. .....V _. V. V 7.....v..........V..V.V_.V. .2. .V.. . V . u . — . .a... . V . V. .V. . .. .- o .. .. .V‘. . . V . . . .0. V I . . V . V v..... . ._ ... . . o r .. V. .. v . ‘0‘ O . .. cl. . . .. _ .. V . . . . .u.. .I . V. V . o. n . . .. _ . . . V . . . _ . .‘ . .. .1 V . . — V. ... . o. . . . p L. . V. . ..V V .v .. V. . . . 0.. . .. o... . . . o. . . _ . V . . < I . u . . .. . . ._-. .. V . V V .. . .. U I . u 4 C, n .0... u . ‘—-.I I 1V._ 0 _ . b . .. . . . V .6 V..- n g.- n n. O . O. . . . w . .- . o. u . . o . . . . O - .. . o . u o — 4. u . a o . . . -. . . .. . .V . . u . V. . . V . . ... V. . '- p . . _ . O . I n. 0 V . o ... . . u . V . . . V - _ . . V V . . . . .V . .o. . .7 .. o . I .u I . Q . . u .. . .o . .. . o . . . V. .O I n I I t o - I n — V l . u o . . V .b cv. . o - u _ - E. USES AN EXAMI‘NA ‘ SOCIAL REINFORCEMENT ~ « -'o‘~.b.-o v-w 9 o 0“..- v. ‘C u m u w .. .V.. c . . O o - . 4 . ... . . ..o . u 4. V. _ V . o u . . V . _-. . . s - . . . _ . v0 . . . . o .. . . u - d n .- . . . u . V . . _ O O . V I o c . l C o . . . Vo. . . . . o . u .1 o ‘ o 1 non- f 'NDIT R G GESTION BEHAVIO G ‘ V . .. ; ‘. . . V a u I” 1.0 V.- V. _ ‘. ...VV( -pCaso.bcp.r Lb. ._.\.-.V.. ~o . Jr}. .w... .3. ..n.... . 1.. a I. . a V. .0 . in f _ .~ .V . i... o V . . . .. .vr. . . . . o . . . . yo . . n . V .. A . . . . . ..1.o ..V . ._ . .. ,0... . f... . .o...VV...Vo. _ a}...v§...§._ In. ..._ .7... .. .«Vv..;_.$.!..... ,- : . . V. ...V 5....- . 0.......8.. .u .11... 1...... :91. 7.1 _ . . .V. 3'.. .. . . : .. _ V 2. V V . o...o».......... V V1.V;I..V.v U.‘ 3J1“.- ’ 'v if \i 6 f. M.VA. gree. 0 STATE UNEVERSITY . u ..oV .02).. .V... ’Ai'w . {~"fc BRWE BER " D SAARI 3:: '35:. 1 ”.0!“ 09(06 he 3'8 - 1975 4,. Thesis for It MECMGAN o o 1 l . ulll .w u . V. . .V 1.... _. v r. 17.1.... . . .h ‘ .¢V...._. ._». ‘.o!_. a....... VI... V I . V V . .. .L _ c . . .V r .. _. . o .... Z. . ~ ..... .u. _ . . ‘ .p. . .m.‘ g..:. f‘... \ fling .. . .V. V .y .¢ . .. .d... .4 _ n . a ..*.o L. .a... ..._....‘w d: . .H"...H.V.u_u..n-v.. #3.. ....4h.:.. .r. NLLLVFHrbfiiw'. a.— an“: .V.}..flhvhunfikfi c t . . I": o _..._3._ ,VV.... .J V. . V . V _ .. VVV.:.' .V.. . MSW”! ”mum.“ > u “/1 Ilfllflfllflfljfllflumgflfiflflmjfifl ; L ,. .9, :23, .,. ”Mr F "" 1 1 ABSTRACT AN EXAMINATION OF THE USE OF SOCIAL REINFORCEMENT TO CONDITION SUGGESTION BEHAVIOR By Bruce Bernard Saari The present research tested the general hypothesis that operant conditioning, using only social reinforcers, could affect the acquisition of influence attempt behavior in problem solving groups. Ninety-six undergraduate psychology students (48 males, 48 females) participated in one hour experimental sessions as members of twenty-four single sex, four person teams. Subjects selected the team in which they would function based on their availability for experimental sessions. During each session a team attempted to solve two multi-solution problems, the first of which provided base line data on influence attempt behavior and the second of which constituted the treatment period. Each team received one of six treatments (contingent praise, punishment, or implementation; non- contingent praise or punishment; and a control condition) based on a random assignment of teams to conditions. The research manipulated the treatments on an individual basis by providing each subject with an individual light display through which the experimenter presented the treatments. Pre-experimental instructions defined the meaning of the lights in the display. Bruce Bernard Saari Results of the study did not support the operant conditioning hypotheses. Of seven specific hypotheses based on operant condition- ing theory, only those of the null form resulted in predicted relationships. Tests of hypotheses which predicted positive effects of reinforcement indicated significant changes from operant levels, but such changes ran in the direction opposite that predicted by operant learning theory. A discussion of these unpredicted results includes an evaluation of alternative explanations for their occurrence. Comment is also made concerning the results of the testing of other hypotheses and a combined interpretation follows. The discussion concludes by noting the shortcomings of this research and by presenting possible avenues of future exploration in light of the present findings. Approved by Thesis Committee: -. /7 /{ O // I I /' 1' _—- "' - A / Dr. Carl Frost, Chairman {/2:;77%C}227* 5L2>/// Dr. Frank Schmidt Dr. Neal Schmitt Dr. John Nakeley AN EXAMINATION OF THE USE OF SOCIAL REINFORCEMENT TO CONDITION SUGGESTION BEHAVIOR By Bruce Bernard Saari A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1975 ACKNOWLEDGEMENTS Patience has often been termed a virtue and truly my instruCtors in this effort have been virtuous. Words of thanks are not enough to express my appreciation and gratitude. In Dr. Carl Frost, my chairman, I have found a perpetual source of patient support and consolation. In Dr. Frank Schmidt I found a patient and thorough critic of methodological applications. In Dr. Neal Schmitt I found a patient consultant in times of data analysis and interpretation. In Dr. John Nakeley I found a patient advocate of correct theoretical thinking and conceptual developments. Along with these examples of academic patience, I must commend John Howland and Alicia Crenshaw for their respective patience in doing the reliability analysis and typing the rough drafts. Finally, some of my finest teachers of patience have been family, friends, and most of all, my wife. She, in particular, could never be given enough credit for her continued love and support, even when problems with the research made me unloveable. Finally, for her efforts in typing and proofreading at many points along the way, I can only say thank you. ii TABLE OF CONTENTS LIST OF TABLES . . . . . . . CONCEPT DEVELOPMENT . . . . . Introduction . . . . . . . Recent Research . . . . . . Personnel Development and Operant Learning Theory Summary of Learning Research . Hypotheses . . . . . . (l) . . . . . . . . (2) . . . . (3) . . . . . . (4) . . . . . . . . (5) . . . . . . . (6) . . . (7) . . . . . . . . (Corollary 7-A) . (Corollary 7-B) . . . . METHODOLOGY . . . . . . . . Subjects . . . . . . . Experimental Method . . . . . iii Page RESULTS . . . . . . . . . . . . . . . , . . 39 Interrater Reliability . . . . . . . . . . . 39 Hypothesis Testing . . . . . . . . . . . . . 40 (l) . . . . . . . . . . . . . . . . 42 (2) . . . . . . . . . . . . . . . . 48 (3—6) . . . . . . . . . . . . . . . . 49 (7, Corollaries 7-A & 7-B) . . . . . . . . . 50 DISCUSSION . . . . . . . . . . . . . . . . . 54 APPENDICIES Appendix A . . . . . . . . . . . . . . . .68 Appendix B . . . . . . . . . . . . . . . 7l BIBLIOGRAPHY . . . . . . . . . . . . . . . . 74 iv LIST OF TABLES Table Page I. Treatment Conditions and Distribution of Subjects . . . 33 2. Repeated Measures ANOVA Design . . . . . . . . . 4l 3. t Tests of Differences in Response Frequency Between Sexes . . . . . . . . . . . . . . . 43 4. Repeated Measures ANOVA Table--Hypotheses l and 2 . . . 45 5. Mean Response Frequencies for Operant and Treatment Periods . . . . . . . . . . . . 47 6. Repeated Measures ANOVA Table--Hypotheses 3-6 . . . . 