EFFECTIVENESS OF ELECTRONIC NEGOTIATION STYLES USING THE THOMAS-KILMANN MODEL

EFFECTIVENESS OF ELECTRONIC NEGOTIATION STYLES USING THE THOMAS-KILMANN MODEL

EFFECTIVENESS OF ELECTRONIC NEGOTIATION STYLES USING THE THOMAS-KILMANN MODEL Renata A. Zaremba1) and Gregory Kersten2) 1) Greenhouse Gas Division, E...

689KB Sizes 0 Downloads 10 Views

EFFECTIVENESS OF ELECTRONIC NEGOTIATION STYLES USING THE THOMAS-KILMANN MODEL Renata A. Zaremba1) and Gregory Kersten2) 1)

Greenhouse Gas Division, Environment Canada Ottawa, Ontario, Canada 2) J. Molson School of Business, Concordia University Montreal, Quebec, Canada

Abstract: With the emergence of the Internet, the dynamics of the negotiation process change. This paper investigates the effects of the five Thomas-Kilmann (TK) negotiation styles (competing, collaborating, accommodating, compromising, and avoiding) on qualitative and quantitative negotiation outcomes in the situation when the negotiations take place over the Internet. A new variable is introduced that measures the difference between the opponents rating of the deal and that of the main negotiator: Individual Outcome Difference (IOD). Multivariate correlation analysis is used first to determine any relationships that may exist between the outcomes and the TK negotiation styles. Since the negotiators often use a combination of the five styles during the negotiation process, exploratory data analysis was done, which allowed us to investigate the simultaneous impact of all the five styles on the negotiation outcome. The methodological approach and the results of tests performed using the INSPIRE system are presented and discussed. Copyright © 2006 IFAC. Keywords: E-negotiations, Thomas-Kilmann model, INSPIRE system, cluster analysis, negotiation styles.

1. INTRODUCTION Although negotiating over the Internet is a relatively new area, it is becoming an important and widespread field. Earlier studies on the difference of Internetbased negotiations and face-to-face negotiations suggest that negotiations over the Internet are different in several respects. One of the most important differences between e - negotiations and face-to-face negotiations is the non-verbal communication. In electronic negotiations non-verbal cues and signals are eliminated: the tone of voice, facial expressions, etc. Since non-verbal cues may relay information more quickly, one possible result is that text messages may take longer to be conveyed than in face-to-face negotiations. Students in (Köszegi and Kersten 2003) research found it harder to understand the intentions of their opponents and more difficult to interpret opponents behaviour. With a better understanding of social cues in face-to- face negotiations, rapport is fostered and the basis of trust is strengthened (Droplet and Morris, 2000). It has been observed that the decision to trust is linked to

the collaborating style as opposed to the decision to dominate for the competing style (Sheffield 1989). Enegotiations may easier lead to increased conflict (Schei and Rognes 2003). Another significant difference observed between electronic negotiations and face-to-face negotiations is the use of tactics. (Galin, et al. 2002) found that in electronic negotiations, as opposed to face-to-face negotiations, more hard tactics were used and less soft and authority related tactics. Soft tactics such as information sharing, are related to collaborative style, while information withholding to the competitive style (Schei and Rognes 2003)). Research also shows that generally joint outcomes are greater for electronic negotiations than for face-to-face negotiations (Perkins et al., 1996; Delaney et al., 1997; Croson 1999).

301

One area of negotiation research has focused on proposing and specifying different approaches and styles of negotiation, and on creating behavioural models to measure these approaches. For instance,

Putnam and Wilson (1982) acknowledged three negotiation styles: solution-oriented, nonconfrontation, and control modes. Rahim et al. (1993) identified five different styles: integrating, obliging, dominating, avoiding and compromising. Thomas and Kilmann (1974) described a negotiator’s orientation towards two basic dimensions: the degree to which a person tries to meet his or her concerns and the degree to which a person tries to meet the concerns of the other party. They developed the Thomas-Kilmann Questionnaire Instrument (TKI) to determine one’s negotiation profile based on five negotiation styles: collaborating, accommodating, competing, avoiding, and compromising. It is widely used in negotiation research (Van de Vliert and Kabanoff 1990) and has been assessed by various studies.

course web site (http://mis.concordia.ca/negocourse).

