Virtual Team Interaction Styles - CiteSeerX

20 downloads 11943 Views 253KB Size Report
technologies (e.g., email and groupware) make them a very relevant subject for information .... text messages to each other on a threaded discussion web on a web site dedicated to that team. .... Group communication was accomplished using an Active Server Page (ASP) ...... International Consumer Marketing, 12(4). Potter ...
Virtual Team Interaction Styles: Assessment and Effects (Running Title: Virtual Team Interaction Styles) Richard E. Potter* Department of Information and Decision Sciences (MC 294) College of Business Administration University of Illinois at Chicago 601 South Morgan Street Chicago, IL 60607-7124 Email: [email protected] Phone: (312) 996-5360 Fax (312) 413-0385 Pierre A. Balthazard School of Management Arizona State University West 4701 W. Thunderbird Road Phoenix, AZ 85306-4908 Email: [email protected] Phone: (602) 543-6120 Fax: (602) 543-6220

*Corresponding author

Virtual Team Interaction Styles: Assessment and Effects Richard E. Potter Department of Information and Decision Sciences (MC 294) College of Business Administration University of Illinois at Chicago 601 South Morgan Street Chicago, IL 60607-7124 Email: [email protected] Phone: (312) 996-5360 Fax (312) 413-0385 Pierre A. Balthazard School of Management Arizona State University West 4701 W. Thunderbird Road Phoenix, AZ 85306-4908 Email: [email protected] Phone: (602) 543-6120 Fax: (602) 543-6220

Summary The virtual team is an increasingly common strategic work unit of many organizations. The virtual team, via various computer-based media (e.g., email, groupware) and noncomputer-based media (e.g., telephone), can interact and collaborate though separated by distance and time. One approach to their study is determining whether factors that drive conventional team performance also exist in the virtual environment. Interaction style has been shown to have a great effect on conventional teams' ability to achieve solution quality and solution acceptance on collaborative decision tasks (Hirokawa, 1985; Hirokawa and Gouran, 1989; Watson and Michaelsen 1988; Cooke and Szumal, 1994). Group interaction styles affect communication and thus team performance by facilitating

2

or hindering the exchange of information among group members. These styles reflect an aggregation of behavioral traits of individual team members, rooted in their individual personalities. The interaction style of conventional teams can be reliably assessed, and from that assessment, performance on collaborative decision tasks can be predicted. This study investigated whether or not virtual teams who collaborate via computer-mediated communication also exhibit similar interaction styles, and whether the styles have the same effects on their decision performance and process outcomes as they do with conventional teams. Members of 42 virtual teams completed an intellective decision first individually and then collaboratively. Post task measures captured individual and team performance data (e.g., solution quality) as well as process perceptions (individual acceptance of the team solution). An additional post task tool was able to accurately capture the teams’ interaction style. Results show that the interaction styles of virtual teams affect both performance and process outcomes in ways that are directionally consistent with those exhibited by conventional face-to-face teams. Implications include recommending the methodology for virtual team management, and suggestions for future research are offered.

INTRODUCTION The team work unit, advanced telecommunications and computer network technologies, and a hypercompetitive business environment have been the catalysts for a new organizational forms—the virtual team (Jarvenpaa and Ives, 1994). Virtual team members have distinct complimentary areas of expertise and are geographically and often temporally distributed, possibly anywhere within (and beyond) their parent organization

3

(Lipnack and Stamps, 1997; Townsend et al., 1998; Duarte and Snyder, 1999). The virtual team, via telephone, email, FAX, teleconferencing, videoconferencing, and some new emerging technologies (e.g., instant messaging) can interact and collaborate though separated by distance and time. This ability gives organizations increased flexibility and responsiveness, permitting them to rapidly assemble dispersed and disparate experts into a virtual team that can work on an urgent project. These advantages drive the increasing use of virtual teams. The fact that much of their interaction is via information technologies (e.g., email and groupware) make them a very relevant subject for information systems research. One approach to studying virtual teams is determining whether factors that drive aspects of conventional team performance also exist in the virtual environment. Interaction style has been shown to have a great effect on conventional teams' ability to achieve solution quality and solution acceptance on collaborative decision tasks (Hirokawa, 1985; Hirokawa and Gouran, 1989; Watson and Michaelsen 1988; Cooke and Szumal, 1994). Group1 interaction styles affect communication and thus team performance by facilitating or hindering the exchange of information among group members. These styles reflect an aggregation of behavioral traits of individual team members, rooted in their individual personalities. The interaction style of conventional teams can be reliably assessed, and from that assessment, certain types of objective decision performance and process outcomes on collaborative decision tasks can be predicted. The present study seeks to determine whether or not virtual teams who collaborate via text-based computer-mediated communication also exhibit similar interaction styles, and whether the styles have the same effects on their performance and

4

process outcomes as they do with conventional teams. Beyond answering this interesting research question, the practical result of this research is to introduce an web-based methodology that virtual team managers can use to assess how a potential virtual team is going to interact, and by extension, how they will share information, and how well they are likely to perform those team functions that require collaborative problem-solving and decision making. The methodology can also be used to diagnose communication-based difficulties that underperforming virtual teams may be having. We offer a brief review of research on interaction styles in conventional face-toface (FTF) teams, and on relevant preliminary research on virtual team communication and performance. We then pose our hypotheses and present a study that 1) validates an instrument that can accurately assess interaction styles of virtual teams collaborating via the Internet, and 2), uses that instrument to answer our research questions. We present the results of that study and then we discuss these results and their implications for virtual team management, and offer suggestions for future research.

