Feedback distractions during computer-mediated group collaboration.
Evaluative feedback is a form of communication that frequently occurs within group discussion. By evaluative feedback, the authors refer to the subjective assessment of another's performance in relation to a self-identified measure of satisfactory performance (Farh et al, 1991; Hebert and Vorauer, 2003). The feedback focused upon in this study is the evaluative feedback provided by peer group members. Openly participating and sharing opinions within groups can bring about apprehension for many people (Burgoon and Hale, 1983), so it is often felt that positive feedback is a useful tool for encouraging group members to contribute. This creates what is referred to by West (1990) as "participative safety" as group members subsequently feel free to express themselves without fear of ridicule or mockery. While it would seem that positive feedback from others would lead to more effective decision-making and improve decision-making performance in groups, much of the research on feedback has shown mixed results (Laurie and Swaminathan, 2009). Examining the influence of negative feedback within CMC groups is equally important. Negative feedback can frequently occur in CMC groups because limited social cues can lead to uninhibited behavior (Kiesler and Sproull, 1992) and a tendency for CMC users to refrain from sugarcoating negative information (Sussman and Sproull, 1999).
A better understanding of how computer-mediated groups handle evaluative feedback, whether positive or negative, and how it might affect task performance is needed in order to better manage group collaboration. Therefore, the purpose of this study is to examine the influence of evaluative feedback, both positive and negative, on the performance of computer-mediated groups. To do so, an experiment was conducted manipulating whether or not feedback was given and the valence of feedback (i.e., positive or negative). A confederate introduced positive and negative feedback distractions to groups. Groups were randomly assigned to one of three conditions: positive feedback condition, negative feedback condition, or control. The study focuses on the process and performance metrics that were produced during a collaborative group task.
Though little research has examined the receptiveness of peer feedback by the rest of the group, there is evidence that it is well received by group members and, at times, perceived as being helpful (Druskat and Wolff, 1999). Within CMC groups, feedback is considered an important part of promoting a shared understanding and for determining areas of agreement between group members (Geister et al., 2006). While feedback from other group members would seem to be beneficial for performance within computer-mediated groups, there may also be other unintended downsides to group member feedback namely groupthink and process losses due to production blocking.
Groupthink occurs when group members are more influenced by a desire to reach an agreed-upon decision rather than the most effective decision. Originally conceptualized by Janis (1972), groupthink is at its most prominent when members of a highly cohesive group become preoccupied with maintaining a consensus at the expense of opening group discussion to critical thought. Some have argued that groupthink is more nuanced than was originally thought, and depending on each group member's perspective, it can be based at times on collective optimism and at other times on collective avoidance of conflict. Others place the blame on group members' needs for maintaining social identity of the group and negating their own low individual self-efficacies (Baron, 2005). Regardless, the outcome of groupthink tends to be the same, and decision-making suffers from group discussion rather than being enhanced by it.
Prior research has indicated that distractions during group discussion tend to create obstacles hindering progress toward decision-making. Distractions can range from ambient noise within the environmental setting (Lowry et al., 2006) to nonessential communication between the group members themselves (Yoo and Alavi, 2001). By their nature, distractions do not require an immediate response because they invade sensory channels which are not entirely necessary for completing the task, allowing individuals the choice to ignore or eliminate the distraction (Basoglu et al., 2009). For example, groups often engage in groupthink to shorten their exposure to the distracting influence (Kruglanski and Webster, 1991). However, the potential distraction of group member feedback and its intersection with groupthink has not been explored in the extant literature. The social identity maintenance explanation may be the most pertinent to feedback, particularly positive feedback. This perspective of groupthink suggests that group members hope to develop and protect a positive image of the group by external others in the face of possible failure. A hallmark of social identity maintenance is the need to seek positive feedback from the in-group before attempting to exhibit competency to people outside the group (Turner and Horvitz, 2001). The possibility of public failure constitutes a threat that the group seeks to avoid, instead using positive feedback to encourage similar viewpoints and reduced flexibility among its members and creating an environment ripe for groupthink (Peterson and Behfar, 2003). By presenting a united front with the aid of positive feedback, the group is able to defend its decisions to others with the power of their unanimity.