Negocourse

2.1 Negotiation Case The experiments involved a case of business negotiations between the representatives of two companies: a producer of bicycle components and a builder of bicycle. The simulated negotiation involved one commodity (a bicycle component). Every participant might terminate one negotiation and request a new one thus making negotiation breakdown possible. The parties discussed only four issues: Price, Delivery, Payment, and Return. The INSPIRE users had to decide their issue priorities and rate the issues and options. The issue ratings had to add up to 100. Each participant was given a clear indication as to the company’s perspective regarding the direction of the issue options. However, there were no mechanisms to check the option ratings and so students might not follow the direction indicated.

The competing style (CTS) involves focusing on one’s own needs and not considering the other party’s needs. The accommodating style (ACS) considers the needs of the other party and not one’s own need. The collaborating style (CLS) considers both the needs of the other party and one’s own needs. The avoiding style (AVS) neither considers the needs of the other party or his/her own needs. The compromising style (CMS) is intermediate in both levels of assertiveness and cooperativeness.

2.2 Participants The negotiation process involved 112 students (56 teams) in total. Students from different universities were paired together. These experiments involved students from Austria, Taiwan, Canada, and Russia. Only 61 students, those from Canada (Ottawa) and Austria, completed the TKI and were part of the analysis.

Although the theoretical five-model framework has been supported by recent research and widely recognized, surprisingly little empirical research compares the effect of the five styles on outcomes in Internet based negotiations. Also, numerous studies suggest that the negotiators may use a combination of the five styles (Munduate, Ganaza et al. 1999). This paper analyses all the five negotiation styles to determine their effect on e - negotiations quantitative outcomes.

2.3 Negotiation process The Pre-negotiation Phase. Before the students start negotiating, they read the case. They then submitted the issue and option ratings, as well as the package ratings. All students rated 12 packages, each containing the 4 issues. An example of a package is shown in Fig. 1. Finally, the students filled out a prenegotiation questionnaire. Some of these questions were used to assess negotiation outcomes and are further discussed in Section 2.5.

Most research has focused on quantitative outcomes and only certain qualitative outcomes, such as satisfaction. One challenge is to determine which qualitative negotiation outcomes are affected by the five negotiation styles. This paper considers both qualitative negotiation outcomes and quantitative outcomes. Also, a new variable was proposed that measures the difference between the opponents rating of the deal and that of the main negotiator IOD (Individual Outcome Difference). In the final part of the paper experiments are described that were carried out using the INSPIRE web based negotiation support system (InterNeg, 2006). The results obtained from statistical analysis and exploratory data analysis are provided and discussed.

Price

Delivery Payment

Returns Rating 75% refund Upon 100 with 10% delivery spoilage

4.12 $

30 days

Figure 1. An example of the 12 rated packages. The actual negotiation. The students had 3 weeks to reach an agreement. Every offer had to encompass the four issues. The system used the issue and option rating indicated by the students to calculate both the offer ratings and the final agreement. During the negotiation the students had access to a History Graph showing both their own offer ratings and those of the opponents (Figure 2). The offer ratings of the opponent were calculated based on the preferences indicated by the negotiator and not the opponent. Therefore, each student was aware of how far his offer and the opponent’s offer were aligned with his own preferences.

