) to determine what dimensions of children's social interaction behav- ior may be related to communication behavior, and (2) to explore the 118 utility of the observational procedure in measuring unique dimen- sions of behavior. If similar relationships exist between offering and receiv- ing, as would be expected from the literature on peer reinformcement and modeling (Mowrer, 1950; Hartup & Coates, 1967) children's fre- quency of response to others may be related to the degree to which they in turn receive responses from others. In this study, communi- cative competence was a measure of behavioral impact on the social environment, thus reflecting the degree to which others responded to the subject's behavior. However, either responses or initiations could elicit communication. As initiations were present in the data in only 31% of the intervals that elicited communication, and since communication occurred in 65% of the intervals, much of the communication that occurred resulted from peer observations of the subject's responses to stimuli elicited elsewhere or to ongoing activity. Thus, it would appear that children's responsivity pat- terns may be as highly related to communicative competence as to initiation patterns. AS a result of the various analyses, it can be said that initiation and responsivity patterns are in themselves highly related (r=.74) but are not Similarly related to communicative competence. Although when the intensity of the information flow is controlled, responsivity is more highly related to communicative competence than to initiative scores, this relationship merely confirms the fact that a large number of responses elicited communi- cation. 119 Initiation patterns are definitely more salient than response patterns in producing communication. The more frequently and more intensely one initiates interaction with either the social or physical environment, the more likely one is to impact on others in proximity and thus complete an act of communication. Further, specific initiations to peers are most likely to be communicated. The other finding of interest in these analyses is that one's level of social behavior is more predictive of and more strongly related to communicative competence than either responsivity or initiative scores. Parten (1932), developed this ordinal scale to represent a measure of children's social involvement in a play setting. It has weathered time and methodological advances in social interaction research and still provides the most precise measure of the quality of children's social involvement. As would be expected, the more proximal and more intimately involved peers are, the more likely it is that communication will transpire. A review of the frequency data confirmed this relation- ship. When children were engaged in unoccupied, solitary, or onlooker behavior, the probability that communication would occur was .24. When children were engaged in parallel play, the proba- bility was .37. This probability of communication increased con- siderably when children were interacting at the associative or cooperative levels of play. At these more involved levels, the probability of communication was .95. However, an interesting contribution of this study is the fact that children are extremely aware of each other's behavior. 120 Even when children were in the socially inactive levels of play, 24% of their behaviors were communicated! Description of the Incidence of Communication A task in this investigation was to describe the frequency with which children's play behavior communicated, whether or not the intent to communicate was present. Hence, a time-sampling framework was chosen. The resulting measures of frequency of com- munication represented a cross-sectional view of children's behavior in a small group setting. Considering this methodology, these fre- quency data may be conservative compared to other studies that use event-sampling procedures and hence specifically observe interaction or intentional exchange. In this study, communication occurred in 65% of all inter- vals. This figure was comparable to the 66% of the time that 3 1/2 to 5-year-old dyads were reported in ''mutual engagements" by Garvey and Hogan (1973). Communication occurred more frequently with age. Whereas communication occurred in 54% of all intervals for three-year-olds, it occurred 68% of the time for four-year-olds. These age trends were Similar to those reported by Honig et al. (1970) and Garvey and Hogan (1973). Of the time-sampled behaviors that communicated, 57% were accompanied by a verbalization. In 43% of the intervals communi— cation occurred through the nonverbal mode. No other studies have similarly noted the incidence of comnunication among children through 121 these two modes of exchange. However, Honig et a1. (1970) reported that more behavior was directed through the verbal mode with three- and four-year-olds than with two-year-olds. In this study, three- and four-year-old children communicated about equally through the verbal mode and only slightly more through the verbal mode than through the nonverbal mode. Although Mueller (1972) only investigated verbal communi- cation, he observed that 85% of all utterances were communicated. In the present study, 91% of all sampled behaviors with verbaliza- tions were communicated. Although the probability was much greater that a verbal behavior would communicate, much nonverbal behavior was also communicated. The probability of communication for a non- verbal behavior was .47. The affective nature of the observed behavior also affected the probability of communication. When a negative affect was expressed, the probability that the behavior would communicate was .94. In contrast, when neutral or positive affect was expressed, the probability of communication was .59, only slightly greater than chance. When the behavior was directed to a peer and contact was established, the probability of communication was high for both negative (.98) or neutral/positive behavior (.93). However, when no contact was established or if the object of the behavior was the materials or environment, behaviors associated with negative affect communicated significantly more frequently than did behaviors associated with neutral/positive affect. 122 Thus, these children were more sensitive to behaviors asso- ciated with negative affect. Although no empirical evidence is available to collaborate these findings, Piaget (1926) suggested that a contribution of peer interaction to cognitive development was the presentation of information dissonant from the child's experience. It would appear from these results that children are more attentive to dissonant events, i.e., events considered less socially appropriate. The fact that three- and four-year-old children are egocen- tric by nature may account for the finding that they are most responsive to dissonant events--events that strongly impact on their attention. However, an equally salient explanation may be that children at this age are generally less able to exercise con- trol over others' behavior. Young children may be limited in their interpersonal integrating skills (Goffman in Sutton-Smith, 1971). Thus aggressive behavior exhibited in the small group play session may be more threatening and therefore more noticeable to these children than such behaVior may be in other settings or for older children. This potential role of communication skills in both facilitating and coordinating peer social interaction has important implications for the socialization process and hence deserves further research attention. But, likewise, the generalizability of this finding of selective attention must be explored to see if it holds in other settings. Its generalizability to children's tele- vision viewing habits is a timely issue to explore. 