alllllllllll llllllllll THB‘ 3 123 25767 \ ._\ \ '\ \ \ This is to certify thatx‘the thesis entitled INFORMATION DISTRIBUTION, SOURCE CREDIBILITY AND CLAIM STRENGTH AS DETERMINANTS OF ORGANIZATIONAL PURCHASE OUTCOMES: AN EXPERIMENTAL APPROACH presented by Robert E. Krapfel Jr. has been accepted towards fulfillment of the requirements for Ph.D. Business Administration degree in /// ///9/)///!//1/1 /// ANIMD » 0-7639 © 1980 ROBERT EMMETT KRAPFEL, JR. ALL RIGHTS RESERVED INFORMATION DISTRIBUTION, SOURCE CREDIBILITY AND CLAIM STRENGTH AS DETERMINANTS OF ORGANIZATIONAL PURCHASE OUTCOMES: AN EXPERIMENTAL APPROACH by Robert E. Krapfel, Jr. A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Marketing and Transportation Administration 1979 ABSTRACT INFORMATION DISTRIBUTION, SOURCE CREDIBILITY AND CLAIM STRENGTH As DETERMINANTS OF ORGANIZATIONAL PURCHASE OUTCOMES: AN EXPERIMENTAL APPROACH by Robert E. Krapfel, Jr. Thoughtful marketers, academic and practitioner alike, have long recognized that the characteristics of organizational customers and the products they purchase necessitate marketing strategies different from those applied in consumer goods markets. Chief among these differences are directness of distribution channels and emphasis on personal sell- ing in promotional effort. Recently many students of organizational purchasing have emphasized behavioral dimensions of buyer-seller relationships to better understand their dynamics. This research focuses on a process approach to group decision making in the buying center. A deductively develOped causal model hypothesizes three vendor influenceable variables, information distribution, source credibility and claim strength, as determinants of the willingness of one or more buying center members to assume an advocate role. Furthermore the model posits that perceptible advocacy behavior influences group vendor selection. The model was tested by means of two experiments. The first was a factorial design employing MBA students as subjects. The second was a crossed and nested repeated measures design employing technical and managerial practitioners. In both experiments groups rather than individuals were the unit of analysis. Multivariate and univariate analysis of variance of multiple-item scale dependent variable data were conducted for the purpose Of model validation. Complete model validation was obtained with the practitioner sample while partial validation and evidence of interaction effects were found with the student participants. Model validation provides evidence of linkage between the independent factors mentioned and problem specific self-confidence through which advocacy behavior and finally vendor choice are influenced. In addition to theoretical conclusions resulting from validation of the model managerial implications are provided as well. Sales represen- tatives may now influence not only what is discussed in group delibera- tions but the manner in which it is discussed as well. In the context Of this new process model vendor representatives who correctly facilitate advocacy behavior by one or more influential buying center members enjoy enhanced probability of success. ACKNOWLE DGMENTS I would like to express sincere gratitude to my parents who taught self-reliance at an early age, to my wife Gabriella who counselled and occasionally prodded, to my committee, and especially Dr. Gilbert Harrell, who allowed me necessary latitude, to Mr. Don Moyer who took extraordinary pains to help make the practitioner sample a reality and to Ms. Scotti Lucasse who taught me several things about the English language. ii TABLE OF Introduction. . . . . . . . . . The Problem . . . . . . . . . . . Nature Of the Problem. . . . Research Purpose . . . . . . Research Questions . . . . . Variable Specification . . .' Model Development. . . . . . Model Validation . . . . . . Research Limitations . . . . Literature Review . . . . . . . . Marketing. . . . . . . . . . Communication. . . . . . . . Social Psychology. . . . . A Causal Model. . . . . . . . . . Model DevelOpment. . . . . . Model Specification. . . . . Model Validation . . . . . . Research Method . . . . . . . . . Experimental Design. . . . . Population . . . . . . . . . Instrument Development . . . CONTENTS Instrument Administration Procedure. Data Analysis. . . . . . . iii Page 10 Io 11 14 14 28 33 38 38 49 56 61 61 69 7o 77 80 Conclusions and Implications. . Theoretical Consequences . Managerial Implications. . Issues in Future Research. Summary. . . . . . . . . Appendices Appendix 1 . . . . . . . . Appendix 2 . . . . . . . . Appendix 3 . . . . . . . . Appendix 4 . . . . . . . . Appendix 5 . . . . . . . . Appendix 6 . . . . . . . . Appendix 7 . . . . . . . . Appendix 8 . . . . . . . Bibliography. . . . . . . . . . iv 108 108 115 118 124 127 128 140 141 146 152 156 158 160 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. LIST OF TABLES Product/Vendor Selection Criteria: 1924 vs. 1974. . . . Example Fixed Effects, Three Factor ANOVA Table. . . . . Example Summary ANOVA Table for An A x (BxS) Design. . . Summary Pre-Test Source Credibility Data . . . . . . . . Student Sample Specific Self-Confidence (YSSC) MANOVA. . Student Sample Perceived Advocacy (YPA) MANOVA . . . . . Student Sample Vendor Evaluation (YVE) MANOVA. . . . . . Student Sample Behavioral Intent (YBI) ANOVA . . . . . . Student Sample Dependent Scale Reliability Coefficients. Student Sample Specific Self-confidence Scale Total (Y “OVA 0 O O O O O O O O I C O O C I O O O O O C C I C . SSCTOT Student Sample Perceived Advocacy Scale Total (Y ) ANOVA PATOT Student Sample Vendor Evaluation Scale Total (Y ) ANOVA VETOT Student Sample Scale Total Dependent Variable Relationships Student Sample Vendor Choice with Perceived Advocacy . . Summary of Student Sample Significant Relationships. . . Practitioner Sample Dependent Variable Scale Reliability Coefficients . . . . . . . . . . . . . . . . . . . . . . Practitioner Sample Specific Self-confidence (CONSCALB) “OVA 0 C O O O C O O C C O O C C O O O C O C C O O O O . Practitioner Sample Perceived Advocacy (ADVSCALE) ANOVA. Practitioner Sample Vendor Evaluation (VENDSCALE) ANOVA. Practitioner Sample Scale Total Dependent Variable Relation- ships. . . . . . . . . . . . . . . . . . . . . . . . . . V Page 65 68 73 86 87 89 90 91 92 93 93 94 95 96 102 102 104 104 105 21. 22. 23. 24. 25. 26. 27. Practitioner Sample Vendor Choice with Perceived Advocacy Contingency Analysis. . . . . . . . . . . . . . . . . . . Summary of Practitioner Sample Significant Relationships. Student Sample Product Features ANOVA Summary . . . . . . Student Sample Maintenance Services ANOVA Summary . . . . Student Sample Delivery ANOVA Summary . . . . . . . . . . Student Sample Total Cost ANOVA Summary . . . . . . . . . Student Sample Vendor Reputation ANOVA Summary. . . . . . vi 106 108 156 156 157 157 157 10. 11. 12. 13. 14. 15. LIST OF FIGURES The Buy-Grid Model. . . . . . . . . . . . . The Locus of Influence Matrix . . . . . . . The Sheth Industrial Buyer Behavior Model . The Modified Hackman-Morris Model . . . . . The Generic Decision Process Model. The Causal Process Model of Group Choice. . The Student Sample Experimental Design. . . The Practitioner Sample Experimental Design Information Distribution by Claim Intensity Effect on YSSC (Item 23). . . . . . . . . . Information Distribution by Claim Intensity Effect on YSSC (Item 25). . . . . . . . . . Information Distribution by Claim Intensity Effect on YSSC (Scale Total). . . . . . . . Information Distribution by Claim Intensity Effect on YSSC (Item 24). . . . . . . . . . Interaction Interaction The Causal Model Based on Student Sample Experiment . . The Causal Model Based on Practitioner Sample Experiment. The Structural Equation Model of Source Credibility and Self-confidence . . . . . . . . . . . . . . vii Page 16 24 41 45 52 54 67 69 98 99 100 101 111 112 124 Chapter I INTRODUCTION Study of organizational buyer behavior has greatly increased in recent years. With this acceleration of activity has come a prolifera- tion of concepts and models designed to explain behaviors exhibited by individuals performing purchasing functions in organizations of various types. Unfortunately, the subject of buyer behavior in general remains an abstruse topic for many marketing practitioners and academicians alike. Explanations for this vary; however, the manner of model development may be suggested as one causal factor. Beginning with motivation research in the late 1950's and continuing through the currently popular multi- attribute decision models, buyer behavior researchers have held out the promise of behavioral prediction. All too often that promise has remained unfulfilled due to an uncritical adOption of theory generated in other disciplines. While some marketing students have been successful in creating "new" theoretical constructs, much of the develOpment process remains dependent on building blocks borrowed from others. Under these circumstances the marketer bears responsibility for making those adaptations required to achieve congruence between model and environmental context. Further, given the state of development of organizational buyer behavior theory, it is inapproPriate to suggest that prediction should be the primary objective of research. In light of these considerations, the research described on the following pages is presented as an example of an apprOpriate approach to the development of deductively generated causal theories Of organiza- tional buyer behavior. In particular, it reports on the develOpment and 1 testing of a model of small group decision-making processes specifically geared to the organizational buying environment. While the particular model under study is of substantive theoretical interest, the procedure used to generate and test it may be of greater long-run significance and so methodological issues receive consideration accordingly. THE PROBLEM Nature of the Problem Thoughtful marketers have long recognized that the characteristics of organizational customers and the products they purchase necessitate approaches somewhat different from those used in the consumer goods sector. While the elements of marketing strategy planning and execution remain fundamentally the same, their relative importance and emphasis shift in moving from consumer to organizational markets. Principal among these differences is the directness of the distri- bution channels employed and the preportion of promotional effort allocated to personal selling. It is widely accepted that personal sell- ing is the predominant element in the typical industrial marketer's communications mix. In many cases this selling is performed by the vendor's own sales force calling directly on users and buyers in customer firms. Owing to the importance of the sales representative's mission, and the time and dollar commitment made in develOping and supporting an effective sales force, industrial marketing managers are continually seeking ways to improve the effectiveness and efficiency of selling Operations. Articles on compensation plans, quota setting, territory definition, motivation and other sales management issues are commonplace in the literature. Recently, in some circles, attention has shifted somewhat as we become more aware of the fact that these buyer-seller linkages are made, not between large monolithic corporate entities, but between individuals who are pursuing a wide range of personal, as well as corporate goals. When vendor representatives call on corporate purchasing or operational managers, they are entering into multi-faceted relationships whose nature is influenced by a vast array of factors. Parties on both sides harbor expectations about the parameters that define these relationships. Furthermore, each is constantly attempting to induce the other to under- take a course of action that will result in accomplishment of certain desired ends. In many respects the buyer-seller relationship is an adversary one. This is true because the marketing concept in its purest form remains largely a textbook notion. This is not to say that firms do not strive for what the concept espouses, rather that devotion to its precepts Often is swept away by the exigencies of day-to-day Operations. Given that the atmosphere between buyers and sellers is not one of complete openness and cooperation, and that each vendor faces competitive pres- sure from others, those sales representatives having a better understand- ing of their customers hold a competitive advantage. A typical representative approaches potential customers armed with a battery of information concerning his product Offerings and, perhaps, those of his competitors. In addition, he usually has in mind a particu~ lar method of approach designed to demonstrate to the customer the benefits accruing from the purchase and use of his Offerings. All too Often, however, his knowledge Of the real needs of that customer are fragmentary at best. And, even if he has called on that account for several years, he may have only a minimal understanding of the policies, procedures and processes which govern vendor selection for a given product at a given point in time. Thus, while he may be very cognizant of the strengths and weak- nesses of his own Offering, he may be very much in the dark concerning the inner workings of a potential client firm. This condition is characteristic of a sales rather than marketing approach and not at all surprising. Many firms purposefully prevent vendor contact with certain individuals and often, even when no formal prohibitions exist, the repre- sentative is unable to identify and approach those most influential in a particular decision. In some instances, job responsibilities are in a state of flux and management inside the customer firm may not know them- selves who is currently responsible for a particular area! Assuming that the one or more relevant individuals are known, the representative yet lacks guidelines as to how he is to present both the product and himself. Much will depend on the particular criteria being employed for vendor selection at this juncture. These criteria are themselves multi-dimensional and can be traced to numerous sources. In the last several decades, considerable industrial marketing research has been devoted to the task of enumerating and cate- gorizing the factors that impinge on product and vendor selection. Melvin COpeland, in 1924,1 produced a list of twenty-five buying and patronage motives commonly appealed to by industrial advertisers. The twenty-five were generated by noting frequency of appearance in trade publications then available. Fifty years later, to the month, Lehman and O'Shaughnessy2 published a study on differences in product/vendor attri- bute importance in various situations. They interviewed purchasing agents in several firms asking them to rate selection criteria by impor- tance in specific decisions. Tabulated below are the ten primary criteria cited in each study. Table l Product/Vendor Selection Criteria: 1924 vs. 1974* COpeland Lehman and O'Shaughnessy Economy in use Technical performance Protection against loss Product reliability Enhancement of plant productivity Price Dependability in use Ease of use Dependability in quality Post-sale service Reliability of seller Supplier reputation Punctuality of delivery Delivery reliability Promptness of delivery Supplier past experience Exact specification fulfillment Technical service Variety of selection Flexibility to meet specifications The purpose in presenting this comparison is not to demonstrate how little things have changed in the past fifty years, but to suggest that, in general, a rather clear picture Of salient selection criteria has emerged and that the task remaining is specification of those subsets of criteria invoked in particular circumstances. Other and yet still funda- mental issues remain unresolved. For example, one of the earliest attempts at categorizing buying motives centered about the supposed dichotomy between rational and emotional factors. All of the entries in Table 1 are what would be called rational criteria. Examples of emotional criteria include buyer- seller friendship, desire for status enhancement, non-functional obsolescence, etc. This rational-emotional split has been a source of extensive debate among industrial marketers. 4 . Discussions by P. D. Converse in 1930, and Maynard, Weidler and * The factors attributed to COpeland appear in Principles of Merchandis- ing published in 1925.3 Beckman in 1939,5 clearly indicate the prevailing attitude of early mar- keting scholars. Rational motives predominate in the purchase of industrial goods; however, when competing suppliers' offerings are approximately equivalent, emotional criteria may swing the balance. Thus their advice to industrial marketers was to emphasize product per- formance and reliability, but to be prepared to appeal to other, lesser motivations if the choice seemed in doubt. Today, we continue to speak of the difference between performance and emotional appeals in our basic texts,6 and task vs. non-task decision elements in more specialized monographs.7 There is today, however, a better appreciation of the idea that these classes of motivations or criteria are not mutually exclusive in any way and that they interact continually as individuals and groups process information. To better understand the interplay of these factors, a more useful dichotomy than rational vs. emotional, is one which distinguishes between attribute variables and process variables. Attribute variables are all those describing what information it is that gets processed. Price, per- formance, quality and reliability are product attributes. Delivery time, stock-out frequency and post-sale service are supplier attributes. Punctuality, appearance and product knowledge are sales representative attributes. To date the bulk of published industrial marketing research has been concerned with attributes and their relative impacts on purchase decisions. It has only been relatively recently though that attention has turned to the question of how information about attributes is utilized, or for that matter, if it is utilized at all. Consider the purchasing agent pro- vided a sales brochure comparing two vendors' offerings in which one of the two is clearly shown to be superior. What determines if that information is to be taken at face value or discounted in some way? Further, once assimilated individually, what conditions need be present for that individual to pass this information on to others, and under what conditions will that passage be as forceful advocacy rather than mere giving Of information? That is, process variables are those which determine how much infor- mation is passed along, when it is shared and in what manner. Examples of process variables include participation frequency and quality in group discussion. . Finally, it may be assumed that both process and attribute variables are, to a great extent, situation specific. That is, buyers might well emphasize different criteria in moving from a routine inventory replen- ishment buy to a new capital equipment acquisition and conceivably rules regarding number of participants and depth of inquiry might change as well. Perhaps in certain situations most of the uncertainty associated with decision outcome can be traced solely to attribute variables, in which case processes take on little importance. However, it is a principal assumption in this study that in many cases the incremental variance attributable to processes is significant and worthy of explana- tion. Research Purpose This research has three primary objectives: - Identification of factors relevant to group decision-making processes in an organizational purchasing context. - Development of a causal model relating certain of those factors to vendor selection and other outcomes. - Empirical evaluation of that model utilizing laboratory experimentation. Research Questions As mentioned above this research is an inquiry into decision-making processes and information utilization in small purchasing groups. The sales/marketing manager and field representative alike require a better understanding of the factors that impact on vendor selection. More specifically, each might conceivably raise the question, "What, if any- thing, can I do to influence the manner in which members of the buying center share information in the course of their deliberations?" This perspective represents somewhat Of a departure from the tradi- tional interpersonal influence view of buyer-seller relations. In the traditional approach the task of the vendor is to persuade the buyer of the desirability of the vendor's offering. Several dyadic models now exist which alternatively conceptualize buyer-seller relations as bargain- ing games or social exchange processes. However, in those situations where vendor selection is truly a group effort, the salesperson's task might better be conceptualized as a need to persuade one or more persons inside the buying center to advocate choice of a particular brand or product. Now the task is no longer to simply convince one person, but to convince one person (or persons) in such a way that they would be willing and able to convince others inside the buying center. The presumption of course being that movement to what might be called this extended interpersonal influence model would require an alteration of selling tactics.8' 9' 10 Implementation of this approach necessarily presumes that other avenues of gaining adoption have been or are being investigated and that management is willing to explore possibilities other than the traditional marketing approaches, e.g., product modification, redirection of promo- tional message emphasis, etc. In this case, the question arises as to how vendor firm representatives might purposefully alter the internal workings of the customer firm buying center. Do the bodies of theory mentioned above suggest guidelines as to whether or not this is feasible and if so, which approaches might be most fruitful in diverse circumstances? The answer is a qualified yes. Research evidence suggests that when hierarchical position or status in the group is not externally imposed,11 those individuals who participate more in small group discussions are given higher leadership and influence ratings by their peers when com- pared to those who participate less. Furthermore, certain externally influenceable variables do appear to relate to the amount and quality of participation in discussion. For example, it has been demonstrated that an individual given sole posses- sion of task relevant information receives higher ratings on both number and quality of contributions to the decision.12 1 These findings must be tempered by the caveat that they are the product Of tightly controlled experimental settings. Nevertheless, they suggest an approach to the task of develOping a model of group processes tailored to the problem initially posed. Variable Specification Group decision-making processes and outcomes have been studied from several points of view by individuals in numerous disciplines. These include communication theorists, social psychologists, counseling and training specialists, organizational behaviorists and so on. Each, in 10 turn, tends to emphasize variable sets peculiar to their research goals. For example, organizational behaviorists seeking better understand- ing of group decision-making effectiveness have examined the impact of perceived role and authority structures on norm formation and rate of movement toward consensus. On the other hand, a communication student may wish to artificially restrict amount, timing or direction of verbal messages and measure the impact Of this on time to completion of a task. The variables selected for the present investigation are ones pre- vious studies have shown to influence the extent and manner of use of task relevant information in group Choice situations. These are drawn principally from the communication and social psychology literature. Model Development The second research objective is to develOp a causal model which outlines the manner in which the variables described above relate to each other. This has been accomplished through a deductive reasoning process grounded in both group discussion and group choice theories. The model presented here incorporates elements and concepts originally developed in other disciplines and contexts, that have been adapted to study group decision making in an organizational purchasing environment. It is important to note that the model is more normative than empir- ical in foundation. That is, it suggests an explanation of how individuals should act based upon certain deterministic theories. Model Validation Inasmuch as the model is presented as being of the causal variety, validation is presently possible through either the use of experiments or structural equations analysis. As the latter technique is still in its 11 infancy and its efficacy has not been conclusively demonstrated, a deci- sion was made to conduct this phase of the research utilizing laboratory experiments. While experimental research is encumbered with difficul- ties Of its own, primarily external validity, these have been accepted in order to provide evidence concerning the existence and direction of causal relationships. Too few of the current organizational buyer behavior models have been subjected to rigorous testing of this nature. And it is precisely because questions concerning model validity remain unanswered that potential users of these models often approach them with deserved skepticism. Research Limitations The principal limitations encountered in research of this type can be placed in either of two categories, i.e., scope of investigation and quality of Operationalization. To a great extent these areas are synonymous with concerns over validity and reliability common to virtually all behavioral studies. Research can be undertaken with any of several purposes in mind. Perhaps the most common taxonomy of research goals is one which presents the investigator as seeking to: 1. Describe observable behavior in a systematic fashion. 