A S T U D Y OF C O M P U T E R - S U P P O R T E D N E G O T I A T I O N : M O D E OF C O M M U N I C A T I O N , I N C E N T I V E S C H E M E , A N D DECIS ION SUPPORT T O O L S by Y V O N N E K I N W I N G C H A N B. Sc., University of London, 1990 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF M A S T E R OF S C I E N C E I N B U S I N E S S A D M I N I S T R A T I O N in T H E F A C U L T Y OF G R A D U A T E STUDIES (Management Information Systems Division, Commerce and Business Administration) We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y OF BR IT ISH C O L U M B I A November 1994 © Yvonne K i n Wing Chan, 1994 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of flltoAHM&l lArftXMATiOAJ SjVlE^, FALUlTj of COMMtfCt A*d> The University of British Columbia Vancouver, Canada Date -[Of* btOVfntBL*, Iffi DE-6 (2/88) ABSTRACT This paper examines the effects of mode of communication (computer-mediated versus face-to-face), incentive scheme (divisional versus mixed), and the availability of a decision support tool that calculates the optimal solutions on the outcome and process of negotiations. An experiment was conducted in which subjects were involved in two-person bilateral bargaining in a transfer pricing setting. Subjects played the role of either the Marketing Manager (the buyer division) or the Production Manager (the seller division); their task was to negotiate the transfer price and quantity of an intermediate commodity. Thirty-two negotiation dyads were randomly assigned to one of the experimental treatments so that there was an equal number of dyads in each condition. A negotiation support system (NSS) was used in all the experimental sessions. The system provides subjects with various decision support tools as well as an electronic linkage which allows the transmission of both textual and graphical information. Three main dependent variables were measured: (1) negotiation outcome, (2) communication and process variables, and (3) perception on communication efficiency and co-operation. Results showed that in the presence of an NSS, the difference in experimental treatments led to no differences on the negotiation outcome. More exchanges of remarks were found in the face-to-face condition but there was no evidence that the communication medium resulted in a more efficient and co-operative negotiation. A mixed-incentive scheme, however, was found to enhance communication efficiency and promote co-operative behaviour in negotiations. ii T A B L E OF CONTENTS Abstract i i Table of Contents i i i L ist of Tables v i L ist of Figures v i i i Acknowledgement ix Introduction 1 Chapter 1 Literature Review 1.1 Negotiation 4 1.2 Negotiation and Decision Making 6 1.3 Transfer Pricing 7 1.4 Negotiation Support Systems 9 1.5 GDSS and Communication 11 1.6 GDSS , Computer-mediated Communication, and Group Performance 13 1.7 Computer-mediated versus Face-to-face Communication on Groups 14 Chapter 2 Hypothesis Development 2.1 Computer-mediated versus Face-to-face 17 2.2 Divisional-incentive Scheme versus Mixed-incentive Scheme 18 2.3 Availabil ity of a Decision A i d to compute "Optimal" Solutions 19 2.4 Interaction Effects 19 Chapter 3 The Negotiation Support System 3.1 TransferPricing Game 21 3.2 Decision Support Tools - A n Overview 22 3.3 Panels 3.3.1 Market Demand Curve 23 3.3.2 Average Cost Curve 24 3.3.3 Profit Table 24 3.3.4 Offer 25 3.3.5 Message 26 3.4 Profits Calculations in T G 27 i i i Chapter 4 Experimental Design 4.1 The Independent Variables 30 4.2 The Task Environment 4.2.1 Task Description 30 4.2.2 Computer-mediated Communication 31 4.2.3 Face-to-face Communication 31 4.2.4 Divisional-incentive Scheme 31 4.2.5 Mixed-incentive Scheme 32 4.2.6 Optimal Solutions 32 4.3 The Dependent Variables 4.3.1 Outcome Measures 3 3 4.3.2 Communication and Process Variables 34 4.3.3 Perception Measures • 36 4.4 Additional Measures - Ranks 37 4.5 Subjects 37 4.6 Experimental Procedure 38 Chapter 5 Results 5.1 Outcome Measures 40 5.2 Communication and Process variables 44 5.3 Perception Measures 46 5.4 Ranks 47 5.5 Experimental Data 47 Chapter 6 Discussion and Conclusion 6.1 Summary of Results 51 6.2 Discussion and Implications 52 6.3 Extensions 57 Bibliography 61 Tables 65 Figures 102 Appendix A Log Files 111 Appendix B Advertisement 116 Appendix C Consent Form 117 Appendix D Demographic Data Questionnarie 119 iv Appendix E Task Description Sheets 120 Appendix F Post-experiment Questionnarie 126 Appendix G Profit Schedules Example 130 v LIST OF TABLES Table 1 Outcome measures - Joint outcome ( A N O V A ) Table 2 Outcome measures - Joint outcome (Cell means) Table 3 Outcome measures - Difference between profits ( A N O V A ) Table 4 Outcome measures - Difference between profits (Cell means) Table 5 Outcome measures - Deviation from the Nash solution ( A N O V A ) Table 6 Outcome measures - Deviation from the Nash solution (Cell means) Table 7 Outcome measures - Willingness to share information ( A N O V A ) Table 8 Outcome measures - Willingness to share information (Cell means) Table 9 Outcome measures - Time to reach agreement ( A N O V A ) Table 10 Outcome measures - Time to reach agreement (Cell means) Table 11 Information and Outcome measures ( A N O V A ) Table 12 Information and Outcome measures (Cell means) Table 13 Information and Outcome measures ( M A N O V A ) Table 14 Outcome measures ( M A N O V A ) Table 15 Communication/Process - Number of offers ( A N O V A ) Table 16 Communication/Process - Number of offers (Cell means) Table 17 Communication/Process - Number of remarks ( A N O V A ) Table 18 Communication/Process - Number of remarks (Cell means) Table 19 Remarks - General ( A N O V A ) Table 20 Remarks - General (Cell means) Table 21 Remarks - Requests for information ( A N O V A ) Table 22 Remarks - Requests for information (Cell means) Table 23 Remarks - Discussion of offers ( A N O V A ) Table 24 Remarks - Discussion of offers (Cell means) Table 25 Remarks - Task-irrelevant ( A N O V A ) Table 26 Remarks - Task-irrelevant (Cell means) Table 27 Communication and process measures ( M A N O V A ) Table 28 Perception measures - Communication efficiency ( A N O V A ) Table 29 Perception measures - Communication efficiency (Cell means) Table 30 Perception measures - Co-operation ( A N O V A ) Table 31 Perception measures - Co-operation (Cell means) Table 32 Perception measures ( M A N O V A ) Table 33 Ranks - Marketing ( A N O V A ) Table 34 Ranks - Marketing (Cell means) Table 35 Ranks - Production ( A N O V A ) Table 36 Ranks - Production (Cell means) Table 37 Ranks ( M A N O V A ) LIST OF FIGURES Figure 1 Market Demand Curve Panel Figure 2 Market Demand Curve Figure 3 Production Cost Curve Panel Figure 4 Production Average Cost Curve Figure 5 Create 10x10 Profit Table Panel Figure 6 Offer Panel Figure 7 Joint Graphs Figure 8 Optimal Solutions Table Figure 9 Message Panel v i i i Acknowledgement I would like to thank my thesis supervisor Professor Izak Benbasat for his support, time, and valuable advice at all states of this project. The constructive comments of committee members Professor Keith Murnighan and Professor Carson Woo are vital in bringing about the final form of this thesis and are gratefully appreciated. M y thanks also go to Doctor Andrew Trice who advised me on the initial design of the computer system. Last but not least, I would also like to thank my sister Barbara for helping me with the literature search on Dialog when she was a student in the School of Library, Archival and Information Studies, University of British Columbia. i x INTRODUCTION The Management Information Systems academics have spent a substantial amount of research effort on studying Group Decision Support Systems (GDSS). These computer systems, which combine communication and decision support tools, are developed to overcome "process lossess" in group meetings and to support group problem formulation and solving (DeSantis and Gallupe 1987; McGrath and Hollingshead 1993). The impacts GDSS has on group decision process is mainly due to the changes in the pattern of interpersonal communication that is brought about by the technology (DeSantis and Gallupe 1987). Extensive research, therefore, has been carried out to compare the benefits of computer-mediated communication (CMC) over face-to-face communication (FTF) on group performance. In this study, an experiment was carried out to investigate the impacts computer support has on the outcome and process of negotiations carried out in a transfer pricing setting. Three independent variables were manipulated: (1) mode of communication: computer-mediated versus face-to-face, (2) incentive scheme: divisional versus mixed, (3) decision aid to compute optimal solutions: available versus not available. Thirty-two negotiation dyads were formed and were randomly assigned to each of the eight conditions. Dependent variables such as the joint outcome, time to reach agreement, differences between individual outcomes, as well as subjects' perception on the efficiency of the negotiation were measured. This study differs from other computer-supported negotiation studies in the setting and the task involved. While the computer systems used in previous studies only provided facilities for sending and receiving electronic messages and that subjects in the 1 face-to-face condition were not given any form of computer support, all subjects in this study negotiated using a Negotiation Support System (NSS). This system provides negotiators with tools to calculate their profits at different transfer price and transfer quantity combinations. Negotiatiors can exchange their profit and cost information via the electronic linkage provided. This allows negotiators to see their own and their opponents' profit schedules in both numeric and graphical forms. Moreover, the system keeps track of the offers a negotiator sent and received. Because an electronic linkage between subjects was present in all cases, results of this study do not pertain only to the difference in communication media, but also indicate differences, if any, on the negotiation outcome and process when negotiating parties are able to communicate both electronically and face-to-face. The NSS used in this study is a facilitator (Kersten 1987) as it does not intervene the decision process but provides negotiators with the necessary decision support and information processing. One of the functions provided is the capability of calculating some solutions that give the optimal overall outcome. The human information processing capacity is limited and often in decision making, they will search for a reference point so that judgements can be based upon (von Winterfeldt and Edwards 1986; Montgomery 1989). Another objective of this study is therefore to find out if negotiators will "anchor" their decisions on this extra information and hence, influences the outcome of the negotiation. Transfer pricing takes place between two individual divisions within the same firm (Abdel-khalik and Lusk 1974). An appropriate incentive scheme, is therefore, necessary to ensure that each profit centre is making profit and that the overall firm's profit is maximized. Chalos and Haka (1990) manipulated this variable on two levels: 2 (1) divisional - in which rewards are based solely on individual's profit, and (2) mixed -in which rewards are based on individual's profit and that of the firm's. A similar approach is taken here to manipulate the variable to find out the main effects of incentive scheme and its interaction with the other two independent variables. This paper proceeds as follows. Chapter 1 reviews the literature on negotiation, decision making, transfer pricing, NSS, GDSS, and communication media. Chapter 2 presents the hypotheses. Chapter 3 describes the NSS used in the study, followed by the experimental design in Chapter 4. Chapter 5 presents the results. Chapter 6 discusses the results and concludes the paper. 3 CHAPTER 1 - L ITERATURE REVIEW 1.1 Negotiation Negotiation takes place when there is a conflict of interests. The objective of the process is to find a solution that is mutually acceptable to all involved parties. Specifically, negotiation can be regarded as a series of joint strategies, proposed by each negotiator: one person makes a proposal, the other person makes a counter proposal, and so on. The cycle continues until both parties are satisfied with or agree to a certain outcome. According to Bartos (1974), each negotiator starts with a high demand and then gradually lowers it. As demands decline over time, and payoffs increase (because each offer improves on the previous one), the two finally converge to a point where demands are met and hence, a solution is reached. Kniveton (1989) describes negotiation as a 3-stage process. In stage 1, both negotiators spell out fully their standpoints and by making offers which give them the best possible outcome, they convey a message that their goals may be completely incompatible. In the second stage, extreme positions are put aside and negotiators begin to explore possibilities where agreement can be reached. In the final stage, they find an agreement. Kersten (1988), who views negotiation on the basis of aspiration level, describes a similar model. Decision-makers (negotiators) define aspiration levels at the beginning of the decision process (negotiation). Proposals are then made according to the aspiration levels. When decision-makers see that none of the proposals are simultaneously acceptable to all involved parties, they lower their aspiration levels until possible alternatives that are feasible to all can be found. 4 Two points of view exist as to how offers should be made. One can behave like the "hawks" who never concede but put pressure on the other parties to force them to concede. Conversely, one can act like the "doves" who advocate the soft approach and stimulate mutual concessions throughout. The availability of information about the opponent's payoffs determines what strategy one should adopt. Bartos (1974) points out that "Dovish behaviour is likely when information about the opponent is abundant; hawkish behaviour is likely when it is scarce." In other words, the less information the negotiators have about their opponents, the more rivalistic their behaviour tends to be. Three basic strategies are identified for moving toward agreement (Pruitt 1981): (1) to concede unilaterally regardless of the opponent's behaviour, (2) to act competitively, and (3) to enact coordinative behaviour. While strategy 1 refers to the "dovish" behaviour and strategy 2 the "hawkish", the third strategy lies mid-way between. By being coordinative, negotiators collaborate with each other to search for a mutually acceptable solution. The strategy involves concession exchange and the sharing of information about goals and priorities which leads to an integrative agreement. An agreement is said to be integrative if it reconciles the parties' interests and thus provides high benefit to both of them (Pruitt 1981). In this study, an agreement is considered to be more integrative than another when it provides a higher joint outcome (defined as the sum of the individual outcomes) and a smaller absolute difference between individual outcomes. Not only does abundant information promote "dovish" behaviour (Bartos 1974), it also encourages integrative bargaining (Walton and McKersie 1963). Integrative bargaining requires a maximum flow of accurate information. Each party must faithfully communicate to his or her opponent actual data about his or her preferences and cost 5 structure. In their series of experiments on bargaining, Siegel and Fouraker (1960) found that bargainers who had complete information on each other's payoff structure achieved higher joint payoff and divided it more equally. 1.2 Negotiation and Decision Making Negotiation is regarded as a decision making process (Harnett and Cummings 1980; Pruitt 1981; Kersten 1988) which according to Huber (1989) can be partitioned into a number of sub-processes: (1) the generation of alternatives, (2) the identification of consequences of actions, (3) the selection of dimensions, (4) the test as to whether an alternative satisfies one's requirements, and (5) the selection of an alternative out o f a set of alternatives. In negotiation, sub-processes 1 to 4 are iteratively carried out in the first two stages where offers are made and evaluated. The selection of a final solution (step 5) takes place in stage three of the negotiation process. Montgomery (1989) sees the decision making process as a search for a dominance structure, i.e. a cognitive structure in which one alternative can be seen as dominant over the other. A procedure often used by individuals when searching for a dominant alternative is "anchoring" (von Winterfeldt and Edwards 1986; Montgomery 1989). When people are asked to make an estimate, they w i l l first find an anchor or a reference point and then adjust upward or downward i f they find it necessary to do so. Here is a typical example of anchoring and insufficient judgement. When asked to estimate the starting salary of a newly hired information systems professional who possesses four years experience and good all-round qualifications, people who were told that a secretary who knows nothing about the profession had guessed an annual salary o f $100,000 in general, gave higher estimates than those who were told that the secretary made a guess 6 of $20,000 (Bazerman 1993). Even though people were told that the secretary is an outsider with no relevant knowledge, the given estimate was still used as a reference point upon which judgements were made. Because of their limited information processing capability, human beings often first try to reduce the number of alternatives to consider before they engage in any evaluation activity (Westhoff 1989). In this study, subjects were purposely "biased" with given anchors (optimal solutions) with the objective to investigate whether this will have positive effects on the negotiation process and outcome. The notion of providing subjects with some optiomal solutions which produce the maximum joint outcome is to determine if the extra information will be used as an anchor so that agreements that are close to the optimal will be reached. 