ELOPMENT' OFAT n D ¢>Owg~_‘ .a.g.y._,.a.>.,. 9—” , .V .. r... .n..—..-=~- ..._..\..., L ....Vs.:.- ‘ \ .. 32.3,.3... .V,......¢... E \I" , , 01F REmi‘i‘CRQ ; m » Eegkszganefl Thwié, m {lief ‘ TEACHER ....,:_. .1. .V . 7. . 3:92;....xix.:.rr{.;£52.... 4.. X5.- . 1“ cat-M 3mg gm: ‘ ¥ E77.N\ . I. 1?; A. , ,. . , V . . . . . ,. . I . . . ,. . _ . ($35.? _ . _. . ‘ .. ... . VV. ¢ ...~. . ..‘ .LVH .. _ U . .,. . .. . . V Tswflwlfl ., , ., ”v . ...V.. . i. V ...V ,. : . V U... . A . M‘V'r / V .V _ . m m ,_ .V % . w . . , T .3 . V N. 19:1. mm WWW mm ill b. .. I, . “L V. . _ 52.... .m ... _. V ..........V.. .nv V . .. n V I.,. .. r -3 .S‘Ithuzf . , ‘ Z n .. . . A): . w.. s. f. .. ‘Cmi. mun” <.~ ..,W.\.thLHJ.3Iu4uA«.I—$. 2.... .77.... . .3 A... in};.5.....«...}._:__..‘.. .705? (ti. .5....~L\U#n$g§%%~ LIBRARY Michigan Static, . University w mew“ This is to certify that the thesis entitled THE DEVELOPMENT OF A TECHNIQUE FOR THE CATEGORIZATION AND ANALYSIS OF TEACHER REINFORCEMENT BEHAVIOR presented by PATRICIA RAMSDELL WIDMAYER has been accepted towards fulfillment of the requirements for ph. D. flame in Education Date June 24, 1971 0-7639 ABSTRACT THE DEVELOPMENT OF A TECHNIQUE FOR THE CATEGORIZATION AND ANALYSIS OF TEACHER REINFORCEMENT BEHAVIORS BY Patricia Ramsdell Widmayer The purpose of this investigation was to develop an observation schedule to categorize and analyze teacher reinforcement behavior. The study was designed to provide instrumentation to facilitate the adeption of the precision teaching model in the classroom by providing a technique for (l) describing the teacher's reinforcement repertoire, (2) presenting alternatives for selecting those reinforcers that should be used more frequently, used less frequently or eliminated from the teacher's repertoire (given her behavioral objectives) and (3) recording the teacher's revised behavior for control and analysis. An observation schedule was written for classroom use with multiple, behaviorally de- fined items in three major categories: rewards, feedback and deterrents. Potential items for inclusion were derived from three sources: inter- action analysis, applied behavior analysis and reinforcement literature, review of video-taped classroom episodes from a study in the Lansing, Michigan, Public Schools and classroom observations by the author. The instrument includes twenty-eight categories plus subcategories and examples to illustrate each item. Four experienced teachers were trained to use the instrument. Tests for observer agreement at the conclusion of the training ses- sion derived percentages of .84, .75, .87, .67 and .76. The percentage of observer agreement on recording during the study averaged .82. The testing of the reinforcement observation schedule was limited to suburban elementary school teachers identified as "competent" by the building principals. Ten teachers in five Royal Oak, Michigan, elementary schools were observed for ten randomly-selected half-hour periods over three weeks in April, l970,and their reinforcement be- haviors systematically recorded. The data provided the basis for determining the reliability and validity of the instrument and, simul- taneously, describing the reinforcement behavior of "competent" teachers as a criterion for practitioner performance and further instrument utilization. Analysis of the distributions of the recorded teacher reinforce- ment behaviors was performed through Gerig's Multivariate Extension of the Friedman Test, Kendall's Coefficient of Concordance and per- centage comparisons. The findings indicate that teacher behavior was stable over time and that the teachers had a high level of agreement on category rankings by frequency. The frequency distributions for the ten teachers suggest certain directions for teacher education, but additional comparative data must be obtained using this instrument to describe teacher behavior in other socio-economic settings, grade levels, teaching styles and levels of competence before conclusive statements can be made. The findings do, however, support the conclusion that a teacher reinforcement behavior observation schedule was developed with sufficient reliability and validity to justify further use in teacher education, both pre-service and in-service. Also, optimal procedures are indicated by this study for future instrument development. ".V A a: *‘w THE DEVELOPMENT OF A TECHNIQUE FOR THE CATEGORIZATION AND ANALYSIS OF TEACHER REINFORCEMENT BEHAVIORS BY Patricia Ramsdell Widmayer A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Education 1971 © Copyright by PATRICIA RAMSDELL WIDMAYER 1971 PLEASE NOTE: Some Pages have indistinct print. Filmed as received. UNIVERSITY MICROFILMS To Larry, Carole and Jacquey, whose assistance and encouragement made this research a possibility and whose constant harrassment made its completion a necessity. ACKN OWLE DGEMENTS Several institutions and individuals made significant contributions to this research. Dr. Judith Henderson was a constant source of inspiration and assistance during all phases of this project from the initial formulation of the idea to the final editing of the thesis. The Learning Systems Institute of the College of Education at Michigan State University under the direction of Dr. Ted Ward generously supported and partially funded this research. The ten teachers and five principals of the Royal Oak Public Schools and Mr. Daniel Welch, Director of Elementary Education, cannot be thanked enough for their willing c00peration in this endeavor. And, finally, Mary B. Greenlee, Margaret Z. Valade and Jacqueline B. O'Malley must be recognized for their perspicacity and persistence in maintaining a hectic schedule over many weeks while observing the teachers in their classrooms. To each I wish to extend my sincere gratitude. iv TABLE OF CONTENTS CHAPTER I. INTRODUCTION OF THE PROBLEM . . . . . Theoretical Framework of the Problem Statement of the Problem and Purpose Characteristics of the Instrument . Objectives of the Study . . . . . . II. REVIEW OF RELATED RESEARCH . . . . . . Applied Behavior Analysis Research . Present School Practices and Problems Interaction Analysis Systems . . . . A Definition of Reinforcement and Its III. PROCEDURES AND METHODOLOGY . . . . . . Development of the Instrument . . . The Selection and Training of Observers Selection of the Population . . . . Selection of the Sample . . . . . . Design of the Study . . . . . . . . Analyses . . . . . . . . . . . . . . IV. PRESENTATION AND ANALYSIS OF THE DATA Observer Agreement in Recording . . Stability of Practitioner Behavior . Practitioner Agreement on Category Ranking Descriptive Distributions of Teacher Reinforcement B ehaVi or O O O O O O O O O O O O 0 PAGE 10 17 21 21 3O 32 37 41 41 46 48 51 52 53 SS 55 59 64 68 CHAPTER PAGE V0 DISCUSSION AND CONCLUSIONS 0 O O O C O O O O O C O O O 77 Instrument Reliability and Validity . . . . . . . . . 77 Implications for Teacher Education . . . . . . . . . 81 Suggestions for Further Research . . . . . . . . . . 83 Optimal Procedures for Future Instrument Development 85 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . 87 APPENDIX I. Observation Schedule and Instruction Manual . . 97 APPENDIX II. Calculation of Observer Agreement . . . . . . . 121 vi LIST OF TABLES TABLE 1. Video-Tape Reliability Checks. . . . . . . . . . . . . . 2. Live Classroom Reliability Checks . . . . . . . . . . . 3. Observer Agreement During the Study . . . . . . . . . . 4. Rank-Order of Category Frequencies for T1 . . . . . . . 5. Rank-Order of Category Frequencies for T2 . . . . . . . 6. Rank-Order of Category Frequencies for T3 . . . . . . . 7. Rank-Order of Category Frequencies for T4 . . . . . . . 8. Rank-Order of Category Frequencies for T5 . . . . . . . 9. Rank-Order of Category Frequencies for T6 . . . . . . . 10. Rank-Order of Category Frequencies for T7 . . . . . . . ll. Rank-Order of Category Frequencies for T8 . . . . . . . 12. Rank-Order of Category Frequencies for T9 . . . . . . . l3. Rank-Order of Category Frequencies for Tlo . . . . . . . 14. Mean Rank for Each Category During Each Observation . . 15. Results of Test for the Stability of Practitioner Behavior . . . . . . . . . . . . . . . . . . . . . . 16. The Ranking of Categories by Frequency for Each Teacher and Mean Rank for Each Category for All Teachers . . 17. Total Frequencies in Each Category for Each Teacher . . 18. Mean Frequencies for All Teachers in Each Category and Pr0portion of Total Behaviors in Each Category . . . 19. Summary of Reinforcement Behaviors for T1 . . . . . . . 20. Summary of Reinforcement Behaviors for T2 . . . . . . . xiii PAGE 56 57 58 60 6O 61 61 61 61 62 62 62 62 63 63 67 7O 71 72 72 TABLE PAGE 21. Summary of Reinforcement Behaviors for T3 . . . . . . . . 72 22. Summary of Reinforcement Behaviors for T4 . . . . . . . . 72 23. Summary of Reinforcement Behaviors for T5 . . . . . . . . 73 24. Summary of Reinforcement Behaviors for T6 . . . . . . . . 73 25. Summary of Reinforcement Behaviors for T7 . . . . . . . . 73 26. Summary of Reinforcement Behaviors for T8 . . . . . . . . 74 27. Summary of Reinforcement Behaviors for T9 . . . . . . . . 74 28. Summary of Reinforcement Behaviors for Tlo . . . . . . . . 74 29. Total Reward, Feedback and Deterrent Proportions for Each Teacher and Mean Proportions for All Teachers . . . . . 75 30. Pr0portions of Total Behaviors for Each Teacher in Teacher-Initiated and Student-Initiated Categories and Mean Proportions for All Teachers . . . . . . . . . 76 viii CHAPTER I INTRODUCTION OF THE PROBLEM THEORETICAL FRAMEWORK OF THE PROBLEM Educational research has finally reached a stage where it can offer substantial assistance to teachers as they attempt to bring about intended learning in students, i.e. learning defined as a relatively permanent change in a behavioral tendency resulting from reinforced practice (Kimble and Garmezy, 1963, 133). This has not been a sudden occurrence, but rather a possibility built upon learning research, group dynamics research and philosophical investigation during the last half century. However, although the principles are tested and available, instrumentation must be designed before the practitioner can use these principles in the classroom. During the past fifty years, many experimental programs were instituted and hundreds of instructional innovations initiated in the schools to increase student learning. Yet there was little evidence that a theoretical grounding for educational research might be emerging. Rather, these experimental attempts were executed and evaluated as short-term, univariate programs involving small numbers of students and teachers. In 1963 Elliot and Foshay (233) reached the con- clusion that: "Perhaps the most common complaint is that research knowledge in education does not cumulate in the manner found in such fields as medicine and engineering." But during the 1960's a new area of inquiry emerged which offered greater potential for cumlative 2 knowledge and theoretical frameworks in education: interdisciplinary studies. This marriage was not entered into voluntarily, but was an expedient brought about by increasing demands for reform. Paul Goodman (1962), Edgar Friedenberg (1963), John Holt (1965), George Leonard (1968), Father James Kavanaugh (1967), Herbert Kohl (1967) and others attacked the educational institution as irrelevant to contemporary needs and they gained credence in a time of turmoil. It became apparent that total educational reform was necessary, not just adding new sub- jects or altering certain teacher strategies. But little research evidence existed to provide substance for such a total change. Thus Specialists began to cooperate