2. Understand behavior based on certain theoretically posited relationships. 3. Predict future behavior. It has been suggested that Often the type of research design appro- priate to one goal is ill-suited to the others. Dubin13 has character- ized this as the power-precision paradox. The implication is that if 12 one sets out to develOp a theory having good explanatory power, i.e., one designed to facilitate understanding, then that theory is quite likely to be weak in predictive precision. The principal aim of this research is develOpment and validation of an explanatory model, and so the research design necessarily focuses on only a small number of the variables known to be relevant to organiza- tional buyer behavior. As expected, this narrowness in scope greatly limits the predictive validity of the study since only a small proportion of the variance associated with actual decision choices is likely to be accounted for. The other major area of concern centers on the issue of reliability. Nunallyl4 defines reliability in terms of the amount of random measure- ment error present. Reliability is improved when results are internally consistent and reproducible over time or instrument administrations. In the context of an experimental design such as the one being employed in this research, poor Operationalization can cause a reduction in relia- bility in either of two ways. First, the manner of treatment manipulation may be ambiguous result- ing in a wide range of responses within a given treatment condition. Second, the procedure used to record responses may be ambiguous resulting in a large amount of measurement error. Problems of the first sort can be detected through the use of mani- pulation checks. These are questions designed to determine if the experimental treatment has had the desired effect on the subjects. Problems of the second variety can be partially avoided by taking multiple measures of important dependent variables. The extent of such problems is determined by calculating inter-item correlations among dependent variable measures, which are reported as reliability coefficients. 13 Both manipulation checks and multiple item dependent measures have been incorporated in the research instrument to aid in detecting pro- blems with Operationalization in the model testing portion of the study. 10. 11. 12. 13. 14. REFERENCES COpeland, Melvin T., "Buying Motives for Industrial Goods." Harvard Business Review, 2, (April 1924), 303-318. Lehman, Donald R. and John O'Shaughnessy, "Difference in Attribute Improtance for Different Industrial Products." Journal of Market- ing, 38, (April 1974), 36-42. Copeland, Melvin T., Principles Of Merchandising, Chicago: A. W. Shaw Co. (1925), chapters 5 and 7. Converse, P. D., The Elements of Marketing, New York: Prentice- Hall (1930), 300-301. Maynard, H. H., W. C. Weidler, and T. N. Beckman, Principles of Marketing, New York: The Ronald Press Co. (1939), 285. McCarthy, E. J., Basic Marketing, Homewood, Illinois: Richard D. Irwin, Inc. (1978), 178. Webster, Frederick E., Jr. and Yoram Wind, Organizational Buyer Behavior, Englewood Cliffs, N.J.: Prentice-Hall (1972), chapter 2. Mathews, L. H., D. T. Wilson and J. F. Monoky, Jr., "Bargaining Behavior in a Buyer-Seller Dyad." Journal of Marketing Research, 9, (February 1972), 103-105. Busch, P. and D. R. Wilson, "An Experimental Analysis of a Sales- man's Expert and Referent Bases of Social Power in the Buyer-Seller Dyad.” Journal of Marketing Research, 13, (February 1976), 3-11. Wilson, David T., "Dyadic Interactions: Some Conceptualizations." Organizational Buying Behavior, Thomas V. Bonoma and Gerald Zaltman (Eds.), Chicago: American Marketing Association (1978). Hoffman, L. R., "Group Problem Solving" in Advances in Experimental Social Psychology, L. Berkowitz (Ed.), v. 2, New York: Academic Press (1965), 101. ' Collins, B. and H. Guetzkow, A Social Psychology of Group Processes for Decision Making, New York: John Wiley and Sons (1964), 29. Dubin, Robert, Theory Building, New York: Free Press (1969), 19. Nunnally, J. C., Psychometric Theory, New York: McGraw-Hill Book. Co. (1967), 172-73. Chapter II LITERATURE REVIEW In examining the impact of group discussion processes on choice outcomes in organizational purchase settings, there are primarily three relevant research traditions. These are (1) organizational buyer behavior in marketing, (2) small group discussion in communication, and (3) group problem solving in social psychology. Selected contributions from each are reviewed below. Marketing One perspective on this subject considers organizational buyer behavior as a special case of individual buyer behavior, constrained by organizational policies, structure, goals, etc. Following this line of reasoning leads one to begin with individual behavior models and adapt them to organizational environments through inclusion or deletion of variables and relationships.1 In fact, the Sheth2 model of industrial buyer behavior is one such adaptation, being derived from the Howard- Sheth3 individual model. Another approach begins with a recognition of organizational buying as a separate and distinct focal area with a requirement for unique conceptualizations. One then proceeds to develop models that are con- textually relevant from the very outset. Which viewpoint one adOpts Obviously impacts on both substance and procedure in research. Ultimately the issue revolves around one's loCus of interest and philOSOphical bent. That is, one can concentrate on individual behavior with little to moderate emphasis on organization necessitated modifications, or one can treat the individual as merely 14 15 one component in the larger, and more interesting, organizational system. Just as clearly, the dichotomy presented above does not exhaust the range of research approaches, and in fact the fundamental viewpoint adopted in this research is that it is the interaction of individual and organizational factors that provides the most fertile research ground. The following paragraphs highlight that research most relevant to a small group decision-making perspective on organizational buyer behavior with a special view to integration of individual and organizational influences. Although not behaviorally oriented, Industrial Buying and Creative Marketing4 is widely recognized as a seminal work on industrial buying practices. In it, Robinson, Paris and Wind develOped the Buygrid Model which characterizes stages in the purchasing process according to type of purchasing situation. The eight process stages are termed Buy Phases while the three purchasing situations are called Buy Classes. The Buy- grid Model appears in Figure l. The Buygrid provides a framework for organization of thinking about who would likely be involved, which sources Of information and evalua- tive criteria are most relevant and when various tOpics are taken up, for each Buy Phase-Buy Class combination. For example, a firm in a new task situation that is seeking to determine the characteristics of a needed item is quite likely to involve several members of various departments in the deliberations. Many different sources of information would be consulted and competing vendors should enjoy relatively easy. access to members of the customer firm. This situation can be contrasted with supplier selection in a straight rebuy situation in which the procedure has been routinized to the point where normal vendor-customer contacts are made computer to 16 .3005. o.m0.>3m mi... | p mung". cow-:13 .8. 88%... .o 855.25. . 85:2 EEO ca «O 3.888 . 33:95. .0 823.3 a... «13:9:- eo 53.31am . .1332: to 3.2-:- o... SEES—uni . .859 32.82. 3 8.2.33.2... vi .8 Show . Eu: cocoa: mo >551. to. 3558223 3 823.35 . :3: .832. .0 33:2... .3.- §§t§5 E cot-£6.85 cows-o. .22... o t..- .ooo... E032: - no EOE-382 3 coma-32:3 . >3... 23.-3m >30: 3:32 a... 3.: ammm<40 >Dm 17 computer via an automated restocking plan. In this case new vendors are actively discouraged unless serious problems arise with the present arrangement. Arising simultaneously with the recognition that there existed no such thing as a typical industrial buy, was an awareness that neither was there a typical industrial purchaser. In fact, it had long been known that frequently several individuals cooperated and each either consciously or unconsciously played a certain role in the process. This awareness led to the development of the first of the major orientations to be taken up, namely role theory. A role is typically conceived as a set of norms which guide individ- ual activity with respect to persons, Objects, etc., in social environ- ments. The fundamental constructs around which research is organized include role perceptions or expectations, role enactment and role conflict.s An individual enacting the role, let's say, of organizational buyer harbors certain expectations regarding the manner in which he should perform his duties. Some of these expectations are the direct result of the formal norms that constitute his job description. Others, less rigid, yet extremely important, arise out of the individual's perceptions of what it takes to succeed, or at least remain in his particular organizational unit. Theoretically the person then acts in a way consistent with these expectations. Often, as a result of simultaneous enactment of multiple roles, the person is thought to experience role conflict when the demands of the various roles are at. variance with one another. He then must resolve that conflict in some fashion, commonly by acting in accordance with the norms attached to the more salient role. The single most influential role theory approach to organizational 18 buyer behavior is that proposed by Webster and Wind.6 They developed the concept of the buying center and described the roles of the individuals occupying that center. The buying center can be thought of as all those members of the organization who interact with each other during the buying decision process. The principal feature of this con- cept is that the buying center is the locus of decision-making authority and its boundaries usually cut across formal structural or departmental lines. Membership in the center is dynamic and will shift across stages in the decision process and characteristics of a given buy. Each member of the buying center enacts one or more of the follow- ing roles: user, influencer, buyer, decider, or gatekeeper.7 Users are those individuals who will actually utilize the equipment, material or component part under consideration. They often initiate requests for specification changes and are most often concerned with operational product features. Users may exert influence singly or in combination and commonly are consulted about the performance of existing equipment as one source of information in establishing qualifying criteria. Influencers are any members of the organization who directly or indirectly influence purchase or usage decisions. Influencers may act purposefully out of a vested interest in choice outcomes or may not even be aware of their influence potential. This latter case occurs when, for example, one department head observes the requisitioning behavior of another, whom he desires to emulate, and so patterns his requisitions accordingly. Buyers are those members of the center who are Charged with the formal authority for vendor selection and purchase negotiation. They typically occupy the position of buyer or purchasing agent in func- tionally specialized industrial firms, or may be members of a formal 19 buying committee in reseller firms. The buyer's influence is felt most strongly in the areas of establishing financial criteria and coming to the actual terms of sale. Also the buyer may act to limit the range of qualifiable vendors based on other criteria, for example, logistical capabilities or past customer service record. Deciders are those who, through the use Of either formal or infor- mal power, actually decide what will be bought and from whom. Deciders may come from various activity areas, and it is for this reason that researchers frequently study buyers rather than deciders--the decider being very difficult to identify.8 It is not uncommon for decision authority to be split, i.e., one person or group decides the specifica- tions to be used in vendor qualification and another chooses from among the qualified vendors. Gatekeepers are any and all individuals who control the flow of information into and out Of the buying center. Gatekeepers play a critical role from the marketer's point Of view, since they may act to prevent salesman contact with certain individuals or fail to pass along key information to deciders or others. The buying decision process then is the product of a complex set of social relationships among members of the buying center, whose behaviors are at least partially prescribed by the norms associated with these roles. In a recent study9 among purchasing and non-purchasing execu- tives of manufacturing concerns in Cleveland, Ohio, it was found that role-related perceptual differences were significant in 16 of 65 vendor selection attributes investigated. These perceptual differences are suggested as a source of conflict among members of the buying center. A similar study10 involving firms purchasing chemical intermediates demonstrated that role-related perceptual differences extend to the 20 actual members Of the buying center as well. That is, the perceptions of purchasing agents, management and technical specialists regarding their relative influence on and responsibility for vendor choices were found to be significantly different. More specifically, when the component stages in the procurement process are examined individually, a descriptive study11 involving 76 representatives of 15 Columbus, Ohio, machinery manufacturers suggested that, for example, a far lower percentage of top managers than purchas- ing agents expect functional managers to be responsible for establishing specifications and determining the amount to be purchased. However, there was general consensus that purchasing agents should select suppliers and conduct post-purchase appraisals. An Obvious failing of published role-related research lies in the narrow way in which roles have been conceived. Only the easily defin- able roles provided by job description or title have been used. Neither the perceptual nor behavioral differences suggested by the Webster and Wind taxonomy have been investigated. A potential solution to this problem has been suggested by Calder12 who advocates use of a technique called structural role analysis to integrate individual and organiza- tional determinants of the decision process. When considering organizational buying from a small group perspec- tive additional role concepts are needed, specifically that of group leader. Nomms and behaviors associated with leadership roles are considered in the following section. Leadership in a small group may be obtained, principally, in either of two ways; it may be conferred by the group members out of respect or deference, or it may be appointed through formal organizational rules. The concepts of leadership and power are closely interwoven. Those who 21 lead have power, and those who have power influence decision outcomes. According to a widely employed typology13 there exist fives bases of social power, namely expert, coercive, reward, referent and legiti- mate. Briefly, expert power accrues to those having greater knowledge or comprehension of task relevant information, coercive power results from an ability to force compliance through threat of sanction, reward power comes with ability to provide desired reinforcements, referent power is the product of one person's desire to emulate or be similar to another and finally, legitimate power is the outgrowth of formalized norms indicating that compliance is expected. Traditionally, legitimate power receives the most attention as it results from the formal organizational hierarchy and is therefore most visible. In this case ”leader" is synonymous with manager or supervisor and the salient role expectations revolve around providing direction, choosing among alternatives, and implementing planned for activities. It is only more recently that students of organizational buying have pursued other power-related approaches in their attempts to identify and characterize decision makers in various environments. In a review article probing the question of the mechanisms by which members of a buying center reach joint decisions, Bagozzi14 high- lights two Of the newer approaches. The first, termed Social Judgment Theory, is a model developed to consider human judgment processes in situations involving two or more persons. It is most appropriate in those circumstances in which cognitive conflicts exist among the group members. The type of conflict studied is rather narrowly defined to include only that resulting from differences in models applied to solu- tion of particular problems. These model differences commonly center around the form of the function relating stimulus cues to judgments, cue 22 weighting strategies, information integration rules, etc. Subjects are required to study a stimulus array of some sort, make an individual judgment, discuss the problem with each other, and finally, arrive at a joint decision. Measurements are usually taken on several dependent variables including individual and joint problem solutions and pre- and post-discussion peer evaluations. Empirical research based on the SJT model can be summarized as follows: - subjects do not resolve or change their initial overt judgmental differences - subjects do however alter the models upon which the judgments depend or act in a fashion inconsistent with the original model - changes in judgment models result from social interaction in the discussion phase. The second is a utility-degree of control model of social action develOped by Coleman.15 Operationalization of the model requires gathering data on the degree of control each actor has on the events associated with the collective decision and the level of interest each actor has on the outcomes of those events. When both sets of data are cast in matrix form, one is able to determine: - the value of each event - the overall power of each actor - the outcome of each event. Of particular interest is the ability of the model to specify the extent to which actors have power over each other. This is done by relating various actors' levels of interest in a particular event to their respective degree of control over that same event. Applications of this model are limited in scope and number as it 23 presumes ability to accurately measure the degree of control over and interest in "events." In addition it rests upon the economic assump- tion that each actor is motivated by a desire to maximize control over these events Of interest. Yet a third power-related conceptualization of organizational buyer behavior processes is suggested by Zaltman and Bonoma.16 It is called the Locus of Influence Grid. The grid consists Of a four-celled matrix, reproduced in Figure 2, which isolates both internal and external sources of influence at both the department and corporate levels. This approach suggests areas appropriate for investigation; however, it does not stipulate specific research hypotheses and as yet has not been utilized in published empirical work. In describing the grid the authors note that the bulk of organizational buyer behavior research has focused on areas relevant only to the first cell. The concept of power is Of central interest to students of organi- zational buying. To date most discussions of power have been directed toward identifying those who have power over particular decisions and specifying how that power is vauired.* The present research also considers the role Of power in group purchase decisions, however, in a more narrowly defined way. Specifically, to what extent does the sales representative have an Opportunity to influence the amount of expert power held by one or more members Of the buying center through control of distribution Of task relevant information? Rather than attempt to measure power directly,. the research approach is to measure the consequences of unequal power *The supposition being that once the mechanisms of power acquisition are uncovered we will be better able to predict who will have power in future choice situations. 24 X_m._.4¢2< JEDFJDO + m:.-(> o HEP—.0335.” J<20.h<~.2(0¢0 0 “805.0(5 J8 ...(Om.. O g2w3h¢2u o CNEO a.) ( a.) 02.23082 0 .>. .= agar—i.- .. «hm! 26:.(923380 O uma>a<2< 8a:— 0 «Jun—O! 2°.h(>022_ to zoaauufiu O “we ZQQOWO o «4(2549. + £01m mac... 0 «4wa 2920m0 o 3.4%.89-033 o «CPR-(m A.OZ.<¢F2. «(.— NE 0 .3. .- mmUZmDamZ. muOZqu-Z. Scamp—.2. 2¢I<¢h2= «300.. mh<¢0mmoo mmUZwDJn-Z. J<1mm mm>3m film—.5302. Imem m:... l m mmDOE 82.9.3. 42¢ 2.88... so 2;... .8285. SE... 533.1530 .3 .O 8500 608-5590 8885.5 8.35—5.20 41 5.535 13302!— I 33.5 :5. 9.35.... .v . 228.. n :88“. 58212.3; 2353.5 “a 3.3555 3.3.6.833... 5.3:... N as .5 5.2 a...» carom 5.32.. .. F . c.5203: - - a 89.28.30 85:00 _ .8305 882.. 33.39.... as. 3. , _ «£2. 2.3.5 . 1:33:05 , _ 132...... 4 H .5, , .850 .v 3.58:? - :8: n .352. 8.2.0 v.35 2.8.5.5 ”a 3 5395 85¢ 353.85.. p 88...: 82 _ , - 1 .O a l _ L — .3385 .I I IV 2.9.8833 59.3.35 .2 =02 825 _ if . u . d , _ » “ 5....» as; 32.95.. .3332. I - a... 3.3.3:.xw . .2... 2.. .o . 5...» Fl lllllllllllllllll 5583.6 1.5053 8.8:. s: s: a: 52:3. :83. at... 8.5.85 cos-2.3 8:25.... 18.833 .3 Soc 3338..» 3: .3 - , 42 chapter presents such an exploded or blown-up view of the box labeled Conflict Resolution (3) in the Sheth model. Before proceeding with a detailed description of the concpetual under- pinnings of the model it seems appropriate to comment on the guidelines which shaped it. There exists amongst scientists no small amount of controversy over the "proper" method of theory generation. For example, Glaser and Strauss8 argue strongly for an inductive approach. Elaborat- ing on what they term the discovery of grounded theory, they pointedly remark that "grand theorizing," while esthetically pleasing at worst and vital at best, does not provide direction and fails to close the gap between pure abstraction and a "multitude of miniscule substantive studies." In fact, they contend that the main goal of the social scientist is to develop new theories in a purposeful systematic fashion "from the data of social research."9 That approach has not been adopted here. Rather the new model was grounded in another way. Through careful consideration of previous empirical work, in marketing, buyer behavior and other behavioral sciences, a deductive development process was followed which assures continuity and facilitates integration. The function performed by the inductive approach, namely, assurance of relevance in the sense of external validity, is assumed by the validation procedures described in the following chapter. In addition, it was felt that the new model should meet several specific criteria, in particular it should: - Explicitly indicate hypothesized relationships in a way suitable for empirical test, - Be capable of integration into existing comprehensive models, especially the Sheth model, with a minimum of modification, 43 - Be amenable to and, in fact, motivate research at the group rather than individual level of analysis. As mentioned above, the point of departure for the new model is that portion of the Sheth model labeled Conflict Resolution. He notes: The most important aspect of the joint decision-makingiprocess, however, is the assimilation of information, deliberations on it, and the consequent conflict which most joint decisions entail. According to March and Simon conflict is present when there is a need to decide jointly among a group of people who have, at the same time, different goals and perceptions.10 (Emphasis added) Thus the decision-making process is conceptualized as being synonymous with a conflict resolution process. Emphasis on the manner in which this occurs is apparent, yet Sheth's discussion goes no further than to high- light various general types of processes. That is, when conflict results from disagreements over vendors' offerings or criteria used to evaluate those offerings, what are termed "rational" processes, problem solving and persuasion, are likely to be invoked. Behaviors characteristic of these processes are seeking of additional information, consideration of new suppliers, or heightened conformity pressure on deviant members. On the other hand, when disagreements are more fundamental, when attitudes are more strongly held, or when sources of disagreement are personality or politically based, processes of bargaining and politick- ing come into play. Here discussions are much more likely to become heated, emotional and punctuated with personal attacks. One explanation for existence of these conflicts and their charac- teristic resolution modes, despite supposed common interest in achieve- ment of organizational goals, is disparity among personal goals of group members. Frequently organizational goals are ambiguous, or not immedi- ately relevant to the day-to-day decision-making environment of lower 44 and middle management personnel. Thus, what is perceived as best for the individual and his department or operational area is often simul- taneously perceived as being best for the organization as a whole. Given that individual expectations and attitudes play such an important role in decision-making processes, a question remains as to the nature of mechanisms employed by individuals, in the context of a particular organization and situational environment, in assimilating and subse— quently using information. Insight into~these mechanisms is provided by first considering a generalized input-output model of group processes. The one selected for discussion here is a modified version of one presented by Hackman and Morris,10 and appears in Figure 4. Here interaction processes include all actions engaged in by group members between times 1 and 2. Conceivably they might also include strategies, etc., as well; however, the authors chose to limit discus- sion to observable behaviors. The time period between t1 and t2 is purposefully left unspecified and may be thought of as encompassing a distinct communications "event," for example, a meeting or conference call. The complete decision-making process may include several such events occurring in serial fashion. Examination of the model reveals the process focus. If one syste- matically varies one or more input variables and subsequently observes consistent changes in one or more dependent variables, then the process may be inferred to have played a mediating function. Clues to the mechanism underlying that function are secured through direct observa- tion of differing processes. Alternatively, when direct observation is impractical, participants' post-discussion perceptions may well suffice. Utilizing this model, or similar models, as an organizing framework, 45 .5002 m_mm02.2<2v_o0ao.