2. THE E-NEGOTIATION EXPERIMENT The experiments which are described here were carried out using the INSPIRE system - a web based negotiation support system (InterNeg, 2006) and the 302

Verification of the Pareto-optimality. The utility function of the final agreement is used to determine if Pareto-optimality has been reached. If this has occurred, INSPIRE terminates negotiations. If this has not occurred, INSPIRE then calculates all the packages that are improvements on the agreement and presents up to five of the packages. 2.5 Negotiation outcome Both qualitative and quantitative negotiation outcomes were analyzed. The following quantitative outcomes were considered in the analysis: the main negotiator’s rating of the agreement negotiated (last deal) V1; the opponent’s rating of the agreement negotiated V2; the individual Outcome Difference [V1 - V2 (IOD)] V3; and the absolute value of IOD V4. As well, the qualitative outcomes were considered: determining the relation between 5 ‘Main Negotiator’s characteristics (Informative V6, Persuasive V7, Hones V8, Exploitive V9, and Cooperative V10) and the Five Conflict Styles; assessment of collaboration between the parties (How friendly the opponent found the negotiations V12; if the opponent found the main negotiator Cooperative V11, Informative V13, Exploitative V14, Reliable V15, Likable V16, Rational V17, Trustworthy V18, Kind V19, Fair V20, Flexible V21, persuasive V22, and Honest V23); level of satisfaction of the ‘main negotiator’ (measured by their level of satisfaction of outcome V26 and their performance V27, whether the negotiation outcome matched the negotiator’s expectation V24, and the level of control the ‘main negotiator’ felt he had during the negotiation V25); and the number of agreements suggested by NSS in the post-settlement phase V5.

Figure 2. Negotiator History Graph. Post-Settlement Phase. Once the two students reached an agreement, the INSPIRE system offered up to 5 deals that would improve the rating of both or one of the students. The students could then terminate negotiations or keep negotiating until they reached a better agreement. Finally, upon termination of the negotiation, the participants filled out a post-negotiation questionnaire. Some of these questions were used to assess selected negotiation outcomes and are further discussed in Section 2.5. 2.4 Negotiation Framework The negotiation process used two different kinds of systems: The INSPIRE system and the e-business system Negocourse. Unlike many negotiation support systems, INSPIRE is a web-based negotiation system with thin client - thick server architecture. The system has four basic features: 1. Formation of utility functions; 2. Presentation of negotiation dynamics in the form of a graph. 3. Verification whether Pareto-optimality has been reached; if not, the system suggests better agreements that could be achieved without disadvantaging any of the negotiators; 4. Recording of offers and messages and creation of the negotiation history.

Comparing the difference between the deals achieved by both sides rather than just looking at the actual deals reached by one side may be more relevant. Therefore, how much of a better deal the Main Negotiator negotiates than his Opponent is a better reflection of the effectiveness of one style over the other. Also, by studying whether Pareto optimality has been reached and considering the number of agreements suggested in the post-settlement phase, joint outcome can be analyzed. (Schei and Rognes 2003).

Formation of utility functions. The INSPIRE system elicits preferences (issues and ratings) to determine the relative value of the offers and packages using utility functions. These are determined by means of conjoint analysis, wherein a subset of the possible combinations of issues and ratings, as valued by the user, establish the relative importance of each package or offer. INSPIRE chooses the packages so that the matrix X is orthogonal in the orthogonal design. The ratings of the packages are then used to decompose it into option utilities using the following equation:

2.6. Analytical Approach The approach adopted for the analysis addressed the specific characteristics of the data set available for the analysis: a large number of the negotiation outcomes, a rather small number (61) of the samples of the negotiation process, and a variety of combinations of the negotiations styles. The large variety of the negotiation outcomes creates the problem of deciding on the importance and correlation of the outcomes. A purely statistical approach, producing statistically significant results, cannot always be adopted. A large number of all possible combinations of the negotiation styles raised the issue of the combinatorial explosion, the validity of the tests, and their number. As well, previous research suggests that participants may use a combination of styles (Munduate, Ganaza et al. 1999;

Rating (Pk) = constant + ∑∑ uij xij + error, (1) where Rating (PA) is the total of a package Pk, uij is the utility associated with issue i and option j (j = 1mi; i = 1-n), and xij is a binary variable indicating whether the given option is present in the package. The ratings of the packages are then used to determine uij so as to minimize the errors in linear regression. The utilities of all the possible options are considered. 303

Jain and Solomon 2000). The pattern of the profile, i.e., the combination of the five styles, may predict the negotiators performance in the negotiation outcomes. Therefore, cluster analysis is a preferred method for analysing the configuration of the profiles.

competitive were less likely to rate themselves to be persuasive, exploitive, and cooperative. On the other hand, the greater the compromising style in a student’s profile, the more likely they rate themselves informative, persuasive, honest, and exploitive. Other Negotiation Results. No other statistically significant correlations were observed: this includes the five negotiation styles and IOD. As well, there were no correlations observed between the outcomes measuring satisfaction and the five negotiation styles.