in compiling the research results in their disciplines to hasten payoff research; they began jointly for- mulating, investigating and stating solutions in a problem area. For example, practitioners, curriculum specialists, evaluation special- ists and area experts have combined their expertise in proposing new approaches to classroom instruction (Saunders and Tanck, 1970). As a result of these investigations it is feasible for specialists in combination to purposefully create and reveal synergisms. It is the investigation and manipulation of these interactive effects in the classroom which have made the massive facilitation of learning a possibility (Bandura, 1969). This cumulative research has resulted in an educational model most accurately called precision teaching (also referred to as applied behavior analysis, behavioral counseling and behavior therapy from its clinical background). Meacham and Weisen (1969, 3) describe precision teaching as follows: (It) involves a blending of scientific procedures with the vast problems of modern education . . . . What is developing, then, is a system of education which carefully evaluates its own methods through the framework of science and makes corrections based on data rather than opinion, on facts rather than fiction. In the classroom this new approach is beginning to manifest itself in what is called precision teaching. Precision teaching requires that the environment be programmed for successful learning through the measurement of behavior in the classroom,the setting of explicit behavioral goals, the strength- ening of desirable behavior through planned positive reinforcement and the provision of cues for learning. Thus, precision teaching stands at the crossroads of three very dominant areas of educational research: behavioral science methodology, Skinnerian reinforcement theory and classroom interaction analysis. The melding of these three concerns into one complex, rational approach to classroom manipulation presents a concrete alternative for enhancing both cognitive and affec- tive learning for all children, a goal long sought but seldom realized. The results of precision teaching in modifying behavior have been significant, as will be indicated further in Chapter II. Several basic assumptions are present in precision teaching. A first assumption stems from the old adage that, "It's not what you say, it's what you do that counts." Therefore teacher self reports are not used, but rather the basic data for precision teaching is overt behavior. The unit for observation is a response, a single completed statement or interaction between teacher and student, student and student or student and classroom stimuli which has a clearly observable beginning and end (Meacham and Weisen, 1970, 50), A second assumption is that the most appropriate tactic to produce desirable responses is reinforcement. This is based upon the principles of learning that 4 Kuethe (1968, 57) summarized by stating, "A wealth of experimental data has demonstrated that there is a relationship between the perfor- mance of an organism and the reinforcement that is received for per- formance." A third assumption is that the entire school setting is utilized. Thus, resources for intended learning are not limited to the teacher and the textbook, but any and all stimuli in the environ— ment can be used in the planned changes. A reinforcement approach to teaching points out a new direction which is contrary to much existing classroom practice. Many plans for learning are now based upon the premise that if a person has a high self-concept, a positive attitude toward his environment, and is aware of the consequences that will result from an unsatisfactory school performance (usually negative) he will be likely to achieve at higher levels. However, precision teaching follows B. F. Skinner's contention (1953, 73-75 &242-256) that we do not give a man a sense of achievement, but rather we reinforce a particular action and the knowledge and attitudes that accrue are simply a repertoire of rein- forced behaviors. To reiterate, then, precision teaching deals directly with overt behavior and reinforcement in the school setting. These factors are deemed crucial in that they facilitate measurement and manipulation in a systematic fashion. Several advantages accrue from the precision teaching or behavioral method: 1. The student/teacher relationship is modified to a client/change agent relationship. This latter relationship is a more plastic pairing, necessitating mutual planning and definition of goals. Thus, the rigidity of the cognitive-dominant, teacher- 5 telling strategy is loosened. (Gage & Unruh, 1967, 358) In its place is a prescription for modifying the entire school environment. Also, because of the redefinition of the student/teacher relationship, all environmental stimuli are incorporated into the change agent/client relationship. This setting is very amenable to both the effective utilization of technology (media and computer) and the programming of higher learning experiences (not necessarily through machinery, but structuring of the conditions of learning and its contingencies in the environment). 2. A systematic description of the entry behaviors of both the student and the teacher are required in the categories of defined con- cern. Thus, a record of the overt behaviors of both the client and change agent are available for analysis and change. Two advantages are gained from this procedure: (A) This provides the focus for a determination of the apprOpriateness of present instruction and subse- quent learning goals for the student, i.e. if instruction is directed toward the satisfaction of the individual's specific and unique needs. (B)Such an analysis also provides the vehicle for individual planning. This could answer the criticism voiced by Edmund Gordon (1970, 9) regarding compensatory education: These demonstrations and evaluation efforts reflect a search for generic treatments, a desire to find the program or practice that works for large numbers of peOple: this tendency can be seen as a reflection of the generic nature of research on p0pulation characteristics, which tends to give the impression that we are dealing with a large, homogeneous group with common problems of development. Very little of this research is directed at carefully designed and controlled experimentation or at qualitative analysis of large sample of naturally occurring programs to identify relationships between differential learner characteristics and differential treat- ment characteristics. Questions as to what works for which children under what specific conditions are not heavily reflected in avail- able research to date. Data that would shed light on individual learning style would be collected constantly in precision teaching. 3. An eXPliCit statement of behavioral objectives is re- quired, enabling a clear criterion for assessment to be provided for the student and teacher. This specification has two distinct advantages: (A) By concisely stating the objectives, a specific judgment of the worthiness or goodness of the intended changes can be made prior to the inception of learner behavior modification, and (B) a prior statement of objectives provides the criteria for change agent accountability for intended and unintended learning. 4. The central precept of precision teaching is positive reinforcement for desirable behavior and the withholding of positive reinforcement when undesirable behavior occurs (as defined by the change agent, often in consultation with the client). One advantage is that the emphasis is on reinforcement to the exclusion of punishment. In other words, the potentially negative consequences of the use of punishment (e.q. negative attitudes toward subject matter, teachers and school in general) are diminished. This is in keeping with Leonard's (1968, 43) description of Skinner's research: Skinner has found that, not punishment, but re- ward -- or, in his terms, positive reinforcement -- is the most effective force for shaping or teaching. Positive reinforcement may turn out to be food, money, praise, a kiss, a smile, a fleeting nod of approval. Best of all, it may consist simply of getting a right answer and knowing it's right, of working out a puzzle, of mastering a skill, of finding new beauty and order in words, music, color. . . . Too often, according to Skinner, people have relied on punishment or the threat of punishment to control other people. Or they used gross and insulting positive reinforcement, such as money meted out in a crude piece-work system. Skinner's scientific and humane impulses meet in his desire to substitute reward for punishment, to rely in- creasingly on the more subtle and "beautiful" forms of positive reinforcement. In implementing precision teaching, the overt observable be- havior of the student is the dependent variable and reinforcement is the independent variable. Once a behavioral objective has been stated, a baserate of the applicable student behavior and the teacher reinforcement of that behavior are obtained. Typically this is done by the teacher or an outside observer tallying the student behavior on a desirable-undesirable dichotomy. The teacher's behavior is tallied, usually, as (1) approving desirable, (2) ap— proving undesirable, (3) disapproving desirable, (4) disapproving un- desirable. Second, the teacher determines what is reinforcing for the student. Third, the teacher manipulates her reinforcements, reinforcing appropriate behaviors and withholding reinforcement when inappropriate behaviors occur. Control is exercised over this process by recordkeeping and experimentation with reinforcement techniques until the desirable behavior is exhibited by the student for an extended period of time. Krumboltz and Thoresen (1969, xi) provide evidence that this technique is effective for all age levels in four categories: (1) deficient decision-making skills, (2) in- effective academic skills, (3) inappropriate social skills and (4) self-defeating fears and anxieties. STATEMENT OF THE PROBLEM AND PURPOSE A major problem presently prevents adoption of the precision teaching model in the public schools: there is insufficient instru- mentation in existence for the classroom teacher to categorize and analyze behavior. Some examples are provided in the literature (see Chapter II), but in most cases the categorizations were designed as situation specific. The greatest need, therefore, appears to be for a comprehensive categorization of the reinforcement techniques that are and can be used by teachers. Given the present constraints (lack of time, training and consultation assistance) of the class- room teacher, it is difficult for her to design the necessary instruments, even after she is cognizant of the viability of the model. If reinforcement is the independent variable to effect behavior change, there should be an instrument available that would provide for (1) the assessment of teacher entry behaviors, (2) the description and refinement of reinforcement and feedback and (3) the evaluation of a teacher's performance. Such an instrument would: (1) describe the teacher's reinforcement repertoire, (2) present alternatives for selecting those reinforcers that should be used more frequently, used less frequently or eliminated from the teacher's repertoire (given her behavioral objectives) and (3) record the teacher's revised behavior for control and analysis. An attendant problem in the adoption of the precision teaching model is the limited reinforcement alternatives of most teachers. An extensive listing of reinforcement techniques in the instrument (with behavioral definitions