=.oE:o...>:u ‘ 1:26 b.9330 ; on... umocozaorou. 2225.... :23.— .25....320 mmuUOmm 3:2.0309. nonzero. «:3... 9.290.. _o>o._._a:2>.vc_ =32. 46 examples of relevant research findings include: - Obvious solutions tend to be adopted quickly, while less obvious, though correct, solutions may remain totally undiscovered. - Members generally will give consideration to less obvious solu- tions after having been made aware of them. - Group leaders who are permissive and encourage evaluation of multiple alternatives perform better in helping the group to avoid obvious but incorrect solutions.11 More precise specification of the exact mechanisms of process func- tion should be possible when the entire input-process-output sequence is considered. Using a canonical correlation analysis to link sixteen predictor activity dimensions, e.g., defend, clarify, repeat, to eight group product dimensions, e.g., originality and adequacy, across three intellective task types, Hackman and Morris12 reported several signifi- cant relationships. For problem-solving tasks, which most closely resemble the type of task of interest here, group members' attempts to structure an answer and the sequence of seeking calrification, defending an alternative and repeating were highly positively correlated with creativity, quality and other criterion variables. However, an overall pattern of correlations explaining the entire input-process-output sequence was not found. This result was also obtained in a similar study by Sorenson in which significant relation- ships between input-process and input-output were found but the full sequence was not elaborated.l3 A suggested explanation for this failure centers on the use of what is termed a "molecular" focus to categorizing the mediating process. Anderson14 in discussing impacts of source-message interaction on influence attempts and attitude change in group discussion notes that 47 each statement by a member constitutes a unique source-communication- issue unit. In a molecular analysis one attempts to relate individual units or patterns of units to unique outcomes. A representative example of the molecular approach is the earlier mentioned work undertaken by Vinokurls and others to explain choice shifts following group discussion. The model is of the subjective expected utility type in which each argument relevant to the group deci- sion, in this case alternative choice, is supposed to have measurable impact on group members' preferences by either enhancing or lessening the perceived utility or value of one or more of the alternatives. Mathematically, n SEU a}: P.U.. {1} J. 1. i=1 where SBU is subjective expected utility, Pi is the ith person's subjective probability estimate of an alternative's success, Ui is the ith person's perceived utility associated with an outcome following choice. Thus each alternative will have a group level utility value found by summing over group members the product of their individual prob- ability estimates of an outcome's occurrence and the utility of that occurrence. In the context of the risky shift investigations, choices were limited to two, risky and cautious alternatives, and outcomes were also binary, success vs. failure. To make the model mathematically tractable it is assumed that, U > U > U {2} s c f. Where Us is the utility derived from success of the risky alternative, 48 Uc is the utility derived from the cautious alternative which is certain, i.e., PC a l, Uf is the utility derived from failure of the risky alterna- tive. From which it follows that, SEUr =- p Us + (1 - p)Uf, {3} 515thc .. 1 Uc, {4} and if SEUr : SEUC, {5} then p US + (l - p)Uf 3_l° Uc, {6} so that p 3_Uc - Of, {7} Us - Uf where SEUr refers to the utility associated with the risky alternative and SEUc to the cautious alternative's expected utility. Thus the lowest probability of success a subject will consider acceptable is estimated by quantifying their utilities for the alternatives.16 Relating this back to the Hackman-Morris model, if utility measure- ments are taken at various time periods before and after group discus- sion it can be determined whether a choice shift has occurred. That is, if p2>pl a risky shift has occurred and vice versa. As was mentioned earlier, much research utilizing dimensions other than risk has now taken place; however, the fundamental approach has remained the same. Operationalizing the above model at the molecular level involves acquiring from each group member a written list of arguments they have generated pro and con to each alternative, and utility measures prior to the laboratory discussion session. Based on the number and distribution of arguments across group members, an index called a diffu- sion potential is derived which indicates the average number of people in the group who were not previously aware of a particular argument but 49 were exposed to it during the group discussion. The influence potential of each argument is found by taking the mean of the group members' self- reported persuasiveness scores. Finally, the mean total impact of all of the arguments favoring choice of a particular alternative is given by, k k MTIr = Z Dixi / E Di {8} i=1 i=1 where MTIr is the mean total impact of the arguments favoring the t . r h alternative, D1 is the diffusion potential of the ith argument, Ii is the influence potential of the ith argument. When the alternatives under consideration are the dichotomous risky vs. cautious choices than MTIr becomes MTR and MTC in the Vinokur framework. Furthermore when MTR>MTC a pro-risk choice shift can be expected to occur. From a communication theorist's point of view, the above model incorporates only message factors as independent variables. Another molecular model of group process incorporating source and issue, as well . . 18 . as message, factors is proVided by Anderson. That is, n R = C + Z w. s. {9} l 1 i=0 where R is a numerically scaled response, e.g., an attitude or pre- ference score, C is a scaling constant permitting an arbitrary zero point, w is the psychological importance of the ith stimulus to the subject, th s is the scale value which measures the location of the i stimulus on a judgmental dimension, e.g., persuasiveness. 50 Now, if the influence of source, message and issue factors are permitted to vary independently, equation {9} above becomes, L M N . R=C+Z X Zwlmnslm. {10} 1:0 m=o n=o Where Wlmn is the psychological weight or value of the 1th informa- tional stimulus, along the mth judgmental dimension, from the nth source. S is the scale value or message polarity of that same stimulus on the same dimension. The Anderson derived model can be related to the Vinokur model by equat- ing stimuli in the former to arguments in the latter and responses to utility estimates respectively. Both models stress the impact of informational influence on attitude change, group process and alternative choice, although the Anderson formulation provides somewhat more flexibility in this regard by incor- porating affective elements. Application to the current problem is limited by the constraints imposed by the choice shift framework. Exami- nation of the Vinokur model reveals that choice options are held constant, that is, considered pairwise with a certain alternative, while changes in utility scores across time and individuals in the group are measured. The model becomes conceptually more interesting if options are allowed to vary as well and changes in probability of success estimates used as indicators of adoption likelihood. That is, we are less concerned that Vendor 1, initially preferred by group members prior to discussion is preferred even more so after discussion, even though a group shift has taken place. However, if pre-discussion p2>p1 and post- 51 discussion p1>p2, we may infer that during the discussion arguments were brought before the group which caused a preference swing type of shift. Returning to the Anderson model it is clear that each statement made by a group member constitutes a unique message-source-issue combin- ation. Each such combination favoring a particular option increases its perceived utility to the extent that several conditions hold. These are that the statement itself be persuasive, that the communicator, i.e., group member, be perceived as credible and finally that the message content of the statement be perceived as relevant to the task at hand. Employing a utility maximization rule, the group will select the alterna- tive having the highest SBU, hence lowest p, following assimilation of all of the distinct source—message-issue events which took place. However, if the group employs a choice simplification rule, either consciously or subconsciously, the first alternative whose p falls below some critical value, either consciously or subconsciously developed, will be selected and deliberations cease.19 Model Specification At this point the necessary background has been provided to permit discussion of the process model which serves as the focus of the present investigation. The model actually consists of two distinct but related sub-models each designed for different purposes. The first, the generic decision process model, appears in Figure 5. The generic group process model was derived directly from the Hackman-Morris model described earlier but with modifications to both highlight key features and permit better integration of the specific. model into the Sheth framework. The Hackman-Morris model does not by 52 .3002 mama": ZOE—Duo UEmeG mi... | m NEDGE n. J a. A... .. o. 92; 00000.50. «00000.00. .0.000.00._>00. 3000.0 003.000 3000.0 005000 .0.000.00._>00. 00000.50. 0. 0.0.. 030. 0. 0.0. 030. 000000.00. .00o_.0~_0om.o. 0.00.3300. 9.3.5300. 300.50.00.30. ”.3 00500.0 20.009300. .0000) 3000.. 300. :3 002.000. -tIL 02.00.0090. ,, 000 8000.0. ’1, 000 8000.0. IL 11 000000005. 302300.. 2.03.0 00.3000. 3.03.0 00.0.00... 00:20.20. 0.0002305. «002.20. 0.007000. 0.0.30.0 8030.0 0,002.20. 000200050. .30.. .0 0.200. .0 0.5.00. «00:03.00. 302210. 000.330 52.0 00$0£3< 530 0010::< .002200. H .009...— .00200H_ \ .— n00£00wll C.,... T\ ? 1,5 00.30005 \ 000.0 / s ifi n00»...— L 53 itself adequately provide contextual relevance to group purchasing situations, likewise the Sheth model is most relevant to the purchasing situation yet lacks process orientated detail. The generic process model fills this gap. One function performed by the generic model is to serve as an organizing bridge between the conflict resolution compo- nent of the Sheth model and the substantive conceptual content of the specific model which appears in Figure 6. Another function the generic model performs is to stress certain points important to consideration of research design issues in the group problem-solving and decision-making area. Concerning the latter is is readily apparent that research emphasis must remain with the individual as the unit of data collection. Investigations employing a group or organizational unit as the unit of data collection fail to recognize the significant role played by intra-departmental personal influence. On the other hand, to study the individual in isolation is to deny the existence of social and organizational factors entirely. An acceptable compromise is struck when individuals remain units of data collection; however, analysis is performed at the group level by employing either summated or averaged variables indicating group outputs. This approach is evidenced in the choice shift area as mean pre- vs. post-discussion utility estimates are used to detect shifts. In addition to the emphasis on the distinction between unit of data collection and level of data analysis the generic process model makes explicit the time sequenced nature of organizational purchasing. More specifically, it focuses attention on a face-to-face group discussion as one, albeit very important, event in a series. Thus in many cases the actual process outlined in the buygrid model, i.e., from problem recognition through alternative evaluation to final choice extends over 54 mU.O...o 030m0 “.0 .5002 wmmUOmE 4O-——J—‘.bu . 0.3.. . . 30535500 . u .0: 33303.3. - - .L —| «02303.3. 00.00... "7 3500.00 " - ‘ 3.02.3 4 . _ > 0000—3300 w. 00.3000 .n .023 0010....< L4 F0030:— j _ 30000.0( _1 x .023 310...; 02.x; ZOGUmO 0:080 00 awn—0: 9.30080 ( 00.3533. 00.25133 — 55 several weeks or months. However, it is quite likely that the individ- uals involved may spend only a fraction of their working days in this time period on this particular problem. For this reason, despite the lack of time spent in group meetings, their value is enhanced due to the opportunity provided to express Opinions, air differences, decide on next steps, etc. Inasmuch as individuals do move into and out of tasks over time, public positions taken in group meetings may be made from relatively sparse information bases. Thus vendor induced alterations in those bases may prove fruit- ful. If one pursues a molecular paradigm as described earlier, the building block approach is continued down to the meetings as well and these are viewed as a sequence of distinct source-message-issue events. The other alternative is a molar approach which focuses on influence strategies adOpted by group members rather than observed patterns of interaction. Examining the generic model it is evident that the present inves- tigation more nearly approximates the molar approach, as interaction sequences within the discussion phase are not elaborated. As with the Hackman-Morris work and many other group-oriented studies direct measures are taken only of inputs and outputs while conclusions about processes are drawn inferentially. Despite inability to puruse the molecular approach within the present study's framework, it remains of great value in focusing attention on cognitive processes thought to underlie interaction. Returning to the first function performed by the generic model, it assists in making more plain what is meant by Sheth's discussion of bargaining as a conflict resolution mode. That is, the bargaining 56 process is one characterized by trading of information rather than money or tokens in which participants seek to enhance personal status and simultaneously achieve task-related goals. In this light is is clear that individuals' attitudes about alternatives under consideration are not the only ones that are influential in their behavior. An individual's perceptions of task importance, his own ability to be influential, and others in the group should color his approach to the problem. Following choice, decision participants develOp attitudes about themselves, others and the entire process which are largely a function of the conflict resolution mechanism. In the longer term they receive feedback about the quality of their choice from other organizational members. Thus outcomes experienced by users of newly purchased equip- ment become translated into personal outcomes for decision makers so that the experience base upon which the next decision rests is different from what it was previously. Turning now to the specific model, one can see how the particular variables of interest are Operationalized. Recalling the discussion of communications research in the preceding chapter, it was noted that several factors are thought to influence interaction mechanisms and out- puts. The three factors chosen for investigation in this study are distribution of information among group members, perceived credibility of an attributed source of that information, and finally language intensity or strength of claim made in presenting the information. Individually each has been demonstrated to have some impact on group processes and outcomes, e.g., willingness to contribute and satis- faction. It is not known at present what, if any, interaction effects exist among them. Information distribution and claim strength are the two most vendor influenceable variables, while source credibility plays 57 more of a moderator role. That is, a sales representative can feed information to only one or all group members and may do so using rela- tively neutral or strongly worded language. The extent to which these factors influence individual attitudes is in turn a function of the sales representative's perceived credibility. Based on substantive content of information presented and these three factors, individuals develop attitudes regarding alternatives and also evaluate their own position in the group regarding this decision. That self-evaluation is primarily reflected in specific self-confidence which is conceptualized as a causal factor in the individual's willing- ness to utilize whatever information is available to him. It may also be reflected as commitment to an alternative; however, measurement of commitment prior to discussion constitutes a form of public disclosure which has been shown to influence strength of commitment, and so to avoid an instrumentality threat to internal validity commitment is not expressly tested. The actual influence mechanism is the extent and quality of active verbal advocacy by individuals for and against available alternatives. Both group problem-solving students in social psychology and those following the informational influence school in the choice shift area have noted strongly that verbal advocacy is the key determinant of choice. Following decision individual members once again solidify attitudes about themselves, alternatives and other group members as well since discussion has provided an opportunity for peer as well as self- evaluation. Note that perceptions of other group members are not observed prior to discussion as there is no theory to suggest these 58 factors should influence peer perceptions until after discussion has taken place. Finally, longer term outcomes and consequential feedback mechanisms will occur; however, time limitations imposed by the research design chosen for model validation prevented measurement of their type or extent. Model Validation Examining the time line appearing at the bottom of Figure 6, times at which research events occur are noted. At t1 subjects are given research materials and given time to assimilate information in them, at t2 individual level measures of alternative attribute ratings, over- all preference, behavioral intent, perceived source credibility and specific self-confidence are taken. At t3 post-discussion alternative attribute ratings, overall preference and behavioral intent are repeated and in addition self- and/or peer evaluations over dimensions of liking, perceived advocacy behavior and perceived leadership are taken. Utilizing the scores described above, the following substantive hypotheses are tested. I. a. Information distribution influences pre-discussion individual attitudes, self-confidence and behavioral intent. b. Information distribution influences post-discussion group mean attitudes, behavioral intent and perceived advocacy and leadership. II. a. Perceived credibility of an attributed source influ- ences pre-discussion individual attitudes, self- confidence and behavioral intent. III. IV. 59 b. Perceived credibility of an attributed source influ- ences post-discussion group mean attitudes, behavioral intent and perceived advocacy and leader- ship. a. Strength of claim (language intensity) influences pre- discussion individual attitudes, self-confidence and behavioral intent. b. Strength of claim (language intensity) influences post-discussion group mean attitudes, behavioral intent and perceived advocacy and leadership. The independent variables have a combined effect which is greater than the sum of their individual effects. There exists a relationship between perceived advocacy favoring an alternative and group adOption of that alter- native. These substantive hypotheses are tested by means of the following working hypotheses. H o Ial The main effect for information distribution on individual pre-discussion overall vendor preference is zero. The main effect for information distribution on individual pre-discussion specific self-confidence is zero. The main effect for information distribution on individual pre-discussion behavioral intent is zero. The main effect for information distribution on post- discussion group mean overall vendor preference is zero. The main effect for information distribution on post- discussion group mean behavioral intent is zero. The main effect for information distribution on post- Ib4 6O discussion peer evaluations of subjects' advocacy behavior is zero. The main effect for information distribution on post- discussion peer evaluations of subjects' leadership behavior is zero. Null hypotheses associated with substantive Hypotheses II and III, that is, H o . to H and H0 to Ho 4, are identical to those IIa IIb4 IIIal IIIb enumerated above except that the appropriate factor names are substituted for information distribution. o IVl IV3 1V4 IVS IV6 IV7 The two-way interaction effect of information distribution and source credibility on individual pre-discussion overall vendor preference is zero. The two-way interaction effect of information distribution and source credibility on individual pre-discussion specific self-confidence is zero. The two-way interaction effect of information distribution and source credibility on individual pre-discussion behavioral intent is zero. The two-way interaction effect of information distribution and source credibility on post-discussion group mean overall vendor preference is zero. The two-way interaction effect of information distribution and source credibility on post-discussion group mean behavioral intent is zero. The two-way interaction effect of information distribution and source credibility on post-discussion peer evaluations of subjects' advocacy behavior is zero. The two-way interaction effect of information distribution 61 and source credibility on post-discussion peer evaluations of subjects' leadership behavior is zero. The information distribution by claim strength and source credi- . . o o o bltb 1' t tt- ' ' ' 1 1 y y c aim s reng h wo way interactions , HIVB to HIVl4 and HIVlS o . . . . . . to HIV21 respectively, and the three-way information distribution by . . . , . . . , o 0 source credibility by claim strength interactions , HIV22 to HIV28' null hypotheses are identical in format to HIVl to HIV7 respectively. o Hvl: The Pearson product moment correlation between post- discussion peer evaluations of subjects' advocacy scores and post-discussion group mean vendor evaluation scores for alternatives favored by subjects is zero. o o Ial through HIV28 flow and other details of model validation are the subject of the following The statistical model from which hypotheses H chapter. 10. ll. 12. 13. 14. 15. 16. REFERENCES Scott, Jerome E. and Peter Wright, "Modeling an Organizational Buyer's Product Evaluation Strategy: Validity and Procedural Con- siderations." Journal of Marketing Research, 13 (August 1976), 211-24. Webster, F. E., Jr., "Industrial Buying Behavior: A State of the Art Appraisal," in Marketinggin a Changing World, B. A. Morin (ed.), Chicago: American Marketing Association (1969), 256. Ferguson, Wade, "A Critical Review of Recent Organizational Buying Research." Industrial Marketing Management, 7 (1978), 225-30. Wilson, David T., "Models of Organization Buying Behavior: Some Observations," in Buyer[Consumer Information Processing, G. D. Hughes and D. L. Ray (eds.), Chapel Hill, North Carolina: The University of North Carolina Press (1974), 136-41. Gr¢nhaug, Kjell, "Autonomous vs. Joint Decisions in Organizational Buying." Industrial Marketing_Management, 4 (1975), 265-71. Lambert, D. R., R. J. Dornoff and Jerome B. Kernan, "The Industrial Buyer and the Postchoice Evaluation Process." Journal of Marketing Research, 14 (May 1977), 246-51. Gr¢nhaug, Op.cit., 269. Glaser, B. G. and A. L. Strauss, The Discovery of Grounded Theory: Strategies for Qualitative Research, Chicago: Aldine Publishing Co. (1967), 97. Ibid., 28. Sheth, J. N., "A Model of Industrial Buyer Behavior," Journal of Marketing, 37 (October 1973), S4. Hackman, J. R. and C. G. Morris, "Group Tasks, Group Interaction Process and Group Performance Effectiveness: A Review and PrOposed Integration." Advances in Experimental Social Psychology (1975), 45-610 Ibid., 52. Ibid., 55. .121 56- Anderson, N. H., "Integration Theory and Attitude Change." Psycho- logical Review, 78 (l97l), 196-98. Vinokur, A., "Review and Theoretical Analysis of the Effects of Group Processes Upon Individual and Group Decisions Involving Risk." Psychological Bulletin, 78 (1971), 247. 