3. EVALUATION OF QUANTITATIVE OUTCOMES 3.1. Analysis using correlations Due to many possible relationships between the negotiation styles bivariate correlation analysis was performed. Pearson’s correlation coefficients (CP) were computed, and tests for statistical significance were performed. The 2-tailed test was used to determine statistical significance. Relationships were deemed significant with the 2-tailed test values below .05 and highly significant with 2-tailed test values below .01. The statistically significant relationships are shown in Table 1.

3.2. Analysis using regression

-.32

-.29

.04

.02

.04

Regression analysis was used to determine more quantitative measures of the relationships between specific negotiation styles and the selected outcomes, and partly to confirm or dispute the results obtained from the correlation analysis. As well, since the combination of the five negotiation styles and the negotiation outcomes is very large, only certain negotiation outcomes were tested. The outcomes tested were selected based on the results obtained from the outcomes of the correlation analysis discussed in the preceding section. The trends observed using correlation analysis were confirmed by the regression analysis when considering the significance level below 0.05, as shown in Table 2.

.44

53 -.01 .31

53 .31

53 .19

Table 2. Statistically significant relationships

.01

.95 .02

.03

.02

.17

35

35 54

54

54

54

Table 1. Statistically significant relationships. V1 V22 V21 V7 -.28

CP Signif. N CP Signif. N

N

CTS

-.35 .01 55

CMS

CP Signif.

V1

.36 .04

AVS

N

.3

35

CLS

CP Sig.

V8 V9 V10

Negotiation Results. There were no statistical significant correlations found between the five negotiation styles and IOD. There were no correlations between the competing, collaborating, compromising, and accommodating styles and the last offer. However, students who had a greater avoiding style in the student’s profile negotiated on average a lower deal. This relationship was observed to be highly statistically significant. How the participants were viewed. There was a highly statistically significant correlation was observed between an increase in the student’s compromising style in their profile and the likelihood that the opponents saw these students as pushovers. A statistically significant correlation was also observed between an increase in the student’s collaborating style and the likelihood that the opponents saw these students as rigid. How the participants rated themselves. The competing and the compromising negotiation style correlated with some of the 5 characteristics: Informative, Compromising, Persuasive, Honest, and Exploitive. Those students analysed who were

304

B β Sign. B β Sign. B β Sign. B β Sign.

-32 -0.35 0.01

V3 V22 V21 V5

V7 V8

1.52

-1.40 -0.28 0.04 1.75 1.76 1.79

0.44 0.01

0.29 0.31 0.03 0.02

V9

-1.86 -1.77 -0.38 -0.29 0.03 0.04 2.16

0.30 0.31 0.03 0.02

1.74 0.36 0.03

Regression analysis and correlation analysis demonstrated no statistically significant relationship observed between all five negotiation styles and both IOD and the number of agreements. In addition, none of these styles appear to influence the joint outcome as determined by looking at the number of offers generated by Inspire in the post-settlement phase V5 (i.e., looking at Pareto-efficiency). This is consistent with (Schei and Rognes 2003) observation in terms of the competitive and collaborative styles when dealing with face-to-face negotiations. However, it was observed that the avoiding style correlated positively with the rating of the deal reached by the Main Negotiator alone. Regression analysis also confirmed that the students with a greater compromising style were seen as pushovers. Ironically, these same students were more likely than the other students to rate themselves as persuasive. The compromising style students were also more likely to rate themselves as informative, honest, and exploitive. There were no statistically significant results for honest and informative ratings.

depending on a threshold value of the similarity coefficient S. The number of clusters increased with an increasing range of the similarity coefficient. The range of the similarity coefficient was tested from 0.64 to 0.76. Slight changes in the value of the similarity coefficient do not make any difference in the results. The clusters are shown in Table 4, and a graphical representation of the 6 clusters is shown in Fig. 3.