123 Finally, a social skill developed with experience is the ability to attract the receiver's attention before a communicative exchange. The initiation codes of these observational data were analyzed to determine the differential distribution of communication based on the directionality of the initiations. The observers looked for attention-getting gestures or verbalizations as well as eye or body contact when determining the object of the initiation. Eighty-eightpercent of the initiations were directed to the social environment. As would be expected, generalizing from Mueller's (1972) results, more communication occurred when the initiations were purposefully directed to a Specific peer rather than the gen- eralized group or the materials and environment. Summary Three- and four-year-old children were found to be very sensitive to and aware of each other's behavior. Their behavior communicated frequently, although this depended on the attributes of the communicative exchange. As would be expected by nature of the precision with which verbal information can be transmitted, behavior accompanied by a verbalization communicated more often than did nonverbal behavior. Likewise, initiations directed to specific peers communicated more than did behaviors directed to the generalized physical and social environment. However, interestingly, much generalized behavior was also communicated. It would appear that children attend to a wide variety of behavior elicited by others in the social environment, whether or 124 not such behavior is directed at peers or even intended to be communi- cated. In fact, 39% of all behaviors directed to the physical envir- onment when no peer was interacting with the subject were in fact communicated to others in proximity. Thus, it would appear that children are very open to receive information from their environ- ment. Further, these findings suggest that children are more likely to attend to behaviors associated with negative affect than behaviors associated with neutral or positive affect. Children are, it seems, selective in their attending and responding patterns. They do in fact notice dissonant events more frequently than would be expected by chance. Children's communicative competence scores, reflecting the degree to which they impact on the social environment, are strongly related to the relative proportion of the time they initiate to others and the quality of their play involvement. More active, socially involved children make a stronger impact on others. Communicative competence scores are also related to child- ren's age, experience in group care, and sex. Communicative compe- tence increases with age and experience. Male children appear to have higher communicative competence scores than females. However, socioeconomic group membership, ethnicity, and the number of children in the family are not related to the incidence of differential communicative competence. CHAPTER VI CONCLUSIONS, IMPLICATIONS, AND SUGGESTIONS FOR FUTURE RESEARCH Conclusions The primary contributions of this study are: (l) the objectively derived descriptions of children's play group communi- cations, (2) the identification of the conditions associated with the occurrence of communication, and (3) the identification of factors associated with the relative magnitude of communicative competence scores. These descriptive and comparative data aid in understanding the information-transmission potential of peer con- tact. A secondary contribution of this study is an examination of alime-sampling methodology that was adapted to assess multiperson communication. Children's Play Group Communi- cative Competence l. The data from this study Show that three- and four- year-old children are very open to elicit information and to receive information from the social environment. Young children attend to a wide variety of stimuli emitted from the environment. Even when children are not socially involved with others, much of their behavior is being observed. Thus, young children appear to be extremely attentive and responsive to each other. 125 126 At age three and four, most of the children's initiations (88%) are directed to the social rather than the physical environ- ment. When these initiations are directed to specific peers rather than the generalized group, the probability of communication increases. Likewise, behaviors accompanied by verbalizations communi- cate more than do nonverbal behaviors. However, a sobering finding of this study is that behaviors associated with negative affect communicate more than behaviors associated with neutral or postive affect. Such negative affect would be attributed to disruptive or anti-social behaviors. These results naturally lead to the question of the potential effects of selective attention. How do these response patterns affect children's own behavioral repertoire? If children who produce negative behaviors are reinforced by peer attention, what effect does this have on their social development? These questions need to be answered before the impact of the present findings can be known. But at least the knowledge of such selective attention can alert educators and parents to be more aware of children's inter- actions during unstructured play. 2. In this study, children's communicative competence scores were found to be positively related to age and experience in group care. This finding supports the proposition that communi- cation is a learned skill. Communication can be viewed as requiring social experiences for expansion and successful application. Hence, children's peer interactions are important in developing the social skills necessary to increase the efficiency of communication. 127 Peer contact proVides a unique type of interaction and thus a unique opportunity fer developing social and communication skills. When interacting with adults, children's communications can be less precise and still convey meaning. But when children interact with like-skilled peers, their communication skills are challenged. They cannot rely on others to aid in making themselves and their needs known. And since they must communicate in order to influence the sequence of activity, children's peer interactions are important arenas for experimenting with a variety of ways to initiate and maintain social exchange. In the larger study from which these data were derived, children who received a short-term treatment condition, including systemic guidance for initiation and cooperation behavior, exhibited greater environmental control (a variable similar to communicative competence) than did children in the control condition (Boger 8 Andrews, 1975). Through Specific social experiences children's social communication skill improved. The trial-and-error learning and the irregular reinforcement that occurs in the natural environ- ment of peer interaction may be less efficient than more controlled peer interaction experiences within a classroom or curricular unit. Thus, experimental treatments may be more effective than mere expe- riences with peers in influencing children's communicative competence. Nevertheless, the trend of these results confirms the fact that communication as a social skill is learned and is influenced by social experiences. 128 3. Children's social interaction skills aid social exchange. Social interaction measures are most predictive of communication scores. Regardless of the kinds of experiences that lead to the development of basic social interaction skills, these Skills in themselves are critical in determining the success of children's interactions with peers. Three- and four-year-old children, being egocentric and socially naive, may therefore be most responsive to dissonant events --events that strongly impact on their attention. However, the underlying social skills associated with successful communication are perhaps of more importance and perhaps have a greater impact on children's participation in social groups with increasing age. The social skills associated with gaining the attention of peers, of facilitating common play objectives, and especially of transmitting information through the verbal mode may become increasingly critical to success in peer interaction as the child moves into later stages of development. At this age such behaviors are already recognized as being strongly related to communicative competence and the inci- dence of social exchange. Hence, educators should be sensitive to diagnosing and helping children expand these communication and social interaction skills as early as possible. Methodological Contributions 1. A strength as well as a weakness of this study concerns the use of the time-sampling observational methodology. Communica- tion