1.3 Transfer Pricing Transfer pricing is a bargaining process which allocates resources between two individual divisions within a firm (Abdel-khalik and Lusk 1974). This internal pricing mechanism is set up to foster divisional autonomy while ensuring that overall firm profit is maximized (Abdel-khalik and Lusk 1974; Chalos and Haka 1990). The classical transfer pricing model consists of a buyer division (the distribution or the marketing unit) and a seller division (the manufacturing unit) whose common task is to determine the level of output and the price of an "intermediate product" that yield the largest aggregate profit for the firm as a whole (Hirshleifer 1956). Hirshleifer (1956) analyses the scheme under different market conditions and concludes that no single pair of output and price is optimal to all cases. The market price is the correct transfer price if the good is produced in a perfectly competitive market, while transferring the product at the marginal cost 7 yields the highest firm profit in imperfect competition. Transfer pricing falls under the concept of bilateral monopoly - the situation in which a commodity is sold by a single seller to a single buyer. Negotiation takes place between the two parties and the common goal is to agree on the quantity and the price of the commodity. Economists have developed several theories on bilateral monopoly which broadly categorize the negotiation into two forms (Siegel and Fouraker 1960): (1) Price Leadership, (2) Equal Bargaining Strength. Price leadership refers to the situation where the price is determined by one party and leaves the quantity to be the only negotiable variable. Under equal bargaining strength, neither the price nor the quantity are pre-set. Negotiations usually occur in the form of successive counter offers of price and quantity and are termed as "all-or-none" bids meaning that a negotiator must accept both the price and quantity specified in a bid if agreement is to be made. In this latter case of equal bargaining strength where one's revenue is the other's expense, joint payoff is a function of quantity alone. Joint payoff, therefore, is maximized at a certain quantity while the price remains indeterminate. Joint payoff and the individual payoff to the negotiators are two variables of primary interest (Siegel and Fouraker 1960; Carnevale and Isen 1986; Chalos and Haka 1990; Arunachalam 1991; Mahenthiran, Greenberg, and Greenberg 1993). In particular, the negotiation outcome is measured in terms of the sum of the two parties' payoffs and the difference between individual payoffs (Chalos and Haka 1990; Arunachalam 1991). In this study, the negotiation outcome is also compared against the Nash solution (Nash 1950) which maximizes the joint payoff and distributes it evenly between the two parties. In a decentralized firm where a transfer pricing system exists, the profit of each 8 division is often used as a criterion to motivate and evaluate divisional performance (Abdel-khalik and Lusk 1974; Chalos and Haka 1990). In their experiments, Chalos and Haka (1990) manipulated market conditions and performance evaluation procedures. They found that in general, a divisional profit-based incentive scheme in which payoff depend solely on divisional profit results in a more integrative outcome than a mixed-incentive scheme which takes into account both the overall firm profit and divisional profits. A divisional-incentive scheme produced higher joint profits except under uncertain market conditions, and led to a more equal distribution of joint profits. The explanation given was that a competitive incentive scheme elevated bargainers' aspiration levels and thus, led to higher bargainer profits. (According to the authors, Ackelsberg and Yuki observed more co-operative behaviour when subjects' rewards were based on corporate profits but were unable to demonstrate that mixed-incentive schemes were associated with high corporate profits.) This variable is also included in this study and is manipulated similarly on two levels: (1) divisional-incentive, and (2) mixed-incentive. In contrast to the Chalos and Haka (1990) experiments, the negotiation in this study is a zero-sum game in which there are no market externalities and that one division's gain or loss is the other's loss or gain. A high level of conflict is involved in zero-sum games This study provides a means to investigate whether a mixed-incentive scheme will lower the tension between negotiators and lessen the likelihood of extreme behaviour so that better outcomes (i.e. higher joint profits) can be achieved. 1.4 Negotiation Support Systems A lot of attention has been given to the design and impacts of Group Decision Support Systems (GDSS). These systems, which combine communication, computer, and decision technologies, are developed to overcome "process losses" in group meetings 9 and to support problem formulation (DeSantis and Gallupe 1987; McGrath and Hollingshead 1993). Various GDSS have been developed to support different types of group decision making tasks which according to McGrath (1984), are classified into (1) generating, (2) choosing, (3) negotiating, and (4) executing. Negotiation support systems (NSS) emerged as a special class of GDSS that handle the negotiation and resolution of conflicting interests (Shakun 1985; Jarke, Jelassi and Shakun 1987; Jelassi and Foroughi 1989; Nunamaker et al. 1991; Lim and Benbasat 1993). The notion of providing negotiators with computer support stems from the idea of alleviating the limitations associated with the human information processing capacity as well as supporting the communication process (Lim and Benbasat 1993). Specifically, NSS support people in reaching an agreement in "hard" negotiations in which resources are fixed and each party wants to maximize its own share (Jelassi and Foroughi 1989). To be classified as an NSS, a system must consist of two major components and be able to provide three functionalities (Lim and Benbasat 1993). The two components are: (1) a decision support system (DSS) for each negotiator, and (2) a linkage between the DSSs so that negotiators can communicate electronically. In terms of functionality, an NSS should be able to provide (1) enough support so that resistance points can be defined, (2) support for strategic analysis so that the other party's needs and the joint outcome can be evaluated, and (3) a communication channel so that discussions can be based upon common referents. Depending on the design, an NSS acts as either a facilitator or a mediator in the negotiation (Kersten 1987). In the former role, the system does not intervene in the decision process but only processes information according to the negotiators' wishes. 10 Negotiators are provided with tools to compare and exchange information and evaluate offers. For example in MEDIATOR (Jarke, Jelassi, and Shakun 1987), a negotiation facilitator, negotiators' utility for different alternatives are displayed both graphically and in data matrix form. Contrary to the non-intervening characteristic of the facilitating role, a mediating NSS mediates in the negotiation, pressing negotiators to achieve a compromise. For instance, NEGO (Kersten 1985) plays the mediating role and calculates for each negotiator, his or her best alternative which fulfills the requirements of the other negotiators. 1.5 GDSS and Communication The essence of GDSS is to provide group members with computer-based communication, information exchange support, and structured group work methods and procedures (Nunamaker et. al. 1991; Dennis and Gallupe 1993). Its impact on the group decision process is mainly due to the changes in the pattern of interpersonal communication that is brought about by the technology. Such changes alter the nature of participation within the group which in turn, affects the outcome and the quality of the decision (DeSantis and Gallupe 1987). Four dimensions have been defined to distinguish communication media (Poole and Jackson 1993). (1) Social presence. This refers to the degree to which the medium allows a communicator to establish a personal connection with others. The more non-verbal cues the medium conveys, the higher social presence is experienced by the communicator. (2) Information richness. This refers to the capacity of the medium to facilitate shared meaning amongst communicators and the ability to reduce equivocality and uncertainty. An information rich medium is one which provides communicators 11 with immediate feedback, has the capacity to transmit multiple cues, allows the use of natural language in addition to numbers, and focuses on the individual. (3) Symbolic meaning. A medium can add symbolic meaning to the messages it carries. For example, a hand written note, as opposed to a typed one, conveys extra personal concern. (4) Bindingness. This refers to whether the medium is able to bind time and space. Writing binds time as it records past events. Electronic communication binds space as it is capable of joining geographically separated actors and areas. This study used same time, same place negotiations with either computer-mediated communication (CMC) or face-to-face communication (FTF). The primary goal is to investigate whether the differences between the two communication media on the first three aforementioned dimensions affect the negotiation process and outcome. Electronic communication (or computer-mediated communication) is a less rich medium than face-to-face communication (McGrath 1984; Siegel et al. 1986; Sproull and Kiesler 1986; Nunamaker et al. 1991; Kiesler and Sproull 1992; Weisband 1992; McGrath and Hollingshead 1993). Speech delivers paralinguistic cues such as gestures which are absent in other modes of communication medium (Cherry 1978). Since only written messages (and/or graphics) are conveyed in CMC, social context cues are lost and the exchange of "backchannel" feedback is reduced (Siegel et al. 1986). In particular, asynchronous group meetings (i.e. different time and different place) remove physical presence cues and eliminate the exchange of non-verbal cues such as sounds of movement and breathing (McGrath 1984; McGrath and Hollingshead 1993). To engage in CMC requires additional skills. Although it is not necessary for someone to be an excellent typist to use the computer, those who cannot type or who are not familiar with the position of the keys on the keyboard may find it difficult or even 12 frustrating to type in messages. Secondly, reading is a more elaborate ski l l than speaking (Cherry 1978). It involves both the peripheral process of visual recognition of the words and sentences and the central process of comprehension. Moreover, as the messages are not backed up by intonation, the reader may have to decide for himself or herself, by inference, any hidden meaning the speaker wants to express. The reader may sometimes judge wrongly or misunderstand the message. 1.6 GDSS, Computer-mediated Communication, and Group Performance Whether a G D S S improves group performance is contingent upon factors such as the characteristics of the groups and the members, the nature of the group task, and the operating circumstances (spatial and temporal constraints) under which the task is being performed (McGrath and Hollingshead 1993). To this list of variables, Benbasat and L i m (1993) add contextual factors and technological factors. They propose that rewards and the level of support provided by the GDSS affect group performance. C M C can increase the range, capacity, and speed of managerial communication, but whether the medium w i l l improve managerial or organizational performance depends on the particular circumstances under which it is used (Zack 1993). Information richness requirements for a task guide the choice of the group communication media (Poole and Jackson 1993; Zack 1993). McGrath and Hollingshead (1993) relate group tasks and media for group communication based on the information richness required for the tasks. Idea generating tasks require only the transmission of task-oriented messages. Attention w i l l be distracted from the task i f social cues are present. Negotiations, on the other hand, require not only facts, but also the transmission of information such as commitment and expectations. Social-emotional content conveyed by non-verbal cues is 13 deemed essential in a bargaining situation (Benbasat, L im, and Rao 1993). Being a low richness medium, C M C provides a good fit for tasks in the "generate" category (McGrath 1984) but a poor fit for those in the "negotiate" category. On the contrary, F T F works best in resolving conflicts but not in generating ideas and plans. 1.7 Computer-mediated versus Face-to-face Communication on Groups Extensive research has compared C M C and F T F on group performance (Kiesler and Sproull 1992). While a number of experiments have been conducted to compare performance on the "generate" and "choose" tasks (Siegel et al. 1986; Kiesler and Sproull 1992; Olaniran 1991; Straus 1992; Weisband 1992; Daly 1993), only a few address the "negotiate" task (Arunachalam 1991; Mahenthiran, Greenberg, and Greenberg 1993). More ideas are generated in C M C than in F T F (Olaniran 1991; Daly 1993). When communication is restricted to the computer, distributed groups outperform proximate groups in group idea generation (Valacich et al. 1994). These results provide evidence that C M C is appropriate in task-oriented tasks which require little group co-ordination such as idea generation. It appears that social presence hinders performance as members' focus is distracted from the task towards their public selves (Daly 1993; Valacich et al. 1994). Also, groups in the C M C condition took more time to reach consensus, showed more equal participation, encouraged more uninhibited behaviour, and resulted in more direct arguing and conflict. Communication efficiency is lower in computer-mediation because typing and 14 reading electronic messages is slower than talking and listening in face-to-face conversation (Kiesel and Sproull 1992). According to Siegel et al. (1986), the difficulty of reading social status cues in C M C encourages participation especially amongst the low-status members. A t the same time, the elimination of social cues in C M C depersonalizes the situation and leads to greater uninhibited behaviour. Previous studies on computer-supported negotiations concluded that C M C is not suitable to conduct negotiations. Arunachalam (1991) conducted a three-person transfer pricing negotiation study and found that C M C groups (which did not have visual access to one another and communicated only via electronic messaging) had lower overall outcomes than those negotiated FTF . Joint profits were more unequally distributed between negotiators. In addition, more competitive and flaming behaviour were observed. Mahenthiran, Greenberg, and Greenberg (1993) also found that negotiation groups had lower outcomes in the C M C condition and were less wi l l ing to place offers that were compatible with the integrative solutions. The results of these two studies show that the lack of social cues in the medium does not foster a co-operative environment but rather, places the negotiation in a competitive atmosphere. Despite the fact that C M C is found to be less efficient than F T F in conducting negotiations, the decision support capability available in the computer is believed to facilitate the negotiation process. L i m and Benbasat (1993) hypothesize that negotiation dyads when provided with electronic communication take less time to reach settlement and are more satisfied with the outcome than those not equipped with the technology because of an increase in perceived commitment of one party by the other. Such increase in perceived commitment is brought about by the capability to transmit both textual and graphical information which serves as common referents that negotiators can refer to 15 during the negotiation. Their propositions were supported by the results of the Foroughi, et al. study (Delaney, Foroughi, and Perkins 1994) which showed that NSS groups achieved higher joint outcomes, more balanced contracts, and greater satisfaction. Such inconsistent results can be accounted for by the different experimental setting and level of computer support provided. The NSS used in the Foroughi, et al. study was a level 2 GDSS (DeSantis and Gallupe 1987), i.e. one which provides decision support tools and electronic communication, while Arunachalam's only provided the capability of interactive messaging. In the Foroughi, et al. study, individuals in the NSS condition were not physically separated. Unlike Arunachalam's and other CMC versus FTF studies, face-to-face communication was allowed across all treatments which made the electronic communication an additional channel of communication. Although Lim and Benbasat (1993) have emphasized the necessity of being able to transmit textual and graphical model based information, the electronic linkage employed in the above two studies allowed only the transmission of textual information. Delaney, Foroughi, and Perkins (1994) believe that although NSS groups perform better than non-NSS groups in terms of higher and more balanced distribution of joint outcomes, the DSS support in an NSS should be given more merit than the electronic linkage. They compared a comprehensive NSS (i.e. one provides decision support and an electronic communication channel) with a DSS (i.e. one provides only decision support with no electronic linkage) and found that there were no significant differences between the two on joint outcome, differences between individual profits, and the negotiation time. However, satisfaction with the negotiation was significantly better with the NSS. 16 CHAPTER 2 - HYPOTHESIS D E V E L O P M E N T 2.1 Computer-mediated verses Face-to-face Communication plays a critical role in negotiations. Although anonymity is always associated with computer-mediated communication (CMC), it has little effects on two-person negotiations since the identities of the negotiators are known to each other. However, its inability to transmit rich information has significant impact on the task. As discussed in Chapter 1, computer-mediated communication (CMC) depersonalizes the situation because it lacks the capability of transmitting paralinguistic cues compared to face-to-face communication (FTF) and since negotiators usually do not receive instant feedback on a remark made, it is likely that one may feel detached from the negotation which in turn, may reduce one's commitment to the negotiation as well as that perceived of the other party. This may discourage information exchange and make it more difficult for trust to be built up. The capability for interactive communication where there is a simultaneous and continuous flow of information and instantaneous feedback encourages the building of mutual trust (Zack 1993). Because of a lack of such capability, it is believed that in CMC, a person may be less willing to share truthful information with the other party due to the lack of trust. Based on this reasoning, the following hypotheses are suggested: HI a Given the same negotiation task and setting, negotiation dyads who communicate via the computer are less willing to exchange information than negotiation dyads who communicate face-to-face. Hlb Given the same negotiation task and setting, more deceptions will be seen in conditions where negotiation dyads communicate via the computer than in conditions where negotiation dyads communicate face-to-face. 17 The search for an integrative agreement will be more efficient when negotiators act co-operatively and exchange truthful information (Walton and McKersie 1965; Pruitt 1981). The hindrance to the building of trust causes fewer integrative agreements to be reached in CMC than in FTF. Negotiation outcomes will also be further away from the Nash solution. H l c Joint outcomes reached by negotiation dyads who communicate via the computer will be lower than those reached by negotiation dyads who communicate face-to-face. H id Differences between individual profits when negotiation dyads communicate via the computer will be higher than when negotiation dyads communicate face-to-face. 2.2 Divisional-incentive Scheme versus Mixed-incentive Scheme Chalos and Haka (1990) manipulated incentive scheme and found that a divisional reward scheme does not lead to significant divisional inequities compared to a mixed-incentive scheme and that the former produces higher joint outcomes. Since the setting of their experiment (market externalities were present) was different from that of this study (a zero-sum game), the above results may not be applicable. In a two-person zero sum game negotiators' priorities are completely opposed. However, a mixed-incentive scheme gives negotiators a common goal which is to maximize the joint profit so that individuals' profits can be increased. More co-operative behaviour will then be observed than under a divisional-incentive scheme where negotiators seek to maximize only their own profits. Fewer integrative agreements and outcomes that are further away from the Nash solution are therefore expected when negotiators are rewarded based only on their own profits. 