17. Ibid., 247. 18. Vinokur, A. and E. Burnstein, "Effects of Partially Shared Persua- sive Arguments on Group Induced Shifts: A Group Problem-Solving Approach." Journal of Personality and Social Psychology, 29 (1974), 310-11. 19. Anderson, N. H., op.cit., 197-8. 20. Krapfel, R. "A Decision Process Approach to Modeling Organizational Buyer Behavior," in Proceedings, Educators Conference, Chicago: American Marketing Association (1978), 116-20. Chapter IV RESEARCH METHOD Experimental Design The model described in the previous chapter suggests three indepen- dent variables, information distribution, source credibility, and claim strength as being causally related to four dependent variables, specific self-confidence, perceived advocacy, vendor evaluation and behavioral intent. In order to test independent variable impacts, singly and in combination, two experiments have been designed. After a brief discus- sion of behavioral science experiments, each is described in detail. The purpose of experimentation is to test hypotheses. The general model for a single factor experiment, i.e., only one independent variable, may be represented by: Yij = u + ai + €ij° {1} Where Yij is the observed value of the dependent variable on the jth subject following application of the ith treatment level, u is the effect common to all subjects regardless of treatment conditions or individual differences, ai is the effect due to the independent variable, eij is the error component arising from both random and non- random sources. The null hypothesis tested is: H: (1.30 {2'} The statistical test employed is the F-test in which a calculated F- statistic, given by, 62 63 a _ _ 2 as _ 2 F=S §(Ai-T) |/d£a : E; (Asij-Ai) /dfe {3} is compared with the apprOpriate critical value for a desired confidence level. Thus, treatment means, ii' are compared to the grand mean, T, and if the difference is larger than that which would have been expected on a chance basis alone, the null hypothesis of no treatment effect is rejected. A common variant of this type of design is post-test only with control as described by Campbell and Stanley.1 However, the theoretical group process model involves not one but three independent variables or factors. While any of several designs may have served, a fixed effects three-way factorial design was chosen for initial model validation, followed by a two-factor repeated measures design for cross validation. Factorial design characteristics indicating its choice are: l. The design is ”fully crossed" which is to say that all possible combinations of independent variables are observed. This feature aids in specifying interactions among factors and is a compelling argument for use of a fully factorial design. Only where time and money constraints require, or prior evidence warrants, should incomplete designs be employed, i.e., those not incorporating all_independent variable combinations. 2. The design is efficient in the sense that relatively large amounts of information may be gained from relatively few subjects. That is, several single factor experiments will not provide the same amount of information as a single factorial experiment incorporating the same independent variables and using the same total number of subjects. 3. Factorial designs provide for controls necessary for adequate 64 internal validity without requiring a separate control group to which no treatment is applied. The general model for a three-factor, fixed effects experimental design is2 .. = +. . .. . . .. .. Ylel u al+BJ+yk+oBlj+aylk+ByJk+a8yljk+sljkl {4} where Yijkl is the observed dependent variable value for the 1th subject, ai'Bj'Yk are the main effects for each of the three factors when averaged over all levels of each of the other factors, .. . . t - ' ' 0813,071k,8y3k are the wo way interactions among factors, asYijk is a three-way interaction term, 6.. is t e or com onent. ijkl h err p Several null hypotheses are tested: : 3 . 0: a: 5 Bl oi B3 Yk O { } Ho. (18 :0 =3 8 =3 0 {6)- 2' ij o‘Yik ij Ho- 08 = 0 {7} 3' Yijk The statistical procedure employed to test these hypotheses is three-way analysis of variance (ANOVA). For each of the seven separate variance components identified above, an F-statistic is calculated and compared to a Critical value for the appropriate confidence level. Table 2 is a general example of a fixed effects, three-factor summary ANOVA table. 64 independent factors are non-metric, i.e., do not meet interval level criteria. Figure 7 summarizes the student sample experimental design. The practitioner sample experimental design is a fixed effects, repeated measures two-factor design. In the design described previously, each experimental subject is exposed to only one unique combination of treatment levels. This is not the case in a repeated measures design. Here subjects receive more than one treatment application. Repeated measures designs are often employed in studies with cost or sampling restrictions. In general, non-repeated measures designs are preferred as sequential exposure to multiple treatment combinations introduces a threat to internal validity termed a carry-over effect. That is, if one treatment combination consistently follows another, effects measured in the second may be the result of factors contained in both treatments. This produces confounding in interpretation of the second treatment's effect. The problem is commonly handled by randomizing order of treatment presentation and that procedure was employed here. In addition to moving to a repeated measures design a decision was made to exclude the claim intensity factor from the practitioner experiment. This was done due to limited practitioner sampling frame size and the fact that preliminary analyses of student sample data indicated that claim strength is weakest in explanatory power. While the general model for this type of design is essentially similar to that described above, computational procedures change to accommodate introduction of repeated measures. Furthermore, the design incorporates repeated measures only with respect to one factor, source credibility. That is, each experimental subject was exposed to only one 65 Table 2 Example Fixed Effects, Three-Factor ANOVA Table Source Sums of Squares Mean Square D SSA SSA/de MSA/MSE B SSB SSB/de MS /MS B E C SSC SSC/dfC MSG/MSE AB SSAB SSAB/(de) (de) MSAB/MSE AC SSAC SSAC/(de) (dfc) MSAC/MSE BC SSBC SSBC/(de) (dfc) MSBC/MSE ABC ssABC SSABC/(de) (de) (dfc) MSABC/MSE Error SSE SSE/de For example, if the dependent construct being measured were specific self-confidence, and term A in the table above corresponds to the main effect for information distribution, then at a given confidence level, a calculated F-statistic larger than the corresponding critical value would lead one to conclude that distribution of information among group members does in fact have some impact on problem specific self-confidence at the individual level. These same analyses are repeated for each of the other dependent variables, i.e., perceived advocacy, vendor evalua- tion and behavioral intent. It should be noted that interpretation of the results of such an experiment is severely restricted. That is, in a fixed effects design, null hypothesis rejection for any interaction or main effect only permits conclusions to be drawn with respect to the particular levels of the independent variables actually included in the design. It is imprOper to attempt generalization beyond the range of values tested. This caution- ary note is especially relevant to experiments such as this one in which 66 independent factors are non-metric, i.e., do not meet interval level criteria. Figure 7 summarizes the student sample experimental design. The practitioner sample experimental design is a fixed effects, repeated measures two-factor design. In the design described previously, each experimental subject is exposed to only one unique combination of treatment levels. This is not the case in a repeated measures design. Here subjects receive more than one treatment application. Repeated measures designs are often employed in studies with cost or sampling restrictions. In general, non-repeated measures designs are preferred as sequential exposure to multiple treatment combinations introduces a threat to internal validity termed a carry-over effect. That is, if one treatment combination consistently follows another, effects measured in the second may be the result of factors contained in both treatments. This produces confounding in interpretation of the second treatment's effect. The problem is commonly handled by randomizing order of treatment presentation and that procedure was employed here. In addition to moving to a repeated measures design a decision was made to exclude the claim intensity factor from the practitioner experiment. This was done due to limited practitioner sampling frame size and the fact that preliminary analyses of validation sample data indicated that claim strength is weak- est in explanatory power. While the general model for this type of design is essentially similar to that described above, computational procedures change to accommodate introduction of repeated measures. Furthermore, the design incorporates repeated measures only with respect to one factor, source credibility. That is, each experimental subject was exposed to only one 67 ID m1 1132 SC sc scl sc2 scl sc2 YSSC’YPA 5‘1 YVE 3'31 CI2 Independent Variables (Factors): Information Distribution, two levels, equal=IDl, unequal=ID2. Source Credibility, two levels, lowasc , high=SC2. , strong=CI 1 Claim Intensity, two levels, moderateaCIl 2. Dependent Variables: Specific Self-Confidence, individual level, prediscussion, YSSC Perceived Advocacy, group level, post-discussion, YPA' Vendor Evaluation, group level, post-discussion, YVE' Behavioral Intent, group level, post-discussion, YBI' FIGURE 7 — THE STUDENT SAMPLE EXPERIMENTAL DESIGN 68 level of the information distribution factor but in combination with both levels of the source credibility factor. Technically, subjects repre- sent a third factor, as subject by treatment variance terms are meaning- ful with repeated measures designs. Figure 8 presents the experimental design employed with the practitioner sample. Notationally, this type of design is often referred to as a mixed or split plot type of design and is represented as an A X (Bxs) design.3 Here information distribution corresponds to factor A, source credibility to factor B, and subjects to factor S. Use of this type of design is widespread; however, it is computa- tionally less straightforward than the design described in Figure 7. The apprOpriate summary ANOVA table appears below. Table 3 Example Summary ANOVA Table for an A X (BxS) Design Source Sums of Squares Mean Square E A SSA SSA/de MSA/MsS/A S/A SSS/A SSS/A/df S/A None B SSB SSE/de MsB/MsBxS/A AB SSAB SSAB/deB MSAB/MsBxS/A Bxs/A ssBxS/A SSBxS/A/dexS/A Note that subjects is a nested factor in the non-repeated factor and that in the denominators of the F-statistics MSE is replaced by mean squares based on the nested subjects‘ factor. Given the constraints that suggest use of a repeated measures design initially, the mixed design employed here is preferred for its greater sensitivity within the repeated measures portion. 69 ID ID 1 102 sc sc scl sc2 sc1 sc2 s s YSSC’YPA'Y 3'31 51 51 52 S2 53 53 Sm Sn Independent Variables (Factors): Information Distribution, two levels; equal=ID nonrepeated factor. unequalaID l' 2' Source Credibility, two levels, low=SC , high=SC , repeated factor. 1 2 Subjects, msn levels, nested factor. Dependent Variables Specific Self-Confidence, individual level, post-discussion, Y SSC Perceived Advocacy, group level. post-discussion, YPA' Vendor Evaluation, group level, post-disucssion YVE' Behavioral.Intent, group level, post-discussion YBI' FIGURE 8 - THE PRACTITIONER SAMPLE EXPERIMENTAL DESIGN 70 Population Two distinct populations were sampled to provide a subject pool. The first consists of a combination of undergraduate seniors and first- term MBA students. Students were selected on the basis of enrollment in courses offered at Michigan State University in the 1977-78 academic year. All student participants were volunteers. Permission to use students as experimental subjects was granted on September 13, 1977, by Dr. Henry E. Bredeck, Chairman of the University Committee on Research Involving Human Subjects. A COpy of that approval appears in Appendix 1. No pre-screening was undertaken to attempt matching along dimensions of problem solving or persuasive communication abilities, as variance attri- butable to these sources is controlled through randomization in treatment assignment. The sampling procedure is essentially a convenience one. While probability samples are generally preferred over their non-probability counterparts when ability to generalize back to the pOpulation is of interest, e.g., in parameter estimation, the issue of external validity was not of concern with the student sample. That is, the ability of the model to predict performance among other groups of students assembled under similar conditions is not relevant to the research problem. A total of 176 students participated in the experiment. The second sampling frame consists of business practitioners currently actively engaged in vendor evaluation and selection. Repre- sentatives of this pOpulation were drawn from the ranks of a single firm, a large multi-divisioned Fortune 500 manufacturer located in the Midwest. Many of these individuals work together in natural decision-making groups within their respective Operating units and in these cases the experiment 71 was run with the group at their location. Individuals in this category were generally divisional level office services managers and their staff assistants who are charged with vendor relations responsibility for office equipment. The other category of practitioner subjects was buyers and assis- tant buyers of non-product related materials and supplies. These individuals function in purchasing departments of both the divisions the office services managers were drawn from and others as well. These buyers typically do not function in groups rather each has responsibility for a particular product group. Only a few of these individuals were more than casually familiar with the office equipment area. The experi- ment was run with them at a corporate educational facility where they were attending week-long continuing education purchasing programs. Plant, divisional, and corporate levels of the organization were represented by the 101 persons who participated. The decision to limit participation to members of a single organiza- tion was made to reduce error variance due to differences in organiza- tional procedures likely to be employed by individuals coming from different firms. This has the effect of further reducing generaliz- ability; however, a tradeoff occurs in a gain in power of the experimental design, and since the purpose is model validation rather than prediction of actual purchase behavior, this tradeoff was accepted. Instrument Development Vehicles used to operationalize the experimental designs are two purchasing cases developed expressly for this research. Examples of the cases are provided in Appendices 2 and 5. Different versions of the cases corresponding to experimental design treatment cells were produced 72 by interchanging inserts into the body of each case. That is, each sub- ject received a case which was a combination of a section common to all subjects in all treatment conditions and an insert peculiar to the. particular treatment combination the subject was being exposed to. A preliminary version of the case employed minicomputer hardware as the object of the group vendor selection problem; however, pretesting revealed that a lack of familiarity with the rather technical evaluatory criteria employed caused considerable confusion producing unreliable vendor evaluations. For this reason the case subject was changed to bond paper c0pier equipment which most persons have had experience with, at least as users if not purchasers. In addition, office equipment, like a bond paper copier, constitutes a product category it was felt is most amenable to the research purpose. That is, it is non-routinized and so must be deliberated each time a purchase is contemplated. It usually is a modified rebuy or new task so that more than one person is likely to be involved. The particulars of the purchasing cases were develOped from explora- tory field interviews with both users and sales representatives of bond paper copier equipment currently on the market. These interviews, con- ducted in Fall of 1977, provided sufficient familiarization with bond paper copiers to permit development of realistic case scenarios. These interviews, as well as reviews of trade publications, were also used to develOp the attribute list subjects would base comparisons on. In student cases subjects were first introduced to the problem which was a bottleneck in office communications flow at the fictitious Plasti- Part Corporation. Three corporate managers were introduced setting the stage for a group effort. Subjects were informed these three individuals had already conducted a preliminary investigation coming to the conclusion 73 that additional copiers would be needed and suggesting five salient evaluatory criteria, i.e., product features, maintenance service, delivery time, total cost and vendor reputation. Finally, the case's common portion concluded by pointing out that the field of competing vendors had been narrowed to two and all relevant information available at the time was provided in an attachment on the following page. Manipulations of the three independent variables, information distribution, source credibility, and claim strength, were developed and selected examples are presented below. While treatment levels for infor- mation distribution and claim strength were arrived at judgmentally, information sources used to manipulate source credibility were deter- mined in a pre-test involving 61 undergraduate students. The instrument used in that pre-test appears in Appendix 3, while the results are pre- sented in Table 4 below. Table 4 Summary Pre-Test Source Credibility Data Mean Mean Information Source Trustworthiness Competence Total Rank 1. Consulting firm. 6.13 5.64 ll.77 1 2. Purchasing agent, competing firm. 5.46 5.34 10.80 2 3. Purchasing agent, other industry. 4.74 6.03 10.77 3 4. Friend, user in other firm. 4.90 4.98 9.88 4 5. Friend, inside own firm. 3.93 5.79 9.72 5 6. Sales rep, longstanding association. 4.34 3.98 8.33 6 7. Sales rep, mediocre firm. 4.69 3.56 8.25 7 8. Sales rep, no prior experience, big name. 3.41 3.53 6.94 8 9. Sales rep, no prior experience, new and small. 3.10 2.92 6.02 9 74 While no statistical significance tests were run on differences in means it is readily apparent that meaningful differences in perceived source credibility among the sources presented do exist. Since credi- bility of sales representatives is of primary concern, the salesman representing the firm of long-standing and satisfactory association was chosen to Operationalize the high source credibility treatment level while the salesman representing the new, small firm represents the low credibility level. Referring to Appendix 2, "Bill Williams" represents the low credibility source while "Jim Fisher" represents the high credi- bility source. The two non-experimental or dummy sources are "Andy Day," a friend inside the firm, and "Ray Frankel,” a friend who is a user in another firm, both of which, on the basis of findings in Table 4, are moderate in source credibility being ranked fifth and fourth respectively. Manipulation of information distribution was accomplished by varying numbers of cues supplied in the case inserts. For example, in a three- person student group two students would receive dummy inserts each containing 9 cues. These two inserts were always the same regardless of treatment cell. In the equal information distribution condition, the third student, designated the experimental subject, also received an insert with 9 cues. However, in the unequal information distribution condition the experimental subject received an insert containing 15 cues. These additional cues being in the form of warrants in the support of a claim and generally providing greater detail. Differences are illus-- trated by underlined cues in the following paragraphs excerpted from equal and unequal information distribution inserts respectively. Our copy quality is good, in fact, in blind tests some peOple have been unable to distinguish between our copies and the originals 75 they were made from. The machine is well built and easy to operate. Our copy quality is good, in fact, in blind tests some people have been unable to distinguish between our c0pies and the originals they were made from. This is due to a newly patented process that uses a photo cell just like in expensive cameras to precisely meter exposure time. The machine is well built and subjected to exhaustive quality control checks before you get it. We have engineered it specially to make it easyito operate and refill with paper. Finally, the claim strength or language intensity factor was manipulated by modifier polarity. Referring again to the first paragraph above as an example of the moderate claim strength condition, the contrast with the strong claim strength condition is apparent by focusing on under- lined portions. Our COpy quality is gggd, in fact, in blind tests.