5l CONCEPT DEVELOPMENT Introduction Some authors (e.g., Schein, 1970; Katz and Kahn, l966) consider organizational development necessary for organizational survival in a changing environment. This belief is generally based on the theory that organizations function as open systems which are required to constantly interact with the outside world. This interaction takes many forms and affects almost every phase of an organization's operations. One of the more continual interactions that takes place is the movement of the people who make up the organization from the outside environment into the organization and back out again. Since this movement is one that occurs daily, there is a potential for this interaction to be a major conveyor of environmental change to the organization. If a change in some area of life occurs in the outside environment, this change may be brought to bear on an organization by its members. For example, the increase in personal control over one's life which seems to exist due to recent decisions of the Supreme Court may result in the members of an organization desiring a revision of their organization's structure or formal by-laws to permit similar increases in personal control over appropriate areas of their work life. It is such a change that managerial systems such as participative decision making plans (PDM) are designed to accommodate. Such plans are geared to permit the individual involved in the functional use of a decision to also be involved in the making of that decision. If some segment of an organization decides to change the organization's structure and implement a participative decision making plan in order to help the organization better fit into its environment, it appears necessary that those individuals who will be allowed more personal control or influence over their work must also undergo some type of personal development or change. Thus, it becomes obvious that proper administration of the installation of a PDM plan requires an understanding of the changes necessary in individual members to deal with modifications in the organization's structure. Such a need for understanding yields many researchable questions which, when answered, should permit more effective action on the part of a change agent in implementing a PDM plan. However, to be manageable, such questions must be approached individually. The present research examines one portion of the question of how people in organizations learn to exert influence or control over various areas of their work life. Recent Research Burtnett (l973) addressed a similar question by attempting to demonstrate the existence of causal links between organizational variables assumed to affect the success of PDM and the changes in personal influence patterns in an organization installing a PDM program. In doing so he noted some of the theoretical relationships which may exist between influence, participation, and organizational effectiveness. This discussion is quite germane to the present research. From a review of the pertinent literature on influence, participative decision making, and the Scanlon Plan as a form of PDM, Burtnett developed the following ideas. Extending the influence concepts of Likert (196l) and Tannenbaum (l968), Burtnett asserted that the basic premise underlying his research was that "personal influence is inherent in participation," (p. 43). .Since_other researchers_(e.g., Morse and Reimer, l956; French, et al., l958; Seashore and Bowers, 1963 and l969; Coch and French, 1948; Lawrence and Smith, 1955; and Lewin, et al., l938)_had_a1readydemonstrated- ‘that partjgipaljon can improve organizational effectiveness as measured by suchmthings‘as_productivity,nturnOYer_rates, absenteeism,_ \~__d,,dnfl_.ua_. awe” and communicatjgnwpgtter95,Burtnettzobserved.tbat_increased.in£luencea. should inlpLovg organization effectivenessmmugn.enhanced part1: "gipatign. Ajter defining the Scanlon P1an_as amtype of PDM system, he noted that the_Scanlganl n becomes_a vehicle for the exercise of kpersonglminflugnge. Thus, it becomes clear that the solution of his proposed research question is important in contributing to the body of knowledge surrounding organizational effectiveness and the develop- ment of PDM programs. The same argument can be made for the present research since it deals with the very similar question of how people in an organization learn to influence or control parts of their work life when the organization makes a change toward participative decision making. Hhile providing a rationale for undertaking both studies, the above observations also demonstrate the theoretical connection between influence and participation on which Burtnett's research was based. Since influence and participation are assumed to be so interdependent, it seems correct to hypothesize, as Burtnett did, that organizational variables which control PDM should also control changes in influence patterns. However, Burtnett obtained little empirical support for his hypothesis. It is this problem which has prompted the present research and which will be the focus of the next portion of this discussion. To solve the problem of why Burtnett's research resulted in the ambiguity that it did may not be an easy question to answer. However, some attempt at solving it muSt be made if there is to be an understand- ing of what controls influence patterns or relationships in an organization. The present research is designed as such a study, focusing first on both the logical and methodological quality of Burtnett's research. To reconstruct the theoretical framework developed by Burtnett, previous research and discussion on which he based that framework must be re-examined. The foundation of his framework comes from the definitions of influence and participative decision making. Quoting Cartwright's (l965) definition of influence, Burtnett begins as follows: (a) the agent exerting influence, who for convenience is denoted Q, (5) the method of exerting influence, and (c) the agent subjected to influence, denoted B, When an agent, 0, performs an act resulting in some change in another agent, P, we say that 0 has power over P (p. 4). This definition can be extended to cases where 0 and/or P are groups of people rather than individuals since there is no reason to assume that groups cannot act as a unit in changing the behavior of some other individual or group or vice-versa. Note also that all three parts of this definition of an influence relationship are necessary, but not sufficient, conditions of influence since no single part is capable of resulting in influence by itself. Coupled with this definition of influence, Burtnett presents participative decision making in Lowin's (l968) terms: PDM is a “mode of organizational operation in which decisions as to activities are arrived at by the very persons who are to execute those decisions (p. 69).