Abstract
Faultlines have the potential to significantly disrupt team performance due to the creation of intergroup bias. In geographically dispersed teams, given the combination of dispersed locations and other diversity characteristics, faultlines are potentially a major issue that needs to be more fully understood. This study examines the impact of faultlines on geographically dispersed teams and how problems caused by faultlines can be resolved. An experimental study of 40, four-person student teams finds that perceived faultlines heighten conflict and impair decision process quality. The findings also suggest that self-disclosure via weblogs and task elaboration can repair damage caused by faultlines. However, self-disclosure does not have a direct effect on reducing faultlines; the relationship is moderated by social attraction. That is, as team members disclose personal information to out-group members and out-group members are attracted to such disclosure, perceived faultlines are diminished. This study also finds that even in teams with strong perceived faultlines, team members are still able to exchange and integrate perspectives if they have a better understanding of their out-group members via self-disclosure. The negative consequence of faultlines therefore is eased when task elaboration occurs during task execution. Implications of these coping mechanisms for teams with faultlines in organizations are discussed.
With current information and communication technologies (ICTs) and global collaboration, teams often work across time and space boundaries in geographically dispersed teams (Townsend, DeMarie, & Hendrickson, 1998). In a recent Virtual Teams Survey Report (RW3 CultureWizard, 2010), 80% of employees reported that they are part of a team with people based in different locations. While organizations can leverage dispersed teams to reduce travel time and costs, enable the recruitment of talented employees, and build teams of diverse individuals, dispersed teams have more challenges in managing conflict, making decisions, and expressing ideas than face-to-face teams (e.g., Martins, Gilson, & Maynard, 2004). Subgroups are easily formed in dispersed teams since members often work at different locations as well as have varied demographic attributes and backgrounds. As location and demographic attributes (e.g., nationality) covary together, a faultline between two subgroups can be created (Polzer, Crisp, Jarvenpaa, & Kim, 2006). Faultlines refer to “hypothetical dividing lines that may split a group into subgroups based on one or more attributes” (Lau & Murnighan, 1998, p. 328).
Given the fact that subgroup faultlines can impair team performance outcomes (Bezrukova, Jehn, Zanutto, & Thatcher, 2009; Li & Hambrick, 2005) and the dearth of research on the topic, particularly in the geographically dispersed setting (Thatcher & Patel, 2011), there have been calls for further studies into how the negative effects of subgroup faultlines can be reduced (Lau & Murnighan, 1998). Our study addresses this call and draws on the common in-group identity model rooted in research on intergroup bias in the social psychology area. The model suggests that in-group members change their perceptions of out-group members through personalized interaction (Gaertner & Dovidio, 2000). Personalized interaction can be facilitated by self-disclosure, or “the process of making the self known to others” (Jourard & Lasakow, 1958, p. 91). This study experimentally examines the effects of self-disclosure through the use of social computing technologies (SCTs; weblogs, specifically) on perceived faultlines in dispersed team settings. We extend faultline theory and the common in-group identity model by enhancing an understanding of how perceived faultlines can be reduced among dispersed team members. The findings help managers avoid potential negative effects of faultlines and understand how to better leverage technologies in dispersed teams.
Theoretical Background and Hypotheses
A common definition of diversity is that it refers to “differences between individuals within a team that may lead to the perception that another person is different from self” (van Knippenberg, De Dreu, & Homan, 2004, p. 1008). Researchers generally classify diversity by its characteristics: demographic variables, such as ethnicity and gender, nondemographic variables, such as values, attitudes, and abilities (Harrison, Price, & Bell, 1998), and task-related variables, such as functional expertise, education, and tenure (Webber & Donahue, 2001). The research findings on diversity often show inconsistent effects on team processes and outcomes (van Knippenberg et al., 2004).
To address this inconsistency, research on diversity has been guided by two perspectives. The first perspective suggests it is necessary to further investigate deeper meanings and patterns of diversity. Harrison and Klein (2007) suggest that diversity can be further differentiated into separation, variety, and disparity. Each category has its attributes and should be carefully theorized. Separation diversity, which is proposed on the basis of social categorization theory, is the extent to which team members are separated by perceptions that they are different from others. The maximum separation diversity occurs when two (or more) cliques in a team occupy two ends of a diversity continuum. The process of self-categorization can arouse stereotypes, drive competition for resources, create in-group/out-group biases, and hinder communication (Thatcher, Jehn, & Zanutto, 2003). Accordingly, conflict is higher, and cohesion and knowledge sharing is less. Thus high separation diversity mostly has negative effects. Quite differently, variety diversity, drawing on the information/decision-making perspective, emphasizes that group members have a pool of resources they can contribute, such as knowledge and experience, and is positively associated with decision-making outcomes. Disparity variety is based on the social stratification and equality. This perspective considers diversity as social differences, such as status and power, and affects the social dynamics in a team, such as participation and justice. In this study, the faultlines can be categorized as separation diversity since faultlines cut across a team and create multiple subgroups where subgroups members are tightly coupled to each other in terms of diversity characteristics but are distinctly different from members of other subgroup.
In the second perspective, researchers suggest that research on diversity should pay more attention to group information processes. Particularly, it is suggested to study how and when diversity causes social categorization, triggers intergroup bias, negatively affects team processes, gives rise to elaboration of task-relevant information, and brings positive influences into team processes (van Knippenberg et al., 2004). This article will combine these two perspectives because we believe that the perspectives are complementary. Given that this article focuses on challenges in dispersed teams, we will focus on faultlines (i.e., separation diversity) and how faultlines affect dispersed teams.
Faultline Theory
Faultline theory was developed by Lau and Murnighan (1998) to further explain the relationship between diversity and team performance. Faultline theory suggests that multiple types of differences can combine to create a hypothetical line within a group and that this increases the salience of subgroups (Lau & Murnighan, 1998). Strong faultlines are beneficial to subgroup members—increased satisfaction, improved communication, and higher cohesiveness—but detrimental to the whole group—more conflict between subgroups and decreased team performance (Bezrukova et al., 2009; Li & Hambrick, 2005).