Process Loss Theory
A second theory that may explain the influence of group member feedback is process loss theory. One main goal of group CMC research has involved methods and tools that can improve group productivity, including the identification and elimination of inefficiencies termed "process losses" (Steiner, 1972). Any single behavior or interaction during group discussion can have one of three possible influences on its progress toward an outcome (Bostrom et al., 1993; Hirokawa and Gouran, 1989). Promotive influences (aka "process gains") help facilitate successful meeting outcomes, while disruptive influences (aka "process losses") inhibit progression to an outcome. Process losses are of specific interest to the current study. One of the most commonly-researched process losses in the CMC literature is production blocking (Mejias, 2007). Production blocking occurs when the group dynamics prevent individual members from contributing to the discussion (Lim and Benbasat, 1996). Examples of production blocking include instances when members may be distracted by reading comments put forth by others, or when they divert their concentration from crafting their own comments toward remembering comments submitted by others. It is the authors' contention that evaluative feedback issued during a computer-mediated group discussion produces a form of production blocking as feedback distracts group members from the task at hand and eventually affects outcomes produced in the meeting.
The CMC format is often touted as being advantageous for groups faced with certain tasks because of the lack of distractions transmitted across the medium (McGrath and Hollingshead, 1994). The type of distractions that are filtered out in communication over a less rich medium are usually visual and audio-based cues, with paralinguistic cues (predominantly text-based) among the only ones that are successfully transmitted via CMC (Rao and Lim, 2000). That is not to say that text-based communication is not without distractions and that CMC users have no opportunity to become distracted by said communication. Pinsonneault and colleagues (1999) suggested that the pool of ideas being generated by a CMC group can itself produce process losses by distracting users. The distractions described include the possibility for group members to dedicate too much attention to viewing others' ideas out of interest, as well as a focus on not replicating another group member's ideas out of a desire to contribute something unique (Pinsonneault et at., 1999).
Subsequently, a test of the distracting power of other CMC users' input showed that users who were not exposed to input submitted by others performed consistently better at idea generation than users who did see others' input (Potter and Balthazard, 2004). While the ability to view others' comments and input increases the meeting memory for a CMC group (Dennis and Garfield, 2003), it could also present a distracting influence that would otherwise not be as accessible for traditional work groups. Given that even appropriate, on-task comments submitted by group members can be a distracting influence for others, the authors believe that the evaluative feedback provided by a group member will be equally distracting, if not more so.
Finally, group members are typically uneasy about receiving feedback from others, stemming largely from an apprehension of being evaluated and doubts about fairness (London and Sessa, 2006). It is unlikely that those uneasy feelings about feedback will be completely alleviated within a computer-mediated environment, and apprehensions of receiving feedback will be a potential source for process losses no matter which modality is being used for communication.
The following section develops a series of hypotheses regarding the valence of feedback (i.e., positive or negative) introduced within group discussion in the current study. It is crucial to note that naturally occurring feedback among group members took place in each condition as group members were participating in a decision-making task, and that in the positive and negative feedback conditions a confederate provided feedback (see Appendix) in addition to the naturally-occurring feedback. The study hypotheses are intended to examine whether or not groups differed in outcomes based on their experimental condition.
Valence of Feedback
In terms of valence, evaluative feedback from others typically consists of either a positive or a negative affect (Schiable and Jacobs, 1975). Positive feedback is primarily focused toward reinforcing desired behaviors and performance as well as toward establishing feelings of psychological safety, whereas negative feedback targets undesired behaviors and performance with the hopes of motivating the feedback recipient into corrective action. In terms of its impact on group performance, the valence of feedback could have some meaningful impact on processes and output. Feedback interventions in general are intended to draw the recipients' attention toward the gap between actual performance and the goal or standard that is desired (DeNisi and Kluger, 2000). Whether positive or negative, feedback meets the criteria necessary for a process loss within group collaboration.
The valence of the feedback is a factor that is expected to influence the group discussion and any subsequent output. Negative feedback is expected to trigger the "threat-rigidity effect" in which group members respond to negative feedback with decreased flexibility, more control imposed on certain group members, and restricted information sharing (Peterson and Behfar, 2003). Often, the tendency for computer-mediated communicators is to issue direct, powerful statements of opinion. While such harsh feedback may not be received well by the members of a CMC group, performance does not necessarily suffer as a result. Work on minority influence in groups predicts exposure to differences in opinion will force members to reassess the problem at hand and reconsider alternative solutions, ultimately leading to higher quality decision-making (Smith et al., 2001).