“ Integrating this definition with the definition of in- fluence given above, Burtnett starts his actual framework by asserting that influence can be considered inherent in participation (p. 43). To determine the stability of his framework, the logical accuracy of this premise must be established. Assuming that "inherent" means "essential to," consider the participative process which follows from Lowin's definition. Some member, A, of an organization performs a given function, X. At some point in time, X requires some type of decision if it is to continue as a function of the organization. PDM encourages A to make that decision and then requires the organization to react to X by either retaining it under its new form after the decision or by relinquishing X as a function. The necessity of influence in participation is thus established as almost axiomatic in the definition of participation. From a further review of the literature, Burtnett presented a discussion of the impact of participation on an organization's effectiveness. The literature involved has been noted above in the earlier discussion of Burtnett's research. If participation is as beneficial to the effectiveness of an organization as the literature reviewed indicates, and if influence is inherent in participation, as it has been shown to be, the control of influence parameters through the manipulation of variables which enhance participation would be logically plausible and should lead to enhanced organizational effectiveness. Therefore, the theoretical framework for Burtnett's study of influence in a PDM system appears to be consistent and logically correct. The point of focus of the present study must now shift to an examination of Burtnett's methodology for testing his hypotheses if a solution to the problem of nonsupportive results is to be found. Clearly, the major Operational definition of interest in underStanding his research is that of influence. Since the research method took a survey approach, influence was defined by a series of responses to questions. These questions were in a five-point Likert- type format requiring a respondent to report how much he perceives himself determining the course of something or, how much say he has about some area of the work situation or, how much control or influence he has over a given situation. This operational definition of influence departs from Cartwright's theoretical definition in that the operational definition includes Tannenbaum's (l968) division of influence into an actual component (i.e., this is the way it is) and an ideal component (i.e., this is the way it should be) and also adds an importance measure (i.e., this part of my job is important). From this series of three-part questions Burtnett can arrive at an operational definition of a given type of actual and ideal influence and the importance of the influence to the individual by determining scores on various sets of questions. Another concept which Burtnett was required to operationalize was that of the organizational variables which enhance PDM. Lowin (l968) lists fourteen such conceptualized variables from which Burtnett has chosen five. These five are: l) organizational commitment to PDM; 2) the amount of public feedback in the suggestion system; 3) the degree of difficulty of the problems handled by PDM; 4) the clarity of goals in the organization; and 5) the linkage which exists between suggestions and bonus payments. As in the case of influence, these variables were defined and measured by a series of questions in the questionnaire. The questions were five-point Likert-type items but lacked the three part character (actual, ideal, and importance) of the influence questions. Scores on subsets of these questions could be combined to provide a measure of each variable. The final concept which Burtnett needed to operationalize was that of causality, since he used correlated survey data as the method of research. Drawing on the methodology literature in sociology (e.g., Simon, l954), Burtnett describes a technique called cross- lagged panel correlations as a means of obtaining causality indicators from survey data. The first step in operationalizing causality is the measurement of pertinent variables at two points in time. After this has been done, the theory of cross-lagged correlations can be applied. This theory holds that a causal relationship is asymmetrical with respect to time and thus correlations between cause and effect variables across time should be unequal. Such a method never actually permits proof of complete causality, nor does it completely handle the problem of third variable causality, but the theory does assert that predominant causality can be determined. Therefore, what can be found is an operational form of predominant causality. Use of cross—lag correlations on questionnaire data indicates that the question of interest in determining the adequacy of this method rests in the determination of the psychometric quality of the measures used. A second technique-type question which may also arise deals with the proper use of the cross-lagged correlation technique. In his discussion of the results, Burtnett pointed out the deficiencies of his method in both of these areas by demonstrating that neither was completely adequate. Considering the psychometric analysis of the queStionnaire, it seems logical to agree with his desire for more adequate research in the future. However, before proceeding to describe a methodology which may alleviate some of the problems encountered by Burtnett, one additional area of theoretical discussion, not previously mentioned, should be considered. Since influence is often considered a personal attribute, the study of group patterns, as was performed by Burtnett, may need to be expanded to consider some treatment of personal patterns as well. To do so focuses on the personal changes necessary for the implementa- tion of PDM that were noted in the introduction to this discussion. Such a focus points up a possible link between research like that performed by Burtnett in the interest of organizational development and the research around personnel development techniques which are available to organizations for use in helping their members adjust more completely to their organizational environment. Personnel Development and Operant Learning Theory Although personnel development is often equated with training practices and skills, it may be the case that such development goes far beyond the confines of formal training sessions, and functions on a more individual basis than is assumed in some organizational personnel develOpment plans. J. N. Gardner (l964) intimates such a possibility while addressing the problems faced in developing an organizational or societal immunity to decay. He says, "A society decays when its individuals and institutions lose their vitality (p. 2)." He takes a rather Gestalt-like position and sees the com- ponent parts of the organization (society) as the key to its vitality. Amplifying this, he also notes that the processes which cause an organization to become rigid and lose its vitality are those same processes which result in maturity. Thus, to mature without becoming rigid seems to be the logical and ultimate goal. According to Gardner, such a goal is reached only through the efforts of self-renewing men who are defined as those individuals (emphasis added) who "systematically pursue and explore their full range of potentialities, and for whom the processes of self-discovery never end." He postulates that organizational vitality is a function of individual development, and that this development is a continual process. Such development may be the life-line of the future for the organization while remaining a function of the develOpment of individuals in that organization. Such a theory of individual development being an integral part of an organization's vitality is also consistent with the philosophy of participative decision making as discussed above (which is based ID on McGregor's (l960) Theory Y). Other writers, such as Blum and Naylor (1968) also note this consistency. "In Theory X the emphasis . . . is on the goals of the organization. In Theory Y the emphasis is directed toward the . . . individual (p. 43l)." Therefore, it can be concluded that the emphasis of personnel de- velopment in a participative decision making system should