Arguably, faultlines often exist in dispersed teams due to diversity of location, cultural and/or professional backgrounds, and possibly demographic variables. Empirical study has shown that geographically dispersed subgroups increase conflict and impair trust among team members especially when subgroup members have the same nationality (Polzer et al., 2006). The purpose of our research is to investigate if disclosing personal information and learning about fellow team members can lessen perceived faultlines and to examine whether perceived faultlines impact team processes. Our research model is illustrated in Figure 1, and each hypothesis is explained fully below. We start by explaining why self-disclosure may reduce the perception of faultlines in dispersed teams.
The research model.
The Common In-Group Identity Model and Self-Disclosure
The common in-group identity model proposes that several factors can reduce intergroup bias, such as cooperation, intimate and personal interaction, common tasks, and use of inclusive pronouns (e.g., we; Gaertner & Dovidio, 2000). In-group members may experience a gradual disappearance of intergroup bias via two ways. First, when in-group and out-group members have intensive cooperation and share a common task objective, the perception of a superordinate in-group can be formed. This process is known as recategorization. Second, in contrast to recategorization that creates a new in-group at a higher level, decategorization reduces or eliminates categorization by establishing personalized relationships with out-group members. Out-group members then are seen as individuals rather than members of an out-group.
Self-disclosure has been regarded as an important way to develop interpersonal relationships (i.e., socialization) and to maintain psychological health (Collins & Miller, 1994). People who receive other’s disclosure feel trusted and generally have a positive impression of the discloser (Collins & Miller, 1994). In organizational settings, self-disclosure can play a prominent role in the socialization process. Self-disclosure at the onset of a relationship can trigger informal interactions, which in turn, reinforces more self-disclosure and information sharing. As members are willing to bring their personal lives into the work settings, they feel closer to each other, thereby enhancing cohesion. Therefore, self-disclosure may be able to help reduce the negative effect of faultlines. Intimate interactions will increase tolerance of out-group differences (Allport, 1954). When self-disclosure occurs between subgroups, interpersonal trust may develop and negative bias toward out-groups may reduce. However, it should be noted that self-disclosure could be a double-edged sword. As more information is disclosed, differences between each other are more likely to be recognized. Faultlines could be created and strengthened. Theoretical discussion about how and when self-disclosure exerts positive impacts (i.e., the potential moderating role of social attraction) follows after the next section that discusses self-disclosure possibilities through ICT (information and communication technologies) in dispersed teams.
Socialization Via ICT
Given that dispersed teams communicate to a large extent via ICT and that the disclosure of personal information, thoughts, and feelings is necessary for successful dispersed team development, it is crucial to understand how effective social interaction happens through computer-mediated communication.
Our study focuses on one type of ICT (social computing technologies [SCTs]) and one specific SCT (weblogs). SCTs refer to computing applications in which software serves as an intermediary or a focus for a social relation (Schuler, 1994). SCTs, compared to earlier ICTs, have richer and more interactive user interfaces, provide an architecture of participation, and can be easily accessed ubiquitously by mobile devices (Parameswaran & Whinston, 2007). Some SCTs, such as social networking sites and weblogs, provide mechanisms that encourage people to share personal information, through impression management, profile management, visibility management, and identity management (Dwyer, 2007). It is expected that due to limited face-to-face interaction typical for dispersed team members, SCTs can be used to both establish and maintain social relationships (Qian & Scott, 2007). It is expected that SCTs will become more common in dispersed teams as the younger generation moves into the workforce (Kwai Fun Ip & Wagner, 2008), so study of these technologies in dispersed teams is warranted.
A weblog is one of the most popular SCTs (eBizMBA, 2013). A weblog enables bloggers to update others on activities and whereabouts, express opinions to influence others, seek others’ opinions and feedback, “think by writing,” and release emotional tension (Nardi, Schiano, Gumbrecht, & Swartz, 2004). Weblogs have often been used to disclose personal thoughts (Kaplan & Haenlein, 2010) and have been demonstrated to be a preferred platform to present oneself (Papacharissi, 2007) compared to social networking sites (e.g., Facebook) and microblogging (e.g., Twitter). In addition, weblogs are more commonly adopted in organizations and case studies have shown successful implementations of employee blogging (e.g., Efimova & Grudin, 2007). Therefore, this study focuses on weblogs as a personal disclosure medium (see Figure 2 for a screen shot for a weblog system).
A screen shot of IBM Quickr Weblog System.
Prior research suggests that individuals are more willing to present themselves through online media because they have less concern about being ridiculed, rejected, or condemned due to anonymity (Joinson, 2001). In weblog settings, inconsistent empirical results show that under the context of lower anonymity (i.e., weblogs are not typically anonymous), some bloggers self-disclose less due to concerns about negative consequences, whereas others disclose more personal information to maintain self-identity and shape others’ understanding of themselves (Qian & Scott, 2007). Different degrees of privacy concerns affect the weblog’s publicness. Some weblogs can be accessed by the public while others are considered to be a personal diary and offer limited access. Putting this into the self-disclosure context, the former can be called public self-disclosure and the latter can be referred to as private self-disclosure. The focus in this study is on public self-disclosure because perceptions of faultlines are likely to be affected when team members learn about other team members. Self-disclosure has been regarded as an important interpersonal strategy to develop relationships (Forgas & Fitness, 2008) and provide individualized information, thereby leading to decategorization processes. We postulate that greater public self-disclosure via weblogs can reduce perceived faultlines in dispersed teams. When team members share personal information with their other team members through weblogs, intergroup bias and perceived faultlines are reduced. Thus
Hypothesis 1 (H1): Public self-disclosure via weblogs will diminish perceived faultlines.
The Moderating Role of Social Attraction
Hypothesis 1 suggests that learning more about others has a positive impact. However, it is possible to imagine a situation where gaining personal information about another actually highlights or reinforces differences between the two parties. In this case, public self-disclosure may not reduce perceived faultlines; it could actually make a perceived faultline stronger. Consistent with this logic, Carte and Chidambaram (2004) suggested that teams with high observable diversity should use lean media early in their life to hide the diversity so it does not get in the way of team processes. Before a team develops a sense of team identity, observable differences may contribute to categorization and subgroup formation. Therefore, the potential beneficial effect of self-disclosure on perceived faultlines may be lessened or even reversed. This warrants an exploration of what variables might act as moderators and when. Below we explore this further and identify one possible moderator for study.