3.3 Inter-Related Styles The question arises if the negotiation styles are inter-related and the correlation analysis was done to determine if in fact the styles are inter-related. Table 3 shows the relationships between the styles in the profile of the students. Table 3. Correlations between the styles in the profile of the students. CTS CTS

CLS

CTS

AVS

ACS

-1.9

-.03

-.31

-.52

.159

.796

.18

0

-1.9

-.34

-.38

.16

.01

CP Signif.

CLS

CP Signif.

CTS

CP Signif.

AVS

CP Signif.

ASC

CP Signif.

Table 4. Clusters generated with a 0.70 similarity coefficient Object

CTS

CLS

CMS

AVS

ACS

1

0.34

0.38

0.85

0.25

0.58

2

0.00

0.6 8

0.33

0.34

1.10

-.05

3

0.81

0. 10

0.23

0.44

0.97

.003

.69

4

0.91

0.68

0.33

0.09

0.49

5

0.72

0.20

0.74

0.72

0.19

6

0.00

1.00

0.18

0.64

0.34

-.03

-.34

-.29

-.37

.80

.01

.02

.003

-.31

-.38

-.30

.001

.02

.003

.02

.996

-.52

-.05

-.37

.001

0

.69

.003

.996

Some negative inter-relationships were observed. The collaborating style and the compromising style were observed to be negatively inter-related, as well as the compromising and accommodating style. These correlations show that there may be some interrelationships between the styles, and looking at the whole profile of the negotiator, for instance through cluster analysis, may be a better way of determining the effect of styles on outcomes.

Figure 3. Distribution of clusters for S = 0.7. Table 6 shows the coordinates of the 6 clusters in 5dimensional space expressed, for each group, as a percentage of the negotiation style with respect to the maximum value of the 5 negotiation styles. The greater the percentage value, the greater the influence of the negotiation style in the specific cluster.

4. EXPLORATORY CLUSTER ANALYSIS Dealing with multidimensional analysis is inherently complex. Thus the use of a multivariate regression is in this case limited by the number of cases and the resulting statistical significance of the results. Direct extension of linear regression to the analysis of the effects of the combination of several styles leads to a combinational explosion of the combinations. In this work, exploratory cluster analysis was used to determine the distribution of participants in the 5dimensional space of negotiation styles, and to determine if the participants with similar features correlate with the negotiation results. The cluster analysis was performed using the DIG software developed at the National Research Council of Canada. The software calculates a matrix of the values of similarity between n - dimensional samples, and defines clusters reducing at the same time their dimensionality to 3-D. Clusters are defined

Table 6. Coordinates of the 6 cluster centres in the percentage of the dominant negotiation style. Coordinates of cluster centers (weights of the negotiation styles) Group

305

CTS

CLS

CMS

AVS

ACS

1

40%

45%

100%

29%

68%

2

0%

62%

30%

31%

100%

3

84%

10%

24%

45%

100%

4

100%

75%

36%

10%

54%

5

97%

27%

100%

97%

25%

6

0%

100%

18%

64%

34%

the participants does not remain constant throughout the negotiation. Whether the obtained results translate into other negotiation circumstances and environments, wherein there is a time limit to negotiate and the overall goals of the two negotiating participants are opposite, is an opportunity for further research.