be on a continual, individual basis which will benefit both the individual and the organization. This necessarily presents the problem of understanding the processes of individual development. Such an understanding may then be used in administering personnel development plans. It should be noted, however, that such concern for individual deveTOpment processes has been the focus of the psychology of learning for many years. Thus, knowledge from the field of human learning should be helpful in under- standing and facilitating such processes. This conclusion is the major emphasis of several articles dealing with the theoretical application of learning theory to management theory (Aldis, l96l; Griffing, l965; Nord, l969; and Jablonsky and DeVries, T972). One of the major concepts that has come out of learning research is that behavior is controlled by its consequences. This has come to be the foundation of operant learning theory. Operant conditioning is basically defined by the Law of Effect which states that any behavior followed by pleasant or satisfying effects will be learned while behavior followed by unpleasant effects will not be learned (Thorndike, l932). Those effects which are satisfying or pleasant are called reinforcers while those which are unpleasant are called negative or aversive stimuli. (Of course, there are also neutral ll effects which are neither reinforcers nor negative stimuli, and which do not affect the learning of behavior unless they are associated with reinforcing or aversive stimuli). B. F. Skinner (l938) gives one of the more thorough reports of early research on this t0pic while discussing some of the theory involved in the operation of this law. He rephrases the law somewhat by defining behavior as only those operations performed by some organism on its environment which are observable and definable by other organisms. He also defines learning as the process of increasing the strength or probability of a given behavior through the administration of the proper reinforcers. To operationalize the concept of strength or probability more completely he says, "The strength of an operant (any behavior not under stimulus-response control) is proportional to its frequency of occurrence, (Skinner, l938, p. 2l)." Thus, he views operant learning theory as functional in describing the acquisition of observable, definable behaviors whose frequencies (or rates of emission) can be recorded. It is from such an operational definition that research has proceeded. It will be to such research that the remainder of this literature review will be dev0ted, focusing primarily on cases where the research is aimed at applications in real-life learning situations. Such a focus should demonstrate how results obtained from operant learning research might be useful in personnel development programs. Concerning operant research in general, Kimble (l96l) proposes that such researches have as their aim the delineation of a set of functions describing the manner in which certain parameters of l2 reinforcement control performance in the learning situation (p. 137). This is consiStent with the Law of Effect in that if behavior is controlled by its consequences, those consequences (i.e., reinforcers) should be considered the primary independent variable, while frequency of behavior emission becomes the obvious dependent variable, in any pertinent research. Research using such variables has provided much support for the theory of operant conditioning in a highly controlled situation (Skinner, l96l). It has pointed up the importance of reinforcement in any learning situation that involves operant behavior. Skinner has gone so far as to say that such research has shown that "almost any instance of human behavior involves contingencies of reinforcement . . . (Skinner, l968, p. 333)!‘ Else- where he says, In short, in the field of human behavior as a whole, the contingencies of reinforcement which define operant behavior are widespread if not ubiquitous. Those who are sensitive to this fact are sometimes embarrassed by the frequency with which they see reinforcement everywhere . . . (Skinner, l966, p. 3l). Thus, it appears some researchers assume that reinforcement is an aCtive part of all human learning. Research designed to determine the effectiveness of the principle in field situations will now be discussed in an effort to demonstrate findings which would aid in the conceptual development of personnel programs around operant conditioning theory. A large portion of the research involving human operant condition- ing outside the laboratory has focused on changing behaviors which are considered abnormal. Such behavior modification research has been reported in several places (e.g., Ullmann and Krasner, l965; I3 Ulrich, Stachnik, and Mabry, 1966; Bandura, l969; Holpe, l969; Mikulas, 1972). However, the great majority of this research is not particularly germane to the present research since it deals with modifying behavior in therapy-like situations. It should be n0ted, however, that in individual situations, behavior therapy techniques have been highly successful, resulting in more con- Structive and useful behavior patterns for the patient. The use- fulness of reinforcement principles outside the laboratory begins to be demonstrated by this body of research. A second area of concentration focuses on the use of programmed instruction (PI). This method of education is based on two key principles of operant learning. The first is that reinforcement is most effective when presented immediately after the desired behavior is emitted, while the second is that complex behavioral responses can be built up by the processes of shaping and chaining (Reese, l966). The research surrounding PI and its automated sister, the teaching machine, is quite thorough and very diverse. Many areas of human learning, including formal education and on-the-job training, have been researched to determine the effectiveness of this method in applied settings (Lieb, et al., T967; Mayo and Longo, 1966; Welsh, et al., l965; Lysaught, l96l; and Margulies and Eigen, l962). Although this research demonstrates the effectiveness of operant conditioning outside the laboratory, it does not approach the queStion of the ubiquity of reinforcement principles outside a structured learning context. 14 There is some research which demonstrates the use of operant principles outside the confines of a laboratory setting and apart from the structure of some form of PI training program. Those research efforts are of special interest here because they demonstrate cause for proposing that operant learning is capable of contributing to personnel development within organizations. Ayllon and Azrin (l968) report extended longitudinal research aimed at devel0ping an entire motivating environment through the use of operant conditioning principles. Working in one unit of the Illinois State Hospital System, the researchers conducted several experiments in which they used the hospital patients as subjects. The studies varied in their basic design and immediate objeCtives, but all were intended to test aspects of general "principles" used to develop the motivating environment. The authors state, . previous research . . . provides little information on how to use reinforcement theory in a complex and fairly naturalistic environment. Most studies of reinforcement theory have taken place in laboratory situations or in a room where the individual has been isolated from others . . . The only conclusive way of determining whether the Laws of Reinforcement and Extinction can be used as the basis of designing a complex metivating environment is, of course, to try it (p. 6-7). By the end of the first year of work, the research methods for this study had been broadly outlined on the basis of the laws of operant learning. The remainder of the research time (which ran until the writing of the l968 publication, or about four years) was spent modifying, revising, and developing particular aspects of the procedure. To accomplish this, when procedures were found effective in practice, as measured by an increased frequency of functional behavior emission on the part of the patients, a "general rule" was formulated on the 15 basis of earlier psychological research evidence. Such a formulation then led to Other testable procedures which the researchers claim may have never occurred to them otherwise. "Every rule led to some application to determine its usefulness, and every procedure found useful was the basis of another rule (p. 82)." The iterative process described above resulted in at least twenty—three general rules which were experimentally demonstrated to be useful in increasing m0tivation of the patients. Some examples of these rules are: l) Target Behavior Rule: Describe the desired performance in behavioral terms. 2) Variation of Reinforcement Rule: Use many variations of a known reinforcer to discover new ones. . 3) Multiple Reinforcer Rule: Use many different types of reinforcing stimuli with a given individual. These rules and their derivations are too cumbersome to enumerate here, but all involve some direct application of operant theory to developing a motivating environment. If so many general rules were empirically shown to be useful, the experimentally developed en- vironment was probably, as the authors state, effective in motivating the behavior of hospital patients. The research discussed above is in some ways reminiscent of the behavior modification work discussed earlier. However, since this research demonstrates the possibility of constructing an entire living environment on the basis of operant conditioning principles, it seems too important to be overlooked. Moving from a hospital setting to an elementary classroom, Krumbolz and Goodwin (l966) studied learning in a formal setting 16 without the structure of P1 or teaching machines. To teSt hypotheses concerning the efficiency of operant learning techniques in such a situation, the experimenters attempted to develop more appropriate task-oriented classroom behavior in second grade boys (who normally exhibited disruptive behaviors) by manipulating the amount of social reinforcement the boys received. It was hypothesized that operant conditioning could be effective in increasing the time spent on task-oriented behaviors by inattentive pupils. To test this hypothesis, a sample of boys with problem behav- iors was selected from three schools within one school district in California. The teachers of these boys were then assigned to experimental and control groups, and the experimental group teachers were trained in methods of social reinforcement which they could administer to the boys in their classes. The control group teachers received placebo training designed to maintain their present use of social reinforcers. All teachers and pupils were then observed in their normal classrooms by trained observers who followed the cenditioning process and recorded pertinent information on the administration of reinforcement and the time spent on task-oriented behaviors. The results of this study were, in general, not supportive of the hypothesis. In almost all cases of testing secondary and deriva— tive hypotheses, results actually ran opposite to the predicted direction, based on operant learning theory. That is, in all cases but one, the control group actually showed (statistically) non- significant but greater increases in time spent on task-oriented behaviors than did the experimental groups. The one case where this 17 was not true was in relation to time spent on independent task- oriented behaviors. In this case, the results were also statiSti- cally non-significant, but they were in the hypothesized direction. This lack of clarity in results may be due to ineffective manipulation of the independent variable, as the authors note, but more importantly it points up the necessity of extreme care in conducting such a real-life type of study in this area. Sarbin and Allen (1968) studied another aspect of the operant conditioning of classroom behaviors that comes even closer to approximating the conditioning of operant behavior in normal adults in an organizational setting. Using four member Students in a graduate seminar, the effects of social reinforcement on verbal participation in group discussions were studied. Using Skinnerian principles, four subjects (two with a high operant level of participation and two with a low Operant level) were unknowingly subjected to a contingency program of verbal approval, head nodding, and other supportive actions. The two high operant level §s were deprived of social reinforcers contingent on participation to decrease participation while the two low level §s were given positive contingent reinforcements to increase participation. All seminar sessions were recorded on tape so the percentage of total participation could be figured accurately, but secretly for each_§. The results of the study indicate that although low operant level §s never exceeded high operant level §s in percentage of the group's total participation, there was a definite change in the amount of time spent participating by each group. Low operant §s increased 18 their percent of total participation by about fifteen percent while high operant §s decreased their percent of total by the same amount. Thus, the possibility of operant control of participative behavior using fairly intelligent young adults as subjects seems reasonable. Azrin (T960) approached the problem of operant conditioning of adult humans in a controlled work setting in a direct way. Using rest periods as reinforcement for performing a simple but strenuous work task, and administering them on both fixed interval and fixed ratio schedules, it was demonstrated that rest, when administered on some response contingent schedule, is capable of increasing the rate of work output. However, since this study involved a controlled work setting and little actual learning, along with no interaction between working Ss, its generalizability to personnel development in an actual work setting is somewhat limited. Gupton and LeBow (1971) report a study using operant principles in a real-life work situation. Using two male subjects who were part-time telephone solicitors in a large company, the researchers studied the effects of a contingency program on the rate of perform- ing a low probability behavior, when the opportunity to perform a high probability behavior was made contingent upon performing the low probability behavior. The low probability behavior in question was selling new appliance contracts while the high probability behavior was renewing old contracts. By making the opportunity to sell five renewal contracts contingent on the sale of one new contract, the amount of sales for both types was increased over previous operant levels. A reversal phase in the design strengthened support for the notion that operant control of productivity was attained. To 19 quote the