Face-to-face self-disclosure generally occurs between the actors of dyads. As one discloses personal information, relationships develop and such relationships potentially embody interpersonal attraction and liking. However, if one is not attracted to another subgroup’s members due to interpersonal differences in values, beliefs, and preferences, personalized interaction is not easily established. This decreases the potential for reducing perceived faultlines. Social attraction, a desire to socialize with someone (McCroskey, McCroskey, & Richmond, 2006), is particularly related to building interpersonal relationships. When someone is liked, the cognitive closeness increases and the distinction between self and other is less (Aron, Aron, Tudor, & Nelson, 1991). Decategorization and recategorization are more likely to occur. Therefore, as common in the contact hypothesis (Allport, 1954) and as the similarity-attraction paradigm (Byrne, 1971) would suggest, self-disclosure and social attraction together enhance familiarity and safety and meet the need to belong. This should alleviate categorical perceptions. We suggest that social attraction between members of subgroups is an important moderator between self-disclosure and perceived faultlines.
Hypothesis 2 (H2): Social attraction moderates the relationship between public self-disclosure via weblogs and perceived faultlines. Specifically, public self-disclosure that creates high social attraction will weaken perceived faultlines. In contrast, public self-disclosure that creates low social attraction may increase perceived faultlines.
Impact of Perceived Faultlines on Team Conflict and Performance
We consider the impact of perceived faultlines on two constructs of team functioning—conflict and decision process quality—that have been widely studied in face-to-face teams and dispersed teams (Martins et al., 2004). Team members may have different demographic characteristics, functional backgrounds, and values, stereotypes, and prejudices. Such individual differences make interpersonal conflict more likely. It is argued that when these differences align together and create faultlines, conflict becomes more intense. This argument is consistent with coalition theory (Murnighan & Brass, 1991), which proposes that team members with similar characteristics are more likely to interact and find their interactions pleasant and desirable. The group of people will be more likely to form coalitions. As coalitions form, the boundary salience around coalitions would be strengthened due to stereotyping, in-group favoritism and out-group hostility, thereby evoking intergroup conflict or preventing intergroup communication. The impact on conflict may also be stronger in dispersed teams in that the reduced social cues via computer-mediated communication (e.g., anonymity and less social presence) lead to decreased evaluation apprehension, thereby causing antinormative and disinhibited behaviors (Nunamaker, Dennis, Valacich, Vogel, & George, 1991). Different from prior literature largely focusing on dormant faultlines (i.e., potential faultlines based on team members’ characteristics), this research examines perceived faultlines. We posit more intensive conflict exists when differences become salient (van Knippenberg et al., 2004). If a potential faultline is not perceived by the team member, then it should have little effect. To the authors’ knowledge, only one article has examined perceived faultlines on conflict and they found a positive relationship (Jehn & Bezrukova, 2010). Examination of potential faultlines by many faultline researchers may explain the smaller effective sizes and the inconsistent findings regarding the impact of faultlines on conflict (e.g., Lau & Murnighan, 2005). Our study should provide additional evidence to support the power of activated faultlines on conflict.
Hypothesis 3 (H3): The greater the perceived faultlines, the more team members will experience conflict.
Team members with perceived faultlines create separation diversity based on similarity attraction, social identity, and self-categorization theory. When there are salient subgroups within a team, individuals are more willing to work with in-group members and are less open to sharing ideas with out-group members. Furthermore, strong perceived faultlines are expected to generate attribution errors, especially in dispersed teams. Due to limited cues during computer-mediated communication, individuals will rely more on prototypical beliefs, which can hinder communication with out-group members. Cramton’s (2001) field experiments identified that dispersed projects fail due to delayed feedback, communication misunderstandings, and limited personal information. She further found that colocated team members attributed failures to remote team members’ personalities and characteristics rather than situational factors. Intergroup bias appears strong in dispersed teams.
Negative attitudes and beliefs toward out-group members can be continuously reinforced by the social environment created by other in-group members if there are more similarities among in-group members and more differences with out-group members. In-group identity and norms become stronger and the team disintegration hinders decision-making processes (Kerr & Tindale, 2004). Team members accordingly have difficulty in building shared problem domains, coordinating decision-making processes, and working together as a team to stimulate one another’s ideas. Such decision-making process losses (i.e., low decision process quality) often results in less satisfactory decision outcomes. Therefore, we hypothesize the following:
Hypothesis 4 (H4): Higher perceptions of faultlines within a team will be negatively associated with decision process quality.
The Moderating Effects of Task Elaboration
Task elaboration is defined as the exchange, discussion, and integration of task-relevant information (van Knippenberg et al., 2004). Task elaboration requires deep-level information processing as well as the integration of multiple team members’ perspectives. van Knippenberg et al.’s (2004) categorization–elaboration model (CEM) suggests that task elaboration is how the benefits of diverse perspectives, talents, and experiences in a high information diversity team are obtained. The CEM further highlights how intergroup bias may act as a moderator between diversity and task elaboration. When in-group members hold strong bias toward out-group members, in-group members are less willing to exchange information with out-group members.
The CEM also implies that task elaboration activities exist less in groups with strong faultlines due to intergroup bias. However, top management decision making illustrates that even though a top management team consists of various cliques, these teams can reach an adequate decision. In other words, despite the fact that team members belong to different subgroups, if team members have an accurate understanding of out-group members and achieving a common goal is crucial to them, it appears that team members can find ways to communicate and collaborate with other out-group members. Consistent with this, Huber and Lewis (2010) suggest that shared cognition among team members should not be a necessary condition for effective coordination, communication, and collaboration. Huber and Lewis maintain that cross-understanding, which refers to the extent that group members have an accurate understanding of one another’s mental models, is a more critical factor in team functioning.
Research has shown the positive impact of task elaboration on team performance in a high information diverse team (Homan et al., 2008); however, we know little about the impact of task elaboration in a team with strong perceived faultlines. Arguably, if subgroup members can share task-relevant perspectives and ideas, such sharing implies that team members surpass the hindrance of intergroup bias and that the negative effects of perceived faultlines on decision process quality will be attenuated. Therefore, we hypothesize the following:
Hypothesis 5 (H5): Task elaboration will moderate the effect of perceived faultlines on decision process quality. In particular, the negative effect of perceived faultlines on decision process quality will be weaker when task elaboration is high than when it is low.