§gm§ peOple have been unable to distinguish between our COpies and the originals they were made from. The machine is well built and easy to _operate. Our copy quality is excellent, in fact, in blind tests most peOple have been unable to distinguish between our copies and the originals they were made from. The machine is exceptionally well built and we have taken particular care to make the machine easy to operate. Differences of this type were carried out throughout the experimental inserts with attention being given not to alter the number of cues, as this would confound interpretation of the information distribution factor. A four-part questionnaire is attached to each student's case, a COpy of which appears in Appendix 4. Part I consists of 14 items designed to probe vendor preferences and behavioral intent. More speci- fically items 1-5 request specific attribute vendor evaluations, items 6 and 7 are measures of overall attitude toward vendor (A0), items 8 and 9 measure behavioral intent (BI) and items lO-l4 require saliency 76 responses regarding the five attributes in items 1-5. Although not the primary aim of the investigation, collection of this type of data per- mits explorations in multi-attribute decision models of expectancy- value type. Part II consisted of a battery of eight items that are manipulation checks, that is, measures of the extent to which application of the treatments had the desired cognitive effects. Their principal value come as diagnostic aids should the findings be other than what the model predicts. These are followed by a three-item scale which takes the first dependent construct measure, specific self-confidence. Multiple item scales are used for dependent constructs to permit discussion of instru- ment reliability. These scales were develOped through factor analysis of the 1977 minicomputer case pre-test data. .Inclusion criteria were post-varimax rotation unifactor loadings of 0.40 or higher. Part III replicates Part I exactly and is administered after group discussion to measure discussion induced changes in attitudes and intent. Part IV begins with a five-item scale to measure perceived advocacy, another dependent construct. The remaining items, 45-50, are sociome- tric correlates of communicator influence and are not utilized in the current investigation. Measures employed in items 1-8, 26-33, and 40-50 are Guilford Constant Sum Scales, which have been used primarily for two reasons. In the Summer 1977 pre-test, Likert items were used to measure these characteristics and examination of the data seemed to indicate that student subjects were not actually comparing vendors as had been intended. That is, they seemed to examine each in isolation, often coming up with near identical ratings. Introduction of Guilford scales forces cognitive comparisons while still permitting an identical rating 77 on an attribute if that is what the respondent really intends. The second factor motivating choice of the Guilford scale was the fact that it produces data of interval level of measure whereas, technically, the Likert scale does not. In some instances this matter would not be of great concern; however, the ANOVA model's validity rests on assumptions of homogeneity of within-treatment variance and, in the case of repeated measures, homogeneity of between treatment co-variances. Therefore, ability to compute meaningful variance and co-variance terms is of great importance. Turning to the case employed with business practitioners, several similarities and differences are noteworthy. The most important is that information provided in the experimental inserts to compare vendors is numeric rather than verbal. This change was made possible by elimina- tion of the claim strength factor from the experimental design, and came about as a result of feedback from exploratory interviews conducted Fall 1977 which indicated that the student case might be viewed as exces- sively artificial by practitioners. One attributed source of information, the fictitious Buyers Services Inc., is a derivative of an actual firm, Buyers Laboratories Inc., which provides tabular comparisons of this type based on survey data from users of copier equipment. In fact, several of the attribute ratings are actual ones from competing products rated by Buyers Labs. The common portion of the case was rewritten into two versions, one involving a convenience walk-up copier selection decision and the other a central reproduction facility choice. Two versions were necessitated by the repeated measures design, otherwise the second would always be trivial. Also the body of the case was shortened as two complete replications of the experiment with each group are required and 78 practitioners typically face more strenuous demands on their time than do students. Examining COpies of the cases in Appendix 5, one sees that manipu- lation of information distribution once again is accomplished by alter- ing the number of cues presented going from 8 in the equal conditions to 12 in the unequal condition. This compares with the 9 and 15 respectively found in the student cases. Inasmuch as purchasing managers actually do utilize the services of firms like Buyers Laboratories and Data-Pro to assist them in making evaluations, and in view of the rankings in Table 4, the independent and well-respected Buyers Services Inc. represents the high source credi- bility condition, while the relatively little known Mr. Frank Cassady representing a vendor new to the bond paper COpier business Operationa- lizes the low credibility condition. The questionnaire, reproduced in Appendix 6, remains virtually identical to that used with the student subjects with the exception that the delivery time attribute is replaced by additional product features, the manipulation check items referring to vendor A are removed from Part II, and the sociometric measures are removed from Part IV. These changes were made to shorten the instrument and cut down on administra- tion time and because interviews with persons knowledgable in the copier market indicated that delivery dates are seldom a problem whereas vendor differences in special features like automatic duplexing might be a worthwhile point of comparison. Instrument Administration Procedure A principal criticism of much of social psychological investigation into group processes is that often there exists no group at all. That 79 is, subjects often consist of students who are brought together on the day Of the experiment, perhaps never having met before, interact for an hour or so and go their separate ways. Reference group theory indi- cates that group cohesiveness is a function of the desire of members to remain part Of the group and that norm formation and conformity pressure are in turn functions Of cohesiveness. It can be seen then that experi- ments frequently are conducted with temporary aggregations of individuals, not true groups. As a result behaviors elicited and con- clusions drawn are not characteristic of actual groups in natural settings. To a certain extent all laboratory investigation is similarly suspect; however, in this research, procedures for administration of the experiments, with both students and practitioners, include steps taken specifically to partially resolve this problem. The student participants who volunteered in any one term all came from the same academic course. At the term's beginning, volunteers were assigned to or allowed to self-select three-person groups. These groups remained intact throughout the ten-week term. Each team completed two written case analyses, receiving a team grade, prior to administration Of the experimental case. Thus, group members not only were provided an Opportunity to establish norms, they also had a vested interest in the quality of group output. As was mentioned earlier several practitioner groups were wholly natural, functioning as a group on a day-to-day basis. These teams ranged in size from two to eight and from very informal associations to a formal buying committee. The predominant arrangement, however, was an Office services manager with a staff assistant who usually was more technically aware and took care of day-to-day concerns. A total of eight 80 of these natural groups completed the experiment, seven of which were two person and one Of which was three person. In view Of the fact that the two-person group was most common, the buyers and assistant buyers enrolled in purchasing seminars were arranged into two-person groups. Of all of the participants in the study, these individuals' groups were least natural as the seminars were three days long and they had worked together as partners only once before. Randomization procedures were employed with both students and practitioners as group assignment to treatment condition and selectioniof experimental subjects within groups were accomplished with the aid of random number tables. On the day of the experiment participants were assembled and told that the purpose Of the case and its accompanying questionnaire was two- fold. First the case was being develOped for use in continuing educa- tion seminars and second, baseline data was being collected tO determine "the right answer." Cases were then passed out and instructions for completing the questionnaire reviewed. The instructions stressed several points, namely, Subjects should complete Parts I and II before discussion and go to Parts III and IV only after discussion, Subjects should respond to all questionnaire items individually without consulting each other, especially in Part IV, Subjects should discuss the case until a consensus vendor choice was reached and do so quietly so as not tO disturb other groups in the same room, - Subjects should respond to the cases and make a choice based only on the information provided even if that information was incomplete in their Opinion. 81 The experiment typically took about one hour to complete of which approximately 15-20 minutes was actually devoted to group discussion. Following collection Of the cases subjects were debriefed. On question- ing it was readily apparent that most had ascertained that some dimen- sions of group leadership were being investigated; however, none expressed any awareness of the true purposes or methods of manipulation. These were explained during the concluding portion Of the debriefing. Data Analysis Questionnaires used for data collection appear in Appendices 4 and 6. Initially each item is treated as a separate variable and coded, along with independent factors and subject identification sequences, into a distinct field on a case record. Two alternatives exist for analysis of group level data from the student sample. These are multi- variate analysis Of variance on scores pooled over group members but not scale items versus univariate analysis of variance of scores pooled over group members and scale items. As an example, considering items 40-44 in Part IV of the student questionnaire in Appendix 4 the mathematical model describing the evaluation score of one person on one item is, Yijkl = u4-ai+8j+yk+aBij+aYik+ayjk+aByijk+eijk1 {8} which is the same as equation {4} presented above. Disregarding for a moment the subscripts denoting factor levels, i.e., i, j, k, let us denote member evaluations on items by, Y = ulm+alm+° . °+€lm {9} where l designates group members and m individual questionnaire items. Performance Of multivariate analysis of variance necessitates forming a data matrix pooled over the two dummy group members yet retaining 82 individual item identity, that is, submission of a Y.m score for each group-item combination. On the other hand, univariate analysis of variance may be performed on the scale total score by pooling over both group members and items so that for each group a single Y score is input. When possible the multivariate analysis of variance (MANOVA) proce- dure is clearly indicated in studies involving multiple correlated dependent variables.6 MANOVA first tests for the significance of treatment effects on the entire set of items comprising the scale and if significance is found identifies the item(s) principally responsible for that significance. A side benefit to the use of MANOVA lies in the fact that significance may be Obtained from analysis of the complete variable set even when univariate analyses of the separate items in the set do not produce significance. Use of MANOVA is limited to data Obtained from students as software to perform the analysis on data from a repeated measures experiment was not available. Thus significance testing with the practitioner data is obtained by performing univariate ANOVAs on scale totals. In addition to significance testing for presence of an effect, supplementary analyses are performed to measure the strength of relation- ship between independent and dependent variables. That is, a reported alpha level is sometimes erroneously interpreted as a measure of strength Of association.7 A more appropriate method is to report 02 (omega squared), a statistic discussed by Hays.8 This index isolates the pro- portion Of variability in the population total variance attributable to a particular treatment effect, much as a partial 82 might in the regres- sion model. Use of omega squared in conjunction with the usual F-test for significance is especially recommended when investigation is being 83 undertaken in a new and uncharted area. The general form of the equa- tion for omega squared is given by,9 82 - A2 /62 +62 effect effect effect wg' {10} Finally, the ability of a particular experiment to demonstrate the existence Of significant treatment effects when they do in fact exist is functionally dependent on the power Of that experiment. That is, the more powerful the experiment the greater its sensitivity to treat- ment effects. The power of an experiment is a function of four factors, namely, Alpha level, selection Of a numerically higher alpha, moving from .05 to .10 for example, results in an increase in power, - Replications, the larger the number of observations on each treatment condition the greater the power, - Error variance, power is reduced as variance attributable tO any source other than the treatments increases, - Treatment effect magnitude, the larger the treatment effect the more powerful the experiment. Common practice involves first establishing an acceptable alpha level for significance testing and then taking whatever steps possible to increase power. In the present investigation this involved use of the repeated measures design with the practitioners which has the effect Of reducing variation due tO individual differences, as well as select- ing all practitioner subjects from a single firm to reduce variance attributable to organizational policy differences. In both student and practitioner cases manipulation of treatment levels were realistic in terms of what one actually finds in the marketplace. Exaggerating differences between treatment levels will have the result Of enlarging 84 treatment effects and thus increasing power, however, at the price Of making interpretation nonsensical. The chief benefit of a priori power determination is a check to see whether the experiment is likely to produce significant results; however, to do so requires a priori know- ledge Or estimates of treatment effects. Post-hoc power determination is useful in diagnosing problems should results be found not signifi- cant at a given alpha level. All dependent constructs in both student and practitioner data bases are subjected to significance tests for main and interaction treatment effects. Determination of omega squared and power is per- formed selectively as interpretive requirements warrant. In addition, partial correlations among the dependent constructs are performed to provide validating evidence concerning the serial relationships posited in the model, e.g., between specific self-confidence and perceived advocacy behavior and between perceived advocacy behavior and post- discussion group vendor evaluation. References Campbell, J. C. and D. T. Stanley, Experimental and Quasi- Experimental Designs for Research, Chicago: Rand McNally (1966). Theoretical derivation of the general models and discussion of the terms in the summary ANOVA tables may be found in any of several standard texts including: Keppel, G., Design and Analysis: A Researcher's Handbook, Englewood Cliffs, N.J.: Prentice Hall Inc. (1973) or W. L. Hays, Statistics for the Social Sciences, edition, New York: Holt, Rinehart and Winston Inc. (1973). Keppel, G., Op.cit., 423. Ibid., 439. Ibid., 462-4. Finn, Jeremy, A General Model for Multivariate Analysis, New York: Holt, Rinehart and Winston (1974), 210 and ff. Keppel, op.cit., 548. Hays, Op.cit., 512-14. Keppel, op.cit., 553. Chapter V RESEARCH RESULTS The working hypotheses presented in Chapter 3 provide a framework for discussion Of research findings. Each hypothesis was tested on samples drawn from two distinct populations, students and practitioners. In this chapter, consideration is first given to results from the student sample for all working hypotheses followed by discussion of practitioner sample results. Student Sample As indicated in Chapter 4, apprOpriate first analysis is multi- variate analysis of variance. The analysis sequence first tests the highest order interaction, in this case three way, and subsequently tests lower order interactions and finally main effects. This procedure presumes that the sequence is terminated at the highest order inter- action found significant. In the case of no significant interactions, interaction degrees Of freedom and variance are collapsed into the main effects error term used to determine main effects F-statistics. The MANOVA summary tables which follow present an overall F- statistic for the scale measuring a particular dependent variable. Corresponding F's for individual scale items are also presented when significant. In the multi-variate case the calculated F-statistic is that Of Wilks. Note that test results for several working hypotheses are presented in each summary table and that order Of presentation differs somewhat from that in Chapter 3. For example, Table 5 considers the dependent variable specific self-confidence which is measured by items 23, 24 and 85 25 on the instrument in Appendix 4. are o IV23 O O ' H Vl6' HIv9 O O ' H HIIIaZ’ HIIa2 O , and HIa The working hypotheses summarized 2. Only those hypotheses relevant to dependent variables specifically identified in the causal model, Figure 6, are discussed in the body Of this chapter, while the other hypotheses are presented in Appendix 7. Student Sample Specific Self-confidence (YSSC) MANOVA Effect d.f. F Significance Level ID x SC x CI 3,54 .299 .83 Overall 3,54 .299 .83 ID x SC Overall 3,54 1.105 .35 ID x CI Overall 3,54 1.996 .13 Item 23 1,56 2.745 .10* Item 24 1,56 1.893 .17 Item 25 1,56 5.463 .02* SC x CI Overall 3,54 .915 .44 Item 23 1,56 .877 .35 Item 24 1,56 2.774 .10* Item 25 1,56 1.366 .25 ID Overall 3,54 .958 .51** SC Overall 3,54 .508 .68** CI Overall 3,54 .969 .64** *Effect statistically significant at p5,10 . **Effect not interpretable due to presence of higher order interaction The direct effects of factors and their interactions on each Of the dependent variables is considered first followed by consideration of relationships among the dependent variable set. 87 The figures in Table 5 suggest that while there is no significant three-way interaction, weak evidence supporting separate two-way interactions between information distribution and claim intensity and between source credibility and claim intensity does exist. These results are supportive Of the initial link in the model. 0 O O Iv27' HIv20, HIv13’ Table 6 summarizes the tests for hypotheses H O HO O IV6' IIIb3 O ' H Ib3' H IIb 3, and H These hypotheses relate to the direct effect of independent factors and their interactions on the dependent variable peer evaluations Of advocacy behavior. This construct is measured by means Of a five-item scale consisting Of items 101-105 on the instrument in Appendix 4. Table 6 Student Sample Perceived Advocacy (YPA) MANOVA Effect d.f. F Significance Level IC x SC x CI Overall 5,52 .105 .99 ID x SC Overall 5,52 1.598 .18 ID x CI Overall 5,52 .481 .79 SC x CI Overall 5,52 1.217 .31 ID Overall 5,52 .988 .43 Item 101 1,56 .636 .43 Item 102 1,56 .940 .34 Item 103 1,56 .156 .69 Item 104 1,56 4.423 .04* Item 105 1,56 .975 .33 88 Table 6 (cont'd.) Effect d.f. F Significance Level SC Overall 5,52 1.766 .13 Item 101 1,56 2.195 .14 Item 102 1,56 .151 .70 Item 103 1,56 2.489 .12 Item 104 1,56 6.177 .02* Item 105 1,56 .244 .62 CI Overall 5,52 .324 .90 *Effect statistically significant at p:,10 Based on this data there appears to be little evidence supporting direct linkages between the factors and perceived advocacy. This is not unexpected as the model does not postulate any such linkages. A significant main effect for information distribution and source credi- bility is encountered with item 104 in the perceived advocacy scale which measures persuasiveness specifically. While these effects cannot be dismissed as spurious, evidence of relationship is once again quite weak, as it was detected by only a single item in the scale. Once again referring to the causal model in Figure 6 direct rela- tionships between factors and the next two dependent variables, i.e., post-discussion vendor evaluation and post-discussion behavioral intent, O o o o o are tested by means of hypotheses HIbl’ HIbZ' HIIbl' HIIbZ' HIIIbl’ O O O O O O O IIIbl' HIIIbz’ HIv4' HIvs' HIVll' HIv12 O H IV18' HIv19 O 0 ’ HIvzs’ HIv26' , H Post-discussion vendor evaluation is measured by a two-item scale, items 31 and 32, while post-discussion behavioral intent is measured by item 33 in Appendix 4. Examining Tables 7 and 8 a significant relationship between source credibility and post-discussion vendor evaluation and post-discussion 89 behavioral intent is observed. The significant main effect for source credibility is evident with perceived advocacy, vendor evaluation and behavioral intent. None of these is postulated in the model and their implications are taken up in the concluding chapter; however, rela- tionships with the other two factors are not evidenced which is in accordance with the model. Table 7 Student Sample Vendor Evaluation (YVE) MANOVA Effect d.f. P Significance Level ID x SC x CI Overall 2,55 .255 .78 ID x SC Overall 2,55 .272 .76 ID x CI Overall 2,55 .039 .96 SC x CI Overall 2,55 .406 .67 ID Overall 2,55 1.334 .27 SC Overall 2,55 2.448 .10* Item 31 1,56 2.185 .15 Item 32 4,268 .04* CI Overall 2,55 .017 .98 *Effect statistically significant at pfslo 90 Table 8 Student Sample Behavioral Intent (YBI) ANOVA Effect d.f. F Significance Level ID x SC x CI 1,56 1.841 .18 ID x SC 1,56 .003 .96 ID x CI 1,56 .002 .96 SC x CI 1,56 1.299 .26 ID 1,56 .003 .96 SC 1,56 3.207 .08* CI 1,56 1.299 .26 *Effect statistically significant at p5310 To this point hypotheses have been tested by means of MANOVA as outlined earlier. The alternative procedure, to sum the scale item scores into a scale total or index, will now be considered. In order to perform ANOVA on these totals meaningfully it is first necessary to demonstrate scale homogeneity. This is typically done by means of a reliability coefficient. The purpose of the multiple item scale is two- fold. First, to improve the quality and stability of measurement and second, to permit specification of measurement error. The tOpic of reliability and measurement error is given extensive coverage in a recent special issue of the Journal of Marketing Research. As mentioned above, the presence and extent of measurement error is measured by means of Cronbach's alpha reliability coefficient. The square root of this coefficient is, by definition, the correlation between fallible measurements or observations and the error-free latent or true scores they approximate. Clearly the greater the amount of measurement error present the lower the reliability coefficient and hence the true score correlation. As Nunnally states, All these considerations justify the statement that coefficient alpha is a very important formula in the theory of reliability. 91 It represents the expected correlation of one test with an alter- native form containing the same number of items. The square root of coefficient alpha is the estimated correlation of a test with errorless true scores. It is so pregnant with meaning that it should routinely be applied to all new tests.3 Coefficient alpha and its square root are presented below for each of the dependent variable scales utilized with the student sample data. Note that the scale for vendor evaluation is only two items. The SPSS sub-routine used to calculate coefficient alpha requires scales having a minimum of three items. To circumvent this problem item 8, the pre- discussion behavioral intent measure, was included along with items 6 and 7 for the purpose of the original calculation. However, the reported coefficient is that which results when item 8 is deleted from the scale. Table 9 Student Sample Dependent Scale Reliability Coefficients Scale Items a /a Specific Self-confidence (YSSC) 23,24,25 .819 .905 Perceived Advocacy (YPA) 40 to 44 .727 .853 Vendor Evaluation (YVE) 6,7 .677 .823 These coefficients were determined from the responses of the "dummy" subject(s) in each group collapsed over all cells in the experimental design. In addition, for the vendor evaluation scale, pre-rather than post-discussion items were utilized. Both of these steps were under- taken to eliminate possible confounding in the interpretation of coeffi- cient alpha had experimental subject data been utilized and treatment by item interactions present. In examining the greatly different F- statistics for the various items in the scales presented in Tables 5 92 through 7 it appears that treatment by item interactions may exist. In view of the relatively encouraging values obtained for coeffi- cient alpha it is apprOpriate to combine the scale items into an index for each scale and repeat the analyses. The combinatorial method employed was to simply sum the item scores into a scale total score. Table 10 Student Sample Specific Self-confidence Scale Total (YSSCTOT) ANOVA Effect d.f. Mean Square F Significance ID x SC x CI 1 9.766 .485 .49 SC x CI 1 28.891 1.435 .24 ID x CI 1 66.016 3.280 .08* ID x SC 1 .766 .038 .85 CI 1 4.516 .224 .638** SC 1 3.516 .175 .678 ID 1 19.141 .951 .334** Residual 56 20.127 *Statistically significant at p§,10 **Not interpretable due to presence of higher order interaction Comparing the results in Table 10 with those in Table 5, presence of the information distribution by claim intensity interaction effect is supported while the source credibility by claim intensity inter- action is not. Table 6 and 7 suggested the presence of significant main effects for information distribution and source credibility with perceived advocacy (YPA) and source credibility with vendor evaluation (YVE) respectively. However, the model postulates serial causation among specific self-confidence, perceived advocacy, and vendor choice with no direct linkages between the independent factors and either perceived advocacy or vendor evaluation. The following tables, 11 and 12, based on the scale total scores 93 support the existence of a source credibility main effect in both cases however. Table 11 Student Sample Perceived Advocacy Scale Total (YPATOT) ANOVA Effect d.f. Mean Square F Significance ID x SC x CI 1 1.723 .302 .59 SC x CI 1 3.574 .658 .42 ID x CI 1 6.566 1.152 .29 ID x SC 1 1.129 .198 .66 CI 1 .879 .154 .70 SC 1 16.504 2.895 .09* ID 1 11.816 2.073 .16 Residual 56_ 5.74 63 *Statistically significant at p§,1o Table 12 Student Sample Vendor Evaluation Scale Total (YVETOT) ANOVA Effect d.f. Mean Square F Significance ID x SC x CI 1 1.891 .216 .64 SC x CI 1 6.891 .789 .38 ID x CI 1 .391 .045 .83 ID x SC 1 4.561 .517 .48 CI 1 .016 .002 .97 sc 1 28.891 3.308 .07* ID 1 .391 .045 .83 Residual 56. 8.734 63 *Statistically significant at p5310 To determine relationships among the dependent variables partial correlation and contingency analysis were performed. Partial correla- tion was performed on scale totals while the contingency analysis was performed on individual items in the advocacy scale. 