researchers, The main implication of this study is that costly job analysis designed to increase productivity could perhaps be replaced by observing response rates, designating high and low probability behaviors and instituting the appropriate contingencies. The monetary savings and effectiveness of this behavior management approach make it worthy of pursuing (p. 82). This research demonstrates the possibility of using operant condition- ing in a work setting. However, it does not address the question of conditioning an interactive behavior among workers as posed by the problem of acquiring influence patterns. Summary of Learning Research The last two researches reported demonstrate, to some extent, the direct applicability of operant conditioning techniques to management theory as was suggested earlier. The studies preceeding them are indicative of other areas within a work setting that may be susceptible to Operant conditioning in a fashion similar to that of the last two studies. The final two studies demonstrate techniques from which manage- ment of an organization stands to gain the most or where operant conditioning is used for the benefit of management. The three pre- ceeding studies indicate situations where there may be individual gains in obtaining a more personally useful behavior repertoire while also demonstrating some potential gain for those in charge of the individual's behavior. The gap between the two types of studies is obvious. Research in work settings has not sought mutual gains. It appears necessary that if personnel development is to be understood and conducted on a continual and individual basis, research must be performed which will fill this gap. 20 Another indicator of the need for more applied Operant research is its dearth in the present training literature. Although training research has often focused on the effects of various feedback programs, none of this work has approached the problem from an operant point of view. Rather, such research has primarily tested the potency of degrees of feedback specificity in relation to both positive and neg- ative feedback. J. P. Campbell, in a review of the training literature (1971) points out the fact that training research has not previously pursued the area of operant conditioning and behavior modification in an effective way. After noting the behavior modification research discussed above (e.g., Bandura, 1969) as the major training research in this area, he concludes his review by saying, If we are ever to make training and development a profitable enterprise in terms of important behavior changes we must at least . . . Take an intelligent plunge into the methods and concepts of behavior modification . . . (p. 586). Therefore, with respect to the problem posed by Burtnett's research and the desire to understand the acquisition of influence patterns in an organization, one plausible research approach is to pursue the possibility that operant conditioning could enhance participation in a work setting which allows some form of participative decision making. Such research would demonstrate the consistencies between individual develOpment as the basis of organizational immunity to decay, participative management which focuses on the individual, and operant learning theory which may contribute to both of the above areas. Although the Sarbin and Allen article reported above appears to address this question, it is inadequate in at least two respects. These two areas will be discussed below. 21 Hypotheses The connection between the potential applicability of some learning phenomenon to personnel development and the ambiguity presented by the results of Burtnett's research was initially in- dicated in Burtnett's discussion of his results. He said, It should be pointed out that the exercise of personal in- fluence called for by the Scanlon Plan is typically not found in most organizations and therefore the people at all levels in a firm implementing the Plan must learn appropriate influence styles (p. 87). If this is the case, the learning process involved in obtaining a personal influence pattern must be understood before research can look for variables which might enhance influence. To demonstrate the existence of an operantly conditioned learning process then, is the objective of the present research. The development of such a deomonstration requires two things-~a set of operational definitions of the variables under study and a set of hypotheses on how those variables should be related if such a learning situation exists. Studies of learning generally follow Skinner's definition of behavior as a model for defining their dependent variable as some observable activity which the organism under study emits. To be consistent with previous learning research, and in an effort to avoid the measurement problems encountered by Burtnett, logic seems to diCtate that the concept under study in the present research (i.e., influence) be defined in a behavioral manner. Since no part of Cartwright's three-part conceptual definition of influence is logically stronger than the other two (all three are necessary, but not sufficient conditions for the influence relationship to occur), the best operational definition would be the mOSt behaviorally-based 22 part. This necessitates choosing the act of agent 0 in exerting influence over P as the dependent variable to be examined. Since such an act is not influence by itself, the observed behavior may only be termed an influence attempt. Because it is commonly held by organization developers that the greater the number of influence attempts which exist in a PDM system, the more likely it is that PDM will succeed, such a focus on an individual's influence attempts may not be as limited as it might appear. The second variable of importance in any research is the in- dependent variable being manipulated (most commonly a reinforcer in learning research, or one of its parameters, as noted earlier by Kimble). The measurement problem encountered by Burtnett, and mentioned previously, requires the study of only those independent variables which are observable. In the social context in which influence usually occurs, those independent variables which are assumed to control the Operant behavior of influence attempts can be either reinforcers or aversive stimuli. Therefore, the following social variables have been selected for examination: 1) Individually received praise contingent on individual performance; 2) Individually received implementation contingent on individual performance; 3) Individually received aversive Stimuli contingent on inadequate individual performance; 4) Individually received noncontingent praise; and 5) Individually received noncontingent aversive stimuli. Before actually describing the methods of experimentation used in the present research, it seems appropriate to present a set of hypotheses concerning the relationships between these variables and the 23 dependent variable. These hypotheses are based on the knowledge gained from the learning literature reviewed above. Hypothesis 1 Considering the basic premise of operant conditioning (i.e., behavior which is reinforced is learned or acquired), it is hy- pothesized that individually received contingent praise will cause the frequency of acceptable influence attempt emission to increase significantly above the operant level of emitting accept- able influence attempts. Hygothesis 2 As in the case of contingent praise, individually received contingent