On the other hand, positive feedback from one's group members can also hinder a group's decision-making performance for reasons beyond production blocking. Because group members are generally optimistic, positive feedback received by the group is often received as an affirmation that "everything is going well," and the group subsequently chooses not to make any behavioral changes (De Cremer and van Dijk, 2002). In terms of the long-term impact of positive feedback, continued increases in efficacy due to feedback can eventually lead to overconfidence and complacency (Lindsley et at, 1995). Another possible outcome of positive feedback is that group members who receive support for an idea or comment interpret such feedback as grounds that the idea was complete and requires no further thought (Connolly et al., 1990). The end result is that fewer ideas are submitted, appraised, and discussed with due diligence.
The authors predict that members of groups that are not provided feedback by other group members will be less distracted than members of groups who receive evaluative comments, which serve to inhibit task performance. In terms of valence, the authors predict that group members receiving negative feedback during discussions will perform better on a collaborative task than group members who receive positive feedback due to its decreasing influence on idea generation and additional production blocking. Therefore, it is hypothesized:
Hypothesis 1A: Group members who are in the control group will exhibit better decision-making performance than group members who are in the feedback conditions.
Hypothesis 1B: Group members who are in the negative feedback condition will exhibit better decision-making performance during discussion than group members who are in the positive feedback condition.
Another potential contributor to improved task performance of work groups that has been studied in the CMC literature is choice shift. Simply put, choice shift represents the magnitude of change from individual preferences before group discussion to post-discussion choices (Seibold et al., 1996). Ideally, the choice shift an individual experiences following group discussion would produce better performance or more accurate decision-making along the lines of the adage "two heads are better than one" (Cooper and Kagel, 2005). However, this may not always be the case. Accordingly, researchers have sought to understand common reasons why choice shift occurs and why it may be potentially detrimental to decision-making by individuals serving in groups. Reasons for choice shift include diffusion of responsibility, greater influence by risky group members, greater familiarization with pros and cons of variance alternatives which leads to higher risk, risky behavior being viewed as socially desirable, and shifts influenced by how risky or conservative an individual perceives the group to be (Cecil et al., 1976). Of particular interest to this study, social desirability and impression management appear to influence group members toward making a more extreme choice (i.e., a "risky shift") than they would otherwise take (BarNir, 1998).
While group discussion will inherently produce a measure of choice shift in many people, feedback will have an influence on how beneficial any choice shift will be. Specifically, individuals in groups assigned to a positive or negative feedback condition will be associated with a higher incidence of choice shift than groups in the control condition. This is due to the potential that evaluative feedback produces a process loss since it may distract group members. First, the content of feedback exchanges is likely to be salient in a CMC setting since individuals cannot otherwise rely on nonverbal cues to gauge others' opinions (Bordia, 1997). Also, individuals seek to reduce uncertainty and may attempt to do so through interpersonal interactions in their group, requiring feedback from others (Leary, 1996). Though general feedback is thought to reduce uncertainty (Waldersee and Luthans, 1994), the most unambiguous representation of how well a group member's opinions are aligned with a group is through discussion of the current level of group agreement. In fact, one study found that group members who received statistical feedback (i.e., a metric indicating the current level of agreement) encouraged them to change their individual preferences toward the majority opinion (Hiltz et al., 1991). Group member feedback about others' opinions, while not an objective metric, is similarly expected to encourage group members to modify their own opinions should they find their opinion not among the majority. Following that reasoning, choice shift should especially occur when the evaluative feedback from other group members is negative, a further indication that an individual opinion is not aligned with others. Therefore, it is hypothesized:
Hypothesis 2A: Group members in the positive and negative feedback conditions will be more likely to engage in choice shift than group members in the control group.
Hypothesis 2B: Group members in the negative feedback condition will be more likely to engage in choice shift than group members who are in the positive feedback condition.