is a process; it is made up of linear actions and reactions, 129 but the essence of communiction lies with the consequences of a multitude of these linear transactions. Therefore, observing con- tinuous behavior to detect the consequences of communication transactions would be most desirable. Furthering an understanding of how individuals change and are influenced by sharing information with others is the ultimate goal of communication/socialization research. The present study can be seen to contribute in some important ways to that goal, but as operationalized, it deals with influence at a much more basic level. The strength and a contribution of this study is the objec- tively observed representative description of children's multi- person communication behavior. The time-sampling framework provides a description of spontaneous behavior not limited by instances of intentional communication nor interactions that are strictly con- tained between and among individuals. The operationalization of communication with this time-sampling rating procedure produces a representative cross-section of behavior and then asks: How much of this behavior is communication? Thus, these data provide a norma- tive backdrop for future investigations that can ask more specific questions about the conditions and consequences of specific actions. 2. Additionally, in regard to the measure of children's skill in impacting on others--communicative competence--the time- sampling framework, if anything, could have neutralized the distribu- tion of this variable. Since communication was being assessed whether or not initiations occurred, peer responsivity patterns might have influenced communicative competence scores and thus reduced 130 their differential distribution. However, this did not appear to happen. Communicative competence was strongly related to the sub- ject's active participation in the play group, providing evidence that as a variable it did measure internal attributes of the child rather than dimensions of the responsivity of peers. Likewise, the ability of this variable to detect demographic group differences provides further evidence that its operationalization was sound. Communicative competence, although highly related to meas- ures of social involvement and elements of interaction, measures a dimension of the play group behavior that is greater than either of these more traditional measures. Communicative competence is a more precise measure of openness in the human system. Perhaps the most important contribution of this study is the operational definition of communicative competence: a means to assess how children impact on others. This variable can then be placed in a variety of event-sampling frameworks and continuous recording systems to note the sequential impact of children's behavior on others. Limitations of the Study The limitations of this study and suggestions for future improvements are listed in the following section. 1. Sample--A purposeful sample of three- and four-year- olds representing children from urban day care centers was observed in this study. Thus, although this sample included children with a variety of demographic characteristics, the population to which 131 these results can be generalized is restricted. As an initial inves— tigation using a specific methodology to assess communication, this sample provided adequate opportunity to test the methodological operationalization of variables and to provide normative descriptive and comparative data to investigate relevant research hypotheses. The reader is cautioned, however, to be aware of the sampling limi- tations of this study when generalizing these results. 2. Control over Independent Variables--The sample was primarily selected because of sex and SES distributions. Other independent variables identified for investigation in this study were, therefore, limited to the range of values and subsample size that already existed in the sample. The three factors whose rela- tionships to communicative competence may have been most adversely affected by these limitations were experience in group care, ethnicity, and number of children in the family. Further investi- gations using random samples or samples with more equal distributions across levels of these factors are recommended to provide a more precise test of these relationships. 3. Nonverbal Codes--The Observation of Socialization Behav- ior instrument includes only two scales to assess nonverbal behavior -—contact and behavioral tone. Although these two scales can cover all possible events, they provide little specific information to identify the behavior or even the context of the action. Other scales including action verbs such as runs, hits, and offers could be included to provide a means of determining patterns of responses to specific behaviors as well as to specific peers. 132 Blurton Jones (1972) suggested that children may develop patterns of interacting with others and then replace Specific behav- iors with other behaviors that hold the same symbolic meaning or produce the same results. For example, young two-year-olds who bite others in frustration may replace biting behavior with snatching or hitting behaviors at age three and later develop verbal patterns --all aimed at satisfying the same interpersonal objective. By observing these specific behavioral events and the context of their occurrence, such patterns may be identified. Such diagnostic facility would aid in behavioral therapy and increase the ability to predict later behavior from knowledge of present behavioral ten- dencies. 4. Reciprocity—-A next step in social communication research should be the observation of continuous behavior so that the sequence of a communication transaction can be analyzed. Rating scales similar to the OSB can be used, but the sampling framework and the unit of analysis would need to be redefined. An event sampling of interaction behavior of interest to the observer, accompanied by rapidly sequenced time-sampling intervals, is one alternative. Such a system would identify specific events and then continuously record Short intervals to note the sequence of behavior that occurs. 5. Small Group Setting--The relationships derived from this study must be generalized based on the fact that the data were collected while children were playing in a novel, open field setting. Whether these same relationships would be detected with data derived from natural classroom settings or other small group settings is 133 difficult to project. It would appear that the nature of this set- ting in terms of children's prior experiences may have a Specific effect on children's behavior and perhaps even a differential psy- chological impact on children. These potential effects of the setting on children's behavior need to be explored more fully and will aid in moving toward more ecologically oriented inquiry. For the present study, the results must be generalized in light of the nature of the composition of the small group and the fact that the setting was novel. 6. Assumption of Independence--Although the possibility of violating the assumption of independence of observations did not jeopardize the statistical results of this study, the question remains in terms of the validity of the results. AS discussed earlier, the assumption that these measures were unaffected by the specific peers who were in interaction needs to be investigated. The children in the small group were the environment in this situa- tion, and the effect of this Specific environment on children's behavior needs to be explored before these observational measures can be considered representative of the child's behavior in general. Presently, there is no reason to suspect that these behaviors are not representative. In fact, earlier investigations using the same methodology have noted trends in behaviors that could not have been noted if behaviors varied greatly during individual observations. Thus, these results can be accepted with confidence. Yet the development of empirical estimates of the potential variability in measures resulting from environmental influenceS'hsalso imperative. 