Antecedents of Task Elaboration in the Faultlines Context
Under what situation would in-group members communicate task-relevant ideas and perspectives with out-group members? First, in-group and out-group members need to have cooperative interdependence (Allport, 1954). Common goals and fate increase the potential for cooperation. Second, in-group members choose concepts and words that out-group members can understand and accept so that both in-group and out-group members can communicate on common ground (Huber & Lewis, 2010). The former condition can be shaped by leaders who set clear incentives and sanctions for a team while the latter condition can be developed by interpersonal contact and training (Marks, Sabella, Burke, & Zaccaro, 2002). Our study places an emphasis on the second condition—increasing understanding between team members.
We propose that self-disclosure can enhance in-group members’ knowledge of what out-group members’ beliefs, values, and preferences are, thereby stimulating team members to elaborate task-relevant information and ideas when needed. The proposition is based on shared cognition. According to shared reality theory, shared cognition is developed and reinforced through interpersonal relationships (Hardin & Higgins, 1996). Without such shared cognition, uncertainty, cognitive overload, ambiguity, conflict, and other negative features of work interaction will develop (Schein, 2004), inhibiting task elaboration during task execution. Self-disclosure can serve a central role to enhance understanding of out-group members and increase task elaboration. Research has found that employees who are willing to talk about work and personal matters with each other are more likely to share tacit task-related knowledge and exchange new ideas (Albrecht & Ropp, 1984). We suggest that self-disclosure can similarly impact task elaboration in geographically dispersed team settings and hypothesize the following:
Hypothesis 6 (H6): Public self-disclosure via weblogs can encourage team members to elaborate task-relevant information in teams.
Method
An experimental method was adopted since we needed to control for other factors (to eliminate potential confounding), while being able to create a potential faultline. Prior to the experiments, we conducted six pretests and 18 pilot tests. Participants’ feedback and nonparametric statistical tests were used to refine the procedures and questionnaires to enhance reliability and validity (described below).
Participants
Undergraduate university students (M age = 20) were assigned into 45 four-person teams consisting of two males and two females. Five teams did not complete all tasks because of an insufficient number of participants in their teams, resulting in a working sample of 160 students and 40 four-person teams. Participants earned class credit for their participation. The three top teams additionally received prizes (Can $80, Can $40, and Can $20 per team for the 1st, 2nd, and 3rd places, respectively). Participants were assigned by random stratification on gender and year of study. Since participants were recruited from the same school, it was possible that they knew each other. To control for this possibility, we mixed participants from different years of study. A manipulation check also asked, “Did you know anyone in your team before the study began?” and asked how many people they knew in the collocation or remote group. Futhermore, t test analyses of perceived faultlines and social attraction did not show statistically significant differences between teams that consisted of members knowing other members and teams that did not.
Research Design
To create a potential faultline, each team had two subgroups who worked in different locations; members in each subgroup were the same gender and different from that of the other subgroup in their team. Gender was chosen as an additional faultline dimension (along with location) since it is nearly impossible not to be aware of a person’s gender and people are more likely to create faultlines based on gender (Thatcher & Patel, 2011). Even though only limited visual cues exist in geographically dispersed teams, people by and large can identify one’s gender by names or likely know gender after initial conversations via voice cues or conversation content. To further encourage gender visibility in our study, participants’ names were visible on both the weblogs and the online meeting system (described below). Therefore, we created a potential faultline with gender and location aligning together. Gender differences can also be made salient by a group task (Pearsall, Ellis, & Evans, 2008) so we chose the fallout shelter task. Savicki, Kelley, and Lingenfelter (1996) suggested that since the fallout shelter task is a masculine content, intellective task, which focuses on scientific, outdoor, and survival knowledge and has a correct answer based on experts’ knowledge, males should behave differently from females during task execution.
The task asked participants to imagine that a nuclear war had been announced and that his or her team has access to a small basement fallout shelter. Each team member was asked to choose and rank the 10 most useful items for survival during and after attack. Then, each team was asked to work together to come up with a finalized ranking list for the team. Based on the above team composition and gender-biased task, salient potential faultlines were thus created for each condition. Both a manipulation check (i.e., whether participants perceived the existence of subgroups within their team during their collaboration) and above average mean perceived faultlines (M = 3.92, SD = .98) indicate that in-group and out-group perceptions existed in our study.
The experimental treatment in this study is the public self-disclosure via weblogs. There were 22 teams in the experimental condition 1 and 18 teams in the control condition. To have both the experimental and control condition comparable, we had everyone do blogging. Participants in both conditions were asked to pick one topic from a list of three self-disclosure topics developed by Jourard and Jaffe (1970). A sample topic was “Discuss and explain the types of play and recreation I enjoy.” Then, each member posted his or her answers on an individual weblog (i.e., IBM Quickr was used as the platform—see Figure 2 for an example). For the public self-disclosure condition, each team member reviewed all other members’ weblogs and left comments; for the control condition, team members did not access other members’ weblogs.
Procedures
For the purpose of decreasing public self-awareness during the self-disclosure task, participants arrived at separate rooms at prearranged times. Before the task began, participants completed a paper prequestionnaire, measuring personality. On average, individuals took 90 min for all activities, with everyone taking approximately 10 min to conduct the self-disclosure task. Once the self-disclosure was completed, individuals filled out a questionnaire regarding self-disclosure. Then, for the public self-disclosure condition, team members reviewed other team members’ self-disclosures and left comments on others’ weblogs for 15 min (participants in the control condition skipped this section and proceeded to do the second task 2 ). In the second task, the fallout shelter task, each team member first spent 5 min individually ranking the 10 most useful items for survival. Afterwards, two subgroups were created by separating male and female team members into two physically separate rooms. Teams had 25 min to discuss and achieve consensus using a text-based online meeting system (i.e., IBM Sametime—screenshots are available from the authors on request). While it was possible for members of a subgroup to verbally discuss issues among the two of them, all team members were instructed to use the online system and work with the entire team as much as possible. 3 Each group submitted a final solution of the task as well as justification for their ranking. Participants then completed a postquestionnaire, measuring the constructs described below (except for self-disclosure).