implementation of acceptable influence attempts will result in a significant increase in the frequency of acceptable influence attempt emission above the operant level of emitting acceptable influence attempts. Hypothesis 3 Because operant conditioning relies on the efficacy of positive reinforcement, it is hypothesized that individually received aversive stimuli (contingent on unacceptable influence attempts) will result in no significant change in the frequency of acceptable influence attempt emission from the operant level of acceptable influence attempt emission. Hypothesis 4 Similarly, because operant conditioning also requires that reinforcement be contingent on performance for learning to occur, 24 it is hypothesized that noncontingent praise will result in no significant change in the frequency of acceptable influence attempt emission from the operant level of acceptable influence attempt emission. Hypothesis 5 Likewise, it is hypothesized that noncontingent aversive stimuli will also result in no significant change in the frequency of acceptable influence attempt emission from the operant level of acceptable influence attempt emission. Hypothesis 6 Unless there is some inherent problem in the experimental method, it is hyp0thesized that a control treatment condition (where neither reinforcing nor aversive stimuli are presented) will result in no significant change in the frequency of acceptable influence attempt emission from the operant level of acceptable influence attempt emission. Hypothesis 7 In comparing the treatment conditions to the control group, it is hypothesized that individually received contingent praise and implementation will both cause significant increases in the frequency of acceptable influence attempt emission above that occurring in the control condition, while contingent punishment, noncontingent praise and noncontingent punishment will cause no change from the frequency of acceptable influence attempt emission when compared to the control group. 25 Corollary 7—A It is hypothesized that there will be no significant difference in the frequency of acceptable influence attempt emission for the treatment conditions of contingent praise and contingent implementa- tion. Corollary 7-8 It is hypothesized that there will be no significant difference in the frequency of acceptable influence attempt emission for the treatment conditions of contingent punishment, noncontingent praise, noncontingent punishment, and the control condition. Stated generally, the above hypotheses predict that the treatment conditions will cluster into two groups in terms of their effective- ness. Individually received contingent praise and contingent implementation will be equally effective in conditioning acceptable influence attempt behavior, while contingent punishment, noncontingent praise, noncontingent punishment and the control condition will be equally ineffective in conditioning such behaviors. METHODOLOGY The Sarbin and Allen research reviewed above appears to address a problem similar to the one posed by the present research. However, in testing the hypothesis that operant conditioning controls participation in a group setting, their research encountered certain difficulties which must be appropriately dealt with before drawing the conclusion that influence attempt behavior can be operantly conditioned using social reinforcers. The first of these difficulties is the use of apparently ipsative measures of participation. To imply, as Sarbin and Allen do, that an increase in the percentage of total participation experienced by reinforced §s is independent of the decrease in the percentage of total participation time experienced by non-reinforced §s in the same group appears dubious. Such an implication makes it difficult to conclude that operant conditioning is the only possible cause of changes in participation percentages. Secondly, the research was primarily concerned with increasing verbalization among the reinforced Ss. There was no attempt to manipulate one particular type of participation (e.g., statements of new information), making it difficult to know if all types of participatory behavior were equally controlled by the conditioning process. It appears feasible that what could have occurred was an increase in something such as repeating what someone else had already 26 27 said or simple question asking behavior. Therefore, it is not possible to conclude that influence attempt behavior has been demonstrated to be controlled through operant conditioning proce- dures. A final difficulty with the research is the use of only two subjects in each of the two treatment conditions. Although this is a frequent occurrence in learning studies, it must be recognized as a limiting factor in the generalizability of those studies' results. Some attempt at increasing sample size must occur if generalizations about human learning are to ever be meaningful. A second research effort which should be considered as important to the methodological deve10pment of the present research is a study conducted by Oakes (1962). In this study he attempted to condition response classes, as defined by Bales' Interaction Process Analysis (1950), in a group discussion. Four subjects were seated around a table on which two cardboard screens stood. These screens were placed in a manner to prevent subjects from seeing one anather. Each subject had a light display placed in front of him through which reinforcement was presented. Each such group was given a clinical case study and three possible explanations for the behavior described in the case. Subjects were then instructed to discuss the three explanations in terms of how well and why they explained the described behavior. Subjects were also instructed not to come to a group decision. The group discussion was then viewed through a one-way mirror by an experimenter who administered reinforcements to each §_when the §_exhibited a behavior in the Bales' category which was being reinforced in his group. All twelve of Bales' categories 28 were considered and two groups were reinforced for each category. After the experimental session was over, two other experimenters categorized the individual behaviors into Bales' groupings based on written transcripts of tape recordings of the session and tab- ulated each individual's behavior total for each of the twelve categories. The analysis was then based on this categorization, comparing a category of behavior for the individuals in the twenty- two groups not reinforced for that specific category with the same category for the individuals in the two groups which were rein- forced for that category. Based on such an analysis the author concludes that he found "support" for the conditionability of Bales' category "giving opinion." None of the other eleven categories' conditionability gained empirical support. Ignoring the fact that the "supportive results" are no more than what would be predicted by chance alone, the author's inter- pretation of the results still seems somewhat tenuous in light of the methodology employed. The difficulties with the methodology and their impact on interpretation are as follows. First, no operant levels of behavior were established for any of the categories. Thus, there do not seem to be sufficient grounds to argue that any category of behavior was actually modified. It could be that reinforcement had no effect and that the obtained results are due solely to dif- ferential operant levels within a given