BACKGROUND Interaction Styles in Teams Communication is a fundamental behavior of conventional (FTF) teams (McIntyre, Salas, Morgan, & Glickman, 1989; Morgan, Glickman, Woodard, Blaiwes, & Salas, 1986). But team communication has inherent difficulties. Members of problemsolving teams face two types of pressures in achieving quality solutions and high solution acceptance (Maier, 1963, 1967). On the one hand, there is pressure on each member to contribute unique, and possibly controversial, information to maximize the team's

5

resources. On the other hand, members of teams tend to believe that efficient team problem solving and strong solution acceptance (aspects of satisfaction with the team process) are best achieved through conformity of opinions (e.g., Festinger, 1950; Hoffman, 1979; McGrath, 1984). The way in which a team deals with the conflicting "task" (e.g., satisfactory performance) and "maintenance" (e.g., satisfaction with the process) pressures is reflected in the team's interaction style. Watson and Michaelsen (1988) showed that a team's interaction style affects performance. They identified positive and negative behaviors as components of group interaction style. Three groups of behaviors (expectations of performance and integration, leadership, and cohesiveness) contributed to team performance on an intellective task while one group of negative behaviors (e.g., noninvolvement, withholding of information) detracted. Building on the Watson and Michaelsen typology and others (e.g., Maier, 1967; Hoffman, 1979), Cooke and Szumal (1994) showed that group interaction, aggregated from stable personality factors of the individual group members, can be categorized as constructive, passive, and aggressive styles. The constructive style is characterized by a balanced concern for personal and group outcomes, cooperation, creativity, free exchange of information, and respect for others’ perspectives. The constructive style enables group members to fulfill both needs for personal achievement as well as needs for affiliation. The passive style places greater emphasis on fulfillment of affiliation goals only, maintaining harmony in the group, and limiting information sharing, questioning and impartiality. The aggressive style places greater emphasis on personal achievement needs, with personal ambitions placed above concern for group outcome. Aggressive groups are characterized by competition, criticism, interruptions, and overt impatience.

6

Group interaction style is theorized to affect performance because it can impede or enhance team members' ability to bring their unique knowledge and skills to bear on the task, and the extent to which they develop and consider alternative strategies for approaching the task (Hackman & Morris, 1975). This is particularly critical for groups with heterogeneous levels of expertise, as communication by most expert group members is positively correlated with group performance. Zalesny (1990) found that the most accurate member in interacting groups did not influence performance unless he or she was assertive and confident. Bottger (1984) also found that amount of communication time and expertise were positively correlated with performance, though only with highperforming groups. In their study of estimation methods for individual/team performance differences, Cooke and Kernaghan (1987) found that average individual scores explain an average of 57% of the variance in team scores. They also noted that the expertise of the best member contributes significantly to the team score, above and beyond the average individual score, with both factors together explaining an average 69% of the variance in team score performance. Group performance has usually been found to be inferior to that of the best individual, and typically, groups perform better than the average of their individual members and worse than their best individual member (Burleson, Levine, & Samter,1984; Hill, 1982; Libby, Trotman, and Zimmer, 1987; Yetton and Bottger, 1982). Cooke & Szumal (1994) demonstrated that groups whose interactions are characterized by a dominant style achieve different levels and patterns of performance and process outcomes on intellective decision tasks that require information sharing. Specifically, predominantly constructive groups produce solutions that are superior in quality to those produced by passive groups and superior in acceptance to those produced

7

by either passive or aggressive groups. Groups with predominantly passive styles produce solutions that are inferior in quality to those of constructive (and sometimes aggressive) groups and inferior in acceptance to those of constructive groups. Similarly, groups with predominantly aggressive styles produce solutions that are not as consistently of high quality as those generated by constructive groups but not as consistently of low quality as those produced by passive groups. The solutions produced by aggressive groups generate less overall acceptance than those developed by constructive groups and about the same level of acceptance as those generated by passive groups. Cooke and Szumal (1993) presented a validation of a methodology that can reliably assess interaction styles in FTF teams. The basic methodology (described in greater detail below) places participants in a crisis survival scenario and has them individually solve an intellective decision making problem related to the scenario. They then work collaboratively to create a consensus group solution. Teams with constructive interaction styles will exchange sufficient information so as to formulate a synergistic group solution to the problem that is superior (i.e., is closer to an expert’s solution) to those generated by any individual member. At this point both individual solution quality and team solution quality can be assessed, as well as process outcomes such as individual acceptance of the final team solution. Following this part of the exercise, subjects individually complete a self-report survey that solicits post task perceptions of team interaction (we give a more complete description later). From this posttask questionnaire, the group’s interaction style can be derived, as well as subjective measures of the process, such as solution acceptance.

8

Assessing Interaction Styles in the Virtual World Although in real life virtual teams may communicate via a number of media, including text-based computer messages (e.g., email), in FTF meetings, by telephone, FAX, teleconferencing, videoconferencing, and instant messaging, those teams typically used in laboratory research (such as that mentioned immediately below) communicated and worked exclusively via text-based computer-mediated communication (CMC). Pickering and King (1995) define CMC as “person- to-person communication, often in text or graphic form, over computer networks.” Building on research that examined information exchange in FTF teams (e.g., Stasser and Titus, 1985), Hightower and Sayeed (1996) found information exchange to be positively linked to virtual team performance on an intellective decision task. Tan et al. (2000) found information exchange positively related to virtual team performance on a preference task. Warkentin, Sayeed, and Hightower (1997) found that perceptions of shared norms and expectations of task process were types of relational links positively related to a higher level of team cohesion and information exchange in virtual teams. Mennecke and Valacich (1998) also found information sharing to be positively related to decision quality for groups (using text-based CMC tool) whose members had unique information.

In sum, information

exchange via text-based CMC appears to have similar effects on performance of some tasks (i.e., intellective collaborative decision making) and similar effects on process outcomes of some tasks in virtual teams as it does in FTF teams. This is indirect evidence that the virtual teams in these studies exhibited interaction styles that gave rise to varying levels of information exchange.

9

The present study used the Cooke and Szumal (1993) tools and methodology (described in greater detail below) converted to web-enabled, text-based versions. As with the virtual team research discussed above, for the interactive collaborative portion of the present study, we restricted our participants to CMC (text-based messaging via PCs connected to the Internet). Although other portions of the exercise required the participants to work individually (e.g., completing web-based questionnaires), this portion of the task required collaborative participation, and had team members posting text messages to each other on a threaded discussion web on a web site dedicated to that team. Although we will address the issue in greater depth elsewhere in this paper, we posit here that on balance, we do not believe that team interaction via CMC will significantly interfere with the expression and perception of individual interaction characteristics, nor do we believe that this medium will significantly interfere with team members' ability to accurately assess their team's interaction style. We also have no apriori reason to believe that the interaction styles of virtual teams should have effects on specific objective decision performance outcomes that are not directionally consistent with those exhibited by FTF teams. H1

The interaction style of virtual teams will predict objective measures of team performance. H1a

A constructive group interaction style will negatively predict team error and positively predict gain and synergy. (Error, gain, and synergy are defined in the measures segment of the method section below).

10

H1b

A passive group interaction style will positively predict team error and negatively predict gain and synergy.

H1c

An aggressive group interaction style will not predict team error but will negatively predict gain and synergy (e.g., aggressive personalities may have the best solution).