Measures
Self-disclosure
Self-disclosure is about what individuals reveal about themselves to others. We assessed self-disclosure in terms of accuracy (e.g., “My self-disclosures were completely accurate reflections of who I really am”), intimacy (e.g., “I intimately and fully revealed information about myself”), and positiveness (e.g., “On the whole, my disclosures about myself were more negative than positive”). Hereafter, these three measures are called accurate self-disclosure, intimate self-disclosure, and positive self-disclosure, respectively. The measures were adapted from Wheeless (1978) and were assessed on a 7-point Likert-type scale. The Cronbach α of each measure is .81, .79, and .70, respectively. Although self-disclosure is not a shared property of a team, it is aggregated to represent the level of self-disclosure that existed in a team.
Perceived faultlines
The perceived faultline scale measured the extent to which team members perceived that the team split, using six items modified from Cronin, Bezrukova, Weingart, and Tinsley (2007) on a 7-point Likert-type scale (e.g., “During our discussion, our team broke into two or more subgroups, instead of working as one cohesive group”; Cronbach α = .91). Team level perceived faultlines is the focus of our study.
Social attraction
Participants were asked to consider the interaction during group tasks and whether they were willing to socialize with other subgroup members. Social attraction was measured with eight items modified from McCroskey et al. (2006) on a 7-point Likert-type scale (e.g., “I would like to spend time socializing with my remote team members”; Cronbach α = .88).
Task elaboration
Participants were asked to think about whether team members elaborate task-relevant information during team discussions. We used a three-item measure developed by Homan et al. (2008) on a 7-point Likert-type scale (e.g., “The team members contributed unique information during the team task”; Cronbach α = .67).
Conflict
Participants were asked to respond to Jehn’s (1995) conflict measure on a 7-point Likert-type scale, which is comprised of task conflict (e.g., “How frequently were there conflicts about ideas in your team?”) and relationship conflict (e.g., “How much were personality conflicts evident in your team?”). A factor analysis with principle axis factor extraction and varimax rotation did not clearly differentiate these two dimensions. A lack of discrimination could be due to short-term relationships maintained by participants and the context of faultlines. Without prior knowledge of other team members and insufficient interaction, participants may not be able to separate whether the conflict is caused by interpersonal incompatibilities or differences in viewpoints, ideas, and opinions. The faultlines configured in this study may aggravate the uncertainty issues in such initial interaction. Intergroup categorization leads one to believe conflict arising from personal factors rather than task factors (Pettigrew, 1979). Furthermore, many studies have shown that dispersed teams are prone to such attribution errors (e.g., Cramton, 2001). Given the limitation of our study context for participants to distinguish task conflict from relationship conflict, we decided to measure overall intragroup conflict by combining 4 items from relationship conflict with 4 items from task conflict (Cronbach α = .93).
Decision process quality
Decision quality can be differentiated into two dimensions: behavior and outcome (Beal, Cohen, Burke, & McLendon, 2003). Behavior represents how team members are performing to achieve the goal during the decision-making process while the outcome dimension of decision quality is the consequences of decisions. Since the fallout shelter task is an intellectual task, accurate results require team members with relevant survival knowledge (Jehn, Northcraft, & Neale, 1999). Our study did not recruit participants who have survival expertise, so only the behavior decision quality dimension was assessed. Decision process quality was adopted as a proxy of behavior decision quality dimension and was measured using five questions developed by Green and Taber (1980) on a 5-point Likert-type scale (i.e., the decision process is efficient/inefficient, coordinated/uncoordinated, fair/unfair, confusing/understandable, and satisfying/dissatisfying).
Control variables
Multiple control variables were examined: the big-five personality traits, diversity of language within the team, and diversity of ethnicity within the team. Blau indexes (Blau, 1977) were created to assess language and ethnicity diversity. Personality was measured with Saucier’s (1994) version of Goldberg’s measures of the big five traits. Analysis found similar results with and without these control variables; therefore, they are not discussed further.
Analysis and Results
Preliminary Analysis
To justify the extent to which individual-level data were homogeneous within teams, we first assessed within-unit agreement via Rwg (j). The median values of Rwg (j) for five shared team-level variables—perceived faultlines .77, social attraction .95, task elaboration .83, decision quality .90, and conflict .96—show acceptable team member agreement (Klein & Kozlowski, 2000). Then, ICC (1) was used to evaluate the ratio of between-group variance to the sum of between- and within-group variance, via an F test. If the F test is significant, between-group variance is significantly larger than within-group variance (Klein et al., 2000). The high ICC (1) implies that a team has both acceptable within-unit homogeneity and significant between-group variance. The ICC (1) values were as follows: perceived faultlines .27, social attraction .18, task elaboration .12, decision quality .30, and conflict .44. The test statistics (F ratios) associated with the ICC (1) values of all five variables were statistically significant, also justifying aggregation. We also examined ICC (2), which provides the reliability of the team means. The ICC (2) for perceived faultlines, social attraction, task elaboration, decision quality, and conflict was .60, .47, .36, .64, and .76, respectively. The lower value of ICC (2) is due to small group size (k = 4). ICC (2) is a function of ICC (1) adjusted for group size value; large group size leads to large ICC (2) (Bliese, 1998). Bliese (1998) also indicated that lower value of ICC (2) attenuates team-level correlations. Therefore, our team-level model would be a relatively conservative test.
Analytical Technique
With a mediated research model, structural equation modeling (SEM) was warranted. Partial Least Squares (PLS) was chosen as the SEM tool for this analysis due to its ability to work with relatively small sample sizes and to provide consistent results in comparison with covariance-based SEM (Hulland, Ryan, & Rayner, 2010). PLS tends to overestimate measurement model coefficients and underestimate path coefficients vis-à-vis covariance-based SEM (Barroso, Carrion, & Roldán, 2010). The underestimated path coefficients (i.e., biased estimates) for PLS results in a conservative test for our structural model. The issue of overestimated measurement loadings is less significant given solid reliability and validity of the measurement model (see the next section). PLS uses a combination of principal components analysis, path analysis, and regression to simultaneously evaluate theory and data (Pedhazur, 1982). The path coefficients are standardized regression coefficients, while the loadings can be interpreted as factor loadings (Barclay, Higgins, & Thompson, 1995). SmartPLS was used to conduct PLS analysis (Ringle, Wende, & Will, 2005).