This study also focuses on "beneficial" choice shift in which the group discussion aids individuals in improving their decision-making performance from their initial judgments. The valence of feedback is likely to influence how beneficial choice shift will be. Negative feedback is thought to bring about defensiveness while positive feedback is thought to put people at ease (Stone et al., 1984). However, negative feedback is expected to encourage beneficial choice shift because it is likely to stimulate a more critical discussion of alternate solutions, which in turn has been linked to better performance (Kuhn and Poole, 2000). On the other hand, groups in the positive feedback condition are expected to be more prone to groupthink as they are less likely to engage in a full discussion of potential benefits and disadvantages of each alternative. In fact, groupthink is thought by some to influence choice shift (Kroeck et al., 1987). Therefore, groups in the positive feedback condition are predicted to be more susceptible to a less thorough and critical discussion of alternatives which is likely to result in less beneficial choice shift. Therefore, it is hypothesized:
Hypothesis 3A: Group members in the control group will be more likely to improve their decision-making performance through choice shift than group members in either the positive or negative feedback conditions.
Hypothesis 3B: Group members in the negative feedback condition will be more likely to improve their decision-making performance through choice shift than group members in the positive feedback condition.
Finally, feedback is likely to influence the amount of time groups spend during discussion. Feedback from group members will serve to distract other members, and while feedback should not typically interrupt task completion, it should draw enough attention from the task to temporarily interfere with ongoing work flows. Distraction conflict theory stipulates that as distractions occur during relatively complex tasks, individuals are unable to process the same amount of information as can be processed under normal circumstances (Baron, 1986). Following from cognitive resources allocation theory, reducing attention from the task at hand to attend to another signal (i.e., the feedback) comes at the cost of misallocating cognitive resources necessary for optimally completing the task (Speier et al., 1999). Hence, the presence of feedback will require groups to extend time to make a decision.
While positive feedback results in more production blocking and less effective decision-making than negative feedback, the presence of negative feedback will increase the amount of time groups spend in discussion. Empirical evidence suggests that groups that receive negative feedback during a decision-making process tend to engage in more argumentation, most likely in an effort by group members to support their own argument in the face of criticism (Lam and Schaubroeck, 2009). Even with less-aggressive, conflict-based approaches like the use of devil's advocacy by a group member, groups can require more discussion to reach agreement than is normally needed (Valacich and Schwenk, 1995). By comparison, groups who receive positive feedback will likely require less time to reach consensus by virtue of not requiring the same amount of argumentation. This expectation follows the "consensus difficulty hypothesis," which predicts that groups with a cooperative climate will spend less time reaching a decision than more contentious groups (Hollingshead et al., 2005). Therefore, it is hypothesized:
Hypothesis 4A: Groups in the positive and negative feedback conditions will spend more time discussing a task than groups in the control condition.
Hypothesis 4B: Groups in the negative feedback condition will spend more time discussing a task than groups in the positive feedback condition.
a laboratory experiment manipulating the type of feedback that group members received from a confederate was used. Participants consisted of undergraduate business students at a public university who received course credit for their participation. Subjects were randomly assigned to a group in one of three conditions: groups which received positive feedback from a confederate during discussion, groups which received negative feedback from a confederate (both types of groups were composed of four subjects), and a control condition in which groups of five members received no feedback from a confederate. a "fifth group member" who participated in groups in the positive and negative feedback conditions was in reality a confederate participating from a remote location. After eliminating incomplete responses from subjects, the sample included 167 individual group members and resulted in a total of 39 groups that performed the task. Groups communicated solely through the use of web-based chat software implemented by one of the researchers, and group members' comments were only identifiable by an alias.
Groups were presented with the "Lost in the Desert" survival scenario, which has been used in previous studies on group collaboration (Zhou et al., 2004). In the scenario, subjects are provided a list of fifteen objects from which they must select five that will most likely enable them to survive in the desert. Survival experts have arrived at an appropriate ranking for each item allowing the comparison of the subjects' judgments with that of subject matter experts. Within the Group Task Circumplex (McGrath, 1984), the task in this study is best described as a "decision-making task," in that group members were charged with making a decision featuring a generally-agreed upon answer instead of a demonstrably correct answer, while also comparing their opinion with the competing views of others. Decision-making tasks are recommended for studies investigating phenomena like group conflict and choice shift (McGrath, 1984). Pilot testing indicated that students found the task to be salient and engaging.