and examples) would provide most teachers with a greater number of behavior options than are presently being gleaned 9 from pre-service training. Evidence from several sources supports this contention. First, a perusal of the education curriculum in any college catalog will reveal that usually only one educational psychology course is required for graduation and certification. This allows only minimal time for exposure to and understanding of the literature. Second, during classroom observation periods, when pre- service teachers go into the schools, they see reinforcement practices that seldom go beyond praise and grades. Finally, this author has found that students enrolled in methods courses and supervised teaching are seldom able to generate a list of greater than five reinforcement options. Thus, a listing of reinforcement behaviors is needed that will move teachers out of the present limited model. The advantages of precision teaching, an instructional model based upon Skinnerian reinforcement theory, have been discussed. However, no instrumentation exists beyond textbook and research descriptions to assist the practitioner in adOpting this teaching strategy. There— fore, classroom observation instruments should be devised for recording overt behaviors to assist the practitioner in describing classroom behavior, formulating a procedure for behavior change, carrying out the program and evaluating its effectiveness in terms of behavioral modification. Because the teacher's major resource in implementing precision teaching is reinforcement, it is this construct that should be dealt with first. Thus, it is the purpose of this study to develop a classroom observation schedule to categorize and analyze teacher reinforcement behavior. For optimal utilization, this instrument should be employable by the classroom teacher. 10 CHARACTERISTICS OF THE INSTRUMENT In developing the instrument, behavioral definitions and examples should be devised, since none presently exist. Certain methodological eXplanations have been written which could serve as a partial guide in this endeavor (Medley and Mitzel, 1963: Simon and Boyer, 1967). How— ever, a revised approach must be taken in the deve10pment of the be- havior schedule since the observation instruments previously developed have encountered certain problems: 1. Most observation schedules are used as research and pre- service tools, but are not widely used by classroom teachers as part of inservice training. Classroom observation instruments have enjoyed increasing popularity during the past decade among researchers and university instructors. This is because the paper-and-pencil classroom obser- vation instruments provide systematic data on classoom dynamics with minimal interruption of natural on-going processes. Before the observation instrument, classroom interaction research was limited to anecdotal reporting, static sociogram methods or artificial manipulation of classroom processes in the laboratory setting. However, the classroom observation instruments, most notably Flanders' Interaction Analysis (Flanders, 1970), Medley and Mitzel's OScAR (1963), Bellack's Language of the Classroom (1963) and the Aschner-Gallagher TOpic Classification System (Aschner, 1959) have failed to gain wide acceptance among practitioners, although this has been one of their stated intents. This lack of interest has occurred despite many possible classroom uses as described by Rosenshine (1970, 279): (Three) potential uses of these instruments (for the observation of classroom instruction £_are_/): 11 assessing variability in classroom behavior, assessing whether the teacher's performance agrees with specified criteria describing class- room interaction, and determining relationships between observed classroom behavior and outcome measures. The lack of practitioner enthusiasm, however, may be implied by Rosenshine's very statement: the suggested uses of the instrument are so multiple and non-directional as to be inappropriate for practitioners (Rosenshine, 1970, 293). The classroom teacher has minimal training in behavior modification and evaluation techniques. Thus tools with more explicit parameters are needed. Researchers have failed to develop classroom instruments which have specific applicability to a given behavioral model or instructional strategy. It seems crucial to the writer that the development of the classroom observation instruments should now evolve toward the design of categorization systems within conceptual teaching models. 2. An observation instrument must be developed with more numerous and explicit categories. (Meux, 1967, 549—550) The number, naming and defining of categories in the observation instruments provide very convenient, interpretable data for hypothesis testing of classroom interaction by the researcher. However, most systems are limited to ten or less categories that are intended to include 311_teacher statements and actions. Such is Flanders' Interaction Analysis (Verduin, 1967, 33-38) which is divided into (1) teacher talk-direct influence, (2) teacher talk-indirect influence, (3) student talk and (4) silence or confusion with a total of ten subcategories. Those systems that do isolate specific components of the teaching act, such as the classification of teacher (reinforcing) 12 verbal interactions by Wasik, Senn, Welch and Cooper (1969, 183) into (l)positive, (2) neutral, (3) question, (4) redirection and (5) negative (with behavioral examples), still employ gross categories that give little specific indication of alternative behavioral action that could be taken once the data is compiled. Practitioner use and interpretation is diminished because the finite categories each contain a wide spectrum of unspecified behaviors. A practitioner without extensive consultation, verbal explanation and guidance is unable to employ present observation instruments and alter her behavior on a continuing basis; and one of the requisites for analyzing and changing teacher behaviors is that the teacher is able to select behaviors from known alternatives. (MacDonald, 1966, 3) Medley and Mitzel (1963, 298-303) make an important distinction between observation schedules that are category systems and those that are sign systems. They describe category systems as follows: The approach is to limit the observation to one segment or aspect of classroom behavior, deter- mine a convenient unit of behavior, and construct a finite set of categories into one and only one of which every unit observed can be classified. . . . It is supposed to be exhaustive of behaviors of the type recorded . . . . Category systems have been used more often in studies based on well-deveIOped or elaborated theories purporting to indicated specific behaviors to be looked for. . . . Medley and Mitzel then summarize their findings by stating the basic rules for constructing a category system (1963, 301): There should be a relatively small number of categories, each of which is used an appreciable number of timeswith some behavior or aspect of behavior that is relatively common. If convenient, the tallies should be based on natural units; if not, the tallies should be based on brief time units. The behavioral cues on which the discriminations 13 are to be made should be explicit even to the uninitiated, any discrimination should be easy for the observers to make. This is contrasted with sign systems which are characterized as follows: The . . . approach is to list beforehand a num- ber of specific acts or incidents of behavior which may or may not occur during a period of observation. The record will show which of these incidents occurred . . . and . . . how frequently each occurred (298-299) . . . The evolution of a sign begins with the idea that a certain behavior is symptomatic of some be- havior believed to be important (301) . . . A noteworthy feature of signs is that one re- corder can use a relatively large number of them simultaneously (301) . . . . A sign should pretty well define itself (302) . . . (by having) three characteristics: present tense, positive occurrence and singular number. In making this distinction between category systems and sign systems, the basic characteristics of the classroom obser- vation schedules are delimited. Most observation instruments have been category systems such as Flanders' (as described above). They are limited by definition in the number of classifications of behavior to be used (no more than ten are recommended). The sign systems have also been limited in practice in the number of categories used (see the Wasik, 32,21}, schema above). However, the listing of signs is potentially unlimited and thus offers greater possibilities for the generation of alternative behaviors with descriptive labeling. It is not a requirement of the sign system that all signs occur during a given observation period or in a specified setting. This makes it possible to gather information on what is not happening as well as what i§_happening in the classroom. Because of the condition that all signs do not have to occur in a 14 certain setting, some items can be drawn from behavioral research lit- erature rather than exclusively from classroom experience. Given the distinction between category and sign systems for classroom observation instruments, it is suggested that a sign system would be more suitable to the purpose of this study. This sign system should be developed, however, with many more alternatives and more explicit behavioral definitions than are presently employed in most instruments. The suggested procedure would give more explicit, descriptive alternatives for manipulations. The sign system also permits the analysis to be focused on one component of classroom functioning, teacher reinforcement behavior (which is the primary variable in the precision teaching model) rather than attempting to encompass all interaction as the category system does. Few sign systems exist to provide direction for this development. And even fewer instruments exist which separate a single construct or component in the teaching act such as reinforcement in a sign system (see Chapter II). In order to provide the needed instrumentation for precision teaching, therefore, certain departures must be made from present practice. First, an instrument must be developed to isolate the teacher reinforcement component, which has little precedent. Second, a sign system rather than a category system must be written to contain more numerous descriptive alternatives for analysis and manipulation than are presently being written into observation instruments. The categories must be specified sufficiently to achieve an acceptable level of observer agreement for successful testing of an observation instrument. To achieve an acceptable reliability 15 requires the inclusion of some characteristics of a category system in development: "The behavioral cues on which the discriminations are to be made should be made explicit even to the uninitiated. . ." (Medley and Mitzel, 1963, 301) Also, in order to facilitate teacher use, tallies should be based on natural units as suggested for a category system. This procedure should result in an instrument appropriate for introducing precision teaching in the classroom. 