18 H2a Joint outcomes will be lower under a divisional-incentive scheme than under a mixed-incentive scheme. H2b Differences between individual profits will be higher under a divisional-incentive scheme than under a mixed-incentive scheme. 2.3 Availability of a Decision Aid to compute "Optimal" Solutions As pointed out earlier, joint profits in bilateral bargainings are maximized at a certain quantity which can be determined from the demand and cost schedules. In this study, negotiators will be given some combinations of price and quantity which maximize the sum of the individual profits. These optimal solutions, which are attained at a certain transfer quantity, serve as anchors to "bias" judgements so that agreements reached will result in high overall joint outcome. Moreover, the fixation of the quantity reduces the number of dimensions to be considered which helps speed up the negotiation process. H3a Negotiation dyads who are provided with a decision aid to compute the optimal solutions will achieve higher joint outcome than negotiation dyads who are not provided with such a decision support tool. H3b Negotiation dyads who are provided with a decision aid to compute the optimal solutions will take less time to reach an agreement than negotiation dyads who are not provided with such a decision support tool. 2.4 Interaction Effects The provision of the optimal solutions stands separately from the other two dimensions and so interaction effects are not expected, however, interaction effects 19 between incentive schemes and modes of communication are interesting and are worth examining. Based on the above reasoning, it can be expected that better negotiation outcomes (higher joint outcomes and lower divisional inequities) will be arrived under a mixed-incentive scheme and face-to-face negotiation while poorer outcomes are likely under the other treatments. Cross-cell interaction effects are however, difficult to predict. A divisional-incentive scheme encourages competition while more co-operative behaviour is expected in face-to-face negotiations. CMC encourages uninhibited behaviour while under a mixed-incentive scheme, negotiators will act more co-operatively. These effects, although not hypothesized, will be tested as part of the analysis. 20 C H A P T E R 3 - T H E N E G O T I A T I O N S U P P O R T S Y S T E M This chapter describes the negotiation support system used in this study. The aim of this chapter is to give an overview of the decision support tools that are available as wel l as how subjects can communicate with each other v ia the computer system. The N S S is made up of panels/windows which are described in this chapter under separate sub-headings. 3.1 TransferPr ic ing Game Developed on the N e X T platform, TransferPricing Game (TG) is an NSS that supports both the communication and the decision making process (L im and Benbasat 1993). A l l subjects in this study negotiated using this system. T G is made up of two applications: Marketing (the buyer) and Production (the seller). The two applications are identical in every way except that the market demand curve is displayed and determines the profits in the former, while the same role is played by the average cost curve in the latter. The task of the negotiators is to agree on the transfer price and the transfer quantity of an intermediate good. T G is a zero-sum negotiation game, i.e. Production is the sole supplier and the demand of the intermediate good comes only from Marketing. Forty-five minutes are allowed for the negotiation. The negotiation terminates i f an agreement has not been reached in the time allocated. There is no limit to the number of offers either party can make, i.e. negotiators do not have to wait for a counter offer before they can place a second offer. The only rule imposed in making offers is that an offer must consist of both a transfer price and a 21 transfer quantity and that the first offer must come from Marketing. T G allows electronic messages and both numeric and graphical information (directly provided by the negotiation system or otherwise) to be transmitted (L im and Benbasat 1993). Negotiators can share information on their profits by sending each other their own demand or cost curves and/or profit tables (to be discussed later). Negotiators choose whether to send such additional information. Deception is allowed in TG . Negotiators are given the true market demand curve (average cost curve in the Production application) and a "false" curve. A t their own discretion, negotiators can choose to send the true curve, the false curve, or neither to the other party. However, once a curve is given to the other party, the other curve cannot be sent. In other words, i f a person misinforms, there is no way to change that information. T G captures negotiators' actions and messages with time stamps in log files. This ensures that the data collected is accurate and free from subjective bias so that analysis of the negotiation process can be carried out with ease (see Appendix A for a copy of the log file). T G requires negotiators to enter a user identity code before it is launched. The code is then used to name the log files and the directory in which the files are placed. 3.2 Decision Support Tools - An Overview Several decision support tools are provided in the system. They are designed mostly for facilitating the evaluation of potential offers. Negotiators can enter a transfer price and a transfer quantity into the system which w i l l automatically calculate their own profits as wel l as those of the other party 22 and the company's (when the other party has provided his or her cost information). Whi le this allows potential offers to be evaluated one at a time, the system also provides 'Sensitivity Analysis'. Wi th this function, negotiators can create a table which gives the profits of one hundred different combinations of transfer price and transfer quantity. The negotiators can then determine at what price and quantity are their profits maximized. Graphical information is also delivered. Profit and cost information is displayed in the form of a demand curve (in the Marketing application) and an average production cost curve (in the Production application). Moreover, the two curves are overlaid so that negotiators can determine graphically the region feasible to both parties. Offers are sent from one party to the other v ia the electronic linkage provided in the system. They are displayed in a scrollable list box so that negotiators can review and compare them during the negotiations. 3.3 Panels Decision support tools are placed in different panels which can be assessed by choosing the appropriate options in the main menu. 3.3.1 Market Demand Curve (see Figure 1) Present only in the Marketing application, this panel displays the two pre-defined demand curves. It allows negotiators to determine for each demand curve, the market price associated with a quantity and vice versa. Since the exchange of information about needs and priorities is essential to integrative bargaining, T G allows 23 negotiators to disclose their profit information by providing them means to include the market demand curves in their offers to their opponents. The negotiation w i l l decide whether to send this extra information to the othe party. Either the true or the false demand curve can be sent, however, once a demand curve is selected and sent, the other curve cannot be sent in subsequent offers. Figure 2 displays a market demand curve. 3.3.2 Average Cost Curve (see Figure 3) Similar to the Market Demand Curve panel described in 3.3.1, this panel displays the two average cost curves and lets the Production manager determine for each cost curve, the average cost at a particular level of production. As in the Marketing application, the Production managers can include either the true average cost curve or the false curve in their offers to the Marketing managers. However, i f the actual curve has been sent in an earlier offer, the false curve cannot be sent in later offers and vice versa. Figure 4 illustrates a Production's average cost curve. 3.3.3 Prof i t Table (see Figure 5) Sensitivity Analysis is provided in T G in the form of a profit table. Profits are calculated for 100 pairs of transfer price and transfer quantity. Negotiators create the profit table by specifying the starting transfer price, the increment (i.e. by how much is the next transfer price higher than the previous one), the starting transfer quantity, the increment (i.e. by how much is the next transfer quantity higher than the previous one), and the demand curve (average cost curve in the Production application) under which profits are to be calculated. If a negotiator so wishes, he or she can send the profit table to the other party in his or her offer. Sending a profit table is equivalent to sending a 24 demand curve or an average cost curve since both deliver information on one's profit. To ensure consistency in the information given to the other party, a negotiator is only allowed to send a profit table calculated under the actual (false) demand curve or average cost curve i f the actual (false) demand curve or average cost curve was sent in an earlier offer and vice versa. 3.3.4 Offer (see Figure 6) This panel consists of 5 sections. (a) Offers Logs The logs are two scrollable browsers which record offers placed and received. Displayed are the transfer price and the transfer quantity offered, the profits, and the demand curve (average cost curve in the Production application) and/or the profit table that was sent together with the offer. (b) Place Offer This section is where offers are placed. Each offer is made up of a transfer price and a transfer quantity plus optional information such as the demand curve, average cost curve, and the profit table. (c) Evaluate Offer This section lets negotiators find out how well they, their opponents, and the company as a whole can do at a particular transfer price and transfer quantity. The system calculates the opponents' profit and the company's profit once the opponents' profit information is available (i.e. when the negotiator is given the other party's demand/cost curve or the profit table). The company's profit is the sum of the 25 negotiator's actual profit and the opponent's profit. Since the false curve delivers false information, the calculated opponent's profit may not be the actual profit and so is the company's profit. However, this cannot be told from the offer received. (d) Joint Graphs (see Figure 7) The joint graph displays the market demand curve and the production average cost curve. The area enclosed by the two curves is the region which is feasible or profitable to both parties. (e) Optimal Solutions (see Figure 8) The system calculates the optimal transfer quantity, i.e. the quantity under which the maximum company's profit is attained, and suggests some combinations of transfer price and transfer quantity which bring profit to both parties and at the same time, maximize the company's profit. The first one in the list is the Nash solution that distributes the company's optimal profit evenly between the two negotiators. However, as information given by the other party may not be true, the calculated optimal solutions and the Nash solution may also be different from the actual ones. 3.3.5 Message (see Figure 9) Electronic messages are sent and read in this panel. Negotiators type in messages in the Message To text area and read messages in the Message From text area which always displays the most current message. Once the Send Message button is pushed, the message will be delivered to the other person immediately. To make the communication process parallel to that in face-to-face communication, negotiators are 26 not provided with a historical record of messages sent and received. A n "alert" panel w i l l pop up to inform the negotiators that a new message has arrived. 3.4 Profits Calculations in T G 1 (a) Market Demand Curve Actual: market price = 300-0 .5 * quantity False: market price = 250 - 0.5 * quantity (b) Average Cost Curve Actual: total production cost = 2000 + 200 * quantity - 1.25 * quantity2 + 0.0025 * quantity3 False: total production cost = 2500 + 250 * quantity - 1.56 * quantity2 + 0.0031 * quantity3 average production cost = total production cost / quantity (c) Marketing's Profit marketing's profit = (market price - transfer price) * transfer quantity (d) Production's Profit production's profit = (transfer price * transfer quantity) - total production cost (e) Company's Profit company's profit = marketing's profit + production's profit ^The parameters of the curves were arbitrary. However, they were assigned in a way such that any combination of the demand and cost curve would produce some solutions profitable to both parties. 27 Optimal Transfer Quantity and the Nash solution (i) Optimal Transfer Quantity From (e) company's profit = marketing's profit + production's profit (marketing's revenue - marketing's expense) + (production's revenue - production's cost) Since marketing's expense is production's revenue, company's profit = marketing's revenue + production's cost market price * qty + production's cost Consider the actual market demand curve and the true production cost curve, Company's profit, therefore, depends only on the transfer quantity. The optimal transfer quantity is obtained by differentiating the above expression with respect to qty and set it it zero. 300 - qty + 200 -2.5 * qty + 0.0075 * qty 2 = 0 On solving the above quadratic equation, the optimal transfer quantity is obtained, qty* = 253. (ii) The Nash solution A t the Nash solution, the company's profit is maximized and is evenly divided between negotiators. In other words, the difference between individual's profits is company's profit = market price * qty + production's cost (300 - 0.5 * qty) * qty + (2000 + 200 * qty - 1.25 * qty 2 + 0.0025 * qty3) 28 zero. Difference between individual's profits = marketing's profit - production's profit = (marketing's revenue - marketing's expense) - (production's revenue - production's cost) = [(market price * qty* - (price * qty*)] - [(price * qty*) - production's cost)] = [(300 - 0.5 * qty*) * qty* - (price * qty*)] - [(price * qty*) - (2000 + 200 * qty* - 1.25 * qty*2 + 0.0025 * qty*3)] Equate the above expression to zero and solve for price, price* = 113 The following table displays the optimal transfer price, the optimal transfer quantity, and the associated profits under different demand and cost curves. qty* price* Marketing's Profit Production's Profit Company's Profit 253 113 15,307 15,515 30,822 H+Pf 250 119 14,000 16,813 30,813 Mf+Pt 229 95 20,725 9,484 30,209 Mf+Pf 228 101 19,380 10,777 30,157 M t True market demand curve M f False market demand curve Pt True production cost curve P f False production cost curve 29 CHAPTER 4 - EXPERIMENTAL DESIGN A n experiment was conducted to examine the main effects and the interaction effects of communication medium, incentive scheme, and the availability of a decision support tool to calculate the optimal outcome. This chapter describes the experimental design, the independent and the dependent variables, the experimental task as well as the procedures taken in running the experiment. 4.1 The Independent Variables Three factors, each at two levels, were manipulated in this experiment: (1) mode of communication: computer-mediated versus face-to-face, (2) incentive scheme: divisional-incentive scheme versus mixed-incentive scheme, and (3) decision aid to calculate optimal solutions: available versus not available. Negotiation dyads were formed and were allocated randomly to one of the eight conditions resulting in a 2x2x2 factorial design. 4.2 The Task Environment 4.2.1 Task Description The experimental task was an intra-firm negotiation between a buyer division (Marketing) and a seller division (Production) over the transfer price and transfer quantity of an intermediate good. A l l negotiation dyads negotiated using T G and were given a maximum of forty-five minutes to come to an agreement. This was a single-session negotiation which ended when an agreement was reached or when total time 30 expired. 4.2.2 Computer-mediated Communication A physical barrier was placed between the negotiators so that they were not able to see each other. Negotiators could only communicate by means of sending electronic messages and offers; verbal communication was not allowed. Although there were no constraints on the topic and content of the messages sent and negotiators could discuss and place offers in their messages, only offers that were placed through the Offer panel (see Figure 6) were regarded as formal and could be accepted. 4.2.3 Face-to-face Communication In the F T F condition, negotiators had visual and verbal access to one another. The electronic mail component, i.e. the Message panel in T G was removed. Negotiators could converse, make non-verbal expressions, discuss and place offers, however, as in the C M C condition, only formal offers that were placed through the Offer panel could be accepted. A l l negotiations conducted in this condition were tape-recorded. 4.2.4 Divisional-incentive Scheme Subjects in this condition were given a description of the negotiation task which stated that their performance in the negotiation would be evaluated based solely upon their own divisional profits. Subjects were asked to answer several questions prior to the negotiation to ensure that they understood their task objective and the award structure. The experimenter checked that questions were answered correctly and discussed with the 31 subjects i f mistakes were found. The task description sheet was not taken away from the subjects so that references to the task objective could be made during the negotiation. A copy of the task description is attached in Appendix E. 4.2.5 Mixed-incentive Scheme In this condition, negotiators' performance was evaluated based on the following formula: payoff = 50% * negotiator's profit + 50% * company's profit As in the divisional-incentive scheme condition, subjects were given a description of the task objective and examples on the calculation of payoffs. Subjects were also asked to answer several questions to demonstrate that they fully understood their objective in the negotiation and the award structure. The experimenter reviewed the answers and discussed with the subjects any concerns they had over the award structure. Same as in the other treatment, subjects could refer to the task description sheet for information on the award structure and their task objective during the negotiation. A copy of the task description is attached in Appendix E. It should be noted that subjects' payoffs in the experiment were not monetary payoffs. They were used only as the basis for ranking. 