As noted above, Tan et al. (2000) found information exchange positively related to higher levels of cohesion, collaboration, and satisfaction with decision quality in virtual teams who used their dialogue technique than for those who did not. Solution acceptance and satisfaction with the decision process are functions of the perceptions of process quality and fairness, and are characteristic of teams with a constructive interaction style. While satisfaction with task and amount of time needed by teams to reach a decision are relatively common dependent variables, perceived team efficiency (i.e., the perceived use of time) is not (Fjermestad and Hiltz, 1999). We speculate that perceived efficiency is tautologically related to both satisfaction with performance outcomes and process outcomes. We argue that the medium should have little detrimental effect on the ability of members to discern the group's processes and products. We offer the following hypotheses on process outcomes: H2

The interaction style of virtual teams will predict process measures of team performance. H2a

A constructive group interaction style will positively predict member acceptance of the group’s decision, satisfaction with the process, and the members' perception of time efficiency.

11

H2b

A passive group interaction style will negatively predict member acceptance of the group’s decision, satisfaction with the process, and the member's perception of time efficiency.

H2c

An aggressive group interaction style will negatively predict member acceptance of the group’s decision, satisfaction with the process, and the member's perception of time efficiency.

METHOD Participants Group style, performance and process outcomes data were collected from 186 members of 42 virtual teams who completed the Internet version of the "Desert Survival Situation" (Lafferty & Pond, 1974; Balthazard, 1999), a structured problem-solving exercise used for management development and team building in classroom and corporate settings. Subjects were low- and mid-level managers participating in corporate organizational training and development programs, and were members of an executive MBA class on information technology and strategy.. The median number of participants in each problem-solving virtual team was five and included 7 three-member teams, 12 four-member teams, 21 five-member teams, and 2 six-member teams. Data from seven of 186 original participants was incomplete and not included in the individual-level data set for analysis. The age of the participants ranged from 20 years old to 52 years old with more than half of the sample (55%) in the 20-29 years range. Slightly more than 59% of the participants were male and most were White or Caucasian (approximately 78%). The participants reported some college education

12

(46%), a college degree (43%), or a graduate degree (11%). The exercise was performed for professional credit in the organization's continual education system. Given their background and/or academic standing, all participants were assumed to be highly computer literate. The exercise was conducted approximately midway through the semester. The subjects had been exposed to a great amount of material on IT support of business processes including distributed work environments. Before the exercise, many participants expressed their interest in the general topic of management of virtual teams, and many of them regularly participated in this type of work in their professional lives. We also received many positive comments after the exercise. In addition, subjects were told that we would review all the transcripts of their team’s interaction. We cannot guarantee that all subjects performed to the best of their ability, but given their maturity, their professional interest in the subject, the requirements of the course, and the fact that we reviewed all the text that they generated, we are confident that they were enthusiastically engaged and executed the task very well. Task and Technology The "Desert Survival Situation" (Balthazard, 1999) places teams in a desolate region of the Sonoran Desert in Southern Arizona, USA, in the middle of summer (where their chartered plane has crashed) and challenges them to correctly rank fifteen items they have salvaged in order of their importance to the team's survival. The simulation provides participants with an opportunity to practice their skills in both situational analysis and group decision-making. Szumal (2000) describes this type of problem as a content-free simulation that is likely outside the sphere of expertise within the group but designed to direct attention to overall team problem solving processes and skills.

13

Solutions to the exercise are developed first on an individual basis and then as a group. Individual and team solutions are then compared to an expert's solution. Comparisons between individual solutions and the expert's solution indicate how well participants are exercising their knowledge, experience, and skills with respect to situational analysis and complex problem solving. Comparisons between participants’ individual scores and their team’s score indicate whether they were able to achieve synergy by fully using and building on their collective knowledge and skills (see Cooke & Kernaghan, 1987). In other words, the team's score should be better than any individual score if group synergy were achieved. The task is in its first phase an individual task, and then in it second phase it is primarily an interdependent intellective task (McGrath, 1984; Argote and McGrath, 1993), one which has a single correct answer, and which requires collaboration, coordination, and conflict resolution from its participants. The interdependency — entailing persuasion so that the participants can reach consensus— is also one characteristic of judgment tasks (McGrath, 1984). All materials and mechanisms for group collaboration, including a comprehensive description of the group decision task, pre- and post-task questionnaires, and the tools for group communication were all made available on the Internet for participants to access using standard browsers. The textual descriptions of the desert survival task were enhanced by links to digitized streaming video clips available ondemand. Group communication was accomplished using an Active Server Page (ASP) threaded-discussion system that allowed team members to post new threads, reply to current messages, or search the team's written record of the discussion. Messages were posted using a "most recent at the top" method with replies creating clear hierarchies

14

under each thread. In the web interface, the left-hand side window displayed a scrollable roadmap of the discussion —indicating for each contribution a subject label, the name of the contributor, and date/time of the contribution. The right-hand window was used for participants to view selected messages or to create their own contribution. Each team was given access to its own private (password protected) web discussion. Although the system supported semi-synchronous written communication, the interface was significantly different than that of a chat room: it essentially provided more structure to the discussion and the possibility of multiple and non-sequential sub-discussions.

Procedure In an initial meeting, participants were introduced to the Web-based system. They were randomly assigned to a virtual team and asked to provide a team name and select a team secretary (a participant with the responsibility of providing the group consensus solution for the problem, with no implied leadership role). As individuals (with no interaction yet permitted), they were provided with the URL of the home page for the simulation and asked to complete/submit a pre-task “registration” questionnaire requesting biographical type information (name, team name, age, gender, education, and previous experience with technology). They were then directed to peruse for 5 minutes a different Web page that described the decision task. Each participant was then given 10 minutes to rank the 15 items on an individual basis and submit a personal solution for processing by the web system. The participants used their computers to access these materials and to complete this portion of the exercise. Though collocated, they did not

15

communicate with each other at this time, either verbally (as they were collocated) or via computer. Each team was then given the URL of a password protected threaded discussion web and given 7 consecutive days to interact (exclusively using the threaded discussion) to produce a group solution on a consensus basis—that is, a solution that all members could "live with." Team members could access the discussion web from any location (with a computer, browser, and Internet connection) and at any time but could not be collocated with any other team member. Participants were prohibited from interacting with other team members via any other communication medium during this portion of the task. Team leaders were not formally designated and, therefore, any leadership roles assumed by members emerged informally. Upon achieving a consensus solution, the virtual team's secretary submitted the consensus group ranking and each member independently completed a group process questionnaire (assessing group efficiency, effectiveness, satisfaction, and buy-in) and a questionnaire that assessed group interaction style. The post-task questionnaires were answered after ranking the items as a group but before receiving feedback on the "experts' ranks" or the quality of their own solution.