The experiment was conducted in a large electronic classroom with dozens of personal computers. Upon arriving, subjects were asked to provide an initial selection of the five most important items, thus providing their initial opinions. Subjects were then instructed to log into an assigned chat room and discuss the items with their unknown fellow group members, and upon completing the discussion, to select five items from the list again. Although subjects could see the other participants, they were not aware of which subjects were assigned to their particular group. Groups were not given any time limit for discussion, and group members were under no obligation to make the same item selection as the rest of the group.
During group discussion, the confederate submitted either a positive feedback comment or a negative feedback comment depending on the group's condition assignment, similar to the use of confederates in other CMC group studies (Connolly et al., 1990). A list of comments can be found in the Appendix. To ensure that comments were authentic, actual feedback statements were produced by students in the same population who were part of the sample used during pilot testing. Following task completion, pilot subjects were instructed to draft both positive comments and negative comments and submit them over the chat facility as if the, group were still discussing the task. The resulting comments submitted by the confederate consisted of a powerful language style (Adkins and Brashers, 1995), meaning that the comments were absent hedges, hesitations, and qualifiers. In CMC, a powerful comment tends to be more evaluative by its receivers than powerless comments. The confederate submitted one comment every two minutes, no matter what the current topic of discussion was, in the same order for each group.
The dependent variables in this study (i.e., accuracy, choice shift, time in discussion) were all objectively measured. Task performance was measured in terms of the accuracy of each group member's item selection. The task required each group member to determine the top five items in importance to their survival. Each member performed the item selection twice, once before the discussion (hereafter referred to as the "initial selection") and once after the discussion (the "post-selection"). Each selection was then compared to the five most important items agreed upon by the subject-matter experts. Accuracy was determined as a count of the items correctly identified from the experts' list, thus task performance ranged from zero to five.
Choice shift can either be measured through an overall group decision, which connotes a level of compliance simply due to being a member of the group, or it can be measured individually after the group has been dissolved, which reflects internalization of the group discussion (BarNir, 1998). The latter approach was chosen so that the measure would incorporate each group member's personal internalization of feedback. Thus, choice shift was measured as the number of items in the post-selection that were not retained from the subject's initial selection. Related to that, the change in each subject's task performance due to choice shift was subsequently determined as the difference in accuracy between his or her initial selection and post-selection. Finally, the time spent in discussion was measured as the minutes elapsed between the submission of the first comment and the submission of the final comment, which was retrieved directly from transcripts of each group discussion.
Another potential process loss that could be related to feedback is the fear of negative evaluation, although the ability to submit anonymous input through the use of CMC is claimed to reduce or even eliminate evaluation apprehension (Alavi, 1994). Nevertheless, it was considered as a possible control variable for data analysis. Fear of negative evaluation was measured using the twelve-item version (Leary, 1983) of the original 30-item scale developed by Watson and Friend (1969). The scale demonstrated adequate reliability ([alpha] = 0.88). However, a correlational analysis indicated that fear of negative evaluation did not significantly correlate with any of the dependent variables, so it was not added to the measurement model as a covariate. Further, a Levene's Test of Equality suggested that fear of negative evaluation did not introduce additional process losses in any of the three conditions as it varied equally across groups (p = 0.46).
Demographic,variables were also collected. The average age of the subjects was 23.5 years, and 63 percent of the sample was male. Neither sex nor age was correlated with any of the dependent variables, so they were not included in the final measurement model. A Levene's Test of Equality across treatments indicated that neither sex nor age was inordinately represented in any of the three conditions.
Before conducting the main data analysis, a correlational analysis was performed to determine the relationship between the three dependent variables. The results of the analysis are presented in Table 1 along with the descriptive statistics for each variable. All three dependent variables were found to be significantly correlated. Despite being statistically significant, the correlations between the variables were all below 0.80, a level that can indicate the presence of multicollinearity which does not appear to be a concern with these measures (Tabachnick and Fidell, 1996). Finally, through the use of calculated Mahalanobis distance values, the data were also examined to ensure the assumption of normality was upheld and to check for possible outliers.