4.2.5 Optimal Solutions Two versions of T G were used - one of them includes the Optimal Solutions Table (see Figure 8) while the other does not. The table lists some combinations of price and quantity at which company's profit is the maximum and is displayed only when the 32 other party has provided the negotiator with his or her demand (or cost) information. 4.3 The Dependent Variables Three main dependent variables were measured in this experiment: (1) outcome measures, (2) communication and process variables, (3) perception measures. 4.3.1 Outcome Measures This category of dependent variables included, for each dyad, the following measures: (1) joint outcome (2) difference between profits (3) deviation from the Nash solution (4) willingness to disclose information (5) time to reach agreement (1) Joint outcome The joint outcome is the sum of each individual's actual profits at the agreed transfer price and transfer quantity. If an agreement is not reached in the time allocated, each individual's profit and the joint outcome are zero. (2) Difference between profits This is the absolute difference between individuals' actual profits at the agreed transfer price and transfer quantity. A large absolute value indicates an uneven distribution of overall payoff between the two subjects. 33 (3) Deviation from the Nash solution The variable was measured using the following formula. I Mprofit - M p r o f l t ( N a s h ) | + | P p r o f i t _ P p r o f l t ( N a s h ) | 2 (4) Willingness to disclose information Based on the reasoning that it was worse giving false information than not disclosing information at all, the following scoring scheme was used. Each subject received a score of - 1 i f the false curve was sent, 0 i f he or she failed to send his or her demand (or cost) curve to the other party; 1 i f the actual curve was sent. Individual's score in each dyad were summed up to give the overall score for the dyad. A high score indicates that the subjects were wi l l ing to share truthful information. (6) Time to reach agreement The start and finish times of the negotiation were captured in the log files. The time to reach agreement is the difference between the two times expressed in minutes. 4.3.2 Communication and Process Variables Variables in this category were measured by analysing the log files, the electronic messages, and the transcriptions. Two sub-constructs, measured at the dyad level, make up this variable: ( 1 ) number of offers made (2) number of remarks made Mprofit(Pprofit) - Marketing's(Production's) actual profit at the agreed transfer price and transfer quantity M profit (Nash)(Pp r o f 1 t (Nash)) - Marketing's(Production's) profit at the Nash solution 34 (1) Numbers of offers made This variable refers to the total number of offers exchanged between the two subjects in each experimental session and was obtained from the log files and the transcripts. (2) Number of remarks made A remark refers to a statement defined as anything said by a subject without interruption from the other (Carnevale and Isen 1986; Svenson 1989). In the F T F condition, the total number of remarks made by the subjects was readily available from the transcriptions of the tape recordings. Incomplete statements such as "Uh. . . " , "Mm. . . " , "I see", "You know", etc. were not counted. In the C M C condition, the total number of electronic messages sent was counted. Remarks were categorized into task-oriented and task-irrelevant (Siegel et al. 1986, Weisband 1992). (a) Task-oriented remarks. These are remarks which are related to the context of the negotiation. For example, "200 is the lowest price I can offer.". The coding scheme used by Pruitt (1981) and Carnevale and Isen (1986) was adapted to further sub-categorize task-oriented remarks into the following: (i) General - to act/initiation. For example, "Let's see what we should do." "I'm going to find a solution which is better for both of us." (ii) Requests for information. These are statements made to explicitly ask the other person for information on preferences and profits. For example, "Can you give me your cost information? I want to see your cost curve." "Does my offer give you any profit?" 35 (iii) Discussion of offers. These are statements that relate to a particular offer which may release information on one's profit or position. For example, "Your offer gives me $20,000 profit but it does not give as much profit to the company as the previous one." "This is the lowest price I can offer." (b) Task-irrelevant remarks. These are remarks which are irrelevant to decision making. For example, "Shall we have lunch after the negotiation?". Uninhibited remarks which could be either task-relevant or task-irrelevant were also counted. They include: (i) Threats. These are intents to punish the other party. For example, "Agree or I'll..." (ii) Put downs. These are negative statements that derogate the other person's status and power. For example, "This offer is like leaving a dime tip for your waiter." (iii) Remarks which contain swear words, name calling, and insults. For example, "You're a bluffer." 4.3.3 Perception Measures Each subject was asked to answer a post-experiment questionnaire which contained twenty seven-point scale questions adapted from Hi l tz and Johnson (1990) and Arunachalam (1991) (see Appendix F) to measure: (1) communication efficiency (2) degree of co-operation 36 4.4 Additional Measures - Ranks To examine how different treatments affected individual's performance, subjects were ranked based on their payoffs in the experiment. Payoffs were calculated according to the structure of the incentive scheme, i.e. divisional or mixed. Subjects were ranked separately according to their roles (i.e. Marketing or Production) in descending order. 4.5 Subjects Sixty-four students who responded to the advertisement postings were recruited to participate in the experiment. They were randomly formed into dyads and were randomly allocated to the experimental treatments. Subjects received $10 for completing the experimental task. They were told that they would be ranked separately by their roles based on their payoffs as determined by the award structure (divisional or mixed). Those who ranked at the top 25% were promised and awarded a cash bonus between $10 and $50 depending on their ranking. The following is some statistics on the distribution of the subjects. Status: Fu l l Time Students 56 Part Time Students 8 Level: Undergraduate 40 Masters 18 Doctoral 6 37 Faculty: Commerce 20 Sciences 14 Arts 18 Engineering 7 Education 3 Law 2 Experience using the NeXT: never 40 less than 5 times 12 less than 20 times 9 dedicated N e X T user 3 A l l subjects had had taken courses on microeconomic theory. Forty-nine indicated that they were familiar with Windows applications. 4.6 Experimental Procedure The experiment was made up of 5 phases in the following order. (1) The collection of basic demographic data. Subjects filled in a short questionnaire. A copy of the questionnaire is shown in Appendix D. (2) Tutorial. Subjects were given a six-page tutorial which describes the system and the task. The tutorial contained step-by-step instructions and subjects were asked to try out the system. In the tutorial, the subjects were also informed of their objective in the negotiation, i.e. whether their performance was evaluated using a 38 divisional or a mixed incentive scheme. Subjects, on average, spent thirty minutes on the tutorial. (3) A practice negotiation. Subjects were given fifteen minutes where they were explicitly told to try sending offers and messages. They negotiated in the same way as they would in the real negotiation. Offers and messages made in the practice were not captured in the log files. (4) The negotiation. Subjects entered the game and were given forty-five minutes to complete the negotiation. The demand and cost curves parameters were scaled by a different number so that experience in the practice would not influence the offers and decisions made in the real negotiation. (5) Post-experiment data collection. Subjects were asked to answer questions relating to their satisfaction on the process and outcome of the negotiation. A copy of the questionnaire is provided in Appendix F. 39 CHAPTER 5 - RESULTS The SAS A N O V A procedure was used to analyze the data. This chapter presents the findings which are discussed in Chapter 6. For exposition convenience, the following abbreviations are used to refer to the experimental treatments in the study: C M C : Computer-mediated F T F : Face-to-face D I V : Divisional-incentive scheme M I X : Mixed-incentive scheme OPT : Optimal solutions available N O P . Optimal solutions not available These abbreviations are hyphenated to indicate a particular experimental condition. For example, C M C - D I V - O P T : Computer-mediated, Divisional-incentive scheme, Optimal Solutions available; FTF-MTX-OPT : Face-to-face, Mixed-incentive scheme, Optimal Solutions available; 5.1 Outcome Measures O f the thirty-two negotiation dyads, five were excluded from the analysis due to the following: (1) Two groups, one in FTF-DIV-NOP , the other in C M C - D I V - N O P , did not reach an agreement in the time allocated; 40 (2) 2 other groups employed a usual collusion strategy in which one subject made a $2M profit while the other lost $ 2 M to secure their chance of winning the cash bonus so that they could share it afterwards. Both groups were assigned to the mixed-incentive scheme; one of them belonged to C M C - M I X - N O P and the other was in the F T F - M I X - O P T condition. These cases were obvious outliers. (3) One subject, in C M C - M I X - N O P , purposely sent an offer which gave himself a loss and on seeing a $30,000 profit, the opponent accepted the offer immediately. Since the outcome would have been different i f the subject had acted rationally, the data collected from this dyad were not used in analysing this variable. The A N O V A results did not support the hypotheses discussed in Chapter 2 (see Tables 1, 3, 5, 7 and 9). There were no statistically significant differences between C M C and FTF , D I V and M I X , OPT and N O P in terms of joint outcome, difference between profits, deviation from the Nash solutions, willingness to disclose information, and time required to reach agreement. The following table summarizes the A N O V A results: Variables M E IN O P Joint outcome 0.57 (0.46) 0.26 (0.61) 1.07 (0.31) Difference between profits 0.34 (0.57) 0.30 (0.59) 0.01 (0.92) Deviation from Nash 0.27 (0.61) 0.33 (0.57) 0.01 (0.92) Willingness to share information 0.53 (0.47) 0.53 (0.47) 0.06 (0.81) Time to reach agreement 0.29 (0.59) 0.48 (0.50) 0.46 (0.50) The first numbers are the F values. Figures in parentheses are the p values. M E communication medium 41 IN incentive scheme OP optimal solutions The joint outcomes of all treatments were close to the optimal $30,822 (see Table 2). Interaction effects were not found. An A N O V A was also performed to find out if the type of information exchanged (i.e. the actual or the false curve) had any impact on the negotiation outcomes (data used were from the same twenty-seven negotiation dyads). In this study, information could be exchanged in six different ways: (1) no exchange of information (Nil + Nil) (2) one subject did not disclose information and the other sent the false information (Nil + False) (3) one subject did not disclose information and the other sent the true information (Nil + True) (4) both subjects exchanged the false information (False + False) (5) one subject sent the false information and the other sent the true information (False + True) (6) both subjects sent the true information (True + True) The following table records the type of information exchanged in this study: Type C M C F T F DIV MIX OPT NOP A L L Ni l + Ni l 1 1 1 0 1 1 2 Ni l + True 0 1 1 0 1 0 1 N i l + False 3 4 2 1 2 1 3 42 Type C M C F T F DIV MIX O P T N O P A L L True + False 4 3 3 6 3 6 10 False + False 5 4 5 6 7 4 10 True + True 0 1 1 0 1 0 1 Total 13 14 13 13 15 12 27 C M C computer-mediated F T F face-to-face D I V divisional-incentive scheme M I X mixed-incentive scheme OPT optimal solutions table available N O P optimal solutions table not available A L L number of cases out of the 27 groups The A N O V A results showed there was no statistically significant difference in terms of joint outcome but that the type of information exchanged produced statistically significant different results on the balance between individual profits and their distances from the Nash solution (F=2.93, p<0.04; F=3.04, p<0.03) (see Table 12). Because the optimal transfer quantity and the company's optimal profit attained under different demand and cost curves did not differ by large degree (see section 3.4(f)), whether subjects exchanged the true or the false information did not result in large differences in the joint profits. However, it was obvious from the means that distribution of the joint profit was less balanced and that individual profits were further away from the Nash solution when at least one subject failed to disclose his or her cost information (see Table 12). Negotiation outcomes were the most efficient when both subjects exchanged the true information. Joint profits were highest, differences between individual profits and their deviation from the Nash were the least (see Table 13). 43 5.2 Communication and Process Variables Thirty-one experimental groups were included in the analysis of this dependent variable. Data collected from the group in which one subject made a loss were discarded (see section 5.1(3)). (1) Number of offers No statistically significant differences on the total number of offers made were found due to differences in communication medium and the availability of optimal solutions. There was, however, a significant main effect for incentive scheme (F=4.47, p<0.05) which resulted in significant interaction effects with the other two independent variables (see Table 15). A divisional-incentive scheme encouraged more exchanges of offers (the mean was 11 compared to 8 in the mixed-incentive condition, see Table 16). The effect was more pronounced when subjects communicated face-to-face or when the Optimal Solutions Table was not available (the means were 14 and 13 respectively, see Table 16). (2) Number of remarks There was statistically significant difference between C M C and F T F in terms of the total number or remarks made during the negotiation (F=5.20, p<0.03) (see Table 17). Subjects in the F T F condition, on average exchanged 27 remarks, while those in the C M C condition exchanged 10 (see Table 18). In other words, more time was used in communicating with each other in F T F than in C M C . 44 (a) Task-oriented remarks (i) General There were statistically significant differences among treatments on the number of general remarks made during the negotiation (see Table 19). This category, however, was not very interesting as it did not reveal how efficient the communication was nor information on subjects' attitudes towards the negotiation. (ii) Requests for information There were statistically significant differences between C M C and F T F (F=3.62, p<0.07), and on the availability of an Optimal Solutions Table in terms of the number of remarks subjects made to seek information from their opponent (F=9.39, p<0.006). Interaction effects between these two independent variables were also found to be significant (F=4.07, p<0.06) (see Table 21). Table 22 shows that subjects negotiating F T F as well as when the Optimal Solutions Table was present asked for information more often. (iii) Discussion of offers There was a statistically significant differene between C M C and F T F in terms of the number of remarks subjects made on offers discussion (F=4.10, p<0.05) (see Table 23). The cell means showed that less discussion was carried out in the C M C condition regardless of the performance evaluation scheme that was imposed or whether the additional decision tool of optimal profits calculation was available (on average, this type of remarks were recorded 10 times in the F T F condition whereas 3 were recorded in the C M C condition, see Table 24). Although the A N O V A results did not indicate that a mixed-incentive scheme encouraged more offers discussion, the means showed that 45 remarks of this kind were made more in the F T F - M I X than in the F T F - D I V condition (the former averaged 19 while the latter averaged 6, see Table 24). (b) Task-irrelevant remarks There were no statistically significant difference among treatments in terms of the number of task-irrelevant remarks made during the negotiation. In fact, less than two of such remarks were recorded (see Table 26). Uninhibited remarks No negative remarks were recorded in any of the experimental sessions. 5.3 Perception Measures A l l data collected from the thirty-two experimental sessions were included in this analysis. (1) Communication Efficiency There was no statistically significant difference between C M C and F T F on this variable but different incentive schemes resulted in different perception on communication efficiency (F=3.20, p<0.09). There was also significant interaction effects between communication medium and incentive scheme (F=5.49, p<0.03) (see Table 28). The means showed that when F T F communication was allowed, different incentive schemes did not produce different results (means were 64 for F T F - D I V and 62 for F T F - M I X , see Table 29). However, in the C M C condition, communication was perceived to be more efficient under a mixed-incentive scheme (the mean was 69 compared to 53 in the divisional-incentive condition, see Table 29). 46 (2) Degree of Co-operation There was a statistically significant difference between a divisional-incentive scheme and a mixed-incentive scheme in terms of how co-operative subjects felt the negotiation was (F=11.82, p<0.002) (see Table 30). The means showed that more co-operative negotiations were found under a mixed-incentive scheme than under a divisional-incentive scheme (means were 168 and 144 respectively, see Table 31). The other two independent variables did not have significant effects on this variable. 5.4 Ranks Only twenty-seven groups were included in this analysis. Data collected from the five groups mentioned in 5.1 were discarded. Among these five groups, two did not reach an agreement, two colluded to share the bonus, and in the fifth group, one subject made the wrong move. The A N O V A analysis showed that there were statistically significant difference on both the ranks of the Marketing people (F=10.18, p<0.005, see Table 33) and that of the Production people (F=19.88, p<0.0003) (see Table 35). Ranks, in general, were higher in the mixed-incentive scheme condition. The mean ranks for the Marketing and Production people were 18 and 19 respectively under the mixed-incentive conditon, while in the divisional-incentive condition, both were 9 (see Tables 34 and 36). No other main effects or interaction effects were found to be significant. 