3. A third point should be considered in the development of this instrument: the criteria for selecting the types and numbers of reinforcement techniques that could or should be attempted in the classroom. One might argue that the selection should be based upon the research literature, i.e. precision teaching calls for the use of positive reinforcers to modify behavior. But this does not provide a basis upon which to select among positive reinforcers. Further, certain information-giving techniques, called feedback in the litera- ture (DeCecco, 1968, 253), must be included in any classroom interaction as reinforcers even though they are seldom considered truly positive in nature. Daley (Krumboltz and Thoreson, 1969, 42-45) has suggested giving each student a reinforcement menu from which to select a reinforcer when an appropriate behavior has occurred. But this conscious selection can only comprise a small portion of the reinforcers introduced into a classroom. A pattern of behavior must be ascertained that the practitioner could emulate or compare her performance to in order to make self-evaluations and corrective decisions. A strategy for establishing criterion models has been proposed (Shulman, 1970, 385-386): The strategy would involve initially identifying 16 criterial educators, who, like the native speaker of the language in linguistic studies, are taken to represent some standard of excellence as practitioner of the educational arts. Careful descriptive protocols of the criterial educators verbal and non-verbal behavior would then be gathered and, using behavioral equivalents of the linguistic rule discovery tactics, educational researchers would attempt to write a grammar of their teaching behaviors. This grammar would be a set of rules adequate to account for their functioning. I would hypothesize that once made explicit and cross-validated, such rules could be used to develop instructional procedures to help new students attain the criterial performer's level of competence. This strategy requires that competent teachers be identified and their reinforcement behaviors described as a criterion for practi- tioner performance. A method has been researched (Henderson, 1968) for selecting competent teachers (see Chapter III). This method should be used to identify competent teachers (in a specified setting). The class- room actions of the "criterial performers" should then be described according to the teacher reinforcement behavior instrument. This would provide the interested practitioner with a comparative measure for self-evaluation. To summarize, three specific changes are prOposed in the development of a classroom observation instrument: First, instru- ments should be applicable to a given instructional strategy. Second, a sign system should be developed with more definitive, descriptive items than presently exist. And third, competent teachers should be identified and their behavior described quantita-. tively using the teacher reinforcement behavior instrument to serve as a criterion for the practitioner. 17 OBJECTIVES OF THE STUDY Two major problems are apparent in this introduction. First, precision teaching offers substantial assistance to teachers as they attempt to bring about intended learning in students. But instrumenta- tion is lacking for precision teaching to be generally adopted in the schools. Second, classroom observation instruments, appropriate instrumentation for precision teaching, have become highly sophisticated as research tools. But these instruments have not been designed within an instructional model with (multiple) definitive, descriptive items and identification of criterion performances to hasten their use and effec- tiveness in the classroom. Therefore, the development and testing of an observation instrument to categorize and quantify teacher reinforce- ment behaviors is needed to provide instrumentation central to the adoption of precision teaching in the classroom and, simultaneously, provide a model for the deve10pment of additional instruments incor— porating the required characteristics. It is the purpose of this research to develop a classroom observation schedule to categorize and analyze teacher reinforcement behavior according to the criteria established above. This research investigates in depth the feasibility, reliability and optimal procedures for designing, testing and implementing such an instrument. THERE ARE TWO IMMEDIATE OBJECTIVES OF THIS RESEARCH. THE FIRST IS TO DESIGN A MULTI-SIGN SYSTEM WITH SUFFICIENT OBSERVER AGREEMENT AND STABILITY TO ESTABLISH AN ACCEPTABLE LEVEL OF RELIABILITY. Re- liability is synonymous with dependability, consistency, accuracy and objectivity. As defined by Kerlinger (1965, 430), one approach to 18 reliability is to ask the question: "If we measure the same set of objects again and again with the same or comparable measuring instruments will we get the same or similar results?" Reliability is the accuracy or precision of a measuring instrument. Therefore, one criteria for reliability in this instance is the percentage of observer agreement in the recording of teacher behavior, using the designated instrument. Further, a stable record of teacher performance should be obtained for reliability: the frequency distributions for each teacher should not vary significantly over time. THE SECOND OBJECTIVE IS TO DESIGN A MULTI-SIGN REINFORCEMENT INSTRUMENT WHICH WILL GAIN SUFFICIENT DATA ON AND REFLECT A SIGNIFICANT DIFFERENCE BETWEEN THE REPERTOIRES OF THE SELECTED CRITERIAL PERFORMERS TO ESTABLISH INSTRUMENT VALIDITY. The criteria for validity are defined by Medley and Mitzel in Measuring Classroom Behavior by Systematic Observation (1963, 250): The validity of measurements of behavior . . . depends, then, on the fulfillment of three conditions: (1) A representative sample of the behaviors to be measured must be observed. (2) An accurate record of the observed behaviors must be obtained. (3) The records must be scored so as to faithfully reflect differences in behavior. Medley further defines validity in a later article (Medley in Honig- man, 1967, 24-25): A schedule may be said to be valid to the extent that records obtained with it yield an amount of information about teacher behavior which justifies the cost of collecting the data. . . . The need for validity suggests that the one which relates most meaningfully to teacher effectiveness is the best -- that is, the set of categories in terms of which the changes in behavior a teacher needs to make can be described most intelligibly and economically. 19 In the following chapters is a detailed description of the study that was carried out to achieve these objectives. A review of the literature was made to establish the validity of the pre- cision teaching model and state a behavioral definition of reinforce- ment and its functional components. Also, the interaction analysis literature relevant to the development and testing of the instrument was reviewed. This is reported in Chapter II. An observation schedule was then compiled and synthesized from the descriptive analysis of video tape recordings of teachers in the Lansing, Michigan, Public Schools, a review of the category schedules and suggested reinforcement techniques in the literature and anecdotal records made by the author in Oakland County, Michigan,c1assrooms. Behavioral definitions and examples were provided for each category and subscript. This is reported in Chapter III. The training of the observers and the tests for observer agreement are also reported in Chapter III. A sample of ten competent teachers in five elementary schools in Royal Oak, Michigan, were subsequently selected for observation as criterial performers. The teachers were each observed for ten, randomly selected, one-half hour periods over three weeks in April, 1970, by trained observers who were experienced teachers. Checks for observer agreement were made during 20% of the observation periods. Analysis of the distributions derived was performed through Gerig's Multivariate Extension of the Friedman Test, Kendall's Coefficient of Concordance and percentage comparisons. Population and sample selection of the criterial performers, and the recording of the data in the classroom are included in Chapter III. The summary of the data and the data analysis are incorporated into Chapter IV. Chapter 20 V discusses the inplications of the study for teacher education. CHAPTER II REVIEW OF RELATED RESEARCH The literature in threeemajor areas of educational research have direct bearing upon the deve10pment of a teacher reinforcement observation schedule: applied behavior analysis, interaction analysis and reinforcement research. First, applied behavior analysis as it applies to the efficacy of the precision teaching model will be reviewed to provide support for the position that positive conditioning techniques can be effectively employed in the classroom. Second, precedents and procedures for the development of a teacher reinforcement observation schedule will be cited from interaction analysis research. And, finally, the literature on reinforcement will be summarized to provide an operational definition and functional components for instrument development. APPLIED BEHAVIOR ANALYSIS RESEARCH Today the professional educator is placing greater emphasis on the emotional development of the student than ever before. Bodies of research are being accumulated that give ever-growing support to the conclusion that intellectual learning is dependent upon emotional stability and well-being. Yet to grapple with the concept of emotional stability and well-being is like trying to alter a "sea of putty" for the classroom teacher: its dimensions are nebulous and its content variables infinitely flexible. The clinical psychologist and 21 22 psychiatrist have effectively practiced in this area by training for years to develop constructs that can be manipulated on a one- to-one basis to achieve an acceptable level of emotional stability in their clients. However, for the professional educator, this set of circumstances is an impossibility. There are several reasons (Clarizio and Yelon, 1967): 1. The teacher is not trained to probe the causes of behavior. 2. Most teachers have control only over the school environment and cannot manipulate causes that may extend or originate outside the school. 3. Many maladaptive symptoms may continue to persist after the causes are eliminated. 