Measures Group Interaction Styles To assess a group's interaction style, participants answered 33 questions that focused on the ways in which members of a group might interact with one another and approach their task during a meeting or specific problem-solving session. The items were a subset of the Group Styles Inventory© described in Cooke and Szumal (1994).

16

The items assess three distinct, yet interrelated, group interaction styles introduced earlier—constructive (12 items), passive (9 items), and aggressive (12 items). Each member's scores along each of the three interaction styles are calculated by averaging his or her responses to the items composing each of the respective scales. These items describe specific collective behaviors that might characterize a group to a very great extent (response option 4) or, at the other extreme, not at all (response option 0). For our present data set, the Cronbach alpha coefficients for the three style measures are .90 for constructive, .83 for passive, and .89 for aggressive, indicating that the items composing each scale were answered in a fairly consistent manner by respondents. The interaction styles, although distinct, are expected to be interrelated. For our sample, analysis of the individual-level data shows that the constructive scale correlates negatively and significantly (p < .01) with the passive (r = -.45) and aggressive (r = .22) scales. Further, the aggressive scale correlates positively and significantly (p < .01) with the passive scale (r = .70). Constructive interaction styles appear to suppress the passive and aggressive styles, and the latter styles appear to reinforce one another —the aggressive behaviors of some beget passive behaviors from others. -----------------------------Insert Table 1 about here --------------------------------Although these correlations do not exclude the possibility that the styles are empirically distinct, they suggest that the scale scores cannot be relied on to separate the individual and simultaneous effects of the styles (Darlington, 1990). A principal components analysis with varimax rotation was used to facilitate the source of these

17

effects. By examining responses to the 33 items, the factor analysis should recognize orthogonal factors representing constructive, passive, and aggressive interaction styles. The Kaiser-Meyer-Olkin measure of sampling adequacy was .88, suggesting that the approach is appropriate. Initial statistics showed that 14 factors with eigenvalues greater than 1.0 could be extracted. However, the scree plot suggested that three factors be retained for the varimax rotation, altogether representing 50.29% of the variance in the items. The first factor extracted was clearly the aggressive style with 100% of the 12 aggressive items showing their highest loadings —all above .45— on this factor (see Table 1 for a listing of items and factor loadings). In addition, three passive items loaded positively on the aggressive factor. The second factor represented the constructive style with all of its 12 constructive items loading above .54 on it. Further, none of the passive or aggressive items loaded on it. The third factor reflected the passive style with 6 of 9 items showing their highest loadings on it. Two of the three misbehaving passive items (members overly concerned for gaining full and unanimous support for every decision, members showing too much indecision) loaded on the aggressive scale whereas the third (ideas too readily accepted) showed mixed loadings. In general, the great majority of items (30 of 33 items) loaded onto the proper factor and the incorrect loadings of the 3 passive items make sense. For example, it would seem reasonable that aggressive group members would push their solution to be accepted by everyone; they might act indecisive or refuse to go along with any other alternative. Since factor scores from this analysis are appropriate to assess the independent effects of the three interaction styles, interaction style within each team was computed by

18

averaging the factor scores of individual members. The justification for aggregation at the group level for all our measures is discussed below.

Objective Measures of Performance Three measures of solution quality were derived for each team. The first, "team error," was calculated by summing the absolute values of the numerical differences between the team consensus rank for each item and the rank suggested by the desert survival expert (Lafferty & Pond, 1974; McGrath, 1984). Since it is an error score, a lower score (ideally 0) indicates better performance. Second, "gain" was calculated by (a) deriving an error score for each participant, (b) averaging this score across members of each group to obtain an average individual score, and then (c) subtracting the team error score from the average of individual members' scores. This set of calculations produces a score that reflects the gain (or loss) in solution quality achieved by the group relative to the average quality of members' initial, independently derived, solutions (Cooke & Kernaghan, 1987). Third, "synergy" was operationalized by subtracting the lowest of its individual members' error scores from the team's error score (see Szumal, 2000; Straus, 1996). A synergistic team should perform beyond the capabilities of its "best" member (prior to his or her interaction with the team). Low team error scores along with significant gain and synergy reflect effective performance; in contrast, high team error scores with little gain (and often loss) and no synergy reflect poor team performance and low solution quality.

Measures of Process Performance

19

Four measures of process performance were derived for each team: "Solution Acceptance," "Satisfaction," "Group Commitment," and "Perceived Efficiency." Member acceptance of the group's decision (solution acceptance) was measured by three supplementary questions included in the post-task questionnaire. Respondents were asked to report the extent to which they: (1) were personally committed to the solution proposed by the team; (2) thought the solution generated by the group was better than the one they developed; (3) felt that the solution had been reached on a consensus basis. The questions were adapted from the work of Cooke and Lafferty (1988). Responses to each of these items, which ranged from 1 = not at all to 5 = to a very great extent, were averaged for each team member (alpha=.74). High scores on this scale therefore reflect a high degree of solution acceptance. The overall level of member acceptance of the group's decision (solution acceptance) within each team then was computed by averaging the scale scores of individual members. Satisfaction with the process was assessed by two questions included in the posttask questionnaire. Respondents were asked to report the extent to which: (1) members of the group worked together effectively; (2) the group came up with the best solution possible, given time and geography constraints. The questions were again adapted from the work of Cooke and Lafferty (1988). Responses to each of these items, which ranged from 1 = not at all to 5 = to a very great extent, were averaged for each team member (alpha=.73). High scores on this scale therefore reflect a high degree of satisfaction with the process. The overall level

20

of satisfaction within each team then was computed by averaging the scale scores of individual members. Perceived efficiency with the process were each ascertained using one question. Also from the work of Cooke and Lafferty (1988), the item "to what extent did the group seem to waste time and energy?" was used to assess members' perception of the efficiency of the process. Responses to this item ranged from 1 = not at all to 5 = to a very great extent. We reversed the scoring of this item to retain directional consistency with the other measures. Therefore, high scores reflect a perception of a high degree of efficiency with the process. The overall level of perceived efficiency within each team then was computed by averaging the scores of individual members.