Two models were analyzed for this research. Model 1 tests whether public self-disclosure affects perceived faultlines (Hypothesis 1) and task elaboration (Hypothesis 6), the impact of perceived faultlines on conflict (Hypothesis 3) and decision process quality (Hypothesis 4), and the moderating effect of task elaboration on the relationship of perceived faultlines and decision process quality (Hypothesis 5). Separate analyses were used to test the moderator effect of social attraction on the relationship between self-disclosure and perceived faultlines (Hypothesis 2). This was necessary because participants in the private self-disclosure condition did not see one another’s weblogs, so the data for this analysis was limited to the public self-disclosure condition.
The Measurement Models
The reliability and validity of the measurement models were assessed first, followed by the structural models. Convergent validity was assessed by individual item loadings, the composite reliability, and average variance extracted (AVE) (Barclay et al., 1995). Most of the measurement items have acceptable loadings (> .6) and loading range (< .25) (Chin, 2010). Two items were excluded due to low item loadings: one item from accurate self-disclosure (i.e., “I was not always honest”) and another positive self-disclosure (i.e., “I expressed my ‘good’ feelings about myself”). The composite reliabilities of all constructs were greater than .7 and AVE exceeded .5 (see Tables 1 and 2; Chin, 2010). Discriminant validity was demonstrated in two ways. First, a cross-loading matrix analysis found that all the items loaded highest on their target construct. This suggests that a construct is more strongly related to its measure than with any other constructs. 4 Second, the square root of the average variance extracted was significantly higher than any correlation between the constructs as seen in Tables 1 and 2 (Barclay et al., 1995). That is, all constructs share more variance with their indicators than with other constructs. Overall, the results demonstrate that the measurement model is adequate, allowing for the examination of the structural model.
Composite Reliability, Average Variance Extracted, and Correlation Among Constructs.
Note. Diagonal values are the square root of AVE. AVE = Average Variance Extracted.
Composite Reliability, Average Variance Extracted, and Correlation Among Constructs (Analysis for Hypothesis 2).
Note. Diagonal values are the square root of AVE. AVE = Average Variance Extracted.
The Structural Models
Dummy coding was used for the self-disclosure treatment (the control condition = 0, public self-disclosure = 1). Standardized variables were used to test the interaction of perceived faultlines to reduce multicollinearity issues in the context of modeling moderating effects (Aiken & West, 1991). To evaluate the significance of parameter estimates, bootstrapping (200 samples) was used to reestimate the parameters.
Five out of the six hypotheses were supported. Looking first at the relationship between self-disclosure and perceived faultlines, we did not find statistically significant relationships (β = .08, p = .45; Hypothesis 1). As expected, perceived faultlines is positively associated with conflict (β = .62, p < .001), supporting Hypothesis 3. Hypothesis 4, which predicted a negative relationship between perceived faultlines and lower decision process quality, is supported (β = –.37, p < .001). The interaction effect of perceived faultlines and task elaboration is significant for decision process quality (β = .34, p < .001). This supports hypothesis 5: task elaboration moderates the relationship between perceived faultlines and decision quality. Lastly, the results reveal that self-disclosure was positively related to task elaboration (β = .33, p < .001), supporting Hypothesis 6. The multiple R-squared values given for each dependent variable indicate that the model explained 41.0% of decision process quality, 38.7% of conflict, 10.9% of task elaboration, and 0.6% of perceived faultlines. As suggested in Hypothesis 2, we found that the relationship between self-disclosure and perceived faultlines depends on social attraction. Three types of self-disclosure were examined separately. Positive self-disclosure, accurate self-disclosure, and intimate self-disclosure significantly reduce perceived faultlines when social attraction is high (β = –.37, β = –.50, β = –.40, respectively, p < .001). The R-squared values of perceived faultlines are 59%, 57%, and 53%, respectively. Table 3 summarizes the results of the PLS analysis and Figure 3 illustrates the moderating relationship of social attraction for one type of self-disclosure (plots for the other two types of self-disclosure are similar).
Results of PLS Analysis.
p < .001, two-tailed.
Social attraction moderation of the relationship self-disclosure to perceived faultlines.
Discussion
In the present research, supportive of coalition theory (Murnighan & Brass, 1991), we find that perceived faultlines composed of location and gender was positively associated with conflict and negatively associated with decision process quality (H3 and H4). Higher conflict could be due to miscommunication and misunderstandings that accumulate as both subgroups elaborate their ideas. The strong effect of perceived faultlines on conflict as compared to mixed results from prior research (e.g., Lau & Murnighan, 2005) could also be due to the fact that the faultlines were activated in our study, as compared to being just a potential faultline (Chiu, 2010). Potential faultlines are based on objective demographic attributes of team members or team characteristics. People need to perceive that the differences exist to turn potential faultlines into active faultlines (or perceived faultlines). If faultlines are not activated, subgroups will not be salient, and the team may function as usual (Pearsall et al., 2008). Our results do provide empirical support for the theoretical mechanisms that cause the negative effect of activated faultlines and highlight the need for future studies to focus on activated faultlines.
Weaker decision-making process under strong faultlines is consistent with prior research where faultlines consisting of social categorical differences lead to poor performance (Bezrukova et al., 2009). The physical separation also potentially heightens the perception of interpersonal/intergroup distance and amplifies salience of subgroup differences. Both subgroup members may feel weaker affective bonds and perceive lower group identity. Trust in the group would decrease and intragroup cooperation becomes difficult. Consistent with this logic, prior research has shown that geographical distance leads to increased perceived distance among offshore collaboration partners, which in turn demands more efforts for collaboration (Aubert, Rivard, & Templier, 2011).
Our study examined two potential ways for alleviating the negative effects of faultlines on dispersed teams: public self-disclosure and task elaboration. We found that public self-disclosure via weblogs does not directly alleviate issues of intergroup bias in dispersed teams. This could be due to the nature of self-disclosure in this research. Self-disclosure is a prearranged task rather than a voluntary act. Telling someone something truly personal about yourself implies trust. However, if the self-disclosure behaviors are artificial, recipients of positive and intimate personal information give few credits to disclosers and pay more attention to the content of self-disclosure (Derlega, Metts, Petronio, & Margulis, 1993). The content provides cues for individuals to form impressions. Knowledge of another person could influence categorization of others either way: as in-group members or as out-group members. Supporting this potential moderation, our results suggest that social attraction plays a critical moderating role between public self-disclosure and perceived faultlines. As team members perceive in-depth and positive personal information as attractive, perceived in-group and out-group differences are diminished (see the bottom line in Figure 3—high social attraction—which has a negative slope). Put differently, the categorical beliefs of remote team members are changed as the individuating information provided by self-disclosure is incorporated into the evolving impression. Specific information about each remote team member degrades group boundaries and weakens the salience of out-group identity. Feelings of social attraction to out-group members based on individuating information then minimize the negative consequences of categorization.