134 7. Costs—Benefits--The observational methodology imple- mented in this study is extremely costly in terms of human and material resources. To observe, rate, and encode ten minutes of observational data for each child requires approximately one hour. When considering the time consumed in setting up and taping the play sessions and subsequently in processing the data, a large amount of human time is involved. Equally costly are the video- tape supplies and equipment and the computer programming services required to analyze observational data.‘ Hence the costs are great. The benefits of this methodology although difficult to quantify, are also substantial. Briefly, such a methodology provides a natural, unobtrusive assessment of children's behavior. The objectivity and validity of such data are, therefore, more credible than interview data or retrospective reports or ratings that may introduce instrumentation and observer biases. Likewise, an important feature of this methodology is that the actual behavior is recorded permanently. Errors of measurement connected with the ratings can be systematically evaluated with this ability to recycle processes. A variety of different observational procedures can be implemented and their comparability determined. A variety of types of behaviors can be observed for various purposes. Thus, the versatility in investigating a variety of methodological and substantive questions is great. This methodology also has the potential to be used for clinical diagnosis of behavioral problems and for pedological pur- poses in developing strategies to modify behavior. Such an 135 observational procedure can alert teachers to the social feedback that handicapped children may be receiving in unstructured times with peers. Likewise, systematic observations over a period of the school year could provide evidence of the child's adjustment and progress in developing social relationships. The costs appear to be fixed compared to the diversity of benefits potentially available through such a methodology. However, the cost-benefit issue is real and is an important consideration in determining the utility of this methodology. Implications and Suggestions for Future Research The results of this study illustrate that three- and four- year-old children are extremely socially oriented. Even during this more self-centered, egocentric stage of development, children are very responsive to each other. It would appear that because of children's social orientation and extended contact, the peer group may be a significant influence on children's developing social attitudes, skills, and feelings toward themselves and others. Children model, imitate, and reinforce each other's behavior by their attending and responding patterns. Much incidental learning, therefore, may be occurring during peer interactions. Hence, parents and educators should be concerned about the kinds of behav- iors that are being expressed and accepted in their children's interactions. These are important opportunities for children to try out new behaviors and explore new and old patterns for achieving social influence. Thus some deviant behavior should be expected. 136 But the general tone and substance of play situations can provide important cues about what types of behaviors, in general, children nay be reinforcing for each other. Peer contact may be especially useful in the child's develop- ment of communication strategies. These early preschool years pro- vide important experiences in social confrontation and coordination necessary in learning the underlying skills of effective interper- sonal communication. Play provides numerous occasions for pre- schoolers to practice persuasive communication transactions. These early communicative transactions may be short and even blunt, but they are the precursors of the more elaborate verbal and subtle nonverbal exchanges that become commonplace in adult social behavior. Complex human social interaction behaviors need to be broken down into components and their early identification and evolution traced to provide more definitive information about the ontogeny of communication skill. The present data illustrate that basic inter- personal Skills of gaining the receiver's attention, of coordinat- ing activity with others, and of using verbal exchanges to establish contact are present and functional in young children's behavioral repertoire even at the age of three or four years. How did these skills develop and what conditions favor their expansion? Is there a relationship between early parent-child patterns of reciprocity and children's skills in interacting with peers? These questions need to be explored with younger children. But, likewise, these interpersonal interaction skills need to be followed into later 137 stages of development. Are children's early communicative competence scores related to later skill in interpersonal communication? In this study, social communication Skill has been identi- fied and measured. An important question, therefore, is: How are these measures related to the measure of role taking common in experimental communication research? This investigator has been concerned with the dearth of research in the whole area of communi- cations with children. Perhaps the methodology implemented in this study and especially the operationalization of communication can provide a model for further investigations in this area. This study has shown that children's communications can be identified and the consequences of actions can be measured. Thus, the tools are becoming available with which to explore complex social exchanges. It is hoped communication research can be intensified with such methodological advances, and the development of communication skills in young children can become a more focused area of concern for early childhood educators. APPENDIX INSTRUMENTS 138 INSTITUTE FOR FAMILY AND CHILD STUDY MICHIGAN STATE UNIVERSITY Project Agreement Form 1, the undersigned, as parent or guardian of . a child in attendance at the day care center, by my Signature agree: (1) (2) (3) (4) (5) that my child may participate in the Social Development proj- ect approved and administered by the professional staff of the Institute for Family and Child Study at Michigan State University; that I understand that the Social Development project has been judged by the professional staff to be in no way harm- ful to the children involved and in no way an invasion of the privacy of the families; that I understand that participation in this program will not interfere with the regular program in which my child is enrolled and that no additional benefits or effects are guaranteed; that it is my understanding that each research project in which my child might be asked to participate will be explained to me and that I may withdraw my child from participation at any time if such involvement is unacceptable to me without in any way affecting his enrollment in the preschool program in which he is enrolled; that all results will be treated with strick confidence, that all individual children will remain anonymous in reporting any results, and that all results will be handled in a professional manner. By my Signature I indicate that the research has been explained to me in detail and that I understand that any further questions that I may have about the research project will be answered by the teacher, the research coordinator, or the director of the Institute for Family and Child Study. Date: Signed: Witness: 139 GENERAL INFORMATION SHEET Child's Name Sex: Male Female Birthdate Ethnic Background: Black Month Day Year White Biracial Chicano Indian Other FAMILY INFORMATION Family Status: Two parents together Separated Single parent How many years has child lived in a single parent home? Please list all brothers, sisters, or other children living in household: Does this child attend First Name Age Sex Relationship to child school or day care Yes No Please list all other adults living in household: Approximate Age Sex Number of years residing in household Please fill in the following information about the child's father, stepfather or male in the household acting as a father figure. If no father figure is present, leave this section blank. Father's Age: under 20 over 50 140 141 Father's Educational Background to present: less than 12 years of school less than 12 