Inter-Rater Reliability and Justification for Aggregation Justification for aggregating member's reports on the various scales is summarized in Table 2. Inter-rater reliability and agreement was assessed for appropriate measures by means of the eta-squared statistic (η2), a series of one-way analyses of variance (ANOVAs with group membership as the independent variable and the measure to be aggregated as the dependent variable), and tests based on the multiple-item estimator rwg(j) for scales with moderately skewed distributions (see James, Demaree, & Wolf, 1984, 1993; Lindell & Brandt, 1999; Lindell, Brandt, & Whitney, 1999). -----------------------------Insert Table 2 about here ---------------------------------

21

The η2 statistics indicate that group membership explained 42% of the variance in individual responses to the constructive measure, 62% of the variance for the passive and aggressive measures, 46% of the variance for the solution acceptance measure, 55% of the variance for the satisfaction measure, 48% of the variance for the group commitment measure, and 57% of the variance for the perceive efficiency measure. Similarly, the F ratios suggest that the variance in responses between groups is significant in relation to the total variance for each measure (all significant). The η2 and F ratios therefore support the proportional consistency of variance (Kozlowski & Hattrup, 1992) among the responses of members within the same group as compared to the responses of members across groups (i.e., inter-rater reliability). The multiple-item estimator rwg(j), on the other hand, was used to assess inter-rater consensus or the interchangeability among different members' responses within each group to the items associated with each scale. As a measure of convergence among a group of raters, this estimator is particularly relevant to instruments designed to measure group or organizational-level variables on the basis of individual members' reports (Kozlowski & Hattrup, 1992). For each measure, the amount of variance in members' responses within each group was compared to the expected amount of variance based on moderately skewed distributions (which have been observed for the scales in larger face-to-face group samples). The median rwg(j) estimates (across the 42 groups) indicate that agreement within groups is fairly high and that the amount of within-group variance is small relative to the expected amount of variance in responses. While median coefficients of .70 are considered acceptable (e.g., George, 1990), the estimates for our scales are in most cases significantly larger, with the weakest coefficient (Group

22

Commitment) achieving the .70 threshold. The rwg(j) estimates of inter-rater consistency and consensus, along with the η2 and F statistics, support the statistical aggregation of individual responses to the group level for our analyses. Thus, all analyses beyond the factor analysis were performed at the group level.

RESULTS Correlational Analysis Zero-order correlations were computed among the different measures at the group level of analysis. The correlations provide an indication of the direction and magnitude of the relationship between each of the group interaction styles and the objective and process measures of performance. One-tailed t tests were used to determine whether the correlations were likely to be significantly different from zero in the population. -------------------------------Insert Table 3 about here --------------------------------The predicted negative relationship between the constructive style and team error and the positive relationships between the constructive style and gain and synergy (H1a) are partially supported by the pattern of correlations. Although synergy is the only relationship achieving an acceptable level of confidence, it is the most difficult objective measure of performance to attain in practice. Further, the relationships with team error and gain are directionally consistent with the hypothesis (especially in comparison to the relationships involving the passive and aggressive styles). The

23

predicted positive relationships between the constructive style and the process measures of performance (H2a) are all strongly supported. The predicted positive relationship between the passive style and team error and the negative relationships between the passive style and gain and synergy (H1b) are fully supported by the pattern of correlations. The predicted negative relationships between the passive style and the process measures of performance (H2b) are all strongly supported. As predicted, the aggressive style is not related to team errors, and the pattern of correlations of its relationship with gain and synergy (H1c) suggests a negative association. The predicted negative relationships between the aggressive style and the process measures of performance (H2c) are all supported.

Analysis of Variance Supplementary analyses were carried out to provide a more descriptive and somewhat more qualitative picture of the differences between predominantly constructive, passive, aggressive, and mixed-style teams. To do so, differences along the three interaction style factor scores were used to identify and compare teams whose interactions were characterized by a single predominant style. At the group level of analysis, each team's passive and aggressive factor scores were added together and then subtracted from their constructive score and the 10 (of 42) groups with the largest residual scores were deemed predominantly constructive. Similarly, passive and constructive interaction style factor scores were added together and then subtracted from aggressive scores to identify the 10 most aggressive groups, and aggressive and constructive scores were added together and then subtracted from passive scores to

24

identify the 10 most passive teams. This procedure led to 30 teams being classified as either predominantly constructive, passive, or aggressive, and 12 groups classified as mixed with respect to interaction style. -------------------------------Insert Table 4 about here --------------------------------The average constructive, passive, and aggressive factor scores for each of the four types of teams were computed and compared to check the effectiveness of the classification procedures. As shown in Table 4, the four sets of groups differed significantly (all p < .01) in their average scores along each of the three factors. Consistent with the intent of the statistical manipulation, the mostly constructive teams show the highest factor scores along the constructive style; similarly, the mostly aggressive and mostly passive teams show highest factor scores along the appropriate styles. Mean scores for these four sets of teams were computed for the objective and process measures of performance. F statistics from a series of ANOVAs by type of group and one-tailed t tests from a series of planned comparisons then were used to identify significant differences between predominantly constructive, aggressive, and passive virtual teams in accordance with the hypothesized effects. They are presented in Table 5. -------------------------------Insert Table 5 about here --------------------------------Discussion and Conclusion

25

Our results offer a number of important insights on virtual teams. First, as with FTF teams, virtual teams exhibit constellations of communication behaviors that can be perceived and aggregated into an interaction style. Second, the effects of virtual team interaction styles on a number of objective performance and group process outcomes are very similar to those exhibited by conventional teams. Although we present a detailed explanation of the validation of our web-based version of the Cook and Szumal (1993, 1994) tools and methodology, one study is not adequate to firmly establish that these communication phenomena exist and manifest as we believe they do. This issue rests in part on the question of whether the communication medium used here was sufficiently rich to permit accurate expression and reception of the communication. That is, is there something about textual communication that diminishes the expression or reception of the tone or emotional content of a virtual team communication that would not occur in FTF verbal (largely) communication? As CMC restricts the transmissions of eye contact and many other contextual and social cues (such as facial expressions, posture, and other visual and behavioral attributes of the recipient), it may reduce the salience of the recipient's involvement (social presence) (Short et al., 1976). The reduction of social presence in CMC appears to reduce social inhibitions in communications, and increase the voicing of more radical opinions, equality of participation, and a reduction of status differences between members (Kiesler & Sproul, 1992; Dubrovsky et al. 1991). The effects of media on the emotional or social substance of CMC also appear to be related to the relative difficulty of expression via that medium compared to face-to-face. CMC groups communicate less frequently than FTF groups (Hiltz, Johnson, & Turoff, 1986; Siegel, Dubrovsky, Kiesler,