94 Table 13 below summarizes both zero-order and partial correlation coefficients among the three dependent variable scales. Table 13 Student Sample Scale Total Dependent Variable Relationships Correlation Linkage Coefficient bl Significance YSSCTOT - YPATOT (zero order) -.023 62 .43 YSSCTOT - YVETOT (zero order) -.222 62 .04* YPATOT - YVETOT (zero order) -.005 62 .48 YSSCTOT - YVETOT (YPATOT partialed) -.222 61 .04* YPATOT - YVETOT (YSSCTOT partialed) -.011 61 .47 *Statistically significant at p§,1o These results indicate that by and large no relationships exist among the dependent variables. However, the negative relationship between pre-discussion specific self-confidence and post-discussion vendor evaluation warrants comment. The negative sign on the coeffi- cient is due to the manner in which data were coded for analysis. Recall that since a constant sum scale was employed only the respondents' evaluations of the first of the two vendors was coded. The negative coefficient indicates an inverse relationship between specific self- confidence and evaluation of the first of the two vendors. This implies a positive relationship between specific self-confidence and evaluation of the second vendor. This is true because of the characteristics of a constant sum scale. To provide another measure of the relationship between perceived advocacy and vendor choice a contingency analysis of these two variables was undertaken. Pre- versus post-discussion vendor choice is compared 95 against perceived advocacy (YPA)' Instances in which post-discussion group choice agrees with pre-discussion experimental subject choice are coded "agree" while disagreement is coded "disagree." Correspondingly perceived advocacy scores, i.e., scores of 4-8 out of a possible 8 are coded "high" while scores between 1 and 3 are coded "low." This imbal- ance results from use of modal scores to determine the recode categories, i.e., evaluation of three persons by means of a ten-point constant sum scale very frequently resulted in a 3,3,4 type of pattern, therefore, any value of 4 or higher is coded as a "high" score. Each item in the perceived advocacy scale was tested separately to prevent washing out any weak yet significant relationship which may exist. Table 14 Student Sample Vendor Choice with Perceived Advocacy Item Chi-square Significance 101 .712 .40 102 .681 .41 103 2.213 .14 104 .090 .76 105 3.064 .08* *Statistically significant at p§,1o For Item 105: Pre-, Post-discussion Choice Agree Disagree 16 32 Argument Strength Low 16 _ s°°re High 15 4 19 31 20 51 Phi Eta = .287 96 The numerically low frequency in the bottom right-hand cell indi- cates there were significantly fewer instances in which the experimental subject received a high rating on strength of argument and post- discussion vendor choice was different from his pre-discussion choice. This finding is in agreement with what one would expect to find based on the model. Consideration of the implications of the student sample results is taken up in the concluding chapter; however, to briefly summarize the validation phase data: - Complete validation of the entire model was not obtained, - Several of the individual linkages in the model were validated on either individual items in the dependent variable scales or, in some instances, on the entire scale, - Several additional relationships, not postulated in the original form of the model, appear to warrant further investigation. Before proceeding to the practitioner sample data, Table 11 summarizes those relationships found significant. Hays' omega squared is included to facilitate comparisons of relative strength of relation- ship. Figures 9-12 illustrate the four interactions summarized in Table 15. Table 15 Summary of Student Sample Significant Relationships . . . 2 Relationship Statistic Value Significance W ID x CI with YSSC (Item 23) F 2.745 .10 .03 ID x CI with YSSC (Item 25) P 5.463 .02 .07 ' 3.280 .08 .03 ID x CI with YSSCTOT F SC x CI with Ys (Item 24) P 2.774 .10 .03 SC 97 Table 15 (cont'd.) Relationship Statistic Value Significance cuz ID with YPA (Item 104) P 4.423 .04 .05 SC with YPA (Item 104) P 6.177 .02 .07 SC with YPATOT P 2.895 .09 .03 SC with YVB (Item 32) F 4.268 .04 .05 SC with YVETOT P 3.308 .07 .03 SC with YVI (Item 33) P 3.207 .08 .03 YSSCTOT with YVETOT F 3.115 .08 .03 Yssc'ro'r with YVETOT r - .222 .04 Pre-, post-discussion vendor 2 choice with YPA (Item 105) x 3.064 .08 Practitioner Sample Replication of the experiment with subjects drawn from a practi- tioner population provides an opportunity to verify and further explain those relationships uncovered with the student subjects. It also pro- vides another chance to specify relationships in the event that an unforeseen student by treatment interaction were present. This second question is relevant to the issue of the apprOpriateness of student subjects in business research. As with the student sample results summarizing hypothesis tests concerning independent factor-dependent variable relationships are taken up prior to consideration of dependent variable-dependent variable relationships. First, however, reliability analysis demonstrates homogeneity of dependent variable scale content. domains. 98 A 5 .- CI1 Clz ‘ ‘P I ‘CIz CI1 z 3 -- < NJ 5 .1 a} o 2 d- 1 «II- I L ’ I01 . I02 1 C' 2 T ID ' 1 1 3.430 : 4.063 I - - - - -+ - -- -. I 2 4.025 : 4.000 I L CELL MEANS FIGURE 9 — INFORMATION DISTRIBUTION BY CLAIM INTENSITY INTERACTION EFFECT ON YSSC (ITEM 23). 99 5 «II- 0'1 Clz 4 .. Clz CI1 z 3 .. < LU 2 _l .l g 2 an- 1 .- L 1 b I01 I02 1 c' 2 f I ID 1 1 3.750 : 4.313 I )—-—--—--1---———« I 2 5.000 : 3.750 I 1 CELL MEANS FIGURE 10 - INFORMATION DISTRIBUTION BY CLAIM INTENSITY INTERACTION EFFECT ON Yssc (ITEM 25). lOO A 15 -- Cl1 CI 12 ._ CI2 CI1 z 9 " < LL! 2 _l d U 6 III- 3 d- 1 I > ID1 I02 1 C1 2 r ID ' 1 1 11.563 : 13.063 I IIII———-1-———-- I 2 14.688 : 12.125 I 1 CELL MEANS FIGURE 11 - INFORMATION DISTRIBUTION BY CLAIM INTENSITY INTERACTION EFFECT ON YSSC (SCALE TOTAL). 101 A 5 .- C'z 0'1 CI 4 .. 1 C12 2 3 ‘- < LU S .. A d o 2 a. 1 dr- 1 I > $01 $02 1 C' 2 I "31 I 1 4.4375 ' 4.8750 I I 2 5.000 I 4.1375 I 1 CELL MEANS FIGURE 12 - SOURCE CREDIBILITY BY CLAIM INTENSITY INTERACTION EFFECT ON Yssc (ITEM 24). 102 Table 16 Practitioner Sample Dependent Variable Scale Reliability Coefficients Scale Items 0 Va Specific Self-confidence 19,20,21 .899 .948 Perceived Advocacy 41 to 45 .829 .910 Vendor Evaluation 6,7 .862 .928 Table 17 Practitioner Sample Specific Self-confidence (CONSCALE) ANOVA Effect SS d.f. MS F Significance Specific Self-confidence, Scale Total: ID 136.298 1 136.298 7.821 <.01* S/ID 697.108 40 17.428 SC 56.679 1 56.679 4.017 <.06* ID x SC .298 1 .298 .021 N.S. SC x S/ID 564.381 49_ 14.110 83 ID 10.715 1 10.715 7.252 <.01* S/ID 59.095 40 1.477 SC 9.334 1 9.334 8.012 <.01* ID x SC .047 l .047 .040 N.S. SC x S/ID 46.619 49. 1.165 83 ID 16.298 1 16.298 8.286 <.01* S/ID 78.667 40 1.967 SC 5.250 1 5.250 4.070 <.05* ID x SC .583 l .583 .452 N.S. SC x S/ID 51.619 49_ 1.290 83 ID 19.048 1 19.048 20.704 <.001* S/ID 36.786 40 .920 SC 4.762 1 4.762 1.729 <.25 ID x SC 0.000 1 0.000 0 N.S. SC x S/ID 110.262 49_ 1.290 83 *Statistically significant at p§,1o 103 The relationships between information distribution, source credi- bility and pre-discussion specific self-confidence would not appear quite clear. The lack of an information distribution by source credi- bility interaction can be expected as none was found with the student subjects either. It was of course not possible to examine any effects due to claim intensity as this factor was not included in the practi- tioner design. Possible explanations for the occurrence of a significant main effect for information distribution with the practitioners but not with the students are suggested by the differences in designs. That is, - the repeated measures type of design is more sensitive, more statistically powerful, - the information presented was numeric rather than verbal, - elimination of claim intensity also eliminated masking interac- tion effects with that factor. It is not possible at this time to say that any one of these considera- tions is more important than any other, or if in fact any of them is meaningful at all. The existence of a source credibility main effect is confirmed by its clear presence in both student and practitioner designs. Also, as Tables 17 and 18 below indicate, this effect carries through to the perceived advocacy scale scores but fails to appear with the vendor evaluation scale scores. 104 Table 18 Practitioner Sample Perceived Advocacy (ADVSCALE) ANOVA Effect SS d.f. MS F Significance ID 63.441 1 63.441 1.296 N.S. S/ID 1957.619 40 48.940 SC 14.584 1 14.584 3.230 .10* ID x SC 5.249 1 5.249 1.162 N.S. SC x S/ID 180.667 49_ 4.517 83 *Statistically significant at p5310 Table 19 Practitioner Sample Vendor Evaluation (VENDSCALE) ANOVA Effect SS d.f. MS F Significance ID 28.430 1 28.430 1.484 N.S. S/ID 766.312 40 19.158 SC .049 l .049 .004 N.S. ID x SC 16.194 1 16.194 1.279 N.S. SC x S/ID 506.261 49_ 12.657 83 These results differ from those with the student sample in that the source credibility effect did not carry through to post-discussion vendor evaluation. As with the student phase data respondents' scores on items com- prising the dependent variable scales were summed and a partial correla- tion analysis performed. To prevent confusion, scale labels used with the practitioners were changed to CONSCALE, ADVSCALE and VENDSCALE for the pre-discussion self-confidence scale, post-discussion perceived advocacy scale and post-discussion vendor evaluation scale respectively. , Y and Y with the student sample These correspond to Y PATOT VETOT SSCTOT and consist of identical items. 105 Table 20 Practitioner Sample Scale Total Dependent Variable Relationships Correlation Relationship Coefficient N Significance CONSCALE-ADVSCALE (zero-order) .36 82 .001* ADVSCALE-VENDSCALE (zero-order) .14 82 .10* CONSCALE-VENDSCALE (zero-order) - .26 82 .004* ADVSCALE-VENDSCALE (CONSCALE partialed) .05 81 .31 CONSCALE-VENDSCALE ' (ADVSCALE partialed) ‘ .23 81 .02* 1"Statistically significant at p5310 The data in Table 20 indicate intercorrelation among the dependent variables which seems concentrated in pre-discussion specific self- confidence and post-discussion vendor evaluation. This particular rela- tionship holds up even when common variance with post-discussion perceived advocacy is partialled out. This suggests that mere sharing of information without active advocacy of a particular vendor is suffi- cient to influence vendor choice when problem specific self-confidence levels are high. If true, this result would be at variance with the postulated model; however, the possibility of such an effect has been raised previously by choice shift researchers. This type of relationship is meaningful only if it can be demon- strated that the experimental subject's pre-discussion vendor choice was also the group's post-discussion vendor choice. This issue is taken up in the following paragraphs. Note, however, that the postulated relationship between post-discussion perceived advocacy and post- discussion vendor choice does not hold up when common variance due to 106 self-confidence is eliminated. This does not imply that the relation- ship does not exist, rather that removal of common variance with self- confidence produces a weaker relationship as evidenced by the difference in zero-order and partial correlation coefficients. This latter result directly confirms the linkage posited in the model. As with the student data evidence regarding the perceived advocacy- vendor choice relationship is obtained by contingency table analysis. Variables employed and coding procedure are identical to those used to produce Table 14, with one exception, that being the coding of items 41 through 45 in the perceived advocacy scale. The seven-point Likert type items were coded 1 through 4 a Low and 5 through 7 . High. Table 21 Practitioner Sample Vendor Choice with Perceived Advocacy 2 Item x Significance 41 .012 .91 42 3.003 .08* 43 4.277 .04* 44 8.650 .003* 45 3.027 .08* *Statistically significant at p5,10 For Item 42: Pre-, Post-discussion Choice Agree Disagree Influential in Low 15 8 23 Determining Group . Choice High 52 9 61 67 17 84 Phi = Eta = .222 For Item 43: Persuasive in Presenting Ideas Phi = Eta = .263 For Item 44: Argued Strongly for One Vendor Phi 8 Eta = .332 for Item 45: Leader of Group Phi = Eta a .219 107 Pre-, Post-discussion Choice Agree Disagree Low 10 7 17 High 57 10 67 . 67 17 84 Pre-, Post-discussion Choice Agree Disagree Low 19 12 31 High 48 5 53 67 17 84 Pre-, Post-discussion Choice Agree Disagree Low 29 12 41 High 38 5 ‘ 43 67 17 84 In all cases, and especially for items 44 and 45, the small fre- quency in the lower right-hand cell indicates that very seldom did post-discussion group vendor choice disagree with the experimental subject's pre-discussion vendor choice when that subject was highly rated on the perceived advocacy items. 108 Summarizing the practitioner data: - Complete validation of the reduced model was obtained, - Evidence suggestive of new relationships, not postulated and corroborating student sample findings, was found. Table 22 below summarizes those relationships verified with the practitioner sample. Table 22 Summary of Practitioner Significant Relationships Relationship Statistic Value Significance ID with CONSCALE P 7.821 .01 SC with CONSCALE F 4.017 .06 SC with ADVSCALE F 3.230 .10 CONSCALE with ADVSCALE r .36 .001 ADVSCALE with VENDSCALE r .14 .10 CONSCALE with VENDSCALE r .26 .004 CONSCALE with VENDSCALE (ADVSCALE partialed) r .23 .02 ADVSCALE with Pre-, Post- discussion Vendor Choice 2 Item 42 x2 3.003 .08 Item 43 x2 4.277 .04 Item 44 x2 8.650 .003 Item 45 x 3.027 .08 Unlike Table 15, omega squared is not calculated as the necessary within-groups mean square does not exist in repeated measures designs. REFERENCES Items 101 to 105 do not appear directly on the instrument; however, they are computed by averaging the responses of person 1 and person 2 as they evaluate person 3 (the experimental subject) over the characteristics described in items 40 to 44. Where groups are two- person the averaging procedure is not needed and the non- experimental subject's evaluation is used. Journal of Marketing Research, February 1979, pp. 1-102. Nunnally, Jum C. Psychometric Theory, 2nd Edition, New York: , McGraw-Hill Book Co. (1978), p. 214. Chapter IV CONCLUSIONS AND IMPLICATIONS At the outset of this investigation it was stated that the research purpose was to explore, and if possible substantiate, relationships among certain vendor influenceable inputs to group choice processes and vendor choice outcomes resulting from these processes. A model postu- lating the relationships of interest was developed and subsequently experimentally tested on both a student and a practitioner sample. Examining the results of those tests presented in the previous chapter it would appear that the suggested model is indeed valid and presents a reasonable picture of at least one variety of vendor choice mechanism. The first section of the remainder of this chapter focuses on the theoretical ramifications of these findings, the second on managerial implications, and the last on observations regarding method- ological issues relevant to further work in the area. Theoretical Consequences The role of advocacy behavior in group choice processes has been established and to some extent specified. Its central place in what I have termed an extended interpersonal model of buyer-seller relation- ships has also, at least partially, been confirmed. When two or more individuals gather to choose among alternatives it is in fact possible to alter the tone and content of their communications by manipulation of the distribution of information among group members, by attribution of information to a source perceived as being more or less credible, or by varying the strength of claim made by that source. The results of these tests are potentially quite profound, as they suggest that students and 109 110 modelers of organizational purchasing processes should consider influ- ence strategies directed at group processes as well as those directed at individual members. In many instances this will involve rethinking the relative importance of attribute as Opposed to process variables. On a more micro level the results of this research suggest new avenues of pursuit in organizational buyer behavior. Specifically, the original decision process model, presented in Chapter 3, Figure 6, must now be modified to accommodate alternate linkages. Modified versions, based on both student and practitioner samples are presented in Figures 13 and 14 respectively. While somewhat different as regards specific linkages, the revised models share certain features. They both indicate that the nature of influence is more complex than originally thought. That is, causal paths emanating from the independent factors do not impact only on pre- discussion specific self-confidence as hypothesized but extend to other of the dependent variables as well. The implication being that altera- tion of these independent factors may produce enhanced self-Confidence or may not; however, even in those cases where self-confidence is not consciously increased the individual may act as if it were. A most interesting result is that based on the student data which suggests no linkage between specific self-confidence through perceived advocacy to vendor evaluation. Taken at face value this means that inducement of an enhanced state of self-confidence in experimental subjects may be neither a necessary nor sufficient condition for subse- quent active vendor adequacy on their part. At the same time, specific self-confidence appears directly related to post-discussion group vendor evaluation. Consideration of these two together might lead one to the inference that the induced self-confidence state may result in subject 112 thSEmLXM w4m5h3.mamcu mom—DOM 9.9—. “.O awn—OS 20:.(8m 4<¢=h03¢hm ( .mp maze“. i. .l. .i. m as 125 error of estimation of latent specific self-confidence, n2, is given by C The parameters of the model are estimated by maximum likelihood 2. algorithms and the overall fit of the model is tested by means of a Chi- square statistic. The appeal of the structural equations approach lies in the fact that a statistically significant 8 is strong indication of actual rela- tionship no longer subject to artifactual challenge. Also, in the event that B is not significant the researcher has the capability to perform diagnostic Operations. That is, faulty Operationalization in treatment manipulation is indicated in numerically low values for 1, through 14, whereas excessive measurement error on the dependent variable side would be evidenced by low values for A through 17. 5 Also in models employing more than two latent constructs, alterna- tive causal paths may be tested as rival hypotheses. That is, a postulated serial linkage H-—9© may be tested against a coincident causal model of the form .—-—->.(—-—.. In general, structural equations analysis may represent a true methodological advance- ment. However, one strong caveat remains, namely, the existence or non- existence of causal relationships fundamentally derives from the correctness and logic of the theoretical model not on the characteristics of the technique employed to test that model. Summary The research described in this and preceding chapters has accom-. plished the three objectives originally established. That is, a new causal model of a vendor influenceable group choice process has been deductively generated. This model provides insight into mechanisms of group interaction and suggests sales strategies adOptable by organizational 126 marketers. It is characterized by the role of advocate on the part of a member of the buying center and may be viewed as an extended interper- sonal influence model. The empirical portion of the research has demonstrated that advo- cacy behavior, perceptible to peers in a decision making group, can be stimulated and facilitated by vendor influenceable variables. Each of the three factors selected to operationalize the model have been shown, either singly or in combination, on both student and management popula- tions, to be capable of producing perceptible advocacy behavior. Furthermore it has been shown that behaviors on the part of individuals assuming the advocate role do influence vendor choice. Finally the research method employed has been shown to be a suit- able approach to group focused causal research into organizational purchasing mechanisms. The relative merits of small sample experimental and non-experimental procedures make them eminently apprOpriate for continued research in the field of organizational buyer behavior. REFERENCES Credit must be given to Dr. George Waggenheim of the Michigan State University Marketing faculty for initially raising this question during conversations at the time of the initial concep- tualization of the research. Keppel, G. Design and Analysis: A Researchers Handbook, Englewood Cliffs, N.J.: Prentice Hall Inc. (1973), 525-540. Bagozzi, R. P., "Structural Equation Models in Experimental Research, Journal of Marketing Research, 14, (May 1977), 209-226. APPENDIX 1 127 APPENDIX 1 MICHIGAN STATE UNIVERSITY OFFICE FOR RESEARCH DEVELOPMENT EAST LANSING ' MICHIGAN ° 4.824 258 ADMINISTRATION NILDING September 13. 1977 Mr. Robert E. Krapfel Graduate School of Business Administration Department of Marketing and Transportation Administration Campus Dear Mr. Krapfel: Subject: PrOposal Entitled "Information Distribution, Source Credibility and Evidence Height as Detenminants of Organizational Purchase Outcomes: An Experimental Approach" The above referenced project was recently submitted for review to the University Committee on Research Involving Human Subjects (UCRIHS). We are pleased to advise that this review indicated that the rights and welfare of the human subjects appear to be adequately protected and the Committee, therefore, approved this project at its meeting on September 12, I977. Thank you for bringing this project to our attention. If we can be of any future help, please do not hesitate to let us know. Sincerely, 2;. 7 g‘ 7.1 iCILCL Henry E. Bredeck Chairman, UCRIHS jms cc: Dr. Leo G. Erickson APPENDIX 2 128 APPENDIX 2 Plasti-Part Inc. Plasti-Part is a medium sized manufacturing firm located in a large Mid- western city. They produce a line of injection molded plastic gears used in light machinery, office equipment, toys, etc. The firm has had an upsurge in business over the last three years and this has greatly increased the amount of paperwork and record-keeping the office staff must cope with. Several of the secretaries have complained of being overworked and office efficiency has declined noticeably. Mr. Ronald Brock, the general manager, has asked the office manager, personnel manager and head of the accounting department to look into the problem and make recommendations. That was four weeks ago. The evaluation team uncovered several minor problems and one major bottleneck, the fact that all photocopying, regardless of department, is routed to a centralized office. The decision to use a centralized reproduction office was made several years ago in the interest of cost savings and efficiency. Now, however, the volume of photocopying is so large that peOple often have to wait 2-4 days to get their copies back. When copies are needed in a hurry secretaries often go to an outside copying service, but this has resulted in a high cost per cOpy and is a very inefficient use of secretaries' time. In addition, the 2-4 day wait on regular jobs results in costly delays in order processing, production schedule review, etc. The team decided that the firm should purchase additional copiers to be placed in each of the three departments producing the most paperwork. In order to get a quantity discount, the three units would all be purchased from the same vendor and would be the same model. The team also isolated the following factors as being important in evaluating competing brands. 1. Product features, e.g., COpying speed, ease of use, copy quality, simple construction to prevent paper jamming, etc. 129 2. Ability of the vendor firm to provide dependable maintenance service. 3. Ability of the vendor firm to deliver the units on time and get them Operational quickly. 4. Cost, includes initial purchase price, plus per copy user charges and maintenance fees. 5. Overall reputation of the vendor. Based on preliminary evaluations, two brands are now being considered as alternatives. These are Brand A and Brand B. (Actual names are not used to prevent bias based on personal experience you may have had with different brands of photOCOpiers.) Attachment I contains the information that has been made available to you up until this time. In four days the team must recommend one of the two brands to Mr. Brock; therefore, you must choose one system based on currently available information. 