years and some occupational training High School High School and some occupational training Some college College degree Advanced degree Father's Present Occupation Employer If a student, Name of School and Major: Number of hours worked outside of the home per week Please fill in the following information about the child's mother, stepmother or female in the household acting as a mother figure. If no mother figure is present, leave this section blank. Mother's Age: under 20 20-29 30-39 40-49 over 50 Mother's Educational Background to present: less than 12 years of school less than 12 years and some occupational training High School High School and some occupational training Some college College degree Advanced degree Mother's Present Occupation Employer If a student, Name of School and Major: Number of hours worked outside of the home per week 142 Approximate FAMILY Income per week (take home pay of both parents-- include both assistance and salaries): less than $50 :$50 - $75 :$101 - $125 —$126 - $150 _____ $151 - $175 $176 - $200 $200 Type of Family Dwelling: Single family house_ Apartment_ Duplex_ Trailer_ ”With Relatives_ Type of Transportation to Center (usually): Walk_ Family Car— Public Transport Day Care Center Transport With friend— Approximate time needed to travel from home to the center (circle one): 5 10 15 20 25 30 35 4O 45 50 55 60 minutes. CHILD'S SOCIAL EXPERIENCES Present Day Care Enrollment: 1. How many hours per day does your child attend the center? ___ 2. How many days per week does your child attend the center? 3. How many months per year will your child attend the center?: Past Day Care or Nursery School Experience: 1. How many months has your child been enrolled in Day Care for the full day before September 1, 1973? 2. How many months has your child been enrolled in Day Care for part of the day_before September 1, 1973? 3. How many months has your child been enrolled in Day Care or Nursery School 2 or 3 days per week before September 1, 1973? 4. How many months has your child been cared for in a home situation with a Sitter or Relative during the day before September 1,1973? ' Does your child participate with other children in a group outside of School? Check (70 those activities that he/she participates in. Sunday School Story Hour YMCA Recreation Program #— Lessons (swim, dance, music, etc.) Other 143 The child meets in such groups as above hours(s) per week. Most of the child's playmates at home are: brothers and sisters other relatives friends/neighbors Most often the children that my child plays with at home are: older younger agemates When not at school my child spends approximately (circle one) 8 l 1% 2 3 4 5 6 7 8 hours playing with other children per weekday. OBSERVATION AND SOCIALIZATION BEHAVIOR (REVISED) The present instrument is an adapted version of the original Observation of Socialization Behavior (OSB), an observational rating technique for videotape observation. The original version was devel- oped by Robert P. Boger and Jo Lynn Cunningham, Head Start Research Center, Michigan State University (Boger & Cunningham, 1969). The present version was developed by Robert P. Boger, Jo Lynn Cunning- ham, and Mary Andrews, Institute for Family and Child Study, Michigan State University. General Procedure This observational rating procedure was designed for use in small group free-play (unstructured) situations only. It may be used either with or without a teacher present in the situation. Behavioral ratings of an individual child are made each 20 seconds during the observation. Each frame (representing 20 seconds) is rated as an individual unit. Therefore, the child's behavior at a previous time should not influence the ratings made for any sub- sequent interval, except insofar as the context of a preceding interval must be considered for adequate interpretation of a unit of behavior (primarily verbalization or inferred motivation). Rating of videotaped situations is facilitated if the video- tape unit has an automatic signal tone attachment for recording purposes. Such an attachment may be used to provide an audio signal at the designated ZO-second intervals. 144 145 Form The form developed for use with the videotaped interaction Situations contains two rating frames per 20-second interval. The first frame must be completed as a time sampling of behavior at the signal tone each 20 seconds. The second frame is only completed if no peer interaction occurs in the first frame but subsequently occurs during the 20-Second interval. This second frame is therefore reserved for the first observed peer interaction each 20 seconds. If a level 5 or 6 of social behavior with peers occurs during the first frame--no further observational rating is required during the 20 second interval (frame 2 will be crossed out). Likewise if no peer interaction occurs during the interval, the second frame will remain blank (crossed out). The information included in each frame consists of: 1. Interaction Responses Initiations 2. Object of interaction 3. Level of involvement 4. Peer impact 5. Verbalization 6. Verbal fantasy 7. Voice tone 8. Physical behavior 9. Physical tone 10. Social behavior 11. Autonomy 12. Leadership 13. Social Competency l4. Emotionality The format for recording an observational segment is shown in Figure A-l. Response Initiation 146 Involvement Impact Verbalization hysical Tone 01+ Social Behavior . I once Tone 0 Autonomy Leadership Figure A-l. Social CompetenCy Emotronalrty Inferred Motivation An example of one observational frame. Codes The categories and descriptions for each code follows. One and only one coding category can be ascribed unless otherwise noted. All behavioral events must be classified by one of the available mutually exclusive classifications. acceptance: covert or overt awareness and acceptance 1 - intense overt acceptance 3 - covert or weak acceptance rejection: covert or overt awareness and rejection 1 - intense overt rejection 2 — moderate rejection--withdrawal submission 3 - covert or weak rejection Interaction and Involvement Response A - of another's initiation 2 — moderate acceptance R- of another's initiation N- no awareness of another's initiation, no acknowledgment 147 O - ongoing behavior (behavior continues for at least 5 seconds with no apparent initiation or response to an initiation) l - intense overt behavior 2 - moderate behavior 3 - covert or weak behavior X - behavioral transition--initiation imminent Initiation - introduction of self or change in activity prompted by self l - intense overt initiation 2 - moderate (normal level) initiation 3 - passive initiation, covert or tentative attempt to initiate Object of Interaction (up to three separate codes may be recorded) A-N = letter code of each peer with whom S_is involved (two peers may be recorded) G = group involvement with all three other peers: initiation or response not directed to any specific individual T = adult M = materials. The objects provided specifically for play purposes (including personal articles of apparel on self) E = environment, objects not intended for play but present in the setting (walls, light switches, gate, door, etc.) Impact Codes: The consequence of st behavior as reflected in the behavior of other peers Impact recorded separately for each peer in the group A - acceptance of S's behavior 1 - intense overt acceptance 2 - moderate (normal level) of acceptance 3 - covert or hesitant acceptance N - no impact, no acknowledgment or awareness of st behavior 148 R - rejection of st behavior 1 - intense overt rejection 2 - moderate (normal level) of rejection 3 - covert, mild, or hesitant rejection Verbalizations: The entire verbalization accompanying the behavior SL TR AG SU OP 0R AR AP AS 05 ST AN being observed should be categorized as a unit. Shows solidarity: raises another's status; gives help or reward Tension release: jokes, laughs: squeals, shows satisfaction Agrees: shows passive acceptance: understands, con- curs; compiles Gives suggestions or directions, implies autonomy for others Gives opinion, evaluation, or analyses: expresses feeling or wish Gives orientation or information: repeats, clarifies, confirms Asks for orientation: information, repetition, confir- mation Asks of opinion, evaluation, analyses, expressions