26

& McGuire, 1986). This information suppression (Hollingshead, 1996a, 1996b) may be compensated for by the expression of more polarized or diverse opinions and a higher proportion of task-related messages. Hiltz and Turoff (1993) have shown that some forms of communication that are suppressed in a particular medium (e.g., facial expression cues in e-mail) can replaced with alternative expressions of the same message appropriate to the media. Walther (1992, 1994, 1996) has argued that the "cues filtered out" stance inherent in social presence theory cannot adequately explain many of the conflicting results in CMC research. Rather, he contends that CMC users can and do develop individuating impressions of others (e.g., virtual team members) through accumulated communication. Based on these impressions—both actively and passively promulgated—members initiate relational communication (i.e., more social/emotional in character, rather than simply task oriented, which serves to define relationships between the actors involved). He also argues that the impersonal nature of CMC is more likely and more appropriate for task oriented communication that constitutes the majority of communication for problem solving groups (particularly the one-time-only, time-limited, which typify many virtual teams). Socioemotional or relational content can be expressed in the context of taskoriented communication (e.g., the criticism of a member's suggestion of task process). With some groups and tasks, socioemotional or relational content may become prevalent only if the task and communication continue beyond an artificially brief boundary period (as sometimes used in laboratory experiments on CMC or GDSS) (Rice and Love, 1987). In one study, Walther (1995) noted that perceptions of immediacy/affection were actually stronger and emerged sooner in CMC problem-solving groups compared to FTF groups.

27

Kahai and Cooper (1999) also noted that socioemotional content was higher in their CMC groups than FTF groups, and also found that both positive and negative socioemotional content were positively associated with task-oriented communication. Weisband and Atwater (1999) investigated the relationship of member liking to ratings of others' contribution in face-to-face and virtual groups working on a decision task. They found that ratings of liking contributed less bias to evaluations of contribution for virtual groups than face-to-face groups. Hedlund et al. (1998) found that leaders of CMC decision-making teams were better able to differentiate members on the basis of the quality of their decisions compared to leaders of FTF teams. These findings suggest that assessment of member contribution may be more accurate (or objective) in an environment where visual cues are restricted. To summarize, findings show that although CMC groups may communicate less frequently, they can compensate in various ways. The computer-mediated communication medium may actually be superior to face-to-face communication for objective and accurate evaluation of others' input into teamwork. There also seems to be no fundamental reduction of the human tendency to promulgate relational or socioemotional communication. Nor does extant research suggest that people have difficulty interpreting the emotional tone or other manifestations of personality that are expressed in CMC. Our results are consistent with these findings. Although it may be coincidence, the similarity of our results regarding the performance and process outcomes of virtual teams to those of FTF teams suggests (though cannot confirm) that the CMC medium used here was sufficiently rich to permit expression and reception of

28

communication that was not significantly diminished compared to that displayed commonly by FTF teams. One limitation of the present study is that of our teams. Although our subjects interacted with each other for an entire semester, they were formed into interdependent teams for only the relatively brief duration of our task. A second limitation of the present study

concerns culture. Our research setting was the United States, with a largely male Caucasian sample. Other cultural groups may indeed hold different norms concerning communication behavior via CMC and other aspects of group interaction (Potter and Balthazard, 1999, 2000). A third limitation is that of our specific task and technology. These limitations suggest that further research in this area is needed. One direction is to develop additional theoretical perspectives on virtual team interaction, and a second is to develop additional methodologies and research approaches. Third, there are many opportunities to increase the diversity of the type of task and participant beyond what was employed here. Other cultures may exhibit significantly different group interaction behaviors, and these can include different national cultures, different organizational or professional cultures, or teams of mixed status members. McGrath (1984) and others have identified numerous types of group tasks and each may or may not elicit group interaction styles, with similar or dissimilar effects. Our technology did not include real time audio such as conference calling that real virtual teams sometimes use. We can speculate that this type of channel would permit greater latitude of expression (vocal tone and inflection, for example), likely leading to improved message reception, and to stronger effects of the resulting interaction style on performance and process outcomes. On the other hand, particularly for distributed virtual teams operating

29

in different time zones, asynchronous CMC such as email and the technology used in the present study may be the more popular. The more practical result of the present study is to demonstrate a methodology that virtual team managers can use to assess how a potential virtual team is going to interact, and by extension, how they will share information, and how well they are likely to perform those team functions that require collaborative problem-solving and decision making. The methodology can also be used to diagnose communication-based difficulties that underperforming virtual teams may be having. Given the increasing reliance on virtual teams, we believe that this diagnostic approach will become very important. It also suggests another ripe area for research: If the reality of virtual team life is like that of conventional interdependent teams that face decision tasks, the majority of them will exhibit nonconstructive interaction styles with accompanying performance problems. Beyond diagnosis, how do we improve them? Can the technology shape constructive interaction and diminish nonconstructive interaction? These will surely become important challenges for the interested researcher. NOTES 1. The terms "team" and "group" are used interchangeably in this paper, although they are not strictly synonymous. Hollenbeck et al. (1997) consider groups to be configurations of two or more interdependent individuals who interact over time, and teams to be special cases of groups, whose members incorporate skill differentiation and share a common fate (i.e., similar consequences for all members depending on success or failure at the team level). Brannick and Prince (1997) also distinguish teams from groups by their members having distinct and

30

noninterchangeable functions.

Our subjects met some of these definitional

requirements of teams, but not others. REFERENCES Argote, L., and McGrath (1993). Group processes in organizations: Continuity and Change. In C. L. Cooper and I. T. Robertson (Eds.) International Review of Industrial and Organizational Psychology. (pp.333-389). New York: Wiley. Balthazard, P. A. (1999). Virtual version of the Desert Survival Situation by J. C. Lafferty.and A. W. Pond. .Arlington Heights IL: Human Synergistics/Center for Applied Research. Bottger, P. (1984). Expertise and air time as bases of actual and perceived influence in problem-solving groups. Journal of Applied Psychology, 69, 214-221. Brannick, J. T., & Prince, C. (1997). An overview of team performance measurement. In M. T. Brannick, E. Salas, & C. Price (eds.), Team Performance Assessment and Measurement. Mahwah, NJ: Erlbaum Associates. Burleson, B. R., Levine, B. J., & Samter, W. (1984). Decison-making procedure and decision quality. Human Communication Research, 10, 557-574. Cooke, R. A., and Kernaghan, J. A. (1987). Estimating the difference between group versus individual performance on problem-solving task. Group and Organization Studies, 12(3), 319-342. Cooke, R. A., and Lafferty, J. C. (1988). Group Styles Inventory. Plymouth, MI: Human Synergistics.