130 Attachment 1 The information below was supplied by Bill Williams. Bill is the sales representative for vendor A (Brand A), which is a small firm whose name you are not very familiar with. Bill has only talked to you once before. "I think our machine is comparable to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our COpy quality is good, in fact, in blind tests some peOple have been unable to distinguish between our copies and the originals they were made from. The machine is well-built and easy to Operate. We have a contract with a maintenance company in this area, so a service representative is always close at hand. We deliver within 5 working days after your order is placed. "You certainly can't say our purchase price isn't right since it's $800 less than Brand B's. Now you may be thinking that since Brand B is an older, more established firm that builds solid machines, that they would be a better choice, but I think when you compare the two machines across the board you'll come to the conclusion that you will be better off with ours." 131 Attachment 1 The following information was supplied by Jim Fisher. Jim is the sales representative for vendor A (Brand A), a firm that you have purchased other office equipment from before. Over the years you have been very satisfied with Jim and the company's performance. "I think our machine is comparable to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our copy quality is good, in fact, in blind tests some pe0p1e have been unable to distinguish between our copies and the originals they were made from. The machine is well-built and easy to Operate. "We have a contract with a maintenance company in this area, so a service representative is always close at hand. We deliver with 5 working days after your order is placed. You certainly can't say our purchase price isn't right since it's $800 less than Brand B's. Now you may be thinking that since Brand B is an older, more established firm that builds solid machines, that they would be a better choice, but I think when you compare the two machines across the board you'll come to the conclusion that you will be better off with ours." 132 Attachment 1 The information below was supplied by Bill Williams. Bill is the sales representative for vendor A (Brand A), which is a small firm whose name you are not familiar with. Bill has only talked to you once before. "I think our machine is comparable to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our copy quality is good, in fact, in blind tests some people have been unable to distinguish between our copies and the originals they were made from. This is due to a newly patented process that uses a photo- cell just like in expensive cameras to precisely meter the exposture time. The machine is well-built and subjected to exhaustive quality control checks before you get it. We have engineered it specially to make it easy to Operate and refill with paper. “A local maintenance company services all of our machines so a service representative is always close at hand. We promise to deliver your machines within 5 working days after your order is received; in the past we have actually shipped copiers by air freight to live up to this promise. "Of course the price is right too. Our machine is $800 less than Brand B's and maintenance costs are fixed when you select the annual service contract option. In fact, for our customers with monthly copy- ing volume over 8,000 copies our total monthly costs work out to $4.8¢ per COpy. This compares favorably with our estimate of 5.l¢ per COpy for the comparable Brand B machine. "It's true that Brand B is an older more established firm, but I think when you compare the two machines across the board you'll come to the conclusion that you'll be better off with ours." 133 Attachment 1 The following information was supplied by Jim Fisher. Jim is the sales representative for vendor A (Brand A), a firm that you have purchased other office equipment from before. Over the years you have been very satisfied with Jim and the company's performance. "I think our machine is comparable to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our COpy quality is good, in fact, in blind tests some pe0ple have been unable to distinguish between our COpies and the originals they were made from. This is due to a newly patented process that uses a photocell just like in expensive cameras to precisely meter the exposure time. The machine is well-built and subjected to exhaustive quality control checks before you get it. We have engineered it specially to make it easy to Operate and refill with paper. "A local maintenance company services all of our machines so a service representative is always close at hand. We promise to deliver your machines within 5 working days after your order is received; in the past we have actually shipped copiers by air freight to live up to this promise. "Of course the price is right too. Our machine is $800 less than Brand B's and maintenance costs are fixed when you select the annual service contract option. In fact, for our customers with monthly COpying volume over 8,000 copies our total monthly costs work out to 4.8¢ per copy. This compares favorably with our estimate of 5.l¢ per cOpy for the comparable Brand B machine. "It's true that Brand B is an older more established firm, but I think when you compare the two machines across the board you'll come to the conclusion that you'll be better off with ours." 134 Attachment 1 The information below was supplied by Bill Williams. Bill is the sales representative for vendor A (Brand A), which is a small firm whose name you are not familiar with. Bill has only talked to you once before. "I think our machine is superior to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our COpy quality is excellent, in fact, in blind tests most people have been unable to distinguish between our cOpies and the originals they were made from. The machine is exceptionally well-built and we have taken particular care to make the machine easy to operate. "We have a contract with a maintenance company in this area, so you never have to worry about breakdowns, since a service representative is always close at hand. We guarantee delivery within 5 working days after your order is placed. "You must admit that the price is right since it is $800 less than Brand B's. While it is true that Brand B is an older, more established firm, we are gaining an excellent reputation very fast, and when you compare Brand A with Brand B across the board you'll surely conclude that you will be much better off with Brand A." 135 Attachment 1 The following information was supplied by Jim Fisher. Jim is the sales representative for vendor A (Brand A), a firm that you have purchased other office equipment from before. Over the years you have been very satisfied with Jim and the company's performance. "I think our machine is superior to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our copy quality is excellent, in fact, in blind tests most people have been unable to distinguish between our cOpies and the originals they were made from. The machine is exceptionally well-built and we have taken particular care to make the machine easy to Operate. "We have a contract with a maintenance company in this area, so you never have to worry about breakdowns since a service representative is always close at hand. We guarantee delivery within 5 working days after your order is placed. "You must admit that the price is right since it is $800 less than Brand B's. While it is true that Brand B is an Older, more established firm, we are gaining an excellent reputation very fast, and when you compare Brand A with Brand B across the board you'll surely conclude that you will be much better off with Brand A." 136 Attachment 1 The information below was supplied by Bill Williams. Bill is the sales representative for vendor A (Brand A), which is a small firm whose name you are not very familiar with. Bill has only talked to you once before. "I think our machine is superior to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our COpy quality is excellent, in fact, in blind tests most people have been unable to distinguish between our copies and the originals they were made from. This is due to a newly patented process that uses a photocell just like in espensive cameras to precisely meter the exposure time. The machine is exceptionally well-built and subjected to exhaustive quality control checks before you get it. We have engineered it specially to make it easy to operate and refill with paper. "We have a contract with a maintenance company in the area, so you never have to worry about breakdowns, since a service representative is always close at hand. We guarantee delivery within 5 working days after your order is placed; in the past we have actually shipped COpiers by air freight to live up to this guarantee. "You must admit that the price is right since it is $800 less than Brand B's and maintenance costs are fixed when you select the annual service contract Option. In fact, for our customers with monthly COpying volume over 8,000 COpies our total monthly costs work out to 4.8¢ per copy. This compares quite favorably with our estimate of 5.3¢ per COpy for the comparable Brand B machine. "While it is true that Brand B is an older, more established firm, we‘ are gaining an excellent reputation in this area very fast, and when you compare Brand A with Brand B across the board you'll surely conclude that you will be much better off with Brand A." 137 Attachment 1 The following information was supplied by Jim Fisher. Jim is the sales representative for vendor A (Brand A), a firm that you have purchased other office equipment from before. Over the years you have been very satisfied with Jim and the company's performance. "I think our machine is superior to any other machine in its price class. We have tested them against both Brand B and other manufacturers' machines. "Our COpy quality is excellent, in fact, in blind tests most people have been unable to distinguish between our copies and the originals they were made from. This is due to a newly patented process that uses a photocell just like in expensive cameras to precisely meter the exposure time. The machine is exceptionally well-built and subjected to exhaus- tive quality control checks before you get it. We have engineered it specially to make it easy to Operate and refill with paper. "We have a contract with a maintenance company in the area, so you never have to worry about breakdowns, since a service representative is always close at hand. We guarantee delivery within 5 working days after your order is placed; in the past we have actually shipped copiers by air freight to live up to this guarantee. "You must admit that the price is right since it is $800 less than Brand B's and maintenance costs are fixed when you select the annual service contract Option. In fact, for our customers with monthly cOpying volume over 8,000 copies our total monthly costs work out to 4.8¢ per copy. This compares quite favorably with our estimate of 5.3¢ per COpy for the comparable Brand B machine. "While it is true that Brand B is an older, more established firm, we are gaining an excellent reputation in this area very fast, and when you compare Brand A with Brand B across the board you'll surely conclude that you will be much better off with Brand A." 138 Attachment 1 The information below was supplied by Andy Day. Andy is a good friend of yours and head of the production department at Plasti-Part. "I've read the literature put out by the two different vendors and it seems that Brand A and Brand B are rather similar. It's true that the Brand A machine copies faster and seems easier to operate, but the Brand B machine looks like it will have lower operating and maintenance COStS . "Brand B is a larger firm with a well-known name and that is probably at least partially due to their rugged construction. They claim the machines only have to be serviced once a month and if that is true it's pretty impressive. "On the other hand, from what I've heard, if you want to be able to take delivery on short notice Brand A gives much better service." 139 Attachment 1 The information below was supplied by Ray Frankel. Ray is the purchas- ing agent for HRM Inc., a manufacturer across town, who you learned has purchased two COpiers in the past year. "Brand A is a pretty good machine. It is reasonably reliable and COpy quality is O.K. We haven't had any real serious maintenance problems, although one of their machines did break down 2 months ago and it took them a week and a half to come up with the right replacement part. "As I recall we did get delivery when we were supposed to but the total monthly costs were a surprise. The purchase price of $4,500 was reasonable but their machines need special paper which costs more. "Of course, I've heard Brand B's machines are good too. They have a real fine maintenance record and their costs are low considering the fact that it's a big name outfit. They have had problems with COpy quality though and you don't always get the machines when you want them. "I really don't have strong feelings for or against either one--it's just a matter of which considerations are more important to you." APPENDIX 3 140 APPENDIX Information Source 1. 10. 11. Independent, prestigious con- sulting firm. Purchasing agent in competing firm that bought one three months ago. Purchasing agent in another industry who bought one three months ago. A friend of yours in a firm which bought one three months ago. A friend inside your own company in the accounting department. Sales representative of one of the competing finms. Your firm has done business with them for many years - well satisfied. Same as six above only the company has only a so-so per- formance record. Sales representative of one of the competing firms. You have no prior experience with this firm - they are large with easily recognizable name. Same as eight above only the sales rep is from a small firm whose name you are not familiar with. 3 Evaluation Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence Trustworthiness: Competence VH_:_}_}_;_5_}_YL VH_5_3_5_3_}_5_YL VH_3_3_5_}_3_5_YL VH_3_3_5_5_3_;_YL VH_;_3_5_3_}_}_YL VH_3_5_5_5_}_}_YL VH_;_5_}_3_5_;_YL VH;5_5_5_5_3_3_YL VH_3_3_5_3_3_3_YL VH_;_}_5_5_5_3_YL VH_}_3_}_}_3_3_YL VH_}_5_3_3_;_;_YL VH_5_}_:_3_5_3_YL VH_}_3_5_5_3_3_YL VH_5_3_3_5_3_3_YL VH_5_3_5_5_3_5_YL VH_}_}_3_5_5_}_YL VH_}_3_5_5_5_3_YL VH_5_3_}_5_}_5_YL VH_}_;_3_}_5_3_YL VH_;_,_3_;_3_5_YL VH : : : : : : VL APPENDIX 4 141 APPENDIX 4 Directions Please follow these instructions carefully. Answer all questions as objectively and honestly as you can. Your response will be kept abso- lutely confidential. There are two types of questions you are asked to answer: A. The first requires you to divide ten (10) points between two or three objects or persons. An example of this type is: Smith Jones. West 1. Speaking ability 6 3 1 Here you are indicating that Smith is twice as good a speaker as Jones and Jones is three times better than West. You may divide the ten points anyway you choose but make sure that one (1) point is given to each persOn or thing evaluated. DO NOT GIVE FRACTIONAL POINTS, E.G., 3.5. The second type of question requires you to indicate, by circling a number, the one location on a seven-point scale that best expressed your opinion or judgment. For example: Completely Completely True False 1. All computers do essentially '7 E5 5 4 3 2 1 the same thing . Do not leave out any of the questions. Even if you are not com- pletely sure of an answer, give the answer that is your best guess. l 42 Part I All questions in this section refer to the case itself and the informa- tion contained in it. In making your evaluations consider only information presented in the case, that is, as best you can, disregard any knowledge of the subject you may have gotten from other sources. (Divide 10 points between the two systems for each item.) Product Features Maintenance Services Reliable Delivery Cost Vendor Reputation Overall Acceptability Overall Desirability Likelihood that you would buy Brand A Brand B Which system is your choice as the one Plasti-Part should buy? (Circle one) 1. Brand A 2. Brand B How important was each of the following factors to you in making your (Please circle one number for each factor.) decision? 10. Product Features 11. Maintenance Services 12. Reliable Delivery 13. Cost 14. Vendor Reputation Very High Importance 7 6 5 7 6 5 7 6 5 7 6 5 7 6 5 2 Very.Low Importance l l 143 Part II The following questions concern your personal judgment and feelings about the decision you have just made. Please answer all questions by circling one number only. 15. l6. 17. 18. 19. 20. 21. 22. 23. 24. 25. How trustworthy do you think the source of your information is? Very Not Trust- Trustworthy 7 6 S 4 3 2 1 worthy At All How reliable do you think the source of your information is? Very Not Reliable Reliable 7 6 5 4 3 2 1 At All How knowledgeable do you think the source of your information is? Very Not Knowledge- Knowledgeable 7 6 5 4 3 2 1 able At All How competent do you think the source of your information is? Very Not Competent 7 Competent 6 5 4 3 2 1 At All The information in the attachment indicates that Brand A is a better quality product. Agree Disagree Completely 7 6 5 4 3 2 1 Completely The information in the attachment indicates that Brand A is a better value for the money. Agree Disagree Completely 7 6 S 4 3 2 Completely The information in the attachment indicates that Brand A is a more dependable product. Agree Disagree Completely 7 6 5 4 3 2 1 Completely The information in the attachment indicates that with Brand A comes better service. Agree Disagree 7 Completely 6 5 4 3 2 1 Completely How confident are you that you have made the best choice? Very Not Confident Confident 7 6 5 4 3 2 1 At All How sure are you that you have analyzed the case correctly? Very Not Sure Sure 7 6 5 4 3 2 1 At All How strongly do you feel you have made a correct decision? Very Not Strongly Strongly 7 6 5 4 3 2 1 At All *************t********************** DO NOT GO ON TO PART III UNTIL AFTER YOU HAVE DISCUSSED THE CASE AS A GROUPIIII 144 Part III Based on the discussion your group has just had, please evaluate the two brands again . Brand A Brand B 26. Product Features 27. Maintenance Services 28. Reliable Delivery 29. Cost 30. Vendor Reputation 31. Overall Acceptability 32. Overall Desirability 33. Likelihood that you would buy 34. Which system is your choice as the one Plasti-Part should buy? (Circle one) 1. Brand A 2. Brand B How important was each of the following factors to you in making your decision? (Please circle one number for each factor.) Very High Very Low Importance Importance 35. Product Features 7 6 5 4 3 2 l 36. Maintenance Services 7 6 5 4 3 2 l 37. Reliable Delivery 7 6 5 4 3 2 l 38. Cost 7 6 5 4 3 2 l 39. Vendor Reputation 7 6 5 4 3 2 l 145 Part IV In this section you are to evaluate yourself and the other members of your group. Remember, you must give each person at least one (1) point, the points must add to ten on each line, and a higher number of points means a higher rating on that characteristic. Person 1 Person 2 Person 3 Name Name Name 40. Participation in discussion 41. Amount of influence in determining group choice 42. Talkativeness in discus- sion 43. Persuasiveness in getting group to go along with his/her ideas 44. Strength of argument to get group to choose one system over another 45. Leader of the group 46. Seemed confident that his/ her choice was best 47. Competence, that is, analyzed the case correctly 48. Overall asset in helping group make best choice 49. Likeable and easy to work with 50. Would like to work with again APPENDIX 5 146 APPENDIX 5 General Products Inc. The purchasing agent in a fabrication plant has requested your recomm- endation of a walk-up free access bond paper copier to be used by their office staff. He has also supplied the following information: A monthly volume of 10-12,000 copies is expected, most of which will be 1-3 COpies/original. Roughly 75% of the originals are typed letters, memos, etc., with the remainder equally divided between line drawings and half-tones. There are no special requirements for paper stock; however, reduction capability and automatic two-sided COpying are both desirable if the cost is not too high. Based on a preliminary screening, the choice has been narrowed down to two brands: Vendor A and Vendor B. The purchasing agent has indicated that he is pressed for time which means that you should make a recommendation based on whatever informa- tion is currently available, even if you feel that more information would normally be needed. The information below was given to you by Jim Pierce. He is in charge of an office group consisting of 45 people in product design. He has had first-hand experience with both the A and B machines. "Based on our experience we've found that the A machine is somewhat slower than the B machine but tends to break down much less. Both pieces of equipment produce fairly good copies as long as you stick to nice sharp originals, but with fuzzy originals the B machine performs better. There isn't that much difference in costs as far as we can see, but we have not done any thorough cost studies either. The people who service the B machine generally get here faster after we call but that is not always the case." 147 IMF Inc. Divisional headquarters staff has circulated a memo indicating that a controlled environment, high volume copying/duplicating center is being planned for installation when the staff occupies its new home office building. You have been asked to submit your recommendation for the machine that will best satisfy the requirements listed below. Anticipated monthly volume is 80,000 COpies with an average of 30 copies per original. Approximately 95% of the production will be typed letters and memos, with the other 5% consisting largely of half-tones. The machine selected should be capable of good quality and consistent perfor- mance with the paper currently in use and should perform both reduction and automatic two-sided COpying. These latter requirements may be waived if cost considerations warrant so doing. Downtime is to be avoided by selecting thoroughly reliable equipment. The staff memo was delayed in reaching you. However, after doing some quick checking around you found that only two vendors had equipment worth giving serious consideration to; they are vendors X and Y. Allen Page, who is director of the printing department at the Toledo plant, has passed along the following information about the x and Y machines. "We do a fair amount of high volume work and so we rely on the X machine more because of its two-sided copying capability. Of course, cost is always important and the Y machine looks good there too. The problem is that neither the X or Y machine is all that fast and we would really like to improve on turn-around time. I must say though that the service reps for the X machine really seem to know what they are doing and don't dilly-dally around getting here. Of course I see them more often too and I wish I din't." 148 IMF Inc. Divisional headquarters staff has circulated a memo indicating that a controller environment, high volume copying/duplicating center is being planned for installation when the staff occupies its new home office building. You have been asked to submit your recommendation for the machine that will best satisfy the requirements listed below. Anticipated monthly volume is 80,000 copies with an average of 30 COpies per original. Approximately 95% of the production will be typed letters and memos, with the other 5% consisting largely of half-tones. The machine selected should be capable of good quality and consistent per- formance with the paper currently in use and should perform both reduc- tion and automatic-two-sided copying. waived if cost considerations warrant so doing. avoided by selecting thoroughly reliable equipment. The staff memo was delayed in reaching you. These latter requirements may be Downtime is to be However, after doing some quick checking around you found that only two vendors had equipment worth giving serious consideration to; they are vendors X and Y. Comparison data on the X and Y machines based on surveys of users have been published by Buyers Services Inc., an independent and well respected company. Their findings are summarized below: Feature COpy speed (No. per minute) Copy quality: typed letter half-tone Machine breakdowns per month (average) Total cost per copy including supplies (assumes 75,000/month) Reducing capability Automatic two-sided copying Maintenance service response time Retrofit update capability Noise level @ 3 feet Electricity consumption (in kilowatts) Standby: Operation: Brand Rating» Vendor X 70 Good Good 3.9 2.5¢ yes yes 15 hours yes 79 dba Vendor Y 56 Very good Poor 1.9 2.2¢ yes no 2 hours no 87 dba 149 General Products Inc. The purchasing agent in a fabrication plant has requested your recomm- endation of a walk-up free access bond paper COpier to be used by their office staff. He has also supplied the following information: A monthly volume of 10-12,000 copies is expected, most of which will be 1-3 COpies/Original. Roughly 75% of the originals are typed letters, memos, etc., with the remainder equally divided between line drawings and half-tones. There are no special requirements for paper stock; however, reduction capability and automatic two-sided copying are both desirable if the cost is not too high. Based on a preliminary screening, the choice has been narrowed down to two brands: Vendor A and Vendor B. The purchasing agent has indicated that he is pressed for time which means that you should make a recommendation based on whatever informa- tion is currently available, even if you feel that more information would normally be needed. The comparison data below has been supplied by Mr. Frank Cassady, who is the sales representative for vendor A. This firm has only been in the copier business for a few years and you have spoken to Mr. Cassady only a half dozen times in the past year and a half. Feature Copy speed (first cOpy) Copy quality: typed letter line drawing half-tone Machine breakdowns per month (average) Total cost/COpy including supplies (assumes 10,00/month) Reducing capabil ity Automatic two-sided COpying Maintenance service response time (average) Retrofit update capability Noise level @ 3 feet Electricity consumption (in kilowatts) Standby: Operation: Brand Rating Vendor A 6.0 sec. Very good Fair-good Fair-poor 1.6 4.9¢ yes no 3 hours yes 90 dba Vendor B 4.5 sec. Good-very good Fair-good Fair-good 3.6 S.2¢ no yes 55 minutes yes 74 dba .4 7 150 IMF Inc. Divisional headquarters staff has circulated a memo indicating that a controlled environment, high volume copying/duplicating center is being planned for installation when the staff occupies its new home office building. You have been asked to submit your recommendation for the machine that will best satisfy the requirements listed below. Anticipated monthly volume is 80,000 COpies with an average of 30 COpies per original. Approximately 95% of the production will be typed letters and memos, with the other 5% consisting largely of half-tones. The machine selected should be capable of good quality and consistent per- formance with the paper currently in use and should perform both reduc- tion and automatic two-sided copying. These latter requirements may be waived if cost considerations warrant so doing. Downtime is to be avoided by selecting thoroughly reliable equipment. The staff memo was delayed in reaching you. However, after doing some quick checking around you found that only two vendors had equipment worth giving serious consideration to; they are vendors X and Y. Comparison data on the X and Y machines based on surveys of users has been published by Buyers Services Inc., an independent and well respected company. Their findings are summarized below: Feature Brand Rating Vendor X Vendor Y Copy speed (No. per minute) 70 56 Copy quality: typed letter Good Very good half-tone Good Poor Machine breakdowns per month (average) 3.9 1.9 Total cost per copy including supplies (assumes 75,000/month) 2.5¢ 2.2¢ Reducing capability yes yes Automatic two-sided COpying yes no Maintenance service response time 1% hours 2 hours 151 General Products Inc. The purchasing agent in a fabrication plant has requested your recomm- endation of a walk-up free access bond paper copier to be used by their office staff. He has also supplied the following information: A monthly volume of 10-12,000 copies is expected, most of which will be 1-3 copies/original. Roughly 75% of the originals are typed letters, memos, etc., with the remainder equally divided between line drawings and half-tones. There are no special requirements for paper stock; however, reduction capability and automatic two-sided copying are both desirable if the cost is not too high. Based on a preliminary screening, the choice has been narrowed down to two brands: Vendor A and Vendor B. The purchasing agent has indicated that he is pressed for time which means that you should make a recommendation based on whatever informa- tion would normally be needed. The comparison data below has been supplied by Mr. Frank Cassady, who is the sales representative for vendor A. This firm has only been in the copier business for a few years and you have spoken to Mr. Cassady only a half dozen times in the past year and a half. Feature Brand Rating Vendor A Vendor B Copy speed (first copy) 6.0 sec. 4.5 sec. Copy quality: typed letter Very good Good-very good line drawing Fair-good Fair-good half-tone Fair-poor Fair-good Machine breakdowns per month (average) 1.6 3.6 Total cost/copy including supplies (assumes 10,000/month) 4.9¢ 5.2¢ Reducing capability yes no Automatic two-sided copying no yes Maintenance service response time (average) 3 hours 55 minutes APPENDIX 6 Instructions: 152 APPENDIX 6 Part I Based on the information you now have please evaluate the two brands by dividing 10 points between them for each question. Make sure the points add to 10 on each line and that you give at least one point to each brand. your Opinion of it is. Example: 0. Reliable Delivery 1. Product Features 2. Maintenance Services 3. Reliability 4. Total Cost 5. Additional Features 6. Overall Acceptability 7. Overall Desirability 8. Likelihood you would recommend 9. Which COpier would you recommend? 