of feelings Asks for suggestions, direction, possible ways of acting Disagrees: shows passive rejection or formality: withholds help Shows tension: asks for help: withdraws "out of field" (swearing) Antagonism: deflates other's status: defends or asserts self: name calling: (swearing at someone) Mumbling (unintelligible) No verbalization Fantasy: NF 149 Each verbalization is classified as fantasy or nonfantasy based on the use of imaginary representations. F Fantasy verbalization Nonfantasy verbalization Voice Tone: The rating refers to the delivery, not the content, of the verbalization + II positive affect conveyed by voice 0 II neutral voice: no affect conveyed negative affect conveyed by voice Social Behavior: The social nature of the play involvement 1 Unoccupied behavior: The child apparently is not playing at all, at least not in the usual sense, but occupies himself with watching anything which happens to be of momentary interest. When there is nothing exciting taking place, he plays with his own body, gets on and off chairs, just stands around, follows the teacher, or sits in one spot glancing around the room. Solitary play: The child plays alone and independently with toys that are different from those used by the children within speaking distance and makes no effort to get close to or speak to the other children. His interest is centered upon his own activity, and he pursues it without reference to what others are doing. Onlooker behavior: The child spends most of his time watching the others play. He often talks to the playing children, asks questions, or gives suggestions, but does not enter into the play himself. He stands or sits within speaking distance of the group so he can see and hear all that is taking place. Thus, he differs from the unoccupied child, who notices anything that happens to be exciting and is not especially interested in groups of children. 4 = Parallel play: The child plays independently, but the activity he chooses naturally brings him among other children. He plays with toys which are like those which the children around him are using, but he plays with toys as he sees fit, without trying to influence the activity of the children near him. Thus, he plays beside, rather than with, other children. This activity is characterized by physical prox- imity gpg_similarity of activity with reference to other children. 150 0'! ll Associative play: The child plays with other children. They may be borrowing and lending play materials or following one another with trains or wagons. There are mild attempts to control which children may or may not play in the group. All are engaged in similar, if not identical, activity. There is no division of labor and no organization of activ- ity. Each child acts as he wishes and does not subordinate his interest to the group. There is interaction between children, but no common goal. 0‘ ll Cooperative play: The child plays within a group that is organ- ized for the purpose of making some material product, of striving to attain some competitive goal, of dramatizing situations of adult or group life, or of playing formal games. There is a marked sense of belonging or not belong- ing to the group. The control of the group situation is in the hands of one or two members who direct the activity of others. The goal and the method of attaining it neces- sitates a division of labor, the taking of different roles by various group members, and the organization of activity so that the efforts of one child are supplemented by those of another. The critical distinction is the goal- directedness of the group. Physical Behavior: The nonverbal aspect of the initiation, response, or ongoing activity Contact: (coded in relation to the object of the interaction. Peer interaction takes precedence over involvement with materials or environment) C = Contact: physical contact between subject and object or another peer NC = No physical contact with other peers or objects Behavioral Tone + = behavior which is socially acceptable or positive in connotation (holding hands, patting, sitting side by side) 0 = neutral motion: physical behavior which does not convey either ositive or negative connotations (building, running - = behavior which is not socially acceptable or is negative in connotation (pushing, hitting) 151 Inferred Motivation: The following four categories of inferred socio-emotional state are rated on a 5 point bipolar scale as illustrated below. 1 2 3 4 5 positive negative overt/intense covert/mild neutral covert/mild overt/intense Autonomy (psychological independence) self directed independent dependent patient impatient persistent non—persistent tolerant vulnerable to frustration integrated submissive Social Leadership original activity imitation initiates to others follows dominant compliant Social Competency other directed self centered friendly, open withdrawn empathetic rejecting helpful aggressive affectionate disregards others constructive boasting attention-seeking jealous destructive Emotionality happy, confident anxious eager fearful angry hesitant (rejecting) RecordingyObservations For each frame a code must be applied to each available space. If no verbalization or initiation is observable,an "X" is coded in that 152 position. All other spaces require an observational interpretation of the behavior occurring. The only exception to this rule is the rare case in which the person being observed leaves the scene (is out of camera range). In such cases, "X" for the entire frame or any part thereof is permissible. Coding of each category is done by writing in the appro- priate code (for responses, level of involvement, object of inter- action, impact, autonomy, leadershi , social competence, emotional- ity, verbalization, social behavior or by circling the appropriate code symbols (for fantasy, voice tone, physical behavior, and behav- ioral tone). The rating should commence on the first signal after all of the children are in the visual field and the door has been closed. Each subsequent child's ratings should commence in the same interval. Frame 1 (required) When the signal tone is heard marking a 20 second interval, the behavior occurring immediately after the tone is observed. All observations within a single frame refer to this one "bit" of peer behavior. A "bit” of play behavior is defined as a stretch of play behavior that includes an involvement code of either response, initia- tion, or ongoing activity that is terminated by another response or initiation. Frame 1 must be completed each 20 seconds for the entire play session. fram§_§_(optional depending on interaction) If Frame 1 does not contain a 5 or 6 level of social behavior, then prepare to record the first peer interaction that occurs in the 20 second interval. Frame 2 is only completed if a peer interaction occurs dur- ing the interval, otherwise, an "X" is placed through the entire frame. If a peer interaction occurs, record the behavior as a single interaction with all codes applying to that "bit" of inter- action. (The verbalization, physical behavior, social behavior, inferred motivation and impact are all contingent on the interaction sequence.) Whether the interaction begins as a response or an initiation, it is the total sequence of interaction that is observed and rated. R .......... I ---------- Impact 0 .......... I ---------- Impact X .......... I ---------- Impact 153 Reliability Interobserver reliability is established by two independent observers simultaneously recording the behaviors of the same child in the same intervals on their respective recording forms. Intra- observer reliability is established by a single observer rerating a previously observed tape. Two methods of computing reliability are used, one based on total blanks and the other based on total recorded positions. Each type of reliability should be computed for the entire instrument and also for each separate scale. Minimum suggested reliability indices are given in Table A-1. Points for figuring total instrument reliability are assigned as shown in Figure A-2. Procedures for computation of interobserver reliability are as follows: Total Blanks Count and evaluate the total number of possible codes, regardless of whether anything was recorded within that area for that time interval or not. This method credits the observers with agreements for those instances on which they agree that no record- aboe behavior occurred, i.e., both recorded an "X“ for that category of that interval. Formulas used for figuring reliability by this method are as follows: I Agreements (Number of points) % FEIIabIIIty = Number of frames x 23 154 Total Recorded Positions Count and evaluate only those positions in which one or both observers recorded something other than "X." The formula for figuring reliability by this method is as follows: Agreements (Number of points) Agreements plus disagreements . (Number of points possible for positions in which either observer recorded any code) % reliability = Table A—1 Minimum Suggested Rater Reliability Indices for Observation of Socialization Behavior Type of Reliability Method Inter- Intra- Entire Instrument Total Blanks .85 .90 Total Recorded Positions .65 .75 Individual Scales Total Blanks .80 .85 Total Recorded Positions .60 .70 155 Figure A-2. Assignment of points for OSB rater reliability. 