Because the dependent variables were significantly correlated, the hypotheses were tested using multivariate analysis of variance (MANOVA) with the feedback treatments serving as the fixed factors for analysis. One analysis compared groups receiving confederate feedback to groups receiving no confederate feedback (i.e., control groups), and a subsequent analysis compared groups in the positive feedback condition with groups in the negative feedback condition. Prior to conducting the MANOVA, a t-test was conducted to determine if the subjects in any of the conditions were significantly more accurate with their initial vote than the subjects in the other conditions. A significant difference would have a direct impact on both the task performance variable and the choice shift variable, but there was not a significant difference in those initial opinions. On average, subjects initially selected 1.9 correct items from the list.
The results of the MANOVA are displayed in Table 2. A test of the omnibus model showed a statistically significant difference between the feedback/no feedback conditions on the combined dependent variables (F(2,165) = 2.77, p = 0.02, Wilks' Lambda = 0.92, [[eta].sup.2] = 0.08). Similarly, a significant result was obtained between the positive/negative feedback conditions (F(l,85) = 3.93, p = 0.003, Wilks' Lambda = 0.81, [[eta].sup.2] = 0.20).
The results for the separate dependent variables were examined next. The first pair of hypotheses investigated the differences in feedback on a group member's decision-making accuracy. The results of the MANOVA indicated that group members who were assigned to the no feedback condition were more accurate than the group members who were in a feedback condition, supporting Hypothesis 1A. Likewise, there was a similar significant difference between groups in positive and negative feedback conditions, with the members of groups in the negative feedback condition displaying more accuracy than their counterparts in the positive feedback condition, supporting Hypothesis 1B.
The second pair of hypotheses dealt with choice shift and any resulting changes in accuracy due to the group discussion. The MANOVA results showed a significant difference between the group members in the feedback conditions and the control group, with the control group members improving because of the discussion, but the feedback condition group members actually worsened (-0.31). This result supports Hypothesis 3A. The difference between group members in the positive and negative feedback conditions was significant as the negative choice shift was more pervasive in the positive feedback condition than in the negative feedback condition, supporting Hypothesis 3B. Finally, the fourth pair of hypotheses focused on the differences in feedback on the time groups spent in discussion. Here, the difference between groups in either feedback condition and the control groups was not significant, failing to support Hypothesis 4A. However, the difference in the time of discussion between groups in the positive and negative feedback condition was statistically significant, with groups in the negative feedback condition spending more time in discussion than groups in the positive feedback condition, supporting Hypothesis 4B.
Partial eta squared ([[eta].sup.2]) represents the proportion of the variance explained in the dependent variable, or the strength of association, related to each particular independent variable analyzed via MANOVA. Cohen's (1988) guidelines for interpreting the strength of association using eta squared stipulate that 0.01 is a small effect, 0.06 is a moderate effect, and 0.14 is a large effect. The eta squared values provided by the MANOVA indicate that the presence and the valence of feedback produce at least a moderate effect in decision accuracy and change in accuracy. The valence of confederate feedback alone explains ten percent of the variance in time spent in discussion, which is also a moderate effect size.
To summarize the results, clear differences in output variables between groups that received in the feedback condition and control groups. Group members receiving feedback from the confederate group member, regardless of its valence, performed less accurately than group members who did not receive confederate feedback. In fact, for group members in the feedback condition, the change in performance between their initial selections and their post-discussion selections (i.e., the change due to choice shift) was actually worse, whereas control group members who received no confederate feedback had an increase in performance. While there was no difference in the time spent in discussion between groups in the feedback conditions and control groups, the results here provide evidence that the presence of the positive or negative confederate feedback degraded group member performance.
The results also provided some insight into the effects of the valence of feedback contributed by the confederate group member. Group members who were in the negative feedback condition performed the task more accurately than those who were assigned to the positive feedback condition. Further, the groups that were exposed to the negative feedback from a confederate spent more time in discussion than groups presented with positive confederate feedback; a post hoc Scheffe test indicated that groups assigned to the positive feedback condition also spent significantly less time in discussion than groups that were in a no feedback condition. The difference in performance resulting from choice shift for group members in the positive feedback condition was significant at the [alpha] = 0.10 level. Taken together, the current results suggest that the positive confederate feedback did group members no favors, and they support recommendations like that of an intentional group "carper" (Connolly et al., 1990) whose sole purpose is to disrupt groups partaking in overly convivial discussion.