We also found that weblogs are suitable platforms for self-disclosure as participants are generally able to accurately present personal information. Even within the limited amount of time to disclose, participants feel that their self-disclosure through weblogs were accurate reflections of who they are (M = 5.87, SD = .72) and that intimate information about themselves were revealed (M = 4.82, SD = 1.34). It is generally considered that non-face-to-face social interaction is relatively restricted due to a lack of verbal cues, as suggested by media richness theory (Daft & Lengel, 1986). However, it appears that weblogs have several distinct features that may potentially make disclosure via this medium effective. A weblog is an asynchronous medium and bloggers can decide whether to maintain strong or weak ties with other team members through different appropriations of weblogs. To help future work, we propose three technology characteristics that may explain why self-disclosure via weblogs is potentially more effective than face-to-face self-disclosure: rehearsability, reprocessability, and controllability. These three characteristics can potentially foster one’s understanding of another’s mental models. Rehearsability is “the extent to which the media enables the sender to rehearse or fine tune a message during encoding, before sending” (Dennis, Fuller, & Valacich, 2008, p. 587). Disclosers may spend more time reflecting on their thoughts and craft more clear and persuasive messages. Reprocessability refers to “the extent to which the medium enables a message to be reexamined or processed again, during decoding, either within the context of the communication” (Dennis et al., 2008, p. 587). Visitors of weblogs are able to review and reflect discloser’s messages carefully and assess a discloser’s mental models, which in turn potentially enables more thoughtful comments and feedback. Controllability means that weblogs allow disclosers to control what personal information they are able and willing to share, such as profiles, posts, photos, and videos. These three technology characteristics help one to carefully present a self, deeply reflect on another’s self-disclosure, and provide a wider range of personal information in various formats. Both disclosers and the audience are potentially able to more fully engage in self-disclosure activities. To sum up, weblogs show their potential to help dispersed team members present themselves and even control the impressions other people form. Empirically speaking, self-disclosure via weblogs resembles causal chat in a hallway or before meetings of colocated teams and is beneficial to team collaboration. Future research would be valuable to explore and enhance these features of weblogs.
The second faultline coping mechanism examined is task elaboration. As shown in Figure 4, the results demonstrate that the negative effects of perceived faultlines on decision process quality are lessened when team members are willing and able to exchange task-relevant ideas and perspectives (e.g., under high task elaboration, i.e., +1SD, the slope of the relationship between perceived faultlines and decision process quality is close to zero). The perception of faultlines could be subdued if team members exchange information that builds common ground to communicate and negotiate. Such cross-understanding can help in-group members tailor communication and carefully choose concepts and words that can be accepted by out-group members (Huber & Lewis, 2010). Hence, even though perceived faultlines exist, team communication and coordination appear to be more effective if task elaboration transpires between the in-group and the out-group.
Task elaboration moderation of the relationship perceived faultlines to decision process quality.
Theoretical Implications
Our study examines faultline theory in geographically dispersed teams and supports the negative impacts of perceived faultlines as proposed by Lau and Murnighan (1998). Although previous research has indicated that either gender (Pearsall et al., 2008) or location (O’Leary & Mortensen, 2010) can independently create hypothetical dividing lines that split a group into subgroups, we emphasize that faultline alignment and faultline activation play a critical role in engendering negative consequences. We further expand the prior research’s focus on faultlines to the coping strategies for faultlines, specifically task elaboration and self-disclosure.
Complementary to van Knippenberg et al.’s (2004) CEM model, which proposes task elaboration as a mediator between team diversity (i.e., variety diversity or information diversity) and decision quality, we argue that task elaboration plays a moderating role between faultlines (i.e., separation diversity) and decision quality. Although dispersed team members perceive the existence of faultlines, greater exchange and integration of task-relevant information attenuates the negative consequences of perceived faultlines. We also found that public self-disclosure increases the likelihood of task elaboration. Our study moves beyond prior research on why and when task elaboration could be difficult, such as team members who do not believe values of diversity (Homan, van Knippenberg, van Kleef, & De Dreu, 2007) and have educational faultlines (Jiang, Jackson, Shaw, & Chung, 2012). We suggest that public self-disclosure behaviors can be helpful to promote task elaboration. The willingness to share positive social information facilitates the ensuing task-relevant discussion partly because public self-disclosure through weblogs decreases intergroup anxiety and increases familiarity. In the post hoc analysis, public self-disclosure teams show the existence of higher trust, F(1, 38) = 4.12, p = .05. The proactive characteristic of public self-disclosure shows the potential for changing information elaboration processes in a team.
Our findings also imply that social relationships can be developed even at the early stage of team collaboration via social computing technologies (SCTs). We found that a substantial amount of social information was exchanged through weblogs even though team members did not have a collaboration history. The results are different from previous research, which indicates that less social-related information was exchanged when team members communicate through information and communication technologies and when a team was just formed (e.g., Warkentin, Sayeed, & Hightower, 1997). In our study, participants demonstrated their willingness to share personal matters on weblogs as shown by a relatively high degree of intimate self-disclosure (M = 4.82, SD = 1.34) and strong intent to disclose (M = 5.69, SD = .86). This inconsistency could be attributed to task–technology fit. The multiple functionalities of SCTs for supporting relationship development and the users’ rich SCTs experience may provide a comfort zone for participants in this study exchanging social-related information. Therefore, we believe that high task–technology fit was achieved (i.e., social-purpose technology fits into social activity). We further highlight that social information by itself does not directly lead to beneficial effects for a team, especially in the weblog context. Only when self-disclosure stimulates liking of other team members, does self-disclosure have a positive effect. Otherwise, self-disclosure may induce potential differences and engender stereotypes of members of out-groups.