156 Peer A_ Peer 8 Peer C Name ID i Date Rater Figure A-3. Observation of socialization behavior rating protocol. BIBLIOGRAPHY 157 BIBLIOGRAPHY Adams, D. K. The development of social behavior. In Y. Brackbill (Ed.), Infancy_and early childhood. New York: The Free Press, 1967. Armore, S. J. Introduction to statistical analysis and inference for psychology and education. New York: John Wiley and Sons, Inc., 1966. Arrington, R. Time sampling in studies of social behavior: A critical review of techniques and results with research suggestions. Psychological Bulletin, 1943, 49, 81-124. Auserwald, E. Interdisciplinary versus ecological approach. Family Process, 1968, 202-215. Baldwin, T., McFarlane, F., & Garvey, C. Children's communication accuracy related to race and socioeconomic status. Child Development, 1971, 42, 345-357. Bales, R. F. Interaction process analysis: A method for the study of small groups. Cambridge, Mass.: Addison Wesley Press, 1951. Bandura, A. The role of modeling processes in personality develop- ment. In W. Hartup and N. Smothergill (Eds.), The young child: Reviews of research. Washington, D.C.: National Association for the Education of Young Children, 1967. Bandura, A. Social-learning theory of identificatory processes. In D. Goslin (Ed.), Handbook of socialization theory and research. Chicago: Rand McNally Co., 1969. Bearison, D., & Cassel, T. Cognitive decentration and social codes: Communicative effectiveness in young children from differing family contexts. Journal of Abnormal Psychology, 1975, 11, 29-30. Berlo, D. K. The process of communication. New York: Holt, Rine- hart and Winston, 1960. Berne, E. Social behavior patterns in young children. University of Iowa Studies in Child Welfare, 1930, 4, 158 159 Bernstein, B. Elaborated and restricted codes: Their social origins and some consequences. In J. Gumperz and D. Hymes (Eds.), The enthnomp'aphy of communication. American Anthropologist Special Publication, 1964, 66, 55-69. Bernstein, B. A socio-linguistic approach to social learning. In J. Frost (Ed.), Early childhood education rediscovered. New York: Holt, Rinehart and Winston, 1968. Bertalanffy, L. von. General systems theory. New York: George Braziller, 1968. Blurton Jones, N. Categories of child-child interaction. In N. 4 Blurton Jones (Ed.), Ethological studies of child behavior. London: Cambridge University Press, 1972. Boger, R. P., and Andrews, M. Early social development: Parent and child programs. Final Report. East Lansing, Michigan: Institute for Family and Child Study, Michigan State Uni- versity, 1975. Boger, R. P., and Cunningham, J. L. Development of an observational rating schedule for preschool children's peer-group behavior. Paper presented at annual meetings of the American Educa- tional Research Association, New York, February, 1971. Boger, R. P., and Cunningham, J. L. A longitudinal study of the social development of three and four year old children in a preschool program. Paper presented at the annual meeting of the American Psychological Association, Honolulu, Hawaii, September, 1972. Boger, R. P., and Cunningham, J. L. Observation of socialization behavior. Unpublished instrument description, Head Start Research Center, Michigan State University, 1969. Boger, R. P., Cunningham, J. L., Andrews, M. Observation of social- ization behavior (revised edition). Unpublished Instrument Description, Institute for Family and Child Study, Michigan State University, 1973. Bowlby, J. Attachment and loss (Vol. 1). New York: Basic Books, 1969. Boyd, R., and DeVault, M. V. The observation and recording of behavior. Review of Educational Research, 1966, 36, 529-551. Brannigan, R., and Humphries, 0. Human non-verbal behaviour, a means of communication. In N. Blurton Jones (Ed.), Ethological studies of child behaviour. London: Cambridge University Press, 1972. 160 Brazelton, T. B., Koslowski, B., and Main, M. The origins of reciprocity: The early mother-infant interaction. In M. Lewis & L. Rosenblum (Eds.), The origins of behavior. (Vol. 1): The effect of the infant on its caregiver. New York: John Wiley and Sons, 1974. Bridges, K. A study of social development in early infancy. Child Development, 1933, 4, 36-49. Brophy, J. Mothers as teachers of their own preschool children: The influence of socioeconomic status and task structure on teaching specificity. Child Development, 1970, 41, 79-94. Bruner, J. The act of discovery. In 1. Gordon (Ed.), Human develop- ment. Glenview, Ill.: Scotts, Foresman and Co., 1965. Buckley, W. Sociology and modern systems theory. Englewood Cliffs, New Jersey: Prentice-Hall, 1967. Buhler, C. The social behavior of children. In B. Murcheson (Ed.), A handbook of childspsychology. New York: Russell and Russell Publishing Company, 1933. Caldwell, B. A new approach to behavioral ecology. Minnesota Symposium on Child Psychology, 1969, 2, 74-109. Chapple, E. Measuring human relations: An introduction to the study of the interaction of individuals. Genetic Psychology of Monographs, 1940, 22, 3-147. Charlesworth, R., and Hartup, W. Positive social reinforcement in the nursery school peer group. Child Develppment, 1967, 22, 993-1002. Cherry, C. On human communication. New York: Science Editions, 1961. Chin, R. The utility of system models and developmental modes for practitioners. In W. G. Bennis, K. D. Benne and R. Chin (Eds.), The planning of change (2nd ed.). New York: Holt, Rinehart and Winston, 1969. Clausen, J. A. Family structure, socialization, and personality. In L. Hoffman & M. Hoffman (Eds.), Review of child develup: ment research (Vol. 2). New York: Russell Sage Foundation, 1966. Clausen, J. A. (Ed.) Socialization and society. Boston: Little, Brown and Co., 1968. 161 Cunningham, J. L. A comparison of the didactic interactions of mothers and fathers with their preschool children. Unpub- lished doctoral dissertation, Michigan State University, 1972. Dance, F. E. Toward a theory of human communication. In F. E. Dance (Ed.), Human communication theory: Original essays. New York: Holt, Rinehart and Winston, 1967. Emmerich, W. Preschool personal and social behaviors: Relationships with socioeconomic status, cognitive skills and tempo. Dis: advantaged children and their first school experiences; ETS—-Head Start longitudinal study. Princeton, New Jersey: Education Testing Service, 1973. Erwin-Tripp, 5. Language development. In L. Hoffman and N. Hoff- man (Eds.), Review of child development research. New York: Russel Sage Foundation, 1966. Flavell, J. The development of role-takingsand communication skills in children. New York: John Wiley and Sons, 1968. Gardner, D. B. The child as an open system: Conference summary and implications. In Play: The child strives toward self realization. Washington, D. C. National Association for the Education of Young Children, 1971. Garvey, C., and Hogan, R. Social speech and social interaction: Egocentrism revisited. Child Develgpment, 1973, 55, 562-568. Gellert, E. Systematic observation: A method in child study. Harvard Education Review, 1955, 25, 179-195. Glass, G., and Stanley, J. Statistical methods in education and psychology. Englewood Cliffs, New Jersey: Prentice-Hall, nc., . Glucksberg, S. E., and Krauss, R. M. "What do people say after they have learned to talk?" Merrill Palmer Quarterly, 1967, 12, 309-316. Golembiewski, R. The sma11_grogp: An analysis of research concepts and operations. Chicago: University of Chicago Press, 1962. Goodenough, F. L. Measuring behavior traits by means of repeated short samples. Journal of Juvenille Research, 1928, 12, 230-236. Habermas, J. Introductory remarks to a theory of communicative competence. In H. Dreitzel (Ed.), Recent sociology. London: McMillan, 1970. 