5.5 Experimental Data Subjects' actions during the negotiations were recorded automatically onto log 47 files by the computer system. Statistical analysis was not carried out on these data because they were difficult to be interpreted accurately. One could determine the frequency of use of a particular decision support tool by measuring the number of times it was made the "key" panel; similarly, one could calculate the total time (in seconds) a subject spent on it. Nevertheless, none of these two measures can accurately reflect the subjects' action or motive at that point of time. A descriptive examination of the log files data was therefore carried out to provide a general outlook of how the different tools provided in the computer system were used. There was no difference in the way the tools were used across treatments. The Evaluate Offer section of the Offer panel (see Figure 6) and the 10x10 Profit Table (see Figure 5), which were designed to help subjects find an offer feasible to themselves, were the most frequently used. Subjects "consulted" these tools before they sent an offer and in all cases, the offer that was sent immediately after the "consultation" was one of the entries in the 10x10 Profit Table and/or had been evaluated in the Evaluate Offer section. The requirement that the other party's demand (or cost) information must be available before a joint graph can be drawn explains why the Joint Graphs (see Figure 7) were looked at mostly towards the end of the negotiation. The joint graphs which overlaid the demand and cost curves so that a region feasible to both parties was displayed were used the way they had been intended for. The fact that subjects looked at them while they were searching for a potential offer indicates that the graphs were used to locate a point where they could converge so that both parties would benefit. The Optimal Solutions Table (see Figure 8) was not used as frequently as the other support tools. One reason was that the information displayed in the table was static 48 - it was created when the system received the demand (or cost) curve from the other party and stayed unchanged. Secondly, the joint graphs also provided a means to locate the optimal transfer quantity. The lowest point on the Production's average cost curve is where Production incurs the least cost and is where the optimal transfer quantity (i.e. at which the company's profit is the maximum) lies. Whi le some subjects did not give much attention to the Optimal Solutions Table (it was consulted once or twice), some communicated the information it contained to the other party. The following statements were recorded in two F T F sessions: "I'm just looking at the optimal solutions. It looks that the company makes $30,000 i f we offer 250. Why don't we compromise?" "The quantity is not enough to keep my profits high. Y o u see the optimal solution for quantity is 252. So maybe we make that a consent and concentrate on the price we want to trade at." Individual demand and cost curves (see Figures 2 and 4) were not utilized as much as the other decision support tools mentioned above. This can be accounted for by the fact that the information these curves displayed was not as rich as that delivered by the other tools. Both the demand and average cost curves only indicate individual's position whereas information displayed in the other tools mentioned earlier consists of individual's profits, the opponent's as well as the corporate profits. Furthermore, once the other party's profit information was available, their roles were taken up by the joint graphs. Being able to communicate verbally did not alter the way the decision support tools were used in reaching an agreement. Subjects in the F T F condition also 49 "consulted" the 10x10 Profit Table and the Evaulate Offer section before placing any offers. Although there were verbal discussions on offers during the negotiation in the F T F condition, there were no cases in which an offer was explicitly exchanged verbally without being sent v ia the computer system. FTF groups also relied on the graphics and the calculation provided by the computer system because the information it delivered was rich and task-oriented. It was found that two FTF groups did not exchange remarks at all. Their silence during the negotiation may serve as an indicant that verbal communication was not needed because the system had already provided them with the information and tools that they required to reach an agreement. 50 CHAPTER 6 - DISCUSSION AND CONCLUSION 6.1 Summary of Results The results of the study did not support the hypotheses discussed in Chapter 2. There was no significant evidence that negotiation dyads who communicated solely via the computer exchanged less information than those who were allowed to converse face-to-face during the negotiation. Deceptions were seen in both conditions (see section 5.1) and that statistically, there was no evidence that computer-mediated communication (CMC) encouraged subjects to withhold truthful information from their opponents. Neither higher joint outcomes nor more equal distribution of the joint profit were found in negotiation dyads who negotiated face-to-face. A mixed-incentive scheme did not improve the negotiation outcome in terms of higher joint profits and lower differences between individual profits. The provision of the Optimal Solutions Table did not shorten the time to reach agreement nor did it result in higher joint profits to be achieved. The type of information exchanged did affect the negotiation outcome. Differences between individual profits and their distance from the Nash solution were significant larger when at least one subject failed to send his or her demand (or cost) curve to the other party. More remarks were made in face-to-face communication (FTF) but the communication medium did not encourage more exchanges of offers. Subjects, however, made more offers in the divisional-incentive scheme condition. More time was engaged on discussing offers in the F T F than in the C M C condition. F T F also encouraged subjects to explicitly seek information from their opponents. This effect was also found 51 to be significant in the presence of the Optimal Solutions Table. Subjects' perceptions on communication efficiency and degree of co-operation were not influenced by the mode of communication but by the incentive scheme imposed. A mixed-incentive scheme was found to enhance communication efficiency and induce co-operative behaviour. Higher ranks were also found to be associated with a mixed-incentive scheme. 6.2 Discussion and Implications As in the previous studies that compared F T F and C M C , it was found that more remarks were made in the former. Electronic messaging and verbal conversation are two different processes (Cherry 1978). C M C requires extra skill in typing and slows down the exchange of remarks because of the extra time needed to type and read messages. As a result, fewer remarks resulted. Moreover, since people normally do not immediately respond to electronic messages, they have more time to compose their reply and very often, something that is transmitted as a single message in the C M C will be broken into a number of statements and delivered to the other party at different times in the F T F condition. The inconsistencies of these results with those obtained from previous studies on computer-supported negotiation (Arunachalam 1991; Mahenthiran, Greenberg and Greenberg 1993) may be attributed to the difference in the experimental setting and the kind of computer support available. First of all, subjects in the other studies were engaged in a multiple-item negotiation in which they had to reach agreement on three or more items simultaneously. They were given profit schedules which consisted of five to 52 nine different options for each of these items and were not allow to disclose them to the other party/parties. Since most negotiations contain a mix of co-operative and competitive forces, especially those involving multiple items (Murnighan 1991), the profit schedules used in these studies were designed in a way such that some items are strictly distributive while some were integrative so that reciprocating tradeoffs could be made to obtain an integrative agreement that equalized the distribution of the joint outcome (Carvenale and Isen 1986; Arunachalam 1991)3. Secondly, in all these previous studies, the computer was used only as a means of communication; subjects either conversed face-to-face or by sending electronic messages. The computer support systems used were Group Communication Support Systems in which no decision support tools were provided (Arunachalam and Dilla 1992). Although the experimental task of this study was similar to that of the other studies in that subjects negotiated over more than one item and could tradeoff price for quantity, the experimental setting was different. Here, subjects faced a larger solution set; they were not given pre-defined profit schedules and their decisions were not confined to a finite number of combinations of alternatives. There were virtually no upper limits to the transfer price and transfer quantity a subject could offer. An offer with price 9,999,999 and quantity of 9,999,999 was no different from one with 100 as the price and 200 as the quantity and would be accepted by the computer system. In this study, a Negotiation Support System (NSS) was used to provide subjects with decision support tools and an electronic communication channel. As discussed in section 5.5, offers were made via the computer system the same way in the F T F condition as that in the C M C condition. In other words, all subjects communicated electronically. The verbal and visual access available in the F T F condition served only as an additional channel of communication. Please see Appendix G for an illustration of a profit schedule and how tradeoffs can be made to achive the highest joint outcome. 53 C M C eliminates visual, verbal, and social cues. Because it is a low richness medium, it has been claimed as not as effective as F T F communication in group decision making, especially when tasks are equivocal and when there is no correct solution to the problem. Longer time to reach agreement, lower joint outcome, and more unequal distribution of resources were associated with C M C . Although previous studies have repeatedly confirmed the superiority of F T F over C M C on group decision making, results of this study suggest that in circumstances where the computer supports both decision making and communication, the negotiation outcome may not necessary benefit from the extra channel of being able to see and converse with each other. L i m and Benbasat's (1993) model of an NSS is one that provides each negotiator with his/her own decision support systems which are interconnected with an electronic communication channel. Such electronic channels allow both text and graphical and modelling information to be transmitted, which serve as common elements negotiators can commonly refer to in their bargaining process. This study has shown that in negotiations where an NSS is used to provide negotiators with rich and task-oriented information via its electronic channel, verbal communication has no significant effects on the negotiation outcome. It can be seen from Table 2 that the means of the joint outcome were very close to the optimal (under the Nash solution, optimal company's profit = 30,822, see section 3.4(f)) regardless of the experimental treatments. This may be explained by the fact that the decision support tools provided in TransferPricing Game (TG) were used by the F T F subjects in the same manner as those who communicated via electronic messages (see section 5.5). Company's profit depends only on the quantity transferred. Therefore, to maximize the company's profit, one needs only to find out what the optimal transfer quantity is. The easiest way to locate the optimal quantity is by looking at the 54 Production's average cost curve. Since the lowest point on the curve is where Production incurs the least cost, nearly all negotiation dyads agreed to trade at a quantity that was around that lowest point and as a result, joint profits that were close to the optimal were achieved. 4 In addition, the presence of a historical record of all offers sent and received and the calculations of profits may also reduce the need of having verbal communication to complement this rich and task-oriented information already provided by the system. The above argument also provides a plausible explanation as to why a mixed-incentive scheme was not associated with higher joint outcome and smaller difference between individual's profits. The electronic channel enabled each subject to locate the best position for the company and provided them with complete knowledge on their opponent's profit (although the information could have been false). As there is no difference in the way subjects used the NSS and made decisions, it is reasonable to speculate that when negotiators are provided with tools that facilitate the attainment of an integrative solution, the incentive scheme has no impact on the negotiation outcome. Subjects assigned to the divisional-incentive scheme condition negotiated as efficient as those in the other treatment in terms of joint outcome and its distribution between opposing parties. As discussed in section 5.5, because of the ease to locate the optimal transfer quantity, the usefulness of the Optimal Solutions Table was reduced due to the overlapping of functionality. The availability of the Table therefore did not result in higher joint outcomes. Its presence, however, affected the type of remarks made in the negotiation. The requirement that the other party's profit information must be available 4 It should be noted that although different demand and cost curves produce different optimal transfer quantity, as shown in section 3.4(f), in this study, the optimal transfer quantity under the actual or the false curves did not differ by large degree. 55 for the formation of the table may well explain why more requests for the opponent's demand/cost information were found in the OPT condition. Negotiators see a common goal in a mixed-incentive scheme (MIX) which explains why a higher degree of co-operation was perceived in this condition than in the divisional-incentive scheme (DIV) condition. Such a decrease in competitive feelings against each other increases personal contact and since negotiators have a common target to shoot for (to maximize the joint profit), more communication is expected. This common target, on the other hand, helps reduce the constraints negotiators may have in their communication with each other and explains the interaction effects found between communication medium and incentive scheme. Subjects in the C M C - M I X condition, on average, gave higher scores for communication efficiency than those in the C M C - D I V condition (69 compared to 53, see Table 29). A mixed-incentive scheme, therefore, enhances perceived communication efficiency, especially in C M C when a physical barrier blocks out all non-verbal and social cues. The association between high ranks and a mixed-incentive scheme has two implications. Firstly, a mixed-incentive scheme brings better divisional performance. Secondly, the higher ranks indicate that the co-operative nature of the reward structure encourages negotiators to reach an agreement that gives them mutual benefits. The following statements were recorded during the negotiation. "I'm given the option here of showing you an actual or an altered cost curve. In real practice, I would use just the actual and forget about the altered because priority is to maximize the profit for the company." "... between your offer number 3 and number 4 , we can do something for the 56 mutual profit. So let's try." These statements, both made in the F T F - M I X condition, illustrate that more co-operative behaviour is found under a mixed-incentive scheme. The fact that these were remarks made in the F T F condition at the same time sheds light on the possibility of inducing co-operative behaviour by allowing negotiators to converse face-to-face. Subjects exchanged fewer remarks under the C M C condition and since remarks which reveal one's intention and thoughts as those cited above were not found in the C M C condition, the presence of a physcial barrier, therefore, had created a less open atmosphere. Subjects were less inclined to state explicitly their negotiation strategy or that they would endeavour to find an agreement that was beneficial to both. The more frequent remarks on offers discussion and requests for information also revealed the co-operative manner observed in the F T F condition. Positional commitments and statements about one's profit reveal one's resistance point to one's opponent so that alternatives that are beneficial to both can be found. Integrative bargaining requires a maximum flow of accurate information. It involves an attempt to understand and take into account the other person's utility (Walton and McKersie 1965). Therefore, taking an initiative to seek information from the other party is one other indicator of co-operative behaviour. Although results of this study did not support that F T F negotiations were more co-operative and efficient, merits from the previous studies together with the above observations suggest that further investigation with a different experimental design needs to be conducted. 6.3 Extensions One important independent variable that was not included in this study is Group History. In their meta-analysis of Group Support Systems (GSS), Benbasat and L im 57 (1993) found no difference in the decision quality between ad hoc and established groups and that GSS has a smaller effect on equality of participation on the latter because of the presence of a well-established social order and a lower degree of anonymity. Their findings, however, cannot be directly extended to this study because although negotiation is also a group decision making process, it is different from idea generation and other group activities in the sense that only two parties are involved and each has its own objective which may or may not be congruent to each other. Equality of participation is hard to measure on a one-to-one negotiation unless it refers to whether one person has more bargaining power and dominates the negotiation. However, if each party is made up of more than one person, then three separate group processes are involved: idea generation and alternative selection in each party plus the negotiation itself. Results and findings of previous group studies are then applicable. The question here is not to find out whether established groups5 will arrive at better negotiation outcomes than ad hoc groups. One can expect that established negotiation dyads who have maintained a long-term relationship may have knowledge of each other's negotiation strategies and objectives. It is therefore not surprising to observe different outcomes and processes in these groups. Group history on ad-hoc groups, however, is a more interesting independent variable to manipulate. This study can be extended to find out whether there are interaction effects between communication medium and the number of negotiation sessions on ad-hoc negotiation dyads. Group history was manipulated in previous transfer pricing negotiation studies in terms of multiple negotiation sessions (Chalos and Haka 1990; Arunachalam 1991). Results showed that joint outcomes increased over periods and at the same time, Here, groups refer to negotiation dyads. 