category. Within a group of subjects, all the author can conclude from his data is a relative difference in emission rate for one behavior category as compared to the others. A second problem arises from the fact that the analysis of variance design was required to compare a mean computed on eight 29 people with a mean computed on eighty-eight. Such unequal cell sizes must naturally raise questions about the interpretation of the results. Thirdly, in using the categorization described above, the author reports no reliability estimate between the categorization of a behavior as defined by the first experimenter (who also defined the reinforcement contingencies) and the categorization of that same behavior by the other two experimenters (who provided the data for the analysis). Thus, there is no assurance that a behavior which was categorized as being reinforced in the analysis was in fact subject to a reinforcement contingency in the experimental session. A fourth difficulty is in not allowing the S; to see each other. Such a situation greatly reduces the face validity of the method for research in group problem solving. The final difficulty to be mentioned which weighs heavily on the present research is Oakes' finding in relation to the behavior category called "giving sug- gestions." The methodology employed (i.e., giving Ss a finite set of solutions and asking the §s not to reach a group decision) seems to limit the appropriateness of this type of response. This fact seems to be borne out by the results in that few behaviors actually were categorized as "giving suggestions." Thus, Oakes' design does not seem to permit an adequate test of whether "giving suggestions" or any other behavior category can be modified by operant conditioning. In light of the difficulties encountered in interpreting Sarbin and Allen's and Oakes' results, it seems efficient to develop a methodology which is capable of avoiding such pitfalls and which yields direct answers concerning the conditionability of influence attempt behavior. The methodology presented in the remaining portion of this 30 section was designed with these two primary considerations in mind. A third factor which influenced the design was the set of implications drawn from Burtnett's research. Subjects The subjects used in this research were 96 undergraduate college students (48 males, 48 females) obtained by the researcher through the Human Subjects Pool at Michigan State University during the summer and fall terms of 1974. This pool is made up of all students enrolled in the introductory undergraduate psychology classes for a given term, and all subjects who participate in such experiments earn extra credit toward their class grade. Subjects were asked to participate in an experiment in group problem solving and were each given three extra credit points for their participation. Experimental Method The experimental phase of this research was conducted in two parts. The first phase was a brief meeting of subjects with the experimenter for the purpose of assigning subjects to four-person groups. The second part of the research involved an hour meeting of each four—person group. Each four-person group was made up of four males or four females. No mixed sex groups were used in the research. In the first session, subjects reported to the experimenter's office and were asked to fill out a biographical questionnaire which asked for information on name, age, race, years of college completed, any previous leadership experience, and grade point average. The subjects were instructed that all of these data would be used only in 31 a descriptive manner and that no individual data would be released to anyone but the subject himself. After completing the questionnaire, subjects were asked to sign up for an hour experimental period in which they would participate in group problem solving as members of a four-person group. Since subjects could only sign up for exper- imental periods which were compatible with their schedule, random assignment of subjects to groups was impossible. To correct for this, assignment of four-person teams to treatment and control conditions was on a strictly random basis. The second experimental session involved the actual experimental portion of the study. Combining some of Oakes' method with a method developed by Hoffman, Burke, and Maier (1965) in the study of influence in participative problem solving groups, this hour session was con- ducted as follows. Subjects reported to the experimenter's office and waited there with the experimenter until all four group members arrived (the same experimenter met all groups and conducted all the experimental sessions). Subjects were then conducted to a laboratory and seated around a square table. The team was then instructed to practice group problem solving by attempting to collaborate on a work-related problem for fifteen minutes in an effort to develop a series of rationales for and concomitant solutions to the problem. The complete set of instructions for this part of the experiment are presented in Appendix A along with a copy of the problem used in this practice period. The behavior exhibited in this phase served as a measure of the operant level of each individual's influence attempt behavior. 32 At the end of this fifteen-minute session, the subjects were instructed that they were now going to be asked to work for thirty minutes on another problem which was in need of a solution. Subjects were then presented with the work-related problem and permitted to work on it for thirty minutes. Subjects were also instructed con- cerning the meaning of the light display on the table in front of them, thus defining social praise and punishment. Complete instruc- tions for this period and a copy of the problem used for this thirty- minute session are presented in Appendix 8. During this period, the treatment assigned to the group was administered to each individual in the group (See Table l for a description of assignment of groups to treatment conditions). Upon completion of the thirty-minute period, the subjects were partially debriefed, given credit for the experiment, and dismissed. Before leaving they were also told that if they would like to receive a full debriefing on the experiment, they could receive such feedback by mail by leaving their address with the experimenter. In a more specific sense, this hour experimental session included the following. The two problems the group had to solve are adaptations of two role playing exercises Maier presents in his book Supervisory and Executive Development (1957). The problems are l) "The New Truck Dilemma"--the fifteen-minute practice problem and 2) "The Parasol Assembly Bottleneck"--the thirty-minute problem. Each group was seated around the table in alphabetical order according to last names to reduce the possibility of extraneous proximity effects as much as possible. Before beginning the group activity, participants were told that the nature of the research 33 NNUNNNNN NP mpuwwnam mp mpumnnsm op muuwwnzm op mpomnnzm mp mpumwnam NF 325 N A226“. N E23 N 3:8“. N 32.8 N A228 N N «Na: NV N «Na: NV N «Na: NV N ape: NV N 6_Nz NV N NNNz NV masocw q masocu e maaoeu a masocm e masocm a masoco e NZNNINNZNN NNNNNN NZNNINNZNN zoNNNNZNzNNNzN NNNNNN somezou NZNNzthoozoz NZNNZNNZONZOZ NZNNZNNZON NZNNZNNZON NZNNZNNZON mpumhnzm mo cowp=n_eumwo use mcowpwvcoo pcmspomeh .P MNmppm>gmmcou .v a m w: No mm mueaom acm29mmch Nu, mommguoaxzu-m_nah <>oz< mmczmmmz amummamm .v m4m