4. Being responsible for thirty pupils simultaneously, the teacher must deal immediately with inappropriate behavior in order to establish or maintain a stable classroom environment in which all thirty students can function optimally. Thus the teacher must employ techniques that can be utilized and controlled in the immediate environment to modify student behavior. Those studies reported follow certain steps: (1) identifi- cation of an inappropriate behavior and the alternative behavior desired in the student: (2) assessment of the baserate of student behavior; (3) systematic application by the change agent of a pre- scribed operant conditioning technique; (4) evaluation of student behavior with periodic adjustments in the treatment if necessary. This makes good sense, as Bandura (1969, 16-17) explains: If progress in the understanding of human behavior is to be accelerated, psychological theories must be judged by their predictive power and by the efficacy of the be- havioral modification procedure that they produce . . . . There is little to be gained from condemning delinquents for their history of anti-social behavior, but there is much to be gained from having them experience new response consequences that will help them develop a more effective way of life. 23 Bandura presents evidence (1969, 118-624) that cognitive and affective modifications can be achieved more successfully through planned behavioral change than through attempts to alter internal events directly. Skinner provides evidence (1968, 8—17) for the position that the contingencies of reinforcement can be effectively manipulated in the classroom and cites several areas where improvements can be made. The basis for this evidence, though, is consistently derived from the laboratory. Recent studies in behavior modification have moved from the laboratory to the natural setting, thus providing more apprOpriate evidence for the educator on the use of reinforcement in the classroom. The principles utilized are extensions and applications of earlier research in the techniques of operant conditioning (Skinner, 1953). In this application to the natural setting, it has been found (Bijou, 1970) that psychological research can now offer education . . ."a set of concepts and principles derived exclusively from experimental research, a methodology for applying these concepts and principles directly to teaching practices, a research design which deals with changes in the individual child (rather than inferring them from group averages) and a philOSOphy of science which insists on observable accounts of relationships between individual behavior and its deter- mining conditions." In moving to the natural settings, these concepts and princi- Plsas were initially applied in institutions and special classrooms ‘flheere the variables could most easily be controlled and recorded. Yet: it is important to emphasize that the behavior modification teCdxniques in the studies cited were applied while the c1ient(s) 24 remained within the group setting. He was not singled out for treat— ment on a one-to-one basis but functioned in the natural setting. In many cases more than one person (and, occasionally, the whole group) was treated simultaneously. This obviously provides precedent for the classroom teacher. The attentional deficits and disruptive behavior of a class of retarded boys was reduced through praise and tokens (Zimmerman, Zimmerman and Russell, 1969). Praise and candy evoked more cooperative play from two retarded boys in their play groups (Redd, 1969). In schools for the deaf, appropriate visual attending (Craig and Holland, 1970) and remaining seated in the classroom (Osborne, 1969) were increased through food and tokens and additional free time from school work. In institutional settings, several types of undesirable behaviors were modified. Token procedures were used to reduce the excessive screaming and violence, apathy and general lack of response of psychiatric ward patients (Winkler, 1970). Barton, Guess, Garcia and Baer (1970) modified undesirable mealtime behaviors of a hospital cottage of retardates through contingent timeout procedures. Prompts and cigarettes were employed to produce increases in the rates of social greetings of chronic schiZOphrenics (Kale, Kaye, Whelan and Hopkins, 1968). And brief timeout for disruptive and aggressive behaviors and reinforce- ment for appropriate behaviors were successfully used with patients in a state hospital ward (Bostow and Bailey, 1969). The undesirable behavior of adolescent "pre-delinquents" was also modified through reinforcement procedures. Aggressive statements and: poor grammar were replaced by tidiness, punctuality and increased 25 homework completion in a home-style rehabiliation setting when the boys were given points toward home visitation, watching TV and riding bicycles (Phillips, 1968). In a subsequent study, home privileges contingent upon appropriate study behavior and obeying class rules also altered specified delinquent behaviors (Bailey, Wolf and Phillips, 1970). More closely approximating the circumstances of the public and private school classroom and, thus,providing more adequate evidence for modeling, are the series of studies performed in the pre-school. Teacher social reinforcement, such as praise and attention (Buell, Stoddard, Harris and Baer, 1968), increased a 3 year old's use of play equipment (which was intended as a contribution to her deficient motor skills), improved the c00perative play of a 5 year old (Hart, Reynolds, Baer, Brawley and Harris, 1968), produced a high frequency of verbal behavior in a previously non-verbal four year old (Reynolds and Risley, 1968) and modified the gross behavior disorders in a group of nursery school children (Harris, Wolf and Beer, 1964). Kirby and Toler (1970) induced a pre-school boy with a low rate of interaction with his class- mates to pass out candy as a tactic to increase his rate of interaction with them. Tokens exchangeable for a variety of play activities and snacks were effective in eliciting desired handwriting skills in a Head Start program (Miller and Schneider, 1970.) In another pre-school classroom tokens were acquired by engaging in a variety of study behaviors (Bushell, Wrobel and Michaelis, 1968). The study behavior of each child was altered in the same direction, although at different magnitudes. Risley and Hart (1968) found that correspondence could be developed between children's non—verbal and verbal behavior such that 26 their non-verbal behavior could be altered simply by reinforcing through food snacks the children's verbal reports of their non-verbal class activities. In a related study Hart and Risley (1968) utilized several reinforcing techniques to induce disadvantaged pre-schoolers to use and acquire descriptive adjectives. Access to pre-school materials was found to be the most powerful reinforcer. The main research, however, that must be perused for evidence of reinforcement practices in the classroom is centered at the elementary and secondary level. Based upon the previous experiences in institu- tions, special classrooms and pre-schools, more and more attention is being focused on the "normal" classroom. Many of the studies at the elementary level are focused on the disruptive behavior of selected students. Yet all are manipulated within the class setting. Coleman (1970) developed a procedure for use in a public school classroom where only one of the children needed treatment, sophisticated apparatus was not feasible, personnel were untrained in conditioning techniques, and where disruption had to be minimized. Candy reinforcers were contingent upon the working behavior of each of four children and the candy divided equally among the class. The face-touching, poor posture and voice-loudness of a sixth grade girl were modified through tokens (exchangeable for bracelets, pins, dresses and hairstyling) given during the after-school viewing of video tapes of the child's behavior (Schwarz and Hawkins, 1970.) Broden, Bruce, Michell, Carter and Hall (1970) observed significant increases in the attending behavior of two boys seated at adjacent desks when the teacher systematically increased the amount of attention for appropriate attending behavior. Earning token points toward a model increased the 27 attending behavior of an underachieving nine year old male subject (Walker and Buckley, 1968). Teacher attention following study behavior and the ignoring of disruptive and dawdling behavior resulted in sharply increased study rates for one first grade and five third grade pupils in an investigation by Hall, Lund and Jackson (1968). The disruptive classroom behavior of four first grade children was reduced through the systematic use of teacher attention and positive comments while general reprimand and negative comments were controlled (Ward and Baker, 1968). The non-study behavior of our first grade public school students was successfully modified through the dispensing of tokens for special activities contingent upon appropriate behavior (Surratt, Ulrich and Hawkins, 1969). Also, the academic achievement of the fifth grade student allowed to dispense the reinforcers improved significantly when his participation in the study was made dependent upon this improvement. The appropriate classroom behavior of two second grade girls and a kindergarten boy (Abbott, 1969) was improved through the contingent use of teacher verbal interaction, withholding of social reinforcement and timeout from social reinforcement (Wasik, Senn, Welch and Cooper, 1969). Meacham and Weisen (1969, 147) report the increased correct word pronunciation in a graded word list through teacher manipulation of approving, disapproving and instructional statements for an elementary school girl. Finally, a child with chronic misbehavior problems was sent home according to the contract he had made, termed "systematic exclusion," whenever he got out of his seat, talked etC. (Keirsey, 1969, 89-113). Through this system, his behavior was remarkably modified. Behavior modification and reinforcement techniques were also 28 employed with success when the target behaviors were group functions. Madsen, Becker and Thomas (1968), in a model study, found that ignor- ing inappropriate behavior and showing approval for appropriate behavior were very effective in achieving better classroom behavior. In a subsequent study (Thomas, Becker and Armstrong, 1968) the effects of teacher behaviors on the classroom behaviors of children were investigated by systematically varying approving and disapproving classes of teacher behavior. The subjects were 28 students in a middle-primary school class. It was demonstrated that approving teacher responses served a positive reinforcing function in maintaining appr0priate classroom behaviors. Disruptive behaviors increased each time approving teacher behavior was withdrawn. When the teacher's disapproving behaviors were tripled, increases appeared most markedly in gross motor and noise-making categories of disruptive behavior. The findings indicate the importance of the teacher in producing, maintaining and eliminating disruptive as well as pro-social classroom behavior. A token reinforcement program reduced the frequency of dis- ruptive behaviors in seven second grade children in one class, while a combination of rules, educational structure, praise and ignoring only eliminated the disruptive behavior of one child in a study conducted by O'Leary, Becker, Evans and Saudargas (1969). The "paying—attention behavior" in four elementary school classrooms was gradually increased through group contingencies and token-mediated reinforcement by Packard (1970). Excessive sound-intensity levels and, subsequently, out-of—seat behavior were suppressed in a regular public school classroom through ihiditional gym period time and 2—minute breaks (Schmidt and Ulrich, 196%”. And, finally, out-of—seat and talking-out behaviors were reduced 29 in a fourth grade class through the acquisition and loss of special privileges (Barrish, Saunders and Wolf, 1969), while the mean percentage of instructions followed by a kindergarten class increased to 78% when the teacher began attending to each child if an instruction was followed (Schutte and Hopkins, 1970). At the secondary level, the variables are the most difficult to manipulate since students are in one class (usually) for a maximum of an hour per day. Yet studies have reported successful results and offer appropriate designs that could be followed. Arrangements between teacher, student and counselor have been used to mediate individual problems. Cantrell, Cantrell, Huddleston and Woolridge (1969) prepared written contracts specifying ways in which the child could obtain exist- ing individualized reinforcers contingent upon approximations to desired appropriate behaviors chosen as incompatible with the referral problem behaviors. Initial results have been encouraging. A counselor was also able to establish (Castle, 1969, 33-36) a successful praise-and- ignoring program with a seventh grade boy's teachers to modify the wearing of bizarre hats, using baby talk, not completing assignments and being tardy to class. Two studies have been reported on the modifi- cation of the behavior of an entire class. The learning of history and geography material