Managerial Implications
Although managers can try to compose a team that consists of team members with cross-cutting dimensions to avoid potential faultlines, such a practice may not always be feasible in real-world settings. A team is often formed due to the need for specific knowledge, skills, and abilities as well as the availability of potential candidates. Also, due to the complexity of team projects and the makeup of geographically dispersed team membership, we expect that managers may not always be able to identify which kind of diversity dimensions will invoke perceived faultlines. Therefore, we suggest that managers should encourage self-disclosure within teams to reduce the potential negative effects of any faultlines that is present. Via sharing personal interests, skills, or experience, team members may recognize business and personal things in common. Based on this common ground, team members may gradually develop cross-understanding and learn that differences between each other are compatible rather than irreconcilable (Ensari & Miller, 2002). However, we caution that presently we do not know the empirical effects of self-disclosing negative information. Our study explicitly had people disclose information that was positive or at least neutral (e.g., details about recreational interests, etc.). Future research should examine the effects of more negative disclosure to see if this disclosure can also positively influence perceived faultlines through social attraction. At this point, our advice for managers is to encourage self-disclosure of positive/neutral topics. We also suggest that managers should consider providing training about online etiquette and cross-cultural communication. Even the neutral or positive topics can be detrimental to one’s social attractiveness if messages are not conveyed appropriately (e.g., vulgar language, sarcastic comments). Managers should also provide incentives for team members to post interesting and valuable content regularly and consistently so as to pull in audience. As shown in this study, self-disclosure on weblogs without an audience does not help faultlines issues. Thus weblog management would be another important lesson for organizations.
Managers should also stimulate better exchanges and integration of task-relevant information as this can help to reduce the negative effect of perceived faultlines. One may question the likelihood of exchanging and integrating diverse insights and ideas in geographically dispersed teams due to insufficient mutual understanding (Alavi & Tiwana, 2002). Yet the result that task elaboration can exist along with strong perceived faultlines from this study is encouraging. One potential way to enhance task elaboration in geographically dispersed teams is to increase an accurate understanding of other’s mental models (Huber & Lewis, 2010). Blogging about personal interests and skills can contribute to a better understanding among team members.
Limitations and Future Research
Despite several contributions to the literature and practice, our study has limitations. First, the content of self-disclosure in our current study is primarily nonwork related. In work settings, team members may disclose not only socially relevant information but also task-relevant information. Future research is needed that explores a more precise conceptualization of the information that geographically dispersed team members should share. A framework covering socially relevant understanding (e.g., cultural values, diversity beliefs) and task-relevant (e.g., team members’ knowledge, skills, abilities) understanding could help to clarify the specific information that increases social attraction and task elaboration. Also, while we can think of no reason why using students would give different results than employees, future research on self-disclosure in employee teams is needed to confirm the generalizability of the results.
We followed the logic of prior studies by creating a faultline out of multiple dimensions due to the potential ability to create stronger faultlines via aligned attributes. This does not allow us to investigate the isolated effect of each dimension; doing so, and investigating other types of faultlines (e.g., a faultline in terms of location and social demographic attributes vs. a faultline in terms of location and deep-level diversity attributes, such as values and beliefs) would be valuable future research. Also, the configuration in our research setting is fully balanced (2 males in one location and 2 females in another location and only two locations with equal numbers of participants in each). It is possible that different configurations, such as mixed genders at each location or more locations with varying numbers of people, could pose different challenges for dispersed teams. Future research could follow O’Leary and Mortensen’s (2010) suggestions by examining alternative configurations. In addition to configurational dimensions, we also encourage future research to include temporal dimensions of dispersed teams (e.g., differences in time zones). Temporal differences pose challenges to team collaboration (Espinosa, Nan, & Carmel, 2007) and engender attribution errors (Cramton, 2001), which are likely to increase perceived faultlines. Assessing the psychological distance and investigating what influences this could also contribute to more understanding of differing degrees of dispersion and the effects of dispersion (Wilson, Crisp, & Mortensen, 2012). As more faultline attributes are studied, it is also worth considering limitations potentially created by differences in things like national cultures. Our study was done in North America where there are modest value differences between men and women. Some other national cultures have more distance between these values so it is possible that gender has a stronger contribution to faultline perceptions.
This research utilized a cross-sectional study of temporary geographically dispersed teams. Self-disclosure may have stronger effects at the team formation stage due to initial trust and less entrenched biases. Once an impression of out-group members is formed, it may take fairly extreme emotional or instrumental self-disclosure content to alter perceived faultlines. To understand whether the influence of self-disclosure changes over time and how self-disclosure via SCTs reframes one’s evaluation of out-group members during reciprocal, ongoing interaction, a longitudinal research design focusing on team members’ reaction to other’s self-disclosure would offer an opportunity to investigate this. Based on the existing data, generalization of results to ongoing teams must be cautious.
Future research should also examine whether other media can exert similar effects to that found with weblogs. Given broad adoption of social networking sites (SNS) and configurable privacy settings, SNS such as Facebook may work for disclosing and enhancing relationships.
Conclusion
Our study examined faultline theory in geographically dispersed teams and found support for the negative impacts of perceived faultlines, as proposed by Lau and Murnighan (1998). Teams with stronger perceptions of faultlines experienced more conflict and had a weaker decision process. Our study contributes beyond this by examining two possible ways to manage the potential negative impact of faultlines. First, self-disclosure, where team members share personal information about themselves with the rest of the team, was found to break down perceived faultlines; however, this relationship was moderated by social attraction. If receivers perceive the information favorably, it does appear to break down subgroup formation and perceived faultlines. Second, task elaboration, where team members collaborate and interact, moderated the negative impact of perceived faultlines on decision process; more elaboration reduced the strength of the negative relationship between faultlines and decision quality. Managing the diversity within teams is crucial for effective team functioning, especially as teams become more dispersed with team members from different cultures and backgrounds. While this study provides insights into how to avoid the negative effects of diversity, it also identifies many opportunities where future research can potentially enhance our understanding.
Footnotes
Acknowledgements
An earlier version of this paper titled “The Effect of Personal Disclosure within Teams: Can Faultlines in Geographically-dispersed Teams be Bridged?” appeared in the Proceedings of the 2011 Hawaii International Conference on System Sciences conference, © 2011 IEEE. The authors would like to thank Jacqueline Corbett, Jane Webster, and two anonymous reviewers for their comments and suggestions.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a research grant to Sandy Staples by the Social Sciences and Humanities Research Council of Canada.