31

Cooke, R. A., and Szumal, J. L. (1993). Measuring normative beliefs and shared behavioral expectations in organizations: the reliability and validity of the organizational culture inventory. Psychological Reports, 72, 1299-1330. Cooke, R. A., and Szumal, J. L. (1994). The impact of group interaction styles on problem-solving effectiveness. Journal of Applied Behavioral Science, v. 30(4), 415-437. Darlington, R. B. (1990) Regression and linear models. New York: McGraw-Hill. Duarte, D. L. and Snyder, N. T. (1999). Mastering Virtual Teams. San Francisco: Jossey-Bass. Dubrovsky, V. J., Kiesler, S., & Sethna, B. N. (1991). The equalization phenomenon: Status effects in computer-mediated and face-to-face decision making groups. Human-Computer Interaction, 6, 119-146. Fjermestad, J. and Hiltz, S. R. (1998-1999). An assessment of group support systems experimental research: Methodology and results. Journal of Management Information Systems, 15(3), 7-149. Festinger, L. (1950). Theory and Experiment in Social Communication. Ann Arbor: Research Center for Dynamics, Institute for social Research, University of Michigan. George, J. M. (1990). Personality, affect, and behavior in groups. Journal of Applied Psychology, 75, 107-116.

32

Hackman, J. R., & Morris, C. G. (1975). Group tasks, group interaction process, and group performance effectiveness: A review and proposed integration. Advances in Experimental social Psychology, 8, 45-99. Hedlund, J., Ilgen, D. R., & Hollenbeck, J. R. (1998). Decision accuracy in computermediated versus face-to-face decision-making teams. Organizational Behavior & Human Decision Performance, 76(1), 30-47. Hightower, R. T., & Sayeed, L. (1996). Effects of communication mode and prediscussion information distribution characteristics on information exchange in groups. Information systems Research, 7(4), 451-465. Hill, G. W. (1982). Group versus individual performance: Are N + 1 heads better than one? Psychological Bulletin, 91, 517-539. Hiltz, S. R., Johnson, K., & Turoff, M. (1986). Experiments in group decision making: Communication process and outcome in face-to-face versus computerized conferences. Human Communication Research, 13, 225-252. Hiltz, S. R. & Turoff, M. (1993). The Network Nation: Human Communication via Computer. Cambridge, MA: MIT Press. Hirokawa, R. (1985). Discussion procedures and decision-making performance: A test of a functional perspective. Human Communication Research, 12(2), 203-224. Hirokawa, R., & Gouran, D. S. (1989). Facilitation of group communication: A critique of prior research and an agenda for future research. Management Communication Quarterly, 3(1), 71-92. Hoffman, L. R. (1979). Applying experiemental research on group problem solving to organizations. Journal of Applied Behavioral Science, 15, 375-391.

33

Hollingshead, A. B. (1996a). Information suppression and status persistence in group decision making: the effects of communication media. Human Communication Research, 23, 193-219. Hollingshead, A. B. (1996b). The rank-order effect in group decision making. . Organizational Behavior & Human Decision Performance, 68, 181-193. James, L. R., Demaree, R. G., & Wolf, G. (1984). Estimating within-group interrater reliability with and without response bias. Journal Of Applied Psychology, 69, 85-98. James, L. R., Demaree, R. G., & Wolf, G. (1993). R(Wg) - An assessment of withingroup interrater agreement. Journal Of Applied Psychology, 78, 306-309. Jarvenpaa, S. and Ives, B. (1994) "The Global Network Organization of the Future: Information Management Opportunities and Challenges" Journal of Management Information Systems, 10,. Kahai, S. S., and Cooper, R. B. (1999). The effect of computer-mediated communication on agreement and acceptance. Journal of Management Information Systems, 16(1), 165-188. Kiesler, S., & Sproul, L. (1992). Group decision making and communication technology. Organizational Behavior and Human Decision Processes, 52(1), 96-123. Kozlowski, S. W. J., & Hattrup, K. (1992). A disagreement about within-group agreement: Disentangling issues of consistency versus consensus. Journal of Applied Psychology, 77, 161-167. Lafferty, J. C., & Pond, A. W. (1974). The desert survival situation. Plymouth, MI: Human synergistics.

34

Libby, R., Trotman, K. T., & Zimmer, I. (1987). Member variation, recognition of expertise, and group performance. Journal of Applied Psychology, 72, 81-87. Lindell, M. K., & Brandt, D. J. (1999). Assessing interrater agreement on the job relevance of a test: A comparison of the CVI, T, (rWG(J)), and r*(WG(J)) indexes. Journal Of Applied Psychology, 84, 640-647. Lindell, M. K., Brandt, D. J. & Whitney, D. J. (1999). A revised index of interrater agreement for multi-item ratings of a single target. Applied Psychological Measurement, 23, 127-135. Lipnack, J. and Stamps, J. (1997). Virtual Teams: Reaching Across Space, Time, and Organizations with Technology. New York: John Wiley & Sons. Maier, N. R. F. (1963). Problem-solving Discussions and Conferences: Leadership Methods and Skills. New York: McGraw-Hill. Maier, N. R. F. (1967). Assets and liabilities in group problem-solving: The need for an integrative function. Psychological Review, 74, 239-249. McGrath, J.E. (1984) Groups: Interaction and performance. Englewood Cliffs:PrenticeHall, Inc McIntyre, R. M., Salas, E., Morgan, B., & Glickman, A. S. (1989). Team research in the 80's: Lessons learned. (Tech Rep.). Orlando, FL: Naval Training Systems Center. Mennecke, B. E., and Valacich, J. S. (1998). Information is what you make it: The influence of group history and copmuter support on information sharing, decision quality, and member perceptions. Journal of Management Information Systems, 15(2), 173-197.