in four ninth grade classrooms was significantly improved through a self-determined token reinforcement program (Glynn, 1970). And the effects of teacher praise and disapproval on inapprOpriate talking and turning around were investigated in a high school English Class (McAllister, Stachowiak, Baer and Conderman, 1969). The results ixuiicate that the combination of disapproval for the target behaviors and.praise for appropriate incompatible behaviors substantially reduced 30 the incidence of the target behaviors. Each of these studies in applied behavior analysis provides support for the conclusion that operant conditioning techniques can be employed in the classroom. In each setting, institutional, special education, pre-school, elementary and secondary, systematic investigations were conducted that reduced Specified undesirable behavior, primarily through the contingent application of teacher interaction such as praise and attention, food, tokens and special privileges. In several studies reinforcement was administered outside the classroom. The main conclusion to be drawn from these studies is that ample data is available to illustrate the effectiveness of positive reinforcers in altering student classroom behavior. Coupled with the ignoring of inappropriate behaviors or timeout procedures, these techniques negate the necessity of aversive stimuli in the classroom. PRESENT SCHOOL PRACTICES AND PROBLEMS Despite the evidence presented in the applied behavior analysis studies, Skinner contends that aversive stimuli still dominate the class— room (Skinner, 1968, 15, 93-114): It was a part of the reform movement known as progressive education to make the positive consequences more immediately effective. But any one who visits the lower grades of the average school today will observe that a change has been made not from aversive to positive control, but from one form of aversive stimulation to another. The question must then be asked: Why does this continue to be true in.the face of overwhelming research evidence indicating that the (Rbntrary is more effective? Initial objections to reinforcement Inesearch were based upon the premise that "rat research, coming 31 primarily from animals in the laboratory, was insufficient to be generalized to the classrooms which have human beings in a unique social setting. But how can these objections continue now that data from the classroom is available? It appears that there is objection to the employment of certain forms of extrinsic reinforcement on the grounds that the student will become dependent on them and will be unable to perform without them (Glynn, 1970, 123). Some support can be given to this argument since reversal to the baserate of behavior when treatment ended occurred in several studies (Keirsey, 1969, 89-113). Yet sufficient data is available to indicate that consistent use of a wide range of positive reinforcers should produce significant student behavior modification. But the problem is one of providing the teacher with a sufficient repertoire of reinforcers from which can be selected those most appropriate to a specific teacher style and classroom setting. A second problem also exists. It is derived from the response many teacher have to a description of an applied behavior analysis study: "Why, what is new about that? I've been doing these things in my classroom for years." (Hanley, 1970,597-598). Homme and Tosti (1964, 4), though, succinctly identify the difficulty: The difficulty can be primarily traced to the failure to systematically (emphasis added) apply what is known. It is not only that operant principles are not systematically applied, they are, if applied at all, sporadically applied. Yet, it has been demonstrated (Hall, Panyon, Rabon and Broden, 1968; Rosenthal, Underwood and Martin, 1969; CoOper, Thomson and Baer, 1970; Hosford, 1969, 152-154) that teachers can be trained to systematically thee positive reinforcement practices in the classroom. The problem 32 appears to be in developing a technique whereby teachers can select from a wide range of reinforcers and then monitor their use in a systematic fashion if the results of operant conditioning research are to reach a large number of classrooms. In addition, given present school circumstances, a technique would seem to be needed that requires little expense, a minimum of disruption to the students and can be utilized by_the teacher after brief in-service training. The technique that most nearly suits these requirements while providing data on systematic application of teacher behaviors in the classroom and a repertoire of alternative teacher behaviors is interaction analysis. The interaction analysis technique offers a technology and potential to implement positive reinforcement strategies in the classroom. INTERACTION ANALYSIS SYSTEMS Interaction analysis systems are described (Simon and Boyer, Interaction analysis systems are "shorthand" methods for collecting observable objective data about the way peOple talk and act. They make possible a relatively simple record of what is happening but they do not record what is being said. . . . They differ from each other in a variety of ways, but all of them code some aspect of behavior. These systems are made up of sets of categories of be- haviors . . . . Most of these systems are "content free," that is, they can be used with any subject matter or grade level. Thus, interaction analysis is a coding system for objectively quantifying a certain dimension of classroom behavior into behaviorally defined categories. This quantifying is done by an independent observer trained to identify and categorize reliably the specified 33 behaviors. What results is a description of classroom transactions in the form of a distribution of tallies in the pre-determined categories (Medley and Mitzel, 1963, 253). For the purposes of this investigation, the data can be used to assist a teacher in altering and/or expanding her behavior repertoire by providing an objective description of the teacher's baserate behavior. The description can then be used to define intended changes to be instituted to conform with the operant conditioning model in terms of the behavioral definitions available. The analysis system can then be used to monitor the accomplishment of the stated objectives. During the past ten years, much work has been done in the development of interaction analysis systems, but certain additional modifications must be made in order to suit the purposes of this study. As discussed in Chapter 1, these modifications include: 1. An instrument should be designed that specifically deals with a given instructional strategy (precision teaching). 2. The instrument should provide data that is inter- pretable and manipulable by the classroom teacher. 3. A sign system should be developed with a large number of categories to provide more numerous descriptive alternatives. 4. Items should be derived from both the laboratory and the classroom to provide options not only from what i§_presently occurring in the schools, but also what could be occurring. Several category systems have been devised that deal with certain aspects of reinforcement. Dunn and Rankin (Cartwright and Cartwright, 1969, 2) delimited five categories in a reward preference profile: (1) adult approval; (2) competition; (3) consumable; (4) peer approval; (5) independence. A categorization of rewards to be earned 34 was also explored by Clarizio (undated) which considered (1) tangible rewards; (2) social rewards; (3) responsibility and authority; (4) intrinsic; (5) activities. Zahorik (1969) authored a Teacher Verbal Feedback instrument utilizing twelve categories: (1) praise-confirmation; (2) reproof— denial; (3) positive answer; (4) negative answer; (5) positive explanation; (6) negative eXplanation; (7) response extension: development; (8) response extension: improvement; (9)solicitation repetition: several answers; (10)solicitation repetition: one answer; (11) lesson progression; different topic; (12) miscellaneous feedback. A feedback categorization was also devised by Yelon (1969, 38) which includes: (1) action feedback; (2) learning feedback; (3) intrinsic feedback; (4) augmented feedback. Silberman (1969, 403) designed a four category recording system: (1) teacher contact; (2) positive evaluation; (3) negative evaluation; (4) acquiescence. In another study (Rosenthal, Underwood and Martin, 1969, 371-372), a special recording system was designed to consider certain ethnic needs: (1) verbal approval; (2) gestural approval; (3) physical approval; (4) verbal disapproval; (5) gestural disapproval; (6) physical disapproval; (7) non-interpersonal rewards; (8) sanctions or punishments; (9) verbal solicitation; (10) gestural solicitation; (11) physical solicitation. It was also recorded whether the teacher or the student initiated the response sequence. They reported (Ibid., 407) that, ". . . the pragmatic merits of the method appear to be its strongest recommendation: it required (little time) and no elaborate apparatus; it did not appear to disturb the students or disrupt the teachers who were visited." 35 Madsen, Becker and Thomas's study also uses a reinforcement ana- lysis sfstem (1968, 143) with nine definitive categories: (1) contact; (2) praise; (3) facial attention (approval); (4) holding child; (5) criticism; (6) threats; (7) facial attention (disapproval); (8) time- out procedures; (9) academic recognition. However, each of these category systems is deficient according to one or more specific criteria previously established. The Dunn- Rankin preference profile and the Clarizio categorization focus only on rewards with no information to be gleaned on deterrents functioning in the environment. Zahorik's feedback instrument records only verbal academic feedback, eliminating the potential for recording affective and/or non-verbal behaviors. Further, this technique requires transcripts for analysis which adds considerable cost to any study. Yelon's also includes only feedback, not rewards or deterrents. Silberman's and Madson-Becker—Thomas's systems are too limiting given the objective of expanding teacher behavior repertoires, while the Rosenthal, Under- wood and Martin technique is situation specific. Given the deficiencies of presently existing interaction analysis instruments, a new instrument must be designed which represents a synthesis of previous instruments with corrections made for deficiencies. Thus, the instruments cited can provide the basis for an item pool to be used in compiling a new, expanded instrument. Items and developmental techniques can also be gained from several widely-used instruments which consider, in part, reinforcement techniques. In a highly complex system, Spaulding (1970, 11-12) considers(in the general transaction and social behavior categories) several reinforcers: (l) approval; (2) disapproval; (3) structuring; 36 (4) restructuring; (5) information; (6) listening and observing; (7) timeout; (8) withholds privilege. In The Language 9£_the Classroom (Bellack and Davitz, 1963, 172-173), Bellack includes definitive classifications for (l) substan— tive reactions, (2) rating reactions and (3) procedural reactions which are coded from transcripts. Similarly, Joyce and Harootunian consider reward and punishment in the application of sanctions and the development of procedures. Positive, negative and control functions are included in the Hughes system (1959). In Amidon's Modified Category System (Simon and Boyer, 1967, Amidon-S), categories are included for: (l) accepts feelings; (2) praises; (3) accepts ideas; (4) gives directions; (5) criticizes. The Multidimensional Analysis of Classroom