The six clusters were then tested for various negotiation outcomes. Table 7 shows how the participants belonging to each of the 6 cluster groups fared in the IOD negotiation outcome. Table 7: Averages and medians of the 6 clusters in regards to IOD 1

2

Group 3 4

5

6

All Styles

Average

14

30

-22

17

13

42

Median

22

22

-22

17

10

42

No. of values

34

7

2

2

8

3

56

N+

19

5

1

1

4

2

32

N-

11

0

1

1

1

0

14

No agreement

4

2

0

0

3

1

10

REFERENCES

16

Croson, R. (1999). Look at Me When You Say That: An Electronic Negotiation Simulation. Simulation & Gaming, 30(1). Delaney, M., A. Foroughi, et al. (1997). An empirical study of the efficacy of a computerized negotiation support system (NSS). Decision Support Systems, 20. Droplet,A.L. and Morris, M.W. (2000). Rapport in Conflict Resolution: Accounting for How Face to Face Contact Fosters Mutual Cooperation in Mixed Motive Conflicts. Journal of Experimental Social Psychology, 36, 26-50. Galin, A., M. Gross, et al. (2002). E-Negotiation versus face-to-face negotiation. What has changed - if anything? Working paper No. 8/2002. InterNeg (2006). http://interneg.org/interneg/tools /inspire/index.html. Jain, B. and S. Solomon (2000). The Effect of Task Complexity and Conflict Handling Styles on Computer-Supported Negotiations. Information & Management, 37. Köszegi, S. and G. E. Kersten (2003). On-line/Offline: Joint Negotiation Teaching in Montreal and Vienna. Group Decision and Negotiation, 12(4). Munduate, L., et al. (1999). Patterns of Styles in Conflict Management and Effectiviness. Int. Journal of Conflict Management, 10(1). Perkins, W., et al. (1996). Can a negotiation support system help a purchasing manager? Int. J. of Purchasing & Materials Management, 32. Putnam, L.L., & Wilson, C.E. (1982). Communication strategies in organizational conflicts: Reliability and validity of a measurement scale. In M. Burgoon (Ed.), Communication Yearbook, 6. Sage Publications. Schei, V. and J. K. Rognes (2003). Small Group Do Individualists Exploit Negotiation: Cooperators? Norwegian School of Economics and Business Admin., Report of April 11th. Sheffield, J. (1989). Conferencing Communication Media. Chi'89 Proceedings. Sheffield, J. (1989). The Effects of Bargaining Orientation and Communication Medium on Negotiations in the Bilateral Monopoly Task. ACM SIGCHI Bulletin, 20. Thomas, K.W., & Kilmann, R.H. (1974). ThomasKilmann conflict mode instrument. Tuxedo, NY: XICOM. Van de Vliert, E. and B. Kabanoff (1990). Toward Theory-Based Measures of Conflict Management. The Academy of Management, 33(1). Walton, R.E. (1969). Interpersonal peacemaking: Confrontations and third party consultation. Reading, MA: Addison-Wesley.

It was observed that the group with the highest average and median (Group 6) is primarily characterized by the collaborating style. As well, all 3 groups with a high weight of the collaborating style (Groups 2, 4, and 6) are the groups with the highest average and median values of IOD. An interesting observation is that when the collaborative style was correlated with the combined results of the groups of the clusters and not the individual results, a statistically significant correlation was observed. This shows, first, the usefulness of the clustering process and, second, that combinations of styles are a better predictor of negotiation outcomes than a single style. 5. CONCLUSIONS In this paper, the effect of five negotiation styles, as defined in the Thomas-Kilmann model, on quantitative and qualitative e-negotiation outcomes was studied. A new variable, Individual Outcome Difference (IOD), was introduced. Correlation, regression and cluster analysis were performed. It was observed that the participants with a greater avoiding style in their profile negotiated on average a lower deal. The participants with compromising style in their profile were more likely seen as pushovers, more likely to see themselves as informative, persuasive, honest, and exploitive than other styles. On the other hand, the participants with the collaborating style in their profile were more likely seen as rigid. Those with the competing style were less likely to see themselves as persuasive, exploitive, and cooperative. Finally, using cluster analysis, it was observed that the participants with the collaborating style were more likely to negotiate a better agreement than their opponents. This study is the beginning to the understanding of the impact of the five negotiation styles on negotiation outcomes. Further research is needed to study situations when the Thomas-Kilmann profile of

306