35

Morgan, B. B. Jr., Glickman, A. S., Woodard, E. A., Blaiwes, A. S., & Salas, E. (1986). Measurement of team behaviors in a Navy environment. (Tech Rep.). Orlando, FL: Naval Training Systems Center. Pickering, J. M., and King, J. L. (1995). Hardwiring weak ties. Organization Science, 6(4), July/August. Potter, R. E., and Balthazard, P. A. (1999). Supporting integrative negotiation via computer-mediated communication technologies: An empirical example with geographically dispersed Chinese and American negotiators. Journal of International Consumer Marketing, 12(4). Potter, R. E., and Balthazard, P. A. (2000). Cross-Cultural Issues in Virtual Team Support: Communication Characteristics and Task/Technology Perceptions from Mexican and U. S. Team Members. Journal of International Information Management. Rice, R. E., and Love, G. (1987). Electronic emotion: Socioemotional content in a computer-mediated communication network. Communication Research, 14(1), 85-108. Siegel, J., Dubrovsky, V., Kiesler, S., & McGuire, T. W. (1986). Group processes in computer-mediated communication. Organizational Behavior and Human Decision Processes, 37, 157-187. Short, J., F. Williams, & B. Christie (1976). The Social Psychology of Telecommunications. New York: Wiley and Sons. Stasser, G., and Titus, W. (1985). Pooling of unshared information in group decision making: Biased information sampling during groups

36

discussion. Journal of Personality and Social Psychology, 48, 14671478. Straus, S. G. (1996). Getting a clue: The effects of communication media and information distribution on participation and performance in computer-mediated and face-to-face groups. Small Group Research, 27(1), 115-142. Szumal, J. L. (2000). How to Use Group Problem Solving Simulations to Improve Teamwork. In M. Silberman (Ed.) Team and Organization Development Sourcebook . New York: McGraw Hill Tan, B.C.Y., Wei, K.-K., Huang, W.W. & Ng, G.-N. (2000). A dialog technique to enhance electronic communication in virtual teams. IEEE Transactions on Professional Communication, 43(2), 153-165. Townsend, A., DeMarie, S. and Hendrickson, A. (1998). Virtual Teams: Technology and the Workplace of the Future. Academy of Management Executive, 12. Walther, J. B. (1992). Interpersonal effects in computer-mediated interaction: A relational perspective. Communication Research, 19(1), 52-90. Walther, J.B. (1994). Anticipated ongoing interaction versus channel effects on relational communication in computer-mediated interaction. Human Communication Research, 40, 473-501. Walther, J. B. (1995). Relational aspects of computer-mediated communications: Experimental observations over time. Organization Science, 6(2), 186-203. Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal, and hyperpersonal interaction. Communication Research, 23(1), 3-43.

37

Warkentin, M. E., Sayeed, L., and Hightower, R. (1997). Virtual teams versus face-toface teams: An exploratory study of a web-based conference system. Decision Sciences, 28(4), 975-996. Watson, W. E., & Michaelsen, L. K. (1988). Group interaction behaviors that affect group performance on an intellective task. Group & Organization Studies, 13(4), 495-516. Weisband, S., & Atwater, L. (1999). Evaluating self and others in electronic and face-toface groups. Journal of Applied Psychology, 84(4), 632-639. Yetton, P. W., & Bottger, P. C. (1982). Individual versus group problem solving: An empirical test of a best-member strategy. Organizational Behavior and Human Performance, 29, 307-321. Zalesny, M. D. (1990). Rater confidence and social influence in performance appraisals. Journal of Applied Psychology, 75, 274-289.

38

TABLE 1 Principal Components Analysis of Group Interaction Style Itemsa Factor 1 Factor 2 Factor 3 Items (paraphrased) (Aggressive) (Constructive) (Passive) Constructive items Help one another 0.0232 0.7413 -0.0439 Group helpful 0.0241 0.7286 -0.1376 Thoughtful feedback -0.0784 0.7600 -0.1851 Really "listening" -0.1641 0.7034 -0.1983 Open exchange of thoughts -0.2100 0.6974 -0.1803 Cooperation apparent -0.1720 0.6741 -0.2958 Goals set -0.0912 0.5444 -0.0028 Alternatives analyzed 0.0137 0.7949 -0.1356 Focused on objectives 0.0210 0.5399 -0.2148 Creative approach 0.0953 0.6366 -0.2223 Constructive questions -0.1432 0.7121 0.0120 Problem viewed positively -0.1510 0.6175 0.0248 Aggressive items Atmosphere of conflict 0.5178 -0.0799 0.0548 Defend views 0.6378 0.2281 0.1463 Ideas negated 0.6925 -0.1689 0.1118 Points made aggressively 0.8028 0.0654 0.0839 Overconfident attitude 0.6012 -0.0501 0.1865 Influence greater than knowledge 0.6761 -0.0467 0.3686 Winning "the point" 0.7560 -0.2528 0.1434 Turned into a contest 0.7693 -0.0295 0.2029 Own ideas viewed as best 0.6611 -0.1136 0.3364 "Hung up" on details 0.4577 -0.1069 0.2359 Discussion very serious 0.7391 0.0545 0.1531 Unrealistically precise 0.6147 -0.0611 0.2242 Passive items Side with majority 0.0615 -0.0897 0.1854 Personal need to gain approval 0.6666 -0.0896 -0.0168 Ideas too readily accepted 0.4443 -0.3045 0.4219 Waited for leadership 0.4622 -0.1952 0.5087 Too many followers 0.3026 -0.1439 0.6783 Lack of initiative 0.2503 -0.1317 0.7402 Indecisive 0.5717 -0.2675 0.1720 Reluctant participation 0.2359 -0.1481 0.7576 Lack of interest 0.2437 -0.2330 0.7462 Summary Eigenvaluesb 9.98 4.97 1.64 21.14 18.58 10.57 Variance explained (%) c Cumulative variance explained (%) 21.14 39.72 50.29 a. Items listed in order they were entered into the dataset rather than the order on the questionnaire. They are a paraphrased subset from the Group Styles Inventory by R. A. Cooke and J. C. Lafferty (1988), copyright 1988 by Human Synergistics b. Eigenvalues computed before varimax rotation c. Variance explained computed after varimax rotation

39

TABLE 2 Tests of Inter-Rater Reliability and Agreement Indices of Agreement eta2 (%var.)a 42 62 62

Anovab

Rwg(j)

Interaction Style Constructive 2.40** .94 Passive 5.38** .82 Aggressive 5.39** .93 Outcomes Solution Acceptance 46 2.81** .83 Satisfaction 55 4.99** .84 Perceived Efficiency 57 4.40** .76 a. The percent of variance explained by group membership b. F-ratio indicating variance in responses between groups is significant in relation to total variance. **p