Interaction (Simon and Boyer, 1967, Honigman—4) has categories for: (l) performs emotionally-supportive behavior; (2) designates student performance "acceptable;" (3) praises performance; (4) designates student performance unacceptable; (5) crit— icizes; (6) uses student's ideas; (7) uses student's emotional contribu- tions. In the most frequently used of all interaction analysis systems, Ned Flanders (Medley and Mitzel, 1963, 273) included four specific reinforcing categories: (1) accepts feelings; (2) praises/encourages; (3) accepts/uses ideas; (4) criticizes/justifies. Each of these systems provides some direction in the deve10pment of a new system to categorize and analyze teacher reinforcement behavior. Yet none in- cludes the full, definitive spectrum of potential teacher reinforcement behaviors. Before items can be selected, however, from the item pool for potential inclusion, a definition of reinforcement and its major 37 components must be delimited and defined. A DEFINITION OF REINFORCEMENT AND ITS COMPONENTS Reinforcement was first described in 1911 as the Law of Effect by Thorndike (DeCecco, 1968, 250). Since that time much research has been done on the implications of reinforcement for animal behavior, including man. There have also been many efforts to define reinforcement in order to assist research and practice. Deese (1952, 16) defined reinforcement as " . . . any stimuli event that will increase or maintain the strength of a response or stimulus-response connection associated with it." Bigge (1964, 94) described reinforcement as " . . . a special kind or aspect of conditioning within which the tendency for a stimulus to evoke a response on subsequent occasions is increased by reduction of a need or of a drive stimulus." He further explained (1964, 125) that, "Reinforcement is changing (increasing) the future probability of a response in the same class." Miller and Dollard (1941, 29) defined reinforcement similarly: "Rewards (reinforcement) are events producing a reduction in drive . . . any event known to strengthen stimulus - response connections." Clarizio and Yelon (1967, 269) defined reinforcement as: "Whatever serves to maintain the occurrence or increase the strength of a response." In each of these definitions it is implied or clearly stated that reinforcement is synonymous with the receiving of a reward or pleasurable stimuli following a response. Yet there are more subtle distinctions made in a complete conceptualization of reinforcement. Yelon (1969, 104) most aptly 38 describes three main reinforcement contingencies: positive reinfor- cement, negative reinforcement and punishment. Positive reinforcement is the presentation of attractive stimulus and negative reinforcement is the withdrawal of an aversive stimulus. Punishment is either the withdrawal of an attractive stimulus or the presentation of an aversive stimulus. These dis- tinctions are, in this author's experience, more in keeping with the practitioner's conceptualization of reinforcement as it functions in the classroom. Since this instrument is being designed for practitioner use, definitions should be apprOpriate to their under- standing and use of the term, as well as reflecting research findings. In order to include positive and negative reinforcement as well as punishment in the working definition, reinforcement contingencies are defined for this instrument as: Any teacher action or verbal behavior that alters the future probability of a response in the same class, be either strengthening or weakening the stimulus—response connections. Two major components of reinforcement contingencies are obvious for purposes of further categorization. First, positive reinforcement can be described as rewards, a more commonly used term in practitioner terminology. In keeping with existing definitions, rewards can be defined behaviorally as any pleasurable response or activity given to the student to indicate the desirable nature of a behavior, per- formance or action. Punishment is a second component. But because this term is frowned upon by many practitioners and because it excludes a number of techniques used to weaken certain behavior, a more amenable term, deterrents, has been selected. Deterrents are defined as any discouraging response or activity given to the student to indicate 39 the undesirable nature of a behavior, performance or action. Therefore, reinforcement contingencies are broken down into two major parts for categorization: rewards and deterrents. However, a third component of reinforcement as defined above cannot be overlooked, although it is not often included in formal definitions. This is feedback. DeCecco (1968, 253) stressed the inclusion of feedback: In this procedure the student is given knowledge of his correct responses. Whereas the term reinfor- cement connotes the hedonic aspect of reward, the term feedback stresses the informational aspect of the teacher's function. . . . As yet we have no way of separating the reward function from the informa- tional function. Feedback is generally considered to be the giving of neutral responses as to the correctness or incorrectness of a student's academic action. Since feedback is not usually associated with either pleasant or aversive stimuli, it is not often considered a part of reinforcement. Yet, feedback is being given constantly in any classroom for the purposes of evaluation and management as any classroom teacher knows. The cumulative effect of feedback can have a greater impact than a single reward or deterrent. Zahorik (1969) recognized this when he constructed an interaction instrument for teacher verbal feedback and content development to analyze classroom transcripts. Yelon (1969, 37-38) recognized the role of feedback in school learning also when he enumerated several types of feedback that could be used in evaluation. This vital third component of reinforcement, feedback, is thus included and defined as: giving neutral responses to indicate academic correctness; information 40 function; those verbal statements or behaviors which directly or indirectly reflect on adequacy, appropriateness or correctness of pupil statements in relation to academic subject matter development or classroom management. Given the conclusions, implications and guidelines derived from previous research in applied behavior analysis, interaction analysis and reinforcement, the development and testing of a teacher reinforcement observation schedule was able to proceed. The categories were developed under the three major heading of rewards, feedback and deterrents. cf CHAPTER III PROCEDURES AND METHODOLOGY Described in this chapter are the procedural steps taken to design and test the reliability and validity of a teacher reinforce- ment behavior observation schedule. Also discussed is the identi- fication of "criterial performers." A rationale for the procedures accompanies each section. The investigation was carried out in two phases in order to fulfill the stated objectives. Initially, the instrument was developed from video tapes, classroom visits and research descriptions. This instrument was then refined and tested for observer agreement. The second phase was the selection of a sample of "criterial perform- ers" for systematic observation and the subsequent observation of these teachers. The entire study took place during the academic school year of 1969-1970. DEVELOPMENT OF THE INSTRUMENT In the spring of 1968, Judith Henderson conducted a study (Henderson, 1968) in which twelve competent language arts teachers were identified by their peers, supervisors and administrators through a rating technique. These Lansing, Michigan, teachers at the first, eighth and eleventh grade levels, both inner city and suburban, were subsequently video taped three times over an eight week period. These tapes were utilized in the initial descriptions of teacher 41 42 reinforcement behavior. Every tape from the Henderson study was viewed. Each reinforce- ment behavior displayed was tentatively labeled and briefly described in a running written record. Because the classes were on video tape, the action was frequently stopped, replayed and discussed in order to develop more accurate descriptions. The individual behaviors were then sorted into rewards, feedback and deterrents. In order to get additional behavior descriptions from a normal distribution of classrooms, since an interaction analysis system is designed as "content free" (Simon and Boyer, 1967, l), the author visited classrooms at all grade levels and subjects, keeping the same running record for subsequent sorting. Finally, useful labels and descriptions were derived from the literature to add to the item pool. The Learning Systems Institute at Michigan State University developed an Influence Techniques Instrument (1967, mimeo). The Dunn-Rankin reward preference inventory (Cartwright and Cartwright, 1969) listed adult approval, competition, consumable, peer approval and independence. Clarizio generated a list of rewards to be earned that included tangible rewards, social rewards, responsibility and authority, intrinsic rewards and activities. Madsen, Becker and Thomas (1968, 143) devised an analysis system that included rewarding actions such as contact (teacher), praise and facial attention and deterrents including threats, criticism, facial attention and time-out proce- dures. Silberman (1969, 403-404) suggested acquiescence, defined as granting appeals for assistance, permission or information, as a reward in addition to positive evaluation. 43 In considering feedback, Yelon suggested a four category division (1969, 38): action feedback, learning feedback, intrinsic feedback and augmented feedback. A further feedback item, teacher imposition,was suggested by Hughes (1959). Bandura's work (1969) provided an additional item labeled modeling, defined by Clarizio and Yelon (1967, 269) as a condition where the behavior to be acquired is demonstrated for the learner. Premack's principle as described by Bandura (1969, 229) was added to the item pool as contingency management. Finally, Ausubel's "advance organizer" was included. Deterrent items suggested in the literature included physical restraint, isolation, extra homework and grades (Hughes, 1959) and redirects, defined as any statement with which the teacher attempts to redirect a student from an inappropriate to an appropriate behavior. Further, work by Flanders (1970) and Rosenthal, Underwood and Martin (1969, 371) suggested that valuable data could be gained by recording the response sequences as teacher-initiated or student initiated. The items from these three sources, video tapes, live classroom observation and previous research, were then sorted into major categories, subscripts and examples. The criteria designated for the development of a multi-sign instrument, particularly the listing of multiple, understandable categories with explicit definitions and examples, was followed. The resulting items are as follows (see the complete observation schedule in the appendices): l. 44 Content/Substantive Reaction . Adds new information . Clarifies or synthesizes . Gives answer . Corrects or tells what should have been done Elicits Response Elaboration com» A. Developmental questioning or elaboration B. Assists in developing plans or procedures C. Opens new possibilities/sug- gests alternatives/poses problem Evaluates A. Positive rating B. Negative rating C. Implies accuracy of response by lesson progression or use of response D. Repeats or withholds rating E. Explanation/rationale REWARDS 1. Tangible Rewards 7. A. Consumable B. Token C. Use something of teacher's/ manipulable 2. Special Privilege or Activity A. Activity 8. B. Teacher Concern 3. Affection 4. Signal Approval A. Facial B. Gestural 9. 