162 Hare, A. P. Handbook of small group research. New York: The Free Press of Glencoe, 1962. Hartup, W. E. Peer interaction and social organization. In P. Mussen (Ed.), Carmichael's manual of child psychology (3rd. ed.). New York: John Wiley and Sons, 1970. Hartup, W. E., and Coates, B. Initiation of a peer as a function of reinforcement from the peer group and rewardingness of the model. Child Development, 1967, 22, 1003-1016. Hess, R. D. Social class and ethnic influences upon socialization. In P. Mussen (Ed.), Carmichael's manual of child psychology (3rd. ed.). New York: John Wiley and Sons, 1970. Hess, R. D., and Shipman, V. C. Early experience and the socializa- tion of cognitive modes in children. Child Development, 1965, §fl, 869-886. Hicks, D. J. Imitation and retention of film-mediated aggressive peer and adult models. Journal of Personal and Social Psychology, 1965, 2, 97-100. Honig, A., Caldwell, B., and Tannenbaum, J. Patterns of information processing used by and with young children in a nursery school setting. Child Development, 1970, 11, 1045-1065. Hook, N. C., and Paolucci, B. The family as an ecosystem. Journal of Home Economics, 1970, 62, 315-318. Isaac, S., and Michael, W. Handbook in research and evaluation. San Diego, Claifornia: Robert Knapp, Publisher, 1971. Jersild, A., and Markey, F. Conflicts between preschool children. Child Development Monographs, 1935, 21, Kerlinger, F. N. Foundations of behavioral research (2nd ed.). New York: Holt, Rinehart, and Winston, 1973. Kohlberg, L. Stage and sequence: The cognitive developmental approach to socialization. In D. Goslin (Ed.), Handbook of socialization theory and research. Chicago: Rand McNally, 1969. Krauss, R., and Glucksberg, S. Socialization of communication skills. In R. Hoppe, G. A. Milton, and Simnel, E. (Eds.), EarLy experiences and thesprocess of socialization. New York: Academic Press, 1970. 163 Krauss, R., and Rotter, S. Communication abilities of children as a function of status and age. Merrill Palmer QuarterLy, 1968, 14, 161-173. Larson, 0., and Hill, R. Social structure and interpersonal communi- cation, American Journal of Sociology, 1958, 22, 497-505. Laszlo, E. (Ed.). The relevance of general systems theory, New York: Braziller, 1972. Levison, C. Use of the peer group in the socialization of the iso- lated child. Paper presented at the meetings of the Society for Research in Child Development. Minneapolis, Minnesota, 1971. (ERIC Document Reproduction Service No. ED 054 858). Lewis, M., and Rosenblum, L. (Eds.). Origins of Behavior. (Vol. 1). New York: John Wiley, and Sons, 1974. Lin, N. The study of human communication. New York: Bobbs, Merrill Co., 1973. Maudry, M., and Nekula, M. Social relations between children of the same age during the first two years of life. Journal of Genetic Psychology, 1939, 24, l93-215. McQuire, C., and White, G. The measurement of social status: Research paper in Human Development, no. 3 (revised). Department of Educational Psychology, University of Texas, March 1955. Mead, G. H. Mind, self and society, Chicago: University of Chicago Press, 1934. Miller, D. R., and Swanson, G. E. Expressive styles: III. Two styles of expression: motoric and conceptual. In D. R. Miller and G. E. Swanson (Eds.), Inner conflict and defense. New York: Holt, Rinehart, and Winston, 1960. Miller, G. R., and Steinberg, M. Between peop1e: A new ana1ysis of interpersonal communication. Chicago: Science Research Associates, Inc., 1975. Miller, J. G. Living systems: Basic concepts. Behavioral Science, 1965, 19, 193-201. Mowrer, O. H. Learning_theo:y and1persona1itysdynamics. New York: Ronald Press, 1950. Mueller, E. The maintenance of verbal exchanges between young children. Child Development, 1972, 42, 930-938. 164 Nie, N., Bent, D., and Hull, C. H. Statistica1_package for the social sciences. New York: McGraw-Hill, 1970. Olson, W. C., and Cunningham, E. M. Time-sampling techniques. Child Development, 1934, 2, 41-48. Parten, M. Social participation among preschool children. Journal of Abnormal Social Psychology, 1932, 21, 243-269. Paulson, L. Time versus televised observations of social behavior in preschool children. Paper presented at the Annual Convention of the American Psychological Association, September, 1972. (ERIC Document Reproduction Service No. ED 071 735) Pena, 0. Analysis of social skill development in the Appalachia preschool education program, 1971. (ERIC Document Repro- duction Service No. ED 062 021) Pena, D., and Miller, G. Analysis of children's social skill development and their reactions to a preschool television program, 1971. (ERIC Document Reproduction Service No. ED 057 884) Piaget, J. The lapguage and thopght of the child. New York: Harcourt Brace, 1926. Piaget, J. The psychology of intelligence. New York: Harcourt, Brace and World, Inc., 1950. Ralph, J. B., Thomas, A., Chess, 5., and Karn, S. J. The influence of nursery school on social interactions. American Journal of Orthopsychiatry, 1968, 22, 144—152. Reuter, J., and Yunik, G. Social interaction in nursery schools. Developmental Psychology, 1972, 2, 319-325. Richards, M. (Ed.). The integration of the child into a social world. London: Cambridge University Press, 1974. Ryan, J. Early language development: Towards a communicational analysis. In M. Richards (Ed.). The integration of the child into a social world, 1974. London: Cambridge University Press. Schachter, and others. Interpersonal language of preschoolers differing in intrapersonal linguistic effectiveness. Report of research project. New York: Bank Street College of Education, 1970. (ERIC Document Reproduction Service No. ED 056 768) 165 Scheifley, V., and Schmidt, W. Jeremy D. Finn's multivarance-univari- ate and multivariate analysis of variance, covariance, and regression modified and adapted for use on the CDC 6500. Occasional paper no. 22. Office of Research Consultation, College of Education, Michigan State University, 1973. Shantz, C. U. Communication and egocentrism module: Training communications skills in young children. Unpublished research report, Institute for Family and Child Study, Michigan State University, 1970. Shatz, M., and Gilman, R. The development of communication skills: Modifications in the speech of young children as a function of listener. Monographs of the Society for Research in Child Development, 1973, 22, Sims, L., Paolucci, B., and Morris, P. A theoretical model for the study of nutritional status: An ecosystem approach. Ecology of Food and Nutrition, 1972, l, l97-205. Smith, P., and Connolly, K. Patterns of play and social interaction in pre-school children. In N. Blurton Jones (Ed.), Etho- logical studies of child behavior. London: Cambridge Uni- versity Press, 1972. Sutton-Smith, B. Boundaries. In R. E. Herron and B. Sutton-Smith (Eds.), Child'ssplay. New York: John Wiley and Sons, 1971. Sutton-Smith, B., and Rosenberg, B. G. The sibling. New York: Holt, Rinehart and Winston, 1970. Thayer, L. Communication and communicationssystems. Homewood, 111.: Irwin, 1968. Vickers, G. The multivalued choice. In L. Thayer (Ed.), Communica- tion concepts and_perspectives. New York: Spartan Books, 1967. Vygotsky, L. S. Thought and language. New York: John Wiley and Sons, 1962. Watzlawick, P., Beavin, J., and Jackson, 0. Prggmatics of Human Communication. New York: Norton, 1967. Weinstein, E. The development of interpersonal competence. In D. Goslin (Ed.), Handbook of socialization theory and research. Chicago: Rand McNally Co., 1969. Whorf, B. L. Language, thought and reality, New York: The Tech- nology Press and John Wiley and'Sons, 1956. 166 Wright, H. F. Observational child study. In P. Mussen (Ed.)., Handbook of research methods in child development. New York: John Wiley and Sons, 1960. "I11111111111111.1111“.