1. Vendor X 2. Vendor Y The more points you give to one brand the better Vendor X Vendor Y .___.3___ __2____ (Circle one) How important was each of the following factors to you in making your decision? 10. ll. 12. l3. 14. Product Features Maintenance Services Reliability Total Cost Additional Features (Please circle one number for each factor.) Very High Very Low Importance Importance 7 6 5 4 3 2 l 7 6 5 4 3 2 l 7 6 5 4 3 2 l 7 6 5 4 3 2 l 7 6 5 4 3 2 l 153 Part II The following questions concern your personal judgment and feelings about the decision you have just made. Please answer all questions by circling one number only. 15. How trustworthy do you think the source of your information is? Very Not Trust- Trustworthy 7 6 5 4 3 2 worthy At All 16. How reliable do you think the source of your information is? Very Not Reliable Reliable 7 6 5 4 3 2 1 At All 17. How knowledgeable do you think the source of your information is? Very Not Knowledge- Knowledgeable 7 6 5 4 3 2 1 able At All 18. How competent do you think the source of your information is? Very Not Competent Competent 7 6 5 4 3 2 1 At All 19. How confident are you that you have made the best choice? Very Not Confident Confident 7 6 5 4 3 2 1 At All 20. How sure are you that you have analyzed the problem correctly? Very Not Sure Sure 7 6 5 4 3 2 1 At All 21. How strongly do you feel you have made a correct decision? Very Not Strongly Strongly 7 6 5 4 3 2 1 At All ************************************ DO NOT GO ON TO PART III UNTIL AFTER YOU HAVE DISCUSSED THE CASE AS A GROUP!!! 154 Part III Instructions. Based on the information you now have please evaluate the two brands by dividing 10 points between them for each question. Make sure the points add to 10 on each line and that you give at least one point to each brand. The more points you give to one brand the better your opinion of it is. Example: Vendor X Vendor Y 0. Reliable Delivery 8 2 22. Product Features 23. Maintenance Services 24. Reliability 25. Total Cost 26. Additional Features 27. Overall Acceptability 28. Overall Desirability 29. Likelihood you would recommend 30. Which COpier would you recommend? (Circle one) 1. Vendor X 2. Vendor Y How important was each of the following factors to you in making your decision? (Please circle one number for each factor.) Very High Very Low Importance Importance 31. Product Features 7 6 S 4 3 2 l 32. Maintenance Services 7 6 5 4 3 2 l 33. Reliability 7 6 5 4 3 2 l 34. Total Cost 7 6 5 4 3 2 1 35. Additional Features 7 6 5 4 3 2 l 155 Part IV Please evaluate by circling the one number for each question that best expresses your opinion. Agree Disagree Completely» Completely 36. Very talkative in discus— 7 6 5 4 3 2 1 sion just now. 37. Very influential in deter- 7 6 5 4 3 2 l mining group choice. 38. Very persuasive in present- 7 6 5 4 3 2 1 ing ideas. 39. Argued strongly for choice 7 6 5 4 3 2 1 of one vendor over the other. 40. Leader of the group in 7 6 5 4 3 2 l today's discussion. APPENDIX 7 156 APPENDIX 7 The causal model does not specify relationships between independent factors and individual attribute vendor evaluations. However it does postulate relationships with overall vendor preference. Since these overall preferences are based on the attribute evaluations, we may look at these to detect sources of difference. The dependent variables are respectively, product features, mainte- nance services, delivery, cost and vendor reputation. Tables 23 through 27 present summary ANOVA tables for the impact of information distribu- tion, source credibility and claim strength on each of these dependent variables. Table 23 Student Sample Product Features ANOVA Summary _— Effect Mean Square F Significance ID x SC x CI 0.000 0.000 .99 SC x CI 0.374 0.533 .47 ID x CI 0.125 0.178 .68 ID x SC 0.000 0.000 .99 CI 3.248 4.631 .04* SC 1.004 1.432 .24 ID 3.430 4.891 .04* *Significant at p§310 Table 24 Student Sample Maintenance Services ANOVA Summary Effect Mean Square F Significance ID x SC x CI 0.020 0.012 .91 SC x CI 0.072 0.043 .84 ID x CI 1.692 1.021 .32 ID x SC 0.423 0.255 .62 CI 4.071 2.456 .12 SC 1.430 0.863 .63 ID 3.678 2.219 .14 157 Table 25 Student Sample Delivery ANOVA Summary Effect Mean Square F. Significance ID x SC x CI 0.058 0.062 .81 SC x CI 1.512 1.610 .21 ID x CI 0.508 0.541 .47 ID x SC 3.164 3.370 .07* CI 0.026 0.028 .87 SC 0.846 0.901 .35 ID 7.360 7.839 .01** *Significant at pi.10 **Uninterpretable due to presence of significant interaction p— Table 26 Student Sample Total Cost ANOVA Summary Effect Mean Square F Significance ID x SC x CI 0.081 0.095 .76 SC x CI 0.076 0.090 .77 ID x CI 0.000 0.000 .99 ID x SC 0.592 0.700 .41 CI 0.981 1.160 .29 SC 3.363 3.976 .05* ID 0.000 0.000 .99 *Significant at p§,lo Table 27 Student Sample Vendor Reputation ANOVA Summary Effect Mean Square F Significance ID x SC x CI 0.254 0.140 .71 SC x CI 2.732 1.509 .23 ID x CI 3.652 2.018 .16 ID x SC 0.093 0.051 .82 CI 0.135 0.075 .79 SC 14.319 7.912 .01* ID 3.872 2.139 .15 *Significant at p:,lo APPENDIX 8 158 APPENDIX 8 The primary quantitative measure of quality of Operationalization is a design's power. Power is a positive function of the number of subjects, the magnitude of treatment effects and the size of the alpha selected for significance testing. It is also an inverse function of uncontrolled variance. Power is the probability of correctly rejecting the null hypothesis and is usually represented as, Power a l - B {l} where 8 is the probability of a Type II error. A Type II error occurs when one fails to reject the null hypothesis of no treatment differences when real differences do exist. Assuming that estimates of treatment effect sizes are available the principal use of the power equation is an a priori determination of the sample size needed to be able to detect those effects. While estimates of treatment effects were not available prior to this investigation they are now available for future reference. An example is worked out below for a single factor experiment. Power is determined by first calculating a . 2 non-centrality parameter, 0 . 2 S|Z(u.-u)2I2/a {2} fl = l A 2 0 error Where 0: = the parameter for an effect A, S = the sample size, U = the treatment cell means, u = the overall mean, a = the number of treatment levels 2 I o O 0 error = Wlthln cells error variance. 159 The value of 02 may be altered by substituting various sample sizes into the above equation. These 0 values are then compared to power functions determined by the chosen a level and numerator and denominator degrees of freedom. The 0 value intersecting a minimal level of power will then indicate a minimal sample size. In the present case, the student sample data for item 32 reveals that, given u = 4.64 “l = 3.75 u2 = 5.38 2 a - 2 0A - 2.29 Then 0: = s|(.89)2 + (.74)2[/2 {3}. 2.29 2 ”A = (S)(.292) {4} a O ”A 54 V51 {5} Now assuming that the desired O = .05 and power = .80, could that level of power be obtained by choosing S = 18? When S a 18, ”A = 2.3 and power 8 .63 and so the desired level of power could not be obtained with that sample size, in fact to achieve power a .80 the sample size under these conditions would have to be at least 23. If three person groups were the level of analysis then a total of 69 subjects would be needed. For item 32, vendor desirability, the main effect for source credi- bility was significant at o I .04 and with a sample size of 64 groups and ”A a .43 the actual power of the design was >.99. Thus if the effect size were reliably known in advance a much smaller sample size could have been used. BIBLIOGRAPHY BIBLIOGRAPHY Norman H. Anderson, "Integration Theory and Attitude Change." Psycho- logical Review, 78 (May 1971), 171-206. Norman H. Anderson and Cheryl C. Graesser, "An Information Integration Analysis of Attitude Change in Group Discussion." Journal of Personality and Social Psychology, 34 (August 1976), 210-22. Solomon E. Asch, "Effects of Group Pressure Upon the Modification and Distortion of Judgements." in Readings in Social Psychology, E. Maccoby (ecl), New York: Holt, Rinehart & Winston (1958). Richard P. Bagozzi, "Exchange and Decision Processes in the Buying Center." in Organizational BuyingyBehavior, T. V. Bonoma and G. Zaltman (eds.), Chicago: American Marketing Association (1978). Richard P. Bagozzi, ”Structural Equation Models in Experimental Research.” Journal of Marketing Research, 14 (May 1977), 209-26. R. F. Bales, F. L. Strodtbeck, T. M. Mills and M. E. Roseborough, "Channels of Communication in Small Groups." American Sociological Review, 16 (1951), 461—8. Raymond A. Bauer, "Consumer Behavior as Risk Taking." in Proceedings, R. S. Hancock (ed.), Chicago: American Marketing Association (1960). George D. Bishop and David G. Myers, "Informational Influence in Group Discussion." Organizational Behavior and Human Performance, 12 (1974), 92-104. M. Burgoon and M. Ruffner, Human Communication, New York: Holt, Rinehart & Winston (1978). M. Burgoon and D. Stewart, "Empirical Investigations of Language Intensity: I. The Effects of Sex of Source, Receiver and Language Intensity on Attitude Change." Human Communications Research, 1 (1975). 244-8. Peter J. Burke, "Participation and Leadership in Small Groups." American Sociological Review, 39 (1974), 832-43. P. Busch and D. T. Wilson, "An Experimental Analysis of a Salesman's Expert and Referent Bases of Social Power in a Buyer-Seller Dyad." Journal of Marketing Research, 13 (February 1976), 3-11. 160 161 Bobby J. Calder, "A Methodology for Research on Organizational Buyer Behavior." in Organizational Buying Behavior, T. V. Bonoma and G. Zaltman (eds.), Chicago: American Marketing Association (1978). J. C. Campbell and D. T. Stanley, Experimental and Quasi-Experimental Designs for Research, Chicago: Rand McNally (1966). Richard N. Cardozo and James Cagley, "Experimental Study of Industrial Buyer Behavior." Journal of Marketinquesearch, 8 (August 1971), 329- 34. James S. Coleman, The Mathematics of Collective Action, Chicago: Aldine Press (1973). Barry E. Collins and Harold Guetzkow, A Social Psychology of Group Processes for Decision Making, New York: John Wiley & Sons (1964). Paul D. Converse, The Elements of Marketing, New York: Prentice-Hall (1930). James R. Cooley, An Analysis of Relative Power Among Participants in the Purchase of Industrial Components, Unpublished DBA Thesis, Arizona State University (1976). James R. Cooley, Donald W. Jackson and Lonnie L. Ostrom, "Analyzing the Relative Power of Participants in Industrial Buying Decisions." in Proceedings, Educator's Conference, Chicago: American Marketing Associa- tion (1977). Melvin T. Copeland, "Buying Motives for Industrial Goods." Harvard Business Review, 2 (April 1924), 303. Melvin T. COpeland, Principles of Merchandising, Chicago: A. W. Shaw Co. (1925). Donald F. Cox, Risk Taking and Information Handling in Consumer Behavior, Boston: Harvard University Press (1967). Joseph F. Dash, Leon Schiffman and Conrad Berenson, "Risk and Persona- lity-Related Dimensions of Store Choice." Journal of Marketing, 40 (January 1976), 32-9. Morton Deutsch and Harold Gerard, "A Study of Normative and Informational Social Influences upon Individual Judgement." Journal of Abnormal and Social Psychology, 51 (1955), 629-36. Robert Dubin, Theory Building, New York: Free Press (1969). W. Erbe, "Gregariousness, Group Membership and the Flow of Information." American Journal of Sociology, 67 (1962), 502-16. N. Feldman and R. N. Cardozo, "Industrial Buying as Consumer Behavior." in Proceedingsy Educators' Conference, Chicago: American Marketing Association (1976). 162 Wade Ferguson, "A Critical Review of Recent Organizational Buying Research." Industrial Marketing Management, 7 (1978), 225-30. Jeremy Finn, A General Model for Multivariate Analysis, New York: Holt, Rinehart & Winston (1974). B. Aubrey Fisher, Small Group Decision Making: Communication and the Group Process, New York: McGraw-Hill Book Co. (1974). J. D. Ford, Jr., "Some Factors Affecting the Behavior of Emergent and Designated Leaders in Smaller Problem-Solving Groups." American Psychologist, 16 (1961), 389. John R. P. French, ”A Formal Theory of Social Power." The Psychologi- cal Review, 63 (1956), 181-94. John R. P. French and Bertram Raven, "The Bases of Social Power." in Studies in Social Power, D. Cartwright (ed.), Ann Arbor: Institute for Social Research (1959). B. G. Glaser and A. L. Strauss, The Discovery of Grounded Theory: Strategies for Qualitative Research, Chicago: Aldine Publishing Co. (1967). Ronald M. Gorman, "Role Conception and Purchasing Behavior." Journal of Purchasing_and Material Management (February 1971), 57-71. Gary Griksheit and W. J. E. Crissy, "Improving Interpersonal Communica- tions Skill." MSU Business TOpics, 21 (Autumn 1973), 63-8. Kjell Grpnhaug, ”Autonomous vs. Joint Decisions in Organizational Buying." Industrial Marketing Management, 4 (1975), 265-71. Kjell Grpnhaug, "Exploring Environmental Influences in Organizational Buying." Journal of Marketing Research, 13 (August 1976), 225-9. J. R. Hackman and C. G. Morris, "Group Tasks, Group Interaction Process, and Group Performance Effectiveness: A Review and PrOposed Integration." in Advances in Experimental Social Psychology, v.8, New York: Academic Press (1975). Hakan Hakansson and Bjorn Wootz, "Supplier Selection in an International Environment: An Experimental Study." Journal of Marketing Research, 12 (August 1975), 46-51. W. L. Hays, Statistics for the Social Sciences, 2nd ed., New York: Holt, Rinehart & Winston (1973). J. K. Hemphill and C. B. McConville, "The Effects of Communication Restraints upon Mutual Problem-Solving Behavior." American Psychologist, 17 (1962), 335. L. R. Hoffman, "Group Problem Solving." in Advances in Experimental Social Psychology, v.2, New York: Academic Press (1965). 163 Carl I. Hovland and Walter Weiss, "The Influence of Source Credibility on Communication Effectiveness." Public Opinion Quarterly (Winter 1951-52), 635-50. John A. Howard and J. Sheth, The Theory of Buyer Behavior, New York: John Wiley & Sons (1969). James Hulbert and Noel Capon, "Interpersonal Communication in Marketing: An Overview." Journal of Marketing Research, 9 (February 1972) 27-34. Micheal D. HUtt, The New Product Selection Process of Retail Buying Committees: An Analysis of Group Decision Making, Unpublished Ph.D. Thesis, Michigan State University (1975). Daniel Johnson and Robert Andrews, "Risky Shift Phenomenon Tested with Consumer Products as Stimuli." Journal of Personality and Social Psychology, 20 (December 1971), 382-5. David Johnson and Larry Ridener, "Self-Disclosure, Participation and Perceived Cohesiveness in Small Group Interaction." Psychological Reporter, 35 (1974), 361-2. H. H. Kelley and John Thibaut, "Group Problem Solving." in The Hand- book of Social Psychology, G. Lindzey and E. Aronson (eds.), v.4, Reading, Massachusetts: Addison-Wesley Publishing Co. (1969). G. Keppel, Design and Analysis: A Researcher's Handbook, Englewood Cliffs, N.J.: Prentice-Hall (1973). G. E. Kiser, C. P. Rao and S. R. G. Rao, "Vendor Attribute Evaluations of 'Buying Center' Members Other than Purchasing Executives." Industrial Marketing Management, 4 (1975), 45-54. Robert Xrapfel, "A Decision Process Approach to Modeling Organizational Buyer Behavior." Proceedings, Educator's Conference, Chicago: American Marketing Association (1978). D. R. Lambert, R. J. Dornhoff and J. B. Kernan, "The Industrial Buyer and the Postchoice Evaluation Process." Journal of Marketing Research, 14 (May 1977), 246-51. D. R. Lehman and John O'Shaughnessy, "Difference in Attribute Impor- tance for Different Industrial Products." Journal of Marketing, 38 (April 1974), 36-42. ' Theodore Levitt, "Communications and Industrial Selling." Journal of Marketing, 31 (April 1967), 15-21. Theodore Levitt, Industrial Purchasing Behavior: A Study of Communica- tion Effects, Boston: Harvard University, Graduate School of Business Administration (1965). John A. Martilla, "Word of Mouth Communication in the Industrial Adoption Process." Journal of Marketing Research, 8 (May 1971), 173-8. 164 H. L. Mathews, D. T. Wilson and J. F. Monoky, Jr., "Bargaining Behavior in a Buyer-Seller Dyad." Journal of Marketing Research, 9 (February 1972), 103-5. H. H. Maynard, W. C. Weidler and T. N. Beckman, Principles of Marketing, New York: The Ronald Press (1939). Micheal Mazis, "Decision Making Role and Information Processing." Journal of Marketing Research, 9 (November 1972), 447. E. J. McCarthy, Basic Marketing, 5th ed., Homewood, Illinois: Richard D. Irwin (1978). W. J. McEwen and Bradley S. Greenberg, "The Effects of Message Intensity on Receiver Evaluation Of Source, Message and TOpic." Journal of Communication, 20 (1970), 340-50. Joseph E. McGrath and I. Altman, Small GroupyResearch: A Synthesis and Critique of the Field, New York: Holt, Rinehart & Winston (1966). James R. McMillan, "Role Differentiation in Industrial Buying Decisions." in Combined Proceedings, Chicago: American Marketing Association (1973). James R. McMillan, "The Role of Perceived Risk in Industrial Marketing Decisions.” Proceedings, Chicago: American Marketing Association (1973). James R. McMillan, "Industrial Buyer Behavior as Group Decision Making." Paper presented at Nineteenth International Meeting of the Institute of Decision Sciences (April 1972). Cyril P. Morgan and John Aram, "The Preponderance of Arguments in the Risky Shift Phenomenon." Journal of Experimental Social Psychology, 11 (January 1975), 25-34. Jum C. Nunnally, Psychometric Theory, New York: McGraw-Hill Book Co. (1967). Jum C. Nunnally, Psychometric Theogy, 2nd ed., New York: McGraw-Hill Book Co. (1978). Lyman E. Ostlund, "Role Theory and Group Dynamics." in Consumer Behavior: Theoretical Sources, Scott Ward and Thomas Robertson (eds.), Englewood Cliffs, N.J.: Prentice-Hall (1973). B. R. Patton and K; Giffin, Problem-Solving Group Interaction, New York: Harper & Row (1973). Gerry Philipsen and T. J. Saine, "The Effect of Reward Criteria on Verbal Participation in Group Discussion." Speech Monographs, 40 (1973), 151-3. Dean G. Pruitt, "Conclusions: Toward an Understanding of Choice Shifts in Group Discussion." Journal of Personalityyand Social Psychology, 20 (1971), 495-510. 165 Peter H. Reingen, "Risk-taking by Individuals and Informal Groups with the Use of Industrial Product Purchasing Situations as Stimuli." Journal of Psychology, 85 (November 1973), 339-45. Peter H. Reingen, "The Risky-Shift in Ad Hoc and Natural Consumer Groups: A Test of the Polarization Hypothesis and a Majority-Rule Explanation." Advances in Consumer Research, 5 (1978), 87-92. H. W. Riecken, "The Effect of Talkativeness on Ability to Influence Group Solutions of Problems." Sociometry, 21 (1958), 309-21. Patrick J. Robinson, Charles Paris and Yoram Wind, Industrial Buying and Creative Marketing, Boston: Allyn & Bacon (1967). Jerome Scott and Peter Wright, "Modeling an Organizational Buyer's Product Evaluation Strategy: Validity and Procedural Considerations." Journal of Marketing Research, 13 (August 1976), 211-24. M. E. Shaw, "Some Factors Influencing the Use of Information in Small Groups." Psychology_Reporter, 8 (1961), 187-98. Jagdish Sheth, "A Model of Industrial Buyer Behavior." Journal of Marketing, 37 (October 1973), 50-6. Robert Spekman and Bobby J. Calder, "Narrowing the Conceptual-Empirical Gap in Organizational Buyer Behavior." Advances in Consumer Research, F. F. Stephan and E. G. Mishler, ”The Distribution of Participation in Small Groups." American Sociological Review, 17 (1952), 608. E. J. Thomas and C. F. Fink, "Models of Group Problem Solving." Journal of Abnormal Social Psychology, 63 (1961), 53-63. Amiram Vinokur, "Review and Theoretical Analysis of the Effects of Group Processes upon Individual and Group Decisions Involving Risk." Psychological Bulletin, 76 (1971), 231-50. Amiram Vinokur, Y. TrOpe and E. Burnstein, "A Decision-Making Analysis of Persuasive Argumentation and the Choice Shift Effect." Journal of Experimental Social Psychology, 11 (1975), 127-48. Amiram Vinokur and Eugene Burnstein, "Effects of Partially Shared Persuasive Arguments on Group Induced Shifts: A Group Problem-Solving Approach." Journal of Personality and Social Psychology, 29 (1974), 305-15. M. A. Wallach, N. Hogan and D. J. Bem, "Group Influence on Individual Risk Taking." Journal of Abnormal and Social Psychology, 65 (1962), 76-86. Frederick E. Webster, Jr., "Industrial Buyer Behavior: A State of the Art Appraisal." Proceedings, Chicago: American Marketing Association (1969). 166 Frederick E. Webster, Jr., "Informal Communication in Industrial Markets." Journal of Marketing Research, 7 (May 1970), 186-9. Frederick E. Webster, Jr., "Interpersonal Communication and Salesman Effectiveness." Journal of Marketing, 32 (July 1968), 7-13. Frederick E. Webster, Jr., "Modeling the Industrial Buying Process." Journal of Marketing Research, 2 (November 1965), 370-6. Frederick E. Webster, Jr., "On the Applicability of Communication Theory to Industrial Markets." Journal of Marketing Research, 5 (November 1968), 426-8. Frederick E. Webster, Jr. and Yoram Wind, Organizational Buyer Behavior, Englewood Cliffs, N.J.: Prentice-Hall (1972). R. E. Weigand, "Identifying Industrial Buying Responsibility." Journal of Marketing Research, 3 (February 1966), 81-4. David T. Wilson, "Dyadic Interactions: Some Conceptualizations." in Organizational Buying Behavior, T. V. Bonoma and Gerald Zaltman (eds.), Chicago: American Marketing Association (1978). David T. Wilson, "Models of Organization Buyer Behavior: Some Observa- tions." in Buyer/Consumer Information Processing, G. D. Hughes and D. L. Ray (eds.), Chapel Hill, North Carolina: The University of North Carolina Press (1974). David T. Wilson, "Research Approaches to Multi-Participant Decision Processes." Advances in Consumer Research, 5 (1978), 641. Yoram Wind, Industrial Buying Behavior: Source Loyalty_in the Purchase of Industrial Components, Unpublished Ph.D. Dissertation, Stanford University (1966). Yoram Wind, "Industrial Source Loyalty." Journal of Marketing Research, 7 (1970), 450-7. Yoram Wind, "Recent Approaches to the Study of Organizational Buyer Behavior." Proceedings, Chicago: American Marketing Association (1974). Arch G. Woodside, "Informal Group Influence on Risk Taking." Journal of Marketing Research, 9 (May 1972), 223-5. Gerald Zaltman and T. V. Bonoma, "Organizational Buying Behavior: Hypotheses and Directions." Industrial Marketing Management, 6 (1977), 53-60.