58 difference between individual profits decreased. Firm profits were significantly larger in the final multi-period session than in the single-period negotiation session. However, interaction effects with the mode of communication were not found to be significant (Arunachalam 1991). Although results of this study showed that when subjects were engaged in a single negotiation session, neither did the added F T F communication improve the outcome nor did it induce feelings of communication efficiency and co-operation, excerpts from the transcriptions cited earlier suggest that one can expect that when subjects are engaged in a multiple negotiation session exercise, the mode of communication will make a difference on the negotiation outcome and process. The building of trust is a long term process. In this study, subjects had no knowledge on whom they would negotiate with prior to the experiment and were not introduced to each other. Trust, therefore, could hardly be built up given the short period of time even in the F T F condition when one had direct personal contact with one's counterpart. The lack of trust and the reliance on the decision support tools provided in the computer system, help explain the rejection of the hypotheses relating to communication medium. At least two questions need to be answered: (1) Wil l the negotiation outcome improve with the number of sessions held? (2) If so, will there be differences between C M C and F T F (in the presence of an NSS) across periods in terms of the negotiation outcome and the subjects' perceptions on communication efficiency and co-operation? "Reciprocity" explains question 1 (Chalos and Haka 1990). In a single-period negotiation, negotiators will be tempted to capture as much of the available payoff as possible, regardless of the impact on the negotiating partner. This type of opportunistic 59 behaviour, however, may lead to an unprofitable strategy in a multi-period setting since the disadvantaged negotiating partner may take on a revenge in later negotiations. This type of strategy is known as "Tic-for-Tat" - always follow the other player's choice (Murnighan 1991). It immediately retaliates against every non-cooperative choice and to avoid punishment, less rivalistic behaviours will be observed in multi-period sessions and hence, more integrative outcomes will be obtained in both C M C and F T F conditions. Previous studies have concluded that F T F is more efficient than C M C in that it delivers paralinguistic cues and social presence. The instantaneous and rich information flow in F T F should promote the building of trust and mutual understanding between negotiators which are essential to co-operative and integrative negotiations. One can expect that in a multi-period setting, more integrative outcomes will be found in F T F than in C M C and that communication will be more efficient. Nevertheless, such speculations are yet to be proven. 60 Bibl iography Abdel-khalik, A . Rashad, and Edward J . Lusk. "Transfer Pricing - A Synthesis". The Accounting Review 49. 1 (Jan. 1974): 8-23. Arunachalam, Vairavan. "Decision Aiding in Multi-Party Transfer Pricing Negotiation: The effects of Computer-mediated Communication and Structured Interaction". Ph.D. diss., University of Illinois at Urbana-Champaign, 1991. Arunachalam, Vairam, and Wi l l iam Di l la . "Computer-Mediated Communication and Structured Interaction in Transfer Pricing Negotiation". Journal of Information Systems. Fa l l 1992, in press. Bartos, Otomar J . Process and Outcome of Negotiations. New York: Columbia University Press, 1974. Bazerman, Max H . Judgement in Managerial Decision Making. 2nd ed. N.Y. : John Wi ley and Sons, 1993. Benbasat, Izak, Francis J . L i m , and V . Srinivasan Rao. " A Framework for Communication Support in Group Work with Special Reference to Negotiation Systems". Group Decision and Negotiation. 1995, forthcoming. Benbasat, Izak, and Lai-Huat L im . "The Effects of Group, Task, Context, and Technology Variables on the Usefulness of Group Support Systems - A Meta-Analysis of Experimental Studies". Small Group Research 24, 4 (Nov. 1993): 430-462. Carnevale, Peter J . D., and Al ice M . Isen. "The Influence of Positive Affect and Visual Access on the Discovery of Integrative Solutions in Bilateral Negotiation". Organizational Behavior and Human Decision Processes 37, 1 (Feb. 1986): 1-13. Chalos, Peter, and Susan Haka. "Transfer Pricing Under Bilateral Bargaining". The Accounting Review 65, 3 (July 1990): 624-641. Cherry, Col in. On Human Communication: A Review, A Survey, and A Criticism. 3rd ed. Cambridge: M I T Press, 1978. Daly, Bonita L. "The Influence of Face-To-Face versus Computer-mediated Communication Channels on Collective Induction". Accounting, Management and Information Technologies 3. 1 (1993): 1-22. Delaney, Michael M . , Abbas Foroughi, and Wi l l i am C. Perkins. " A n Empirical Study of the Efficacy of a Computerized Negotiation Support System (NSS)". Decision Support Systems. 1994, forthcoming. 61 Dennis, A lan R., and R. Brent Gallupe. A History of Group Support Systems Empirical Research: Lessons Learned and Future Directions. Edited by Leonard M . Jessup, and Joseph S. Valacicb. Group Support Systems - New Perspectives. N.Y.: Macmil lan, 1993. DeSantics, Gerardine, and R. Brent Gallupe. " A Foundation for the Study of Group Decision Support Systems". Management Science 33, 5 (May 1987): 589-609. Eisenson, John, J . Jeffery Auer, and John V . Irwin. The Psychology of Communication. N.Y.: Appleton-Century-Crofts, 1963. Harnett, Donald, L., and L. L. Cummings. Bargaining Behavior: A n International Study. Houston, Tex.: Dame Publications Inc., 1980. Hi l tz , Starr Roxanne, and Kenneth Johnson. "User Satisfaction with Computer-Mediated Communication Systems". Management Science 36, 6 (June 1990): 739-764. Hirshleifer, Jack. "On the Economics of Transfer Pricing". Journal of Business 28 (July 1956): 172-184. Huber, Oswald. Information-Processing Operators in Decision Making. Edited by H . Montgomery and O. Svenson. Process and Structure in Human Decision Making. Chichester: John Wiley and Sons, 1989. Jarke, Matthias, M . Tawfik Jelassi, and Me lv in F. Shakun. " M E D I T A T O R : Towards a negotiation support system". European Journal of Operational Research 31, 3 (1987): 314-334. Jelassi, M . Tawfik, and Abbas Foroughi. "Negotiation Support Systems: A n Overview of Design Issues and Existing Software". Decision Support Systems 5, 2 (June 1989): 167-181. Kersten, Gregory E. " N E G O - Group Decision Support System". Information and Management 8, 5 (May 1985): 237-246. Kersten, Gregory E. "On Two Roles Decision Support Systems can play in Negotiations". Information Processing and Management 23. 6 (1987): 605-614. Kersten, Gregory E. " A Procedure for Negotiating Efficient and Non-Efficient Compromises." Decision Support Systems 4, 2 (June 1988): 167-177. Kiesler, Sara, and Lee Sproull. "Group Decision Making and Communication Technology". Organizational Behavior and Human Decision Processes 52, 1 (June 1992): 96-123. Kniveton, Bromley. The Psychology of Bargaining. Aldershot: Avebury, 1989. 62 L i m , Lai-Huat, and Izak Benbasat. " A Theoretical Perspective of Negotiation Support Systems". Journal of Management Information Systems 9, 3 (Winter 1993): 27-44. Mahenthiran, Sakthi, Penelope Sue Greenberg, and Ralph H . Greenberg. "The Impact of Computer-mediated Communication on the Process and Outcomes of Negotiated Transfer Pricing". Accounting, Management and Information Technologies 3, 4 (1993): 229-248. Montgomery, Henry. From Cognition to Action: The Search for Dominance in Decision Making. Edited by H . Montgomery and O. Svenson. Process and Structure in Human Decision Making. Chichester: John Wiley and Sons, 1989. McGrath, Joseph E. Groups: Interaction and Performance. Englewood Cliffs, N.J. : Prentice-Hall, 1984. McGrath, Joseph E. , and Andrea B. Hollingshead. Putting the "Group" Back in Group Support Systems: Some Theoretical Issues About Dynamic Processes in Groups with Technological Enhancements. Edited by Leonard M . Jessup, and Joseph S. Valacicb. Group Support Systems - New Perspectives. N.Y.: Macmil lan, 1993. Murnighan, J . Keith. The Dynamics of Bargaining. N.J. : Prentice Hal l , 1991. Nash Jr., John F. "The Bargaining Problem". Econometrica. 18 (1950): 155-162. Nunamaker Jr., J . F., A lan R. Dennis, Joseph S. Valacich, and Douglas R. Vogel. "Information Technology for Negotiating Groups: Generating Options for Mutual Gain". Management Science 37. 10 (Oct. 1991): 1325-1346. Olaniran, Bolanle Abodunrin. "Computer-mediated Communication in Small Group Decision Stages". Ph.D. diss., University of Oklahoma, 1991. Poole, M . Scott, and Michele H . Jackson. Communication Theory and Group Support Systems. Edited by Leonard M . Jessup, and Joseph S. Valacicb. Group Support Systems - New Perspectives. N.Y.: Macmil lan, 1993. Pruitt, Dean G. Negotiation Behavior. N.Y.: Academic Press, 1981. Siegel, Jane, Vitaly Dubrovsky, Sara Kiesler, and Timothy W. McGuire. "Group Processes in Computer-mediated Communication". Organizational Behavior and Human Decision Processes 37. 2 (Apr. 1986): 157-187. Siegel, Sidney, and Lawrence E . Fouraker. Bargaining and Group Decision Making. N.Y.: McGraw-Hi l l , 1960. 63 Straus, Susan Gail . "Does the medium matter: an investigation of process performance and affect in computer-mediated and face-to-face groups (computer conferences, group dynamics)". Ph.D. diss., University of Illinois at Urbana-Champaign, 1992. Sproull, Lee, and Sara Kiesler. "Reducing Social Context Cues: Electronic M a i l in organizational communication". Management Science 32. 11 (Nov. 1986): 1492-1512. Svenson, Ola. "Illustrating verbal protocol analysis: Individual decisions and dialogues preceding a joint decision". Edited by H . Montgomery and O. Svenson. Process and Structure in Human Decision Making. Chichester: John Wi ley and Sons, 1989. Valacich, Joseph S., Joey F. George, J . F. Nunamaker, Jr., and Douglas R. Vogel. "Physical Proximity Effects on Computer-mediated Group Idea Generation" Small Group Research 25, 1 (Feb. 1994): 83-104. von Winterfeldt, Detlof, and Ward Edwards. Decision Analysis and Behavioral Research. Cambridge: Cambridge University Press, 1986. Walton, Richard E. , and Robert B. McKersie. A Behavioral Theory of Labor Negotiations. N.Y.: M c G r a w H i l l , 1965. Weisband, Suzanne P. "Group Discussion and First Advocacy Effects in Computer-mediated and Face-to-Face Decision Making Groups". Organizational Behavior and Human Decision Processes 53, 3 (Dec. 1992): 352-380. Westhoff, Kar l . Expectations and Decisions. Edited by A . Upmeyer. Attitudes and Behavioral Decisions. N.Y.: Springer-Verlag, 1989. Zack, Michael H . "Interactivity and Communication Mode Choice in Ongoing Management Groups". Information Systems Research 4. 3 (Sept. 1993): 207-239. 64 Table 1 Outcome measures - Joint outcome ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value Pr >F M E 1 2292701.76 2292701.76 0.57 0.4606 IN 1 1068880.89 1068880.89 0.26 0.6130 M E * I N 1 3409759.19 3409759.19 0.84 0.3699 OP 1 4327683.27 4327683.27 1.07 0.3138 M E * O P 1 1977763.63 1977763.63 0.49 0.4927 IN*OP 1 418330.97 418330.97 0.10 0.7512 M E * IN* OP 1 4110799.72 4110799.72 1.02 0.3259 M E communication medium IN incentive scheme OP optimal solutions 65 Table 2 Outcome measures - Joint outcome Cel l means N Mean SD C M C 13 29739.1538 2612.67146 F T F 14 30322.3571 885.76309 D I V 14 30233.2857 989.67698 M I X 13 29835.0769 2591.70071 C M C - D I V 7 30269.1429 1044.54734 C M C - M I X 6 29120.8333 3771.10092 F T F - D I V 7 30197.4286 1013.94541 F T F - M I X 7 30447.2857 797.12812 OPT 15 29683.4667 2475.07457 N O P 12 30489.1667 626.18743 C M C - O P T 8 29192.2500 3277.30079 C M C - N O P 5 30614.2000 355.79587 FTF -OPT 7 30244.8571 1036.22302 F T F - N O P 7 30399.8571 782.39535 D IV -OPT 8 29965.6250 1246.17093 D I V - N O P 6 30590.1667 324.05889 M I X - O P T 7 29361.0000 3500.67908 M I X - N O P 6 30388.1667 856.24002 C M C - D I V - O P T 4 30109.7500 1406.50000 C M C - D I V - N O P 3 30481.6667 432.79595 C M C - M I X - O P T 4 28274.750 4564.92887 C M C - M I X - N O P 2 30813.0000 0.00000 FTF -D IV -OPT 4 29821.5000 1260.91118 F T F - D I V - N O P 3 30698.6667 199.76570 FTF-MTX-OPT 3 30809.3333 4.04145 F T F - M L X - N O P 4 30175.7500 1020.50392 Table 3 Outcome measures - Difference between profits ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 16399875.8 16399875.8 0.34 0.5692 IN 1 14545385.1 14545385.1 0.30 0.5918 ME*IN 1 4077181.3 4077181.3 0.08 0.7759 OP 1 532732.4 532732.4 0.01 0.9179 ME*OP 1 6249969.5 6249969.5 0.13 0.7246 IN*OP 1 67697688.7 67697688.7 1.38 0.2538 ME*IN*OP 1 8265101.1 8265101.1 0.17 0.6855 M E communication medium IN incentive scheme OP optimal solutions 67 Table 4 Outcome measures - Difference between profits Cel l means N Mean SD C M C 13 10022.0769 6840.47743 F T F 14 8462.2857 6003.67988 D I V 14 9920.5714 6709.48043 M I X 13 8451.6154 6101.30891 C M C - D I V 7 11096.4286 8147.94632 C M C - M I X 6 8768.6667 5397.50975 F T F - D I V 7 8744.7143 5284.11966 FTF -MTX 7 8179.8571 7070.19555 OPT 15 9338.9333 6712.17872 N O P 12 9056.2500 6143.73408 C M C - O P T 8 9646.8750 6904.30289 C M C - N O P 5 10622.4000 7498.19374 FTF-OPT 7 8987.0000 7017.06719 F T F - N O P 7 7937.5714 5311.68984 D IV -OPT 8 8692.7500 6773.73496 D I V - N O P 6 11557.6667 6868.73225 M I X - O P T 7 10077.4286 7099.34689 MTX-NOP 6 6554.8333 4565.47292 C M C - D I V - O P T 4 10141.0000 8789.81433 C M C - D I V - N O P 3 12370.3333 8889.08861 C M C - M I X - O P T 4 9152.7500 5772.05927 C M C - M I X - N O P 2 8000.5000 6629.12607 FTF -D IV -OPT 4 7244.5000 4920.11060 FTF -D IV -NOP 3 10745.0000 6078.77693 FTF-MTX-OPT 3 11310.3333 9860.88405 FTF-MTX-NOP 4 5832.0000 4242.47070 68 Table 5 Outcome measures - Deviation from the Nash solution A N O V A (medium x incentive x optimal) Source DF Anova SS Mean Square F Value Pr > F M E 1 12943597.9 12943597.9 0.27 0.6061 IN 1 15335578.8 15335578.8 0.33 0.5748 M E * I N 1 1887878.6 1887878.6 0.04 0.8434 OP 1 448300.9 448300.9 0.01 0.9233 M E * OP 1 6562505.1 6562505.1 0.14 0.7130 IN*OP 1 65862344.9 65862344.9 1.40 0.2515 M E * I N * O P 1 6283002.8 6283002.8 0.13 0.7189 M E communication medium IN incentive scheme OP optimal solutions 69 Table 6 Outcome measures - Deviation from the Nash solution Cel l means N Mean SD C M C 13 9923.0000 6678.78809 F T F 14 8537.2857 5919.32653 D I V 14 9930.7143 6574.65728 M I X 13 8422.3846 5961.28167 C M C - D I V 7 10912.5714 8039.46449 C M C - M I X 6 8768.5000 5150.23505 F T F - D I V 7 8948.8571 5174.37589 F T F - M T X 7 8125.7143 6981.91802 OPT 15 9319.7333 6426.42718 N O P 12 9060.4167 6216.42865 C M C - O P T 8 9539.5000 6735.62295 C M C - N O P 5 10536.6000 7325.21510 FTF -OPT 7 9068.5714 6579.98553 F T F - N O P 7 8006.0000 5653.47362 D IV -OPT 8 8710.5000 6432.58944 D I V - N O P 6 11557.6667 6992.83724 M I X - O P T 7 10016.0000 6857.78448 M I X - N O P 6 6563.1667 4598.91037 C M C - D I V - O P T 4 9926.5000 8598.38502 C M C - D I V - N O P 3 12227.3333 8857.93759 C M C - M I X - O P T 4 9152.5000 5615.02416 C M C - M I X - N O P 2 8000.5000 6022.42846 FTF -D IV -OPT 4 7494.5000 4321.32406 F T F - D I V - N O P 3 10888.0000 6514.67674 FTF-MTX-OPT 3 11167.3333 9503.43424 F T F - M I X - N O P 4 5844.5000 4592.83707 70 Table 7 Outcome measures - Willingness to share information A N O V A (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 0.85978836 0.85978836 0.53 0.4783 I N 1 0.85978836 0.85978836 0.53 0.4738 M E * I N 1 0.09259259 0.09259259 0.06 0.8130 OP 1 0.09074074 0.09074074 0.06 0.8149 M E * O P 1 0.00000000 0.00000000 0.00 1.0000 IN *OP 1 1.52830688 1.52830688 0.95 0.3421 M E * I N * O P 1 0.07883598 0.07883598 0.05 0.8272 M E communication medium I N incentive scheme OP optimal solutions 71 Table 8 Outcome measures - Willingness to share information Cel l means N Mean SD C M C 13 -1.00000000 0.91287093 F T F 14 -0.64285714 1.33630621 D I V 14 -0.64285714 1.27744594 M I X 13 -1.00000000 1.00000000 C M C - D I V 7 -0.85714286 0.89973541 C M C - M I X 6 -1.16666667 0.98319208 F T F - D I V 7 -0.42857143 1.61834719 F T F - M I X 7 -0.85714286 1.06904497 OPT 15 -0.86666667 1.30201309 N O P 12 -0.75000000 0.96530730 C M C - O P T 8 -1.00000000 0.92582010 C M C - N O P 5 -1.00000000 1.00000000 FTF -OPT 7 -0.71428571 1.70433621 F T F - N O P 7 -0.57142857 0.97590007 D IV -OPT 8 -0.50000000 1.51185789 M I X - O P T 7 -1.28571429 0.95118973 D I V - N O P 6 -0.83333333 0.98319208 M I X - N O P 6 -0.66666667 1.03279556 C M C - D I V - O P T 4 -0.75000000 0.95742711 C M C - D I V - N O P 3 -1.00000000 1.00000000 C M C - M I X - O P T 4 -1.25000000 0.95742711 C M C - M I X - N O P 2 -1.00000000 1.41421356 FTF -D IV -OPT 4 -0.25000000 2.06155281 F T F - D I V - N O P 3 -0.66666667 1.15470054 F T F - M I X - O P T 3 -1.33333333 1.15470054 F T F - M I X - N O P 4 -0.50000000 1.00000000 72 Table 9 Outcome measures - Time to reach agreement A N O V A (medium x incentive x optimal) Source DF Anova SS Mean Square F Value Pr >F M E 1 19.8095238 19.8095238 0.29 0.5946 OP 1 32.2666667 32.2666667 0.48 0.4980 M E * OP 1 10.4297619 10.4297619 0.15 0.6989 I N 1 31.4304029 31.4304029 0.46 0.5036 M E * IN 1 18.0934066 18.0934066 0.27 0.6109 O P * I N 1 4.3743590 4.3743590 0.06 0.8019 M E * O P * IN 1 67.6792125 67.6792125 1.00 0.3296 M E communication medium LN incentive scheme OP optimal solutions 73 Table 10 Outcome measures - Time to reach agreement Cel l means N Mean SD C M C 13 29.0000000 6.3245553 F T F 14 30.7142857 8.6329296 D I V 14 30.9285714 6.0187436 M I X 13 28.7692308 8.9736080 C M C - D I V 7 29.2857143 5.9361684 C M C - M I X 6 28.6666667 7.3120904 F T F - D I V 7 32.5714286 6.0788470 F T F - M I X 7 28.8571429 10.7924136 OPT 15 30.8666667 6.8646784 N O P 12 28.6666667 8.4027413 C M C - O P T 8 29.6250000 7.1900228 C M C - N O P 5 28.0000000 5.2440442 FTF -OPT 7 32.2857143 6.7259271 F T F - N O P 7 29.1428571 10.5107655 D IV -OPT 8 32.2500000 7.2456884 D I V - N O P 6 29.1666667 3.7638633 M I X - O P T 7 29.2857143 6.5755681 M I X - N O P 6 28.1666667 11.8560814 C M C - D I V - O P T 4 29.0000000 7.3484692 C M C - D I V - N O P 3 29.6666667 4.9328829 C M C - M I X - O P T 4 30.2500000 8.0983537 C M C - M I X - N O P 2 25.5000000 6.3639610 FTF-D IV -OPT 4 35.5000000 6.3508530 FTF -D IV -NOP 3 28.6666667 3.2145503 F T F - M I X - O P T 3 28.0000000 5.1961524 F T F - M I X - N O P 4 29.5000000 14.6173413 74 Table 11 Outcome measures MANOVA (medium x incentive x optimal) MANOVA statistics: Wilks' Lambda, Pillai's Trace, Hotelling-Lawley Trace, Roy's Greatest Root Degress of freedom: (5, 15) Source F Pr >F M E 0.3329 0.8852 IN 0.7054 0.6283 ME*IN 0.3557 0.8707 OP 0.2729 0.9209 ME*OP 0.1856 0.9636 IN*OP 0.2780 0.9180 M E * IN* OP 0.6452 0.6694 M E communication medium IN incentive scheme OP optimal solutions 75 Table 12 Information and Outcome measures ANOVA Joint outcome Source DF Anova SS Mean Square F Value Pr >F INFO 5 22038337.3 4407667.5 1.28 0.3100 Difference between profits Source DF Anova SS Mean Square F Value P r > F INFO 5 426635848 85327170 2.93 0.0367 Deviation from the Nash solution Source DF Anova SS Mean Square F Value P r > F INFO 5 421840804 84368161 3.04 0.0320 INFO types of information 76 Table 13 Information and Outcome measures Cell Means Joint Outcome N Mean SD Nil + Nil 2 29407.5000 1990.50559 Nil + True 1 29992.0000 -Nil + False 3 27688.0000 5412.65877 True + False 9 30682.8889 213.73257 Fales + False 11 30208.4545 972.49796 True + True 1 30820.0000 -Difference between profits N Mean SD Nil + Nil 2 10395.0000 9043.89573 Nil + True 1 13883.0000 -Nil + False 3 19604.6667 4964.45700 True + False 9 8189.5556 4339.63014 Fales + False 11 7040.9091 5734.72476 True + True 1 5116.0000 -Deviation from the Nash solution N Mean SD Nil + Nil 2 10824.0000 9043.89573 Nil + True 1 13454.0000 -Nil + False 3 19461.6667 4475.40181 True + False 9 8242.7778 4381.89468 Fales + False 11 6845.9091 5535.54367 True + True 1 5545.0000 -Nil + Nil no exchange of information Nil + True one subject sent the True information while the other did not disclose information Nil + False one subject sent the False information while the other did not disclose information True + True one subject sent the True information and the other sent the False information False + False both subjects sent the False information True + True both subjects sent the True information 77 Table 14 Information and Outcome measures MANOVA Statistic Value F D F Pr > F Wilks' Lambda 0.28890016 2.0013 15, 52.85 0.0329 Pillai's Trace 0.89033246 1.7725 15, 63 0.0593 Hotelling-Lawley Trace 1.85619478 2.1862 15, 53 0.0189 Roy's Greatest Root 1.45890281 6.1274 5,21 0.0012 NOTE: F Statistic for Roy's Greatest Root is an upper bound. 