5. Praise/Encouragement A. Positively evaluates deportment/ideas/response/ task B. Challenge or pep talk 10. C. Recognition from authority figures 6. Verbal Acquiescence 11. A. Reacts to or acknowledges the students comments or appeals 12. B. Accepts or utilizes the students's ideas or eXperiences FEEDBACK 8. 9. 10. 11. Intrinsic Rewards A. Accomplishment/mastery of the task/completion of academic activities B. Gaming C. Removes evaluation sanctions Competition A. Measures or models for others or against an absolute B. Reads or posts student's work as best Independence A. Given responsibility or authority B. Given choice or selects activity Incentive/Conditional Promise A. Teacher assistance B. Reward for performing task Peer Approval A° Class selection/recognition B. Praise or cooperation Unclear/Muddled Structuring/Action Feedback A. Gives clue to expected response set B. Regulation of details C. Advance organizer D. Contingency management Modeling/Response Facilitation A. Task B. Behavior C. Enthusiasm for task or content Teacher Imposition A. Performs task for student B. Relates experiences/moralizes C. Admits mistake or apologizes Change Environment A. Change goal B. Change content C. Change procedures Augmented Feedback Testing Delays Feedback or Promises Unclear/Muddled 45 DETERRENTS l. Punishment 4. Threatens A. Extra tasks 5. Time-out Procedure B. Lowering grades A. Waits/lights out C. Reordering environment B. Keep in at recess or D. Physical restraint/corporal after school punishment C. Deprives of social C. Deprives of privilege interaction 2. Verbal Deterrents D. Withholds positive A. Criticism/sarcasm reinforcement B. Reprimands or negates 6. Redirects 3. Signal Deterrent A. Poses question to student A. Facial B. Reminds of attending B. Gestural behavior or deportment C. Proximity control 7. Peer Disapproval 8. Unclear/Muddled Also included were: No Response and Miscellaneous Reviews of the video tapes and additional classroom visits were then made by the author to determine that all reinforcement behaviors observed in the classroom could be tallied within the category system. It was also determined that judgments could be made as to the source of the response sequence (teacher or student). An additional decision was made at this time concerning coding. Many elaborate techniques have been designed for coding according to the dominant action in a specified time unit. Also combinations of numbers and letters and even plus and minus are used. However, as Simon and Boyer state (1967, 12): "Coding complexity is one of the reasons why the use of many of these systems has never extended beyond the researcher who developed them." Since one of the objectives of this investigation is to design an instrument for teacher use, it was decided that the natural reponse sequence, defined as any single teacher reaction to a student statement or action g£_any teacher be- havior giving the student information about the accuracy or value of a 46 future response, would be the unit of analysis. This means that each time the teacher responds to a student question or comment or gives a set of instructions, etc., it is recorded in the apprOpriate category, regardless of the time involved. Any complete thought or message, whether it is one sentence or a discussion of a few minutes duration, receives one tally. The only requisite is that the response be a message, thought or instruction conveying a single idea. If, in the middle of a response however, the teacher incorporates a second type of reinforcer, it should be tallied separately. A simple tally or slash would indicate a teacher-initiated response sequence and an abbreviated S or curved line would designate a student initiated response. THE SELECTION AND TRAINING OF OBSERVERS Again, since the instrument is primarily intended for inservice teacher training, experienced teachers were recruited as observers. One observer had five years experience as a high school speech and drama teacher, the second taught elementary music for two years and the third spent one year teaching junior high school social studies. The author, also working as an observer, has had two years experience in high school social studies. This team worked together for two weeks of training before observations began. During the two week training period, approximately 20 hours were spent either reviewing video tapes or visiting class- rooms. This is in keeping with Spaulding's exPerience (1970, 9) that observers can be trained in two to three weeks. The first 47 three three-hour sessions were spent looking at video tapes of classroom behavior, discussing the items and calling out behaviors as they occurred. It was at this time that several definitions were modified and refined at the suggestion of these experienced classroom teachers. The fourth session the four observers visited classes in a local elementary school to make a "trial run." During this session they acclimated themselves to live classroom coding. The observers went in teams so that they could discuss any problems they encountered in coding. It was at this time that the final revisions were made in the instrument. For the duration of the training period the observation schedule was assumed as a given. The fifth session was devoted to intensive video tape viewing and mastery of the observation schedule. The sixth and seventh sessions were devoted to reliability checks. Two methods were used for ascertaining observer agreement. First, three twenty-minute video tape segments, selected for audio and visual clarity, were coded simultaneously by the observers with- out interruption. Second, the teams of observers visited a class— room and coded the teacher interaction simultaneously for two twenty minute periods. As suggested by Bijou, Peterson and Ault (1968, 184), Spaulding (1970, 79) and Madsen, Becker and Thomas (1968, 141), reliability for the three video taped segments and the two live class- Ifoom segments was calculated by scoring each interval as agree or dis- agree (match or mismatch) and dividing the total number of agreements bY the total number of agreements plus the total number of 48 disagreements or: # of agreements # of agreements + # of disagreements An example of instrument scoring and calculation of agreement is included in the appendices. Further, periodic reliability checks were made during the study by assigning two observers to 20% of the classroom sessions visited (Rosenthal, Underwood and Morris, undated, 2). Reliability was calculated as described above. SELECTION OF THE POPULATION Several major decisions were then made in selecting the popula- tion for study. Initially, it was hoped that both inner city and suburban schools would be included for comparative analysis. How- ever, the inner city schools accessible to the researcher, specifically Detroit and Pontiac, Michigan, Public Schools, were not permitting outside research to be conducted at the time of this study. It is suggested that future research using this instrument be designed to include a description of the inner city schools (see further discussion in Chapter V). However, this study was necessarily limited to a sub- urban population. Second, this investigation was limited to elementary school. Financial considerations made it imperative that only one school level be included in this initial study. The elementary level was selected for three reasons: (1) precision teaching has its most immediate application at the elementary school level as indicated in Chapter II; (2) the Henderson study indicates that the highest 49 percentage of agreement on the identification of competent teachers is at this level; (3) a teacher's behavior could be sampled from throughout the day at the elementary school level and she would be interacting at all times with the same group of students. There are two additional advantages in sampling from throughout the day. First, all subjects (e.g. arithmetic, science, language arts, social studies, etc.) are observed, providing a more diverse, "content free," spectrum of behaviors. And second, observing the same class would mean the observer would more quickly be accepted as a "piece of the furniture," thus reducing observer effect. Having limited this initial investigation to suburban elementary schools, Oakland County, Michigan, was selected as a representative suburban area. Located immediately north of Detroit and serving as a "bedroom area" for the urban center, the county is described by the Oakland County Schools Director of Social Studies (Platt, 1966, 14) as a mostly "middle class" area with a per capita income well above the national average. Also, it has a diverse racial, ethnic and religious population with 88.2% urban and suburban and 11.8% rural. Platt also reports that 28.8% of the total county population is presently in grades kindergarten through twelve and the percentage of high school graduates going to college is well above the national average. Persons residing in Oakland County range from the heads of major automotive corporations to several communities with scattered concentrations of welfare recipients. The Transportation and Land Use Study (TALUS), commissioned by the Planning Division of the South- eastern Michigan Council of Governments, reported in August of 1968 (15) that the median family income in Oakland County was $9020. This 50 fl£5$I D. Contingency manage 5. MODELING/RESPONSE FACILITATION A. Task 01 / n l B. Behavior C. Enthus task/content 6. TEACHER IMPOSITION o 0 A. Perform task 1 2 B. Relate exp/moralize C. Admit mistake/apol 7. CHANGE ENVIRONMENT Goal C. Procedures 8. AUGMENTED FEEDBACK O1 02' A 5 B // / C 7f£4 hflfl D E 01 “5 9. TESTING O 10. DELA FEEDBACK [9! g I 021/ I 1. UNCLEAR/MUDDLED 0l 02 DETERRENTS 1. PUNISHMENT Extra task Low grades Reorder.environ . Physical restraint/ punish E. Deprive of privilege 2. VERBAL DETERRENT A. Criticism B. Repr imand/negate 3. SIGNAL DETERREN T A. Facial 02 B. Gestural C. Proximity control 4. THREATENS 01 02 U 0 w W 5. TIME-OUT PROCEDURE 0 O A. Wait/lights __il5 2 l B. Recess/after school C. Deprive social inter- action D. withhold positive reinf 412113229 A. Pose question 0 2Zfl(/ B. Remind behavior/deport LW '9‘)? 02 fl - W WEE 01.2 E E. MISCELLANEOUS 122 On the preceding page is a COpy of the observation schedule with the tallies recorded by two Observers during a single observation session. There are two ways that observer agreement can be derived from these distributions depending upon the intended use of the results. Observer agreement can be calculated for the sum totals in each category or agreement can be calculated using the teacher-initiated/student-initiated differentiation in each category. This explanation gives the agreement for the sum totals in each category. Disagreements are considered to be those tallies recorded by one Observer that do not have a correspond- ing tally on the second observer's record. Thus, for this observation the agreements and disagreements are as follows (deleting unclear/muddled and miscellaneous; combining augmented and testing; and considering no response as Cl before beginning with tangible rewards as C2): Agreements Disagreements Agreements Disagreements Agreements Disagreements c1 0 0 C10 1 0 019 0 0 C2 0 0 C11 0 0 C20 0 0 C3 0 0 012 0 1 C21 1 0 c4 0 1 C13 2 3 022 0 0 C5 1 0 €14 o 2 c23 5 3 c6 1 1 015 13 1 c24 0 0 C7 0 0 C16 17 3 c25 0 1 08 0 0 C17 1 0 026 1 0 09 0 0 C18 0 0 c27 6 1 028 2 0 2 2 34 10 15 5 Putting this into the formula: % of agreement = # of agreements = 51 .75 # of agreements + # of disagreements 51+l7