78 Table 15 Communication/Process - Number of offers ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value Pr >F M E 1 49.686022 49.686022 2.62 0.1190 IN 1 84.736022 84.736022 4.47 0.0455 ME*IN 1 127.515169 127.515169 6.73 0.0162 OP 1 13.936022 13.936022 0.74 0.3999 ME*OP 1 0.458026 0.458026 0.02 0.8778 IN*OP 1 100.318740 100.318740 5.30 0.0308 ME*IN*OP 1 0.000000 0.000000 0.00 1.0000 M E communication medium IN incentive scheme OP optimal solutions 79 Table 16 Communication/Process - Number of offers Cel l means N Mean SD C M C 15 4.27395211 8.4666667 F T F 16 5.80804040 11.0000000 D I V 16 11.3750000 5.74891294 M I X 15 8.0666667 4.06143301 C M C - D I V 8 8.1250000 3.27053949 C M C - M I X 7 8.8571429 5.45980987 F T F - D I V 8 14.6250000 5.99851172 F T F - M T X 8 7.3750000 2.50356888 OPT 16 9.1250000 4.24067605 N O P 15 10.4666667 6.13964479 C M C - O P T 8 7.6250000 5.12521777 C M C - N O P 7 9.4285714 3.15473944 FTF -OPT 8 10.6250000 2.66926956 F T F - N O P 8 11.3750000 8.05228450 D IV -OPT 8 9.0000000 4.00000000 D I V - N O P 8 13.7500000 6.47522752 M I X - O P T 8 9.2500000 4.74341649 MTX-NOP 7 6.7142857 2.87020822 C M C - D I V - O P T 4 5.5000000 1.29099445 C M C - D I V - N O P 4 10.7500000 2.21735578 C M C - M I X - O P T 4 9.7500000 6.89806736 C M C - M I X - N O P 3 7.6666667 3.78593890 FTF -D IV -OPT 4 12.5000000 1.73205081 F T F - D I V - N O P 4 16.7500000 8.30160627 FTF-MTX-OPT 4 8.7500000 2.06155281 F T F - M I X - N O P 4 6.0000000 2.30940108 80 Table 17 Communication/Process - Number of remarks ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 2064.32272 2064.32272 5.20 0.0322 IN 1 212.71022 212.71022 0.54 0.4715 ME*IN 1 120.35705 120.35705 0.30 0.5872 OP 1 1200.82272 1200.82272 3.03 0.0953 M E * OP 1 1426.67312 1426.67312 3.59 0.0706 IN*OP 1 114.48205 114.48205 0.29 0.5964 ME* IN* OP 1 343.15902 343.15902 0.86 0.3621 M E communication medium IN incentive scheme OP optimal solutions 81 Table 18 Communication - Number of remarks Cel l means N Mean SD C M C 15 10.7333333 8.7052256 F T F 16 27.0625000 27.6730163 D I V 16 16.6250000 17.0640949 M I X 15 21.8666667 26.7684961 C M C - D I V 8 10.7500000 8.4978989 C M C - M I X 7 10.7142857 9.6214047 F T F - D I V 8 22.5000000 21.7452786 FTF -MTX 8 31.6250000 33.4746706 OPT 16 25.1875000 27.2793420 N O P 15 12.7333333 12.6743085 C M C - O P T 8 10.5000000 9.6658456 C M C - N O P 7 11.0000000 8.2259751 FTF -OPT 8 39.8750000 31.7509842 F T F - N O P 8 14.2500000 16.0512572 D IV -OPT 8 24.8750000 20.4271353 D I V - N O P 8 8.3750000 7.1501748 M I X - O P T 8 25.5000000 34.3095155 M I X - N O P 7 17.7142857 16.1525466 C M C - D I V - O P T 4 9.2500000 10.6262254 C M C - D I V - N O P 4 12.2500000 7.0415434 C M C - M I X - O P T 4 11.7500000 10.0457288 C M C - M I X - N O P 3 9.3333333 11.0151411 FTF -D IV -OPT 4 40.5000000 14.4798711 FTF -D IV -NOP 4 4.5000000 5.4467115 FTF-MTX-OPT 4 39.2500000 46.2772442 F T F - M I X - N O P 4 24.0000000 17.8325545 82 Table 19 Remarks - General ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 579.82917 579.82917 3.18 0.0877 IN 1 10.46250 10.46250 0.06 0.8128 ME*IN 1 0.11905 0.11905 0.00 0.9798 OP 1 1141.31667 1141.31667 6.26 0.0199 ME*OP 1 1388.62202 1388.62202 7.62 0.0112 IN*OP 1 7.11369 7.11369 0.04 0.8451 ME*rN*OP 1 9.28690 9.28690 0.05 0.8234 M E communication medium IN incentive scheme OP optimal solutions 83 Table 20 Remarks - General Cel l means N Mean SD C M C 15 5.5333333 4.9694304 F T F 16 14.1875000 20.6630709 D I V 16 9.4375000 11.4482531 M I X 15 10.6000000 19.5550505 C M C - D I V 8 5.5000000 4.1403934 C M C - M I X 7 5.5714286 6.1334369 F T F - D I V 8 13.3750000 15.1085359 F T F - M I X 8 15.0000000 26.1752337 OPT 16 15.8750000 19.6362760 N O P 15 3.7333333 5.3780860 C M C - O P T 8 5.0000000 4.7809144 C M C - N O P 7 6.1428571 5.4902511 FTF -OPT 8 26.7500000 23.0883397 F T F - N O P 8 1.6250000 4.5961941 D IV -OPT 8 16.1250000 12.7552510 D I V - N O P 8 2.7500000 3.9910614 M I X - O P T 8 15.6250000 25.7567606 M I X - N O P 7 4.8571429 6.7928534 C M C - D I V - O P T 4 5.5000000 4.7958315 C M C - D I V - N O P 4 5.5000000 4.1231056 C M C - M I X - O P T 4 4.5000000 5.4467115 C M C - M I X - N O P 3 7.0000000 7.9372539 FTF-D IV -OPT 4 26.7500000 7.4554231 FTF -D IV -NOP 4 0.0000000 0.0000000 FTF-MTX-OPT 4 26.7500000 34.4710023 F T F - M I X - N O P 4 3.2500000 6.5000000 84 Table 21 Remarks - Requests for information ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 16.9388441 16.9388441 3.62 0.0697 IN 1 7.3596774 7.3596774 1.57 0.2225 ME*IN 1 0.9468702 0.9468702 0.20 0.6571 OP 1 43.9763441 43.9763441 9.39 0.0055 ME*OP 1 19.0623464 19.0623464 4.07 0.0554 IN*OP 1 4.8915131 4.8915131 1.04 0.3173 ME*IN*OP 1 1.8674155 1.8674155 0.40 0.5339 M E communication medium IN incentive scheme OP optimal solutions 85 Table 22 Remarks - Requests for information Cel l means N Mean SD C M C 15 1.33333333 I.44749373 F T F 16 2.81250000 3.22942203 D I V 16 1.62500000 1.66833250 M I X 15 2.60000000 3.31231468 C M C - D I V 8 1.12500000 1.24642345 C M C - M I X 7 1.57142857 1.71824939 F T F - D I V 8 2.12500000 1.95940953 FTF -MTX 8 3.50000000 4.17475406 OPT 16 3.25000000 3.00000000 N O P 15 0.86666667 1.30201309 C M C - O P T 8 1.75000000 1.58113883 C M C - N O P 7 0.85714286 1.21498579 FTF-OPT 8 4.75000000 3.41216312 F T F - N O P 8 0.87500000 1.45773797 D IV -OPT 8 2.37500000 1.76776695 D I V - N O P 8 0.87500000 1.24642345 M I X - O P T 8 4.12500000 3.79614466 M I X - N O P 7 0.85714286 1.46385011 C M C - D I V - O P T 4 1.00000000 1.15470054 C M C - D I V - N O P 4 1.25000000 1.50000000 C M C - M I X - O P T 4 2.50000000 1.73205081 C M C - M I X - N O P 3 0.33333333 0.57735027 FTF -D IV -OPT 4 3.75000000 0.95742711 F T F - D I V - N O P 4 0.50000000 1.00000000 F T F - M I X - O P T 4 5.75000000 4.85626743 F T F - M I X - N O P 4 1.25000000 1.89296945 86 Table 23 Remarks - Discussion of offers ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value Pr >F M E 1 337.664651 337.664651 4.10 0.0546 IN 1 67.564651 67.564651 0.82 0.3743 ME*IN 1 100.027611 100.027611 1.22 0.2817 OP 1 28.564651 28.564651 0.35 0.5615 ME*OP 1 24.474040 24.474040 0.30 0.5908 IN*OP 1 113.216897 113.216897 1.38 0.2528 ME*IN*OP 1 251.506317 251.506317 3.06 0.0938 M E communication medium IN incentive scheme OP optimal solutions 87 Table 24 Remarks - Discussion of offers Cel l means N Mean SD C M C 15 3.33333333 4.82059080 F T F 16 9.93750000 11.98036592 D I V 16 5.31250000 6.6605180 M I X 15 8.26666667 12.1975798 C M C - D I V 8 3.8750000 6.1513645 C M C - M I X 7 2.7142857 3.0394235 F T F - D I V 8 6.7500000 7.2456884 F T F - M I X 8 13.1250000 15.2262696 OPT 16 5.81250000 6.4520669 N O P 15 7.73333333 12.4296113 C M C - O P T 8 3.5000000 4.2426407 C M C - N O P 7 3.1428571 5.7569833 FTF-OPT 8 8.1250000 7.6799833 F T F - N O P 8 11.7500000 15.5264751 D IV -OPT 8 6.1250000 7.7724330 D I V - N O P 8 4.5000000 5.7569833 M I X - O P T 8 5.5000000 5.3452248 M I X - N O P 7 11.4285714 17.0866141 C M C - D I V - O P T 4 2.7500000 5.5000000 C M C - D I V - N O P 4 5.0000000 7.3936910 C M C - M I X - O P T 4 4.2500000 3.2015621 C M C - M I X - N O P 3 0.6666667 1.1547005 FTF-D IV -OPT 4 9.5000000 8.9628864 FTF -D IV -NOP 4 4.0000000 4.6904158 F T F - M I X - O P T 4 6.7500000 7.2284162 F T F - M I X - N O P 4 19.5000000 19.5021366 88 Table 25 Remarks - Task-irrelevant ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 1.29086022 1.29086022 1.51 0.2316 IN 1 0.17419355 0.17419355 0.20 0.6560 ME*IN 1 1.45199693 1.45199693 1.70 0.2054 OP 1 0.17419355 0.17419355 0.20 0.6560 ME*OP 1 1.45199693 1.45199693 1.70 0.2054 IN*OP 1 0.21152074 0.21152074 0.25 0.6237 ME*IN*OP 1 0.35276498 0.35276498 0.41 0.5270 M E communication medium IN incentive scheme OP optimal solutions 89 Table 26 Remarks - Task-irrelevant Cel l means N Mean SD C M C 15 0.53333333 1.18723368 F T F 16 0.12500000 0.50000000 D I V 16 0.25000000 0.68313005 M I X 15 0.40000000 1.12122382 C M C - D I V 8 0.25000000 0.70710678 C M C - M I X 7 0.85714286 1.57359158 F T F - D I V 8 0.25000000 0.70710678 F T F - M I X 8 0.00000000 0.00000000 OPT 16 0.25000000 0.68313005 N O P 15 0.40000000 1.12122382 C M C - O P T 8 0.25000000 0.70710678 C M C - N O P 7 0.85714286 1.57359158 FTF-OPT 8 0.25000000 0.70710678 F T F - N O P 8 0.00000000 0.00000000 D IV -OPT 8 0.25000000 0.70710678 D I V - N O P 8 0.25000000 0.70710678 M I X - O P T 8 0.25000000 0.70710678 M I X - N O P 7 0.57142857 1.51185789 C M C - D I V - O P T 4 0.00000000 0.00000000 C M C - D I V - N O P 4 0.50000000 1.00000000 C M C - M I X - O P T 4 0.50000000 1.00000000 C M C - M I X - N O P 3 1.33333333 2.30940108 FTF -D IV -OPT 4 0.50000000 1.00000000 F T F - D I V - N O P 4 0.00000000 0.00000000 F T F - M I X - O P T 4 0.00000000 0.00000000 F T F - M I X - N O P 4 0.00000000 0.00000000 90 Table 27 Communication and processs measures MANOVA (medium x incentive x optimal) M A N O V A statistics: Wilks' Lambda, Pillai's Trace, Hotelling-Lawley Trace, Roy's Greatest Root Degress of freedom: (8, 16) Source F Pr >F M E 3.0924 0.0260 IN 1.3806 0.2768 ME* IN 1.4262 0.2592 OP 2.1921 0.0864 M E * OP 2.0188 0.1103 IN*OP 2.8464 0.0357 ME* IN* OP 0.4682 0.8610 M E communication medium IN incentive scheme OP optimal solutions 91 Table 28 Perception measures - Communication efficiency ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 6.125000 6.125000 0.05 0.8218 IN 1 378.125000 378.125000 3.20 0.0862 M E * IN 1 648.000000 648.000000 5.49 0.0278 OP 1 312.500000 312.500000 2.65 0.1169 ME*OP 1 1.125000 1.125000 0.01 0.9231 IN*OP 1 15.125000 15.125000 0.13 0.7236 ME*IN*OP 1 0.500000 0.500000 0.00 0.9487 M E communication medium IN incentive scheme OP optimal solutions 92 Table 29 Perception measures - Communication efficiency Cel l means N Mean SD C M C 16 61.8125000 13.8670773 F T F 16 62.6875000 9.3289424 D I V 16 58.8125000 13.2374154 M I X 16 65.6875000 8.9048208 C M C - D I V 8 53.8750000 15.1980967 C M C - M I X 8 69.7500000 6.0886311 F T F - D I V 8 63.7500000 9.4226172 F T F - M I X 8 61.6250000 9.7532046 OPT 16 65.3750000 12.7377392 N O P 16 59.1250000 9.8310732 C M C - O P T 8 64.7500000 15.9261689 C M C - N O P 8 58.8750000 11.7769448 FTF-OPT 8 66.0000000 9.6510547 F T F - N O P 8 59.3750000 8.2624365 D IV -OPT 8 61.2500000 14.5969664 D I V - N O P 8 56.3750000 12.1999707 M I X - O P T 8 69.5000000 9.7833678 M I X - N O P 8 61.8750000 6.4017297 C M C - D I V - O P T 4 56.0000000 18.8148877 C M C - D I V - N O P 4 51.7500000 13.1497782 C M C - M I X - O P T 4 73.5000000 5.8022984 C M C - M I X - N O P 4 66.0000000 3.9157800 FTF -D IV -OPT 4 66.5000000 8.3466560 FTF -D IV -NOP 4 61.0000000 10.8320512 F T F - M I X - O P T 4 65.5000000 12.1243557 FTF-MTX-NOP 4 57.7500000 5.9090326 93 Table 30 Perception measures - Co-operation ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 166.53125 166.53125 0.41 0.5295 IN 1 4826.53125 4826.53125 11.82 0.0022 ME*IN 1 693.78125 693.78125 1.70 0.2052 OP 1 371.28125 371.28125 0.91 0.3503 ME*OP 1 750.78125 750.78125 1.83 0.1882 IN*OP 1 42.78125 42.78125 0.10 0.7492 M E * IN* OP 1 344.53125 344.53125 0.84 0.3679 M E communication medium FN incentive scheme OP optimal solutions 94 Table 31 Perception measures - Co-operation Cell means N Mean SD CMC 16 158.812500 24.4109777 FTF 16 154.250000 22.9651911 DIV 16 144.250000 20.9841210 MIX 16 168.812500 19.2949691 CMC-DIV 8 141.875000 18.1536104 CMC-MIX 8 175.750000 17.0775542 FTF-DIV 8 146.625000 24.5178506 FTF-MIX 8 161.875000 19.9028892 OPT 16 159.937500 23.7584196 NOP 16 153.125000 23.3491613 CMC-OPT 8 157.375000 29.1054119 CMC-NOP 8 160.250000 20.6172605 FTF-OPT 8 162.500000 18.6394359 FTF-NOP 8 146.000000 25.0428205 DIV-OPT 8 146.500000 20.2272800 DIV-NOP 8 142.000000 22.8660697 MIX-OPT 8 173.375000 19.6900083 MIX-NOP 8 164.250000 19.0394328 CMC-DIV-OPT 4 136.000000 21.0713075 CMC-DIV-NOP 4 147.750000 15.2616076 CMC-MIX-OPT 4 178.750000 17.7270979 CMC-MIX-NOP 4 172.750000 18.5000000 FTF-DIV-OPT 4 157.000000 14.7196014 FTF-DIV-NOP 4 136.250000 29.9819390 FTF-MTX-OPT 4 168.000000 22.6568606 FTF-MIX-NOP 4 155.750000 17.6328292 95 Table 32 Perception measures M A N O V A (medium x incentive x optimal) M A N O V A statistics: Wilks' Lambda, Pillai's Trace, Hotelling-Lawley Trace, Roy's Greatest Root Degress of freedom: (2, 23) Source F P r > F M E 0.3305 0.7219 I N 5.7432 0.0095 M E * I N 2.7006 0.0884 OP 1.3173 0.2873 M E * O P 0.9959 0.3848 IN*OP 0.0796 0.9237 M E * i N * O P 0.4574 0.6386 M E communication medium I N incentive scheme OP optimal solutions 96 Table 33 Ranks - Marketing ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 29.076923 29.076923 0.59 0.4506 IN 1 499.054945 499.054945 10.18 0.0048 ME*IN 1 64.463370 64.463370 1.32 0.2657 OP 1 29.400000 29.400000 0.60 0.4482 M E * OP 1 11.790934 11.790934 0.24 0.6294 IN*OP 1 18.003388 18.003388 0.37 0.5516 ME*IN*OP 1 54.960440 54.960440 1.12 0.3029 M E communication medium IN incentive scheme OP optimal solutions 97 Table 34 Ranks - Marketing Cell means N Mean SD CMC 13 15.0769231 . 8.6067952 FTF 14 13.0000000 7.4420841 DIV 14 9.8571429 7.5534724 MIX 13 18.4615385 5.7534828 CMC-DIV 7 12.4285714 9.3069355 CMC-MIX 6 18.1666667 7.2502874 FTF-DIV 7 7.2857143 4.6445052 FTF-MIX 7 18.7142857 4.7157285 OPT 15 14.9333333 7.7594796 NOP 12 12.8333333 8.3430247 CMC-OPT 8 15.1250000 8.3569902 CMC-NOP 5 15.0000000 10.0000000 FTF-OPT 7 14.7142857 7.6749438 FTF-NOP 7 11.2857143 7.3646517 CMC-DIV-OPT 4 13.2500000 9.2150240 CMC-DIV-NOP 3 11.3333333 11.3724814 CMC-MIX-OPT 4 17.0000000 8.2865353 CMC-MIX-NOP 2 20.5000000 6.3639610 FTF-DIV-OPT 4 9.5000000 4.0414519 FTF-DIV-NOP 3 4.3333333 4.1633320 FTF-MIX-OPT 3 21.6666667 5.0332230 FTF-MIX-NOP 4 16.5000000 3.5118846 98 Table 35 Ranks - Production ANOVA (medium x incentive x optimal) Source DF Anova SS Mean Square F Value P r > F M E 1 92.719780 92.719780 2.69 0.1176 PN 1 685.978022 685.978022 19.88 0.0003 ME*IN 1 0.000000 0.000000 0.00 1.0000 OP 1 240.000000 240.000000 6.96 0.0162 ME*OP 1 6.119505 6.119505 0.18 0.6784 IN*OP 1 0.000000 0.000000 0.00 1.0000 ME*IN*OP 1 0.000000 0.000000 0.00 1.0000 M E communication medium IN incentive scheme OP optimal solutions 99 Table 36 Ranks - Production Cell means N Mean SD CMC 13 12.0769231 7.65355624 FTF 14 15.7857143 8.04963995 DIV 14 9.1428571 6.85966071 MIX 13 19.2307692 5.32531448 CMC-DIV 7 7.2857143 6.31702160 CMC-MIX 6 17.6666667 4.80277697 FTF-DIV 7 11.0000000 7.34846923 FTF-MIX 7 20.5714286 5.74041644 OPT 15 11.3333333 6.97614985 NOP 12 17:3333333 8.07164885 CMC-OPT 8 10.8750000 6.91659495 CMC-NOP 5 14.0000000 9.19238816 FTF-OPT 7 11.8571429 7.55928946 FTF-NOP 7 19.7142857 6.87299754 DIV-OPT 8 6.2500000 3.77018378 DIV-NOP 6 13.0000000 8.43800924 MIX-OPT 7 17.1428571 4.81070235 MIX-NOP 6 21.6666667 5.20256347 CMC-DIV-OPT 4 5.2500000 3.30403793 CMC-DIV-NOP 3 10.0000000 9.16515139 CMC-MIX-OPT 4 16.5000000 4.04145188 CMC-MIX-NOP 2 20.0000000 7.07106781 FTF-DIV-OPT 4 •7.2500000 4.42530602 FTF-DIV-NOP 3 16.0000000 8.18535277 FTF-MIX-OPT 3 18.0000000 6.55743852 FTF-MIX-NOP 4 22.5000000 5.06622805 100 Table 37 Ranks MANOVA (medium x incentive x optimal) MANOVA statistics: Wines' Lambda, Pillai's Trace, Hotelling-Lawley Trace, Roy's Greatest Root Degress of freedom: (3, 17) Source F P r > F M E 1.0073 0.4137 IN 34.1487 0.0001 ME*IN 0.4932 0.6917 OP 2.5918 0.0865 ME*OP 0.1100 0.9531 IN*OP 0.3556 0.7857 ME*IN*OP 0.4775 0.7021 M E communication medium IN incentive scheme OP optimal solutions 101 Figure 1 Market Demand Curve Panel Market Demand Curve .Market Demand Curves Click graph to enlarge You are given 2 demand curvesg -Actual Click, graph to enlarge Points On Curve 200 Enter price: Quantity: 440 Enter quantity: Price:[ 150 562 BSfli Altered Click graph to enlarge Points On Curve Enter price: 120 Quantity: 404 Enter quantity: GO /.price: 1550 102 Figure 2 Market Demand Curve 103 Figure 3 Production Cost Curve Panel Cost Curve Click graph to enlarge : Average Cost Curves^ You are given 2 cost curves: Actual Altered Click graph to enlarge -Actual 1 Points On Curve Enter quantity: J 230 Average cost: 81 -if I' Click graph to enlarge Altered 1 Points On Curve Enter quantity: 1 5 0 153 104 Figure 4 Production Average Cost Curve —— "n 'i • — ^ ^ ^ T " " w:« «<« . ; V Actual,Average Cost CurveyivVpi^lt^g 105 Figure 5 Create 10x10 Profit Table Panel Create 1 ox t o Profit Table Profit Table Parameters Starting transfer price: 350 Transfer price increment: 1° Starting transfer qty; 240 Transfer qty increment: 20 — Demand Curve p Actual r Altered liiiiitiis m Transfer Price -Your Profits -Transfer Quantity™ 240 260 280 300 320 "ST | r * I 350 360 370 330 390 400 24,00.0-" 21,600 l 19^200j 16,800 j 14,400] 12,000 I 19,500 14,000 7,500 16,900 14,300 j 9,100 j 6,500 | 11,200 8,400 4,500 1,500 5,600_ 2,800 .14/499)! (7,499)1 (12,799) 410 420 430 440 ti. 9,600 j 7,200} 4,800 | 2,400 i 3,900 J (2,799)| 1,300] (3,399) I .(. 5,5 (10,499)| (13,499)1 (22,399)' 106 Figure 6 Offer Panel Offer -Production's Offers offer # Price Qty Your Actual Profit Your Altered Profit Prod's Profit Co's Profit View 1 BOO 200 - 2 0 , 0 0 0 - 4 5 , 0 0 0 t i o j o o 90,too Cost Curve | Accept • Production's latest offer offer t Price Qty Y o u r Actual Profit -Your Offers Your Altered Profit Prod's Profit Co's Profit View 200 220 100 42,500 30,000 10,200 52,700 120 45,600 16,173 61,773 De^ard Curve I -Place your offers here Transfer price Transfer qty 220 J120 Take from Evaluate Offer - Include: C liothing C Actual demand C Altered demand C profit table Send Offer Evaluate Offer Transfer price 1100 Transfer qty 200 Your ae tual profit Production's profit 80,000 Altered profit 10,100 Co's actual profit 90,100 Joint Graphs • A L L Actual AUerrf Optimal Solutions 107 Figure 7 Joint Graphs Joint Grapfi ACTUAL W Quantity A Actual Demand OAvxj Cost Curve 108 Figure 8 Optimal Solutions Table r someC^t i i^^ Iu^ons5: v; •rr^::;^Q Co Sc rporate me Opti profit is mal solu optimal at transfer quantity = tions are: *' 293 1 Price Gty ." Voui Your Prod's Actual ' > Altered Profit Profit Profit Co*s Profit 208 137 293 293 51,494 14,869 51,473 57,647 21,022 45,320 102,967 102,967 i l 158 293 63,214 28,bay 39,753 102,967 151 293 68,195 3t,570 34,772 102,967 '1 13b 293 72,c83 36,258 30,084 102,967 109 o n J3UBJ 3§BSS9T^ 6 9 j n S ! J Appendix A Log Files Marketing Log Fi le ~ User: acmnmOl Thu Jun 2 10:32:28 1994 Time practice started: 11:14:36 Time game started: 11:31:06 11:31:20 Offer Panel is key 11:31:21 Demand Curve Panel is key 11:31:22 A L T demand curve window is key 11:31:37 Offer Panel is key 11:31:48 evaluate offer p=120.00 q=250.00 Act+Alt 11:32:12 Offer Panel is key 11:32:53 Offer Panel is key 11:32:56 send offer (1) p=120.00 q=250.00 A l t graph 11:33:09 Offer Panel is key 11:35:49 Profit Table is key 11:35:54 Offer Panel is key 11:36:17 Profit Table is key 11:36:46 create table ps=120.00 pi=10.00 qs=200.00 qi=10.00 source=Alt 11:37:29 Profit Table is key 111 11:37:29 11:37:41 11:38:12 11:38:23 11:38:53 11:39:53 11:41:50 11:42:14 11:43:13 11:43:27 11:44:10 11:44:50 11:44:51 11:45:11 11:45:34 11:45:40 11:46:08 11:47:17 Offer Panel is key A L T joint graph window is key Offer Panel is key evaluate offer p=140.00 q=200.00 Offer Panel is key Message window is key Offer Panel is key evaluate offer p=150.00 q=200.00 Profit Table is key create table ps=200.00 source=AIt create table ps=120 source=Alt Offer Panel is key evaluate offer p=150.00 q=150.00 evaluate offer p=l 50.00 q=200.00 send offer (2) p=150.00 q=150.00 Offer Panel is key evaluate offer p=120.00 q=l 50.00 evaluate offer p=120.00 q=200.00 Act+Alt Act+Alt pi=10.00 qs=250.00 qi=20.00 pi=10.00 qs=200.00 qi=10.00 Act+Alt Act+Alt Act+Alt Act+Alt 112 11:47:29 evaluate offer p=120.00 q=250.00 Act+Alt 11:47:46 evaluate offer p=120.00 q=300.00 Act+Alt 11:47:57 Profit Table is key 11:48:53 Message window is key 11:49:24 Offer Panel is key 11:49:43 Offer Panel is key 11:49:43 Message window is key 11:50:01 Offer Panel is key 11:50:05 A L T j oint graph window is key 11:50:26 evaluate offer p=100.00 q=250.00 Act+Alt 11:51:30 evaluate offer p=l 20.00 q=250.00 Act+Alt 11:52:14 Offer Panel is key 11:52:14 send offer (3) p=120.00 q=250.00 11:52:22 Offer Panel is key 11:52:25 Message window is key 11:52:26 Offer Panel is key 11:52:57 A L T joint graph is key 11:52:57 Profit Table is key 11:52:57 A L T demand curve window is key 11:52:58 Demand Curve Panel is key 113 11:53:00 Production has accepted Marketing's offer. Transfer price is 120.00. Transfer quantity is 250.00 Marketing's profit is 13750. Production's profit is 14062. Corporate profit is 27812. 114 Marketing Messages Fi le — User: acmnmOl Thu Jun 2 10:32:28 1994 11:41:42 We stil l have half an hour left. We still can negotiate the first offer later. 11:49:17 Aren't you going to counter my offer? 115 Appendix B Advertisement INVITATION TO PARTICIPATE IN A NEGOTIATION STUDY We would like to invite you to participate in a study to investigate the effects of using a computerized tool to support business negotiations. The exercise gives you an excellent opportunity to experience the very user-friendly NeXT color interface. You and one other person will represent two different divisions of the same company and will be involved in a transfer pricing negotiation. You will receive $10 for your participation. In addition, 25% of the participants will receive cash prizes ranging from $10 to $50 based on their performance in the negotiation. Participation in this study is strictly VOLUNTARY. You may withdraw from the exercise at any time at your own discretion. The study will last for approximately 1.5 hours and is made up of 5 parts: (1) The collection of some basic demographic data - a short questionnaire (2 minutes); (2) Tutorial - step-by-step instructions will be given to help participants familiarize themselves with the computer system (approx. 30 minutes); (3) A practice negotiation (approx. 15 minutes); (4) The negotiation (maximum 45 minutes); (5) The collection of data regarding the negotiation and the computer system -a short questionnaire (2-5 minutes). The records of the exercise will be collected automatically by the computer. Also, conversation made during the negotiation will be tape-recorded. All of the collected data will be treated in STRICT CONFIDENCE and will only be reported anonymously. Data will be aggregated and will not be reported on an individual basis. Your name will not be used in any report, publication or information connected with the study. Those who wish to find out about their individual performance will be briefed by the researchers. Your participation will be very much appreciated. The study will be scheduled during the months of May to July, 1994. If you have queries concerning the study or if you are interested to take part, please contact Yvonne Chan at 822-8516 or 275-6873. Email: ychan@unixg.ubc.ca. 116 Appendix C Consent Form A STUDY OF COMPUTER-SUPPORTED NEGOTIATION Faculty Advisor: Student investigator: Professor Izak Benbasat Yvonne Chan MIS Division, Faculty of Commerce MIS Division, Faculty of Commerce HA 454 (822-8396) Tel: 275-6873 This study is for an M.Sc. thesis for the Management Information Systems Division. Its objective is to investigate the effects of using a computerized tool to support business negotiations. The exercise gives you an excellent opportunity to experience the very user-friendly NeXT color interface. You and one other person will represent two different divisions of the same company and will be involved in a transfer pricing negotiation. You will receive $10 for your participation. In addition, 25% of the participants will receive cash prizes ranging from $10 to $50 based on their performance in the negotiation. Participation in this study is strictly VOLUNTARY. You may withdraw from the exercise at any time at your own discretion. The study will last for approximately 1.5 hours and is made up of 5 parts: (1) The collection of some basic demographic data - a short questionnaire (2 minutes); (2) Tutorial - step-by-step instructions will be given to help participants familiarize themselves with the computer system (approx. 30 minutes); (3) A practice negotiation (approx. 15 minutes); (4) The negotiation (maximum 45 minutes); (5) The collection of data regarding the negotiation and the computer system -a short questionnaire (2-5 minutes). The records of the exercise will be collected automatically by the computer. Also, conversation made during the negotiation will be tape-recorded. All of the collected data will be treated in STRICT CONFIDENCE and will only be reported anonymously. Data will be aggregated and will not be reported on an individual basis. Your name will not be used in any report, publication or information connected with the study. Those who wish to find out about their individual performance will be briefed by the researchers. 117 STATEMENT OF INFORMED CONSENT I consent to participate in the project outlined above and acknowledge that I have received a copy of this consent form and attachments. Name of student: Signature: Date: 118 Appendix D Demographic Data Questionnaire Negotiation Study ID: Session code: Date: _ Before we start, we would like you to fill in this questionnaire so that we can know more about you. The purpose of collecting this information is to find out the distribution of our sample of participants in terms of their computer literacy and knowledge of economics. The information that you have disclosed will not affect your performance in the exercise nor will it be used in any form of data analysis. Please circle the correct response. 1. I am a part-time / full-time undergraduate / Master / PhD student. Faculty: Department: 2. Are you familiar with Window applications? Yes / No 3. Do you have experience using the NeXT? a) I have never used it before. b) I have used it before, but less than 5 times. I have used it before, but less than 20 times. I am a dedicated NeXT user. 4. Are you familiar with some basic microeconomic theory such as demand and market price, production cost and profit? Yes / No 119 Appendix E Task Descriptions Sheets (1) Divisional-incentive Scheme (Marketing) TransferPricing Game TransferPricing Game is a negotiation exercise between you and the Production manager. Your task as the Marketing manager is to reach an agreement with the Production manager as to the transfer price and the transfer quantity of an intermediate good. In this exercise, your objective is to maximize your own profit. Y o u w i l l be evaluated and receive additional monetary awards ranging from $10 to $50 based on the amount of profit you achieve. If you make $20,000 in the negotiation, then your payoff is $20,000. But i f you do not make a profit (i.e. zero or negative profit), then your payoff w i l l be zero. Similarly, i f you do not reach an agreement with the other party in the negotiation, your payoff is zero. It is important that you understand your objective in the negotiation and the award structure. Y o u should discuss with the experimenter any doubts you may have regarding this before continuing the exercise. Please answer the following questions and hand the paper to the experimenter when you have finished. 1. Your profit is $ 15,000, what is your payoff in this exercise? 2. Y o u breakeven at the agreed transfer price and transfer quantity (i.e. profit is zero), what is your payoff? 3. Y o u make a loss of $10,000, what is your payoff? ID: Session code: Date: 120 (2) Divisional-incentive Scheme (Production) TransferPricing Game TransferPricing Game is a negotiation exercise between you and the Marketing manager. Your task as the Production manager is to reach an agreement with the Marketing manager as to the transfer price and the transfer quantity of an intermediate good. In this exercise, your objective is to maximize your own profit. You will be evaluated and receive additional monetary awards ranging from $10 to $50 based on the amount of profit you achieve. If you make $20,000 in the negotiation, then your payoff is $20,000. But if you do not make a profit (i.e. zero or negative profit), then your payoff will be zero. Similarly, if you do not reach an agreement with the other party in the negotiation, your payoff is zero. It is important that you understand your objective in the negotiation and the award structure. You should discuss with the experimenter any doubts you may have regarding this before continuing the exercise. Please answer the following questions and hand the paper to the experimenter when you have finished. 1. Your profit is $15,000, what is your payoff in this exercise? 2. You breakeven at the agreed transfer price and transfer quantity (i.e. profit is zero), what is your payoff? 3. You make a loss of $10,000, what is your payoff? _ ID: Session code: Date: 121 (3) Mixed-incentive Scheme (Marketing) TransferPricing Game TransferPricing Game is a negotiation exercise between you and the Production manager. Your task as the Marketing manager is to reach an agreement with the Production manager as to the transfer price and the transfer quantity of an intermediate good. In this exercise, your objective is to maximize the combined profits of yourself and the company as a whole. You will be evaluated and receive additional monetary awards ranging from $10 to $50 based on the amount of the combined profits. Your payoff in the negotiation is determined by the following formula: your payoff = 50% of your profit + 50% of the company's profit But if you do not reach an agreement with the other party, your payoff will be zero. Examples: a) Your profit is $25,000 Production's profit is $15,000 Company's profit is $40,000 ($25,000 + $15,000) Your payoff = 50% x $25,000 + 50% x $40,000 = $32,500 b) Your profit is $-20,000 Production's profit is $60,000 Company's profit is $40,000 ($-20,000 + $60,000) Your payoff = 50% x $-20,000 + 50% x $40,000 = $10,000 c) Your profit is $0 Production's profit is $15,000 Company's profit is $15,000 ($0 + $15,000) Your payoff = 50% x $0 + 50% x $15,000 = $7,500 122 It is important that you understand your objective in the negotiation and the award structure. You should discuss with the experimenter any doubts you may have regarding this before continuing the exercise. Please answer the following questions and hand the paper to the experimenter when you have finished. 1. Your profit is $20,000 (A) Production's profit is $10,000 (B) Company's profit is $ (C = A + B) Your payoff = 50% x (A) + 50% x (C) = $ + $ = $ 2. Your profit is $30,000 (A) Production's profit is $30,000(B) Company's profit is $ (C = A + B) Your payoff = 50% x (A) + 50% x (C) = $ + $ = $ 3. Your profit is $0 (A) Production's profit is $10,000 (B) Company's profit is $ (C = A + B) Your payoff = 50% x (A) + 50% x (C) = $ + $ = $ 4. Your loss is $10,000 (A) Production's profit is $60,000 (B) Company's profit is $ (C = - A + B) Your payoff = 50% x (-A) + 50% x (C) = $ + $ = $ ID: Session code: Date: 123 (4) Mixed-incentive Scheme (Production) TransferPricing Game TransferPricing Game is a negotiation exercise between you and the Marketing manager. Your task as the Production manager is to reach an agreement with the Marketing manager as to the transfer price and the transfer quantity of an intermediate good. In this exercise, your objective is to maximize the combined profits of yourself and the company as a whole. You will be evaluated and receive additional monetary awards ranging from $10 to $50 based on the amount of the combined profits. Your payoff in the negotiation is determined by the following formula: your payoff = 50% of your profit + 50% of the company's profit But if you do not reach an agreement with the other party, your payoff will be zero. Examples: a) Your profit is $25,000 Marketing's profit is $15,000 Company's profit is $40,000 ($25,000 + $15,000) Your payoff = 50% x $25,000 + 50% x $40,000 = $32,500 b) Your profit is $-20,000 Marketing's profit is $60,000 Company's profit is $40,000 ($-20,000 + $60,000) Your payoff = 50% x $-20,000 + 50% x $40,000 = $10,000 c) Your profit is $0 Marketing's profit is $15,000 Company's profit is $15,000 ($0 + $15,000) Your payoff = 50% x $0 + 50% x $15,000 = $7,500 124 It is important that you understand your objective in the negotiation and the award structure. Y o u should discuss with the experimenter any doubts you may have regarding this before continuing the exercise. Please answer the following questions and hand the paper to the experimenter when you have finished. 1. Your profit is $20,000 (A) Marketing's profit is $ 10,000 (B) Company's profit is $ (C = A + B ) Your payoff = 50% x (A) + 50% x (C) = $ + $ = $ 2. Your profit is $30,000 (A) Marketing's profit is $30,000 (B) Company's profit is $ (C = A + B) Your payoff = 50% x (A) + 50% x (C) = $ + $ = $ 3. Your profit is $0 (A) Marketing's profit is $ 10,000 (B) Company's profit is $ (C = A + B) Your payoff = 50% x (A) + 50% x (C) = $ + $ = $ 4. Your loss is $10,000 (A) Marketing's profit is $60,000 (B) Company's profit is $ (C = -A + B) Your payoff = 50% x (-A) + 50% x (C) = $ + $ = $ ED: Session code: Date: 125 Appendix F Post-experiment Questionnaire Note: Questions 7, 13, 15, and 17 pertain to communication efficiency while the rest are related to co-operation. Negotiation Study ID: Date: Session code: Thank you for taking part in the study. Before you leave, we would like you to complete this questionnaire to find out how you feel about the negotiation task you have worked on. Please circle the appropriate response. 1. When I made my decisions in this exercise, I considered the other party's perspectives. / 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree The overall corporate profits were important to me in making my decisions. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree I had sufficient information about my opponent's profit. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree 4. I am pleased with my performance in the negotiation. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree 126 5. Rate yourself in terms of how you negotiated with your opponent in this exercise. 1 2 3 4 5 6 7 Very Neutral Very Competitively Co-operatively 6. I had sufficient information about my own profit. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree 7. I was able to express myself fully. 1 2 3 4 5 6 7 Never Sometimes Always 8. Rate your opponent in terms of how he/she negotiated with you in this exercise. 1 2 3 4 5 6 7 Very Neutral Very Competitively Co-operatively 9. M y own profits were important to me in making my decisions. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree 10. I felt I could increase my payoffs by being cooperative with the other party. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree 11. I felt I could increase my payoffs by being competitive with the other party. 1 2 3 4 5 6 7 Strongly Neutral Strongly Agree Disagree 127 12. M y opponent profits were important to me in making my decisions. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree 13. I had a feeling of personal contact with the other party. 1 2 3 4 5 6 7 Never Sometimes Always 14. I think the outcome of the negotiation is 1 2 3 4 5 6 7 Extremely Extremely Unfair Fair 15. I was constrained in my communication with the other party. 1 2 3 4 5 6 7 Never Sometimes Always 16. I found the climate surrounding the negotiation 1 2 3 4 5 6 7 Extremely Extremely Competitive Co-operative 17. M y communication and discussion with the other party was: 1 2 3 4 5 6 7 Extremely Extremely Impersonal Personal 1 2 3 4 5 6 7 Extremely Extremely Frustrating Satisfying 128 1 2 3 4 5 6 7 Extremely Extremely Inefficient Efficient 1 2 3 4 5 6 7 Extremely Extremely Unproductive Productive 18. The other party is 1 2 3 4 5 6 7 Extremely Extremely Untrustworthy Trustworthy 19. When the other party made his/her decisions, he/she considered my perspectives. 1 2 3 4 5 6 7 Strongly Neutral Strongly Disagree Agree 20. The process of the negotiation, on the whole was 1 2 3 4 5 6 7 Extremely Extremely Inefficient Efficient 129 Appendix G Profit Schedules Example Given below are the profit schedules for a buyer and a seller. Buyer's profit schedule Option Item 1 Item 2 Item 3 A 0 0 0 B 100 150 250 C 200 300 500 Seller's profit schedule Option Item 1 Item 2 Item 3 A 500 300 200 B 250 150 100 C 0 0 0 When negotiators' preferences on a particular item are completely opposite to each other, the item is considered as "distributive". Item 2 in this example is a distributive item. Items 1 and 3 are considered as "integrative". Both negotiators can tradeoff Item 1 for Item 3 or vice versa so that individual's profit is the same. The best outcome will be to tradeoff the integrative items and compromise on the distributive, i.e. Item 1 - A Item 1 - C Item 2 - B or Item 2 - B Item 3 - C Item 3 - A 130