These results reinforce the theoretical reasoning behind the hypotheses, namely that feedback from other group members during discussion serves as a distraction and produces a type of process loss. Process loss was most evident in groups assigned to the negative feedback condition. Group members who received negative feedback from the confederate often assumed a peacemaker role. Because the feedback did not target any one group member specifically, group members frequently requested an interpretation or an elaboration of the feedback. This finding supports the idea that individuals seek to reduce uncertainty through interpersonal interactions.
While not necessarily relevant to the task, one can argue that the responses to negative confederate feedback fit a series of interactions that Sarker and Sahay (2003) describe as "dealing with trouble," a vital part of team development. "Trouble," in this context, can be considered as the actions taken by group member that are perceived as purposely contradicting or discrediting. Responses that are undertaken with the intention of repairing the damage of a troublesome group member are crucial for realigning group norms and re-establishing the spirit of cooperation (Sarker and Sahay, 2003). Otherwise, group members who actively avoid the negative interactant become "secondary provokers" and may find themselves incorporating the same behaviors as the original offender, making the group truly dysfunctional (Keyton, 1999). Though responding to the negative distractions may be classified as process losses, for the sake of group development group members are likely well-advised to isolate and distinguish themselves from negative group members.
However, the influence of positive feedback might not be best described as a process loss. Rather, the submission of positive feedback often appeared to prematurely end debate on a particular topic before it could be sufficiently discussed, which is symptomatic of the groupthink phenomenon. Positive comments made by the confederate throughout the discussion also set a convivial tone in the sense that other "group members" appeared to be eager to quickly settle any differences in opinion. Subjects often joined in with their own supportive comments adding to the agreeable tone of discussion which most likely had a direct influence on both the time spent in discussion as well as on the worsened performance suffered by group members in these particular groups. However, future research that examines group member reactions to a positive (or negative) comment at the moment it is received would provide much-needed empirical insight toward these results. Neuroimaging data collection might be useful in this regard.
Finally, the lack of support for Hypothesis 3A requires some reflection. Again, there was no significant difference in the time taken for discussion between groups receiving confederate feedback and groups without confederate feedback. However, the result for Hypothesis 3B indicated a wide difference in discussion time between groups in positive and negative feedback conditions, which suggests that the two types of feedback may have effectively canceled each other out when combining them for the H3A comparison. A post hoc Scheffe test across conditions shows that the no feedback condition is not significantly different in discussion time from either the positive feedback or the negative feedback conditions, placing it squarely in the middle in terms of discussion time, a result that is responsible for the lack of support for Hypothesis 3A.
The results of this study should also be of interest to practitioners, especially those supervising young workers who complete work through group collaboration. Anecdotal evidence indicates that young workers would like to receive feedback that is both constructive and timely (Martin and Tulgan, 2001). As organizations increasingly work to foster the feedback which younger workers crave, it is important for managers to understand the impact of feedback on performance in group collaborative settings. Also, because work in groups and teams has become increasingly prevalent with a shift towards the use of technology for group collaboration in the workplace (Bordia, 1997), there is a need to better understand how feedback affects performance under circumstances which involve group collaboration through the use of computers. Hence, the current study helps provide a better understanding of whether or not feedback enhances performance within the context of group collaboration through computer-mediated communication.
The results of this study should not be assessed without acknowledging its limitations. Undergraduate students served as subjects in the experiment, which can limit the generalizability of any findings to other segments of the population. However, a student sample is entirely valid for testing the hypotheses examined in this study. First, students, who use internet technology as a means of communication, are frequently required to complete group projects or assignments within the business school. Second, these activities are not unlike those they will encounter in their future careers. Third, students can be suitable subjects for other studies involving communicative processes and outcomes (Greenberg, 1987). Another limitation to this study was the type of decision-making task that was used. The task was closer to the judgmental side of the intellective-judgment task continuum (Smith et al., 2001) which may be a concern for some. Kluger and DeNisi (1996) proposed that subjective tasks can moderate the effects of feedback on performance, primarily because the cognitive resources needed for completing the task are less likely to be allocated toward digesting the feedback. However, the results of this experiment did not appear to show any moderation due to task characteristics.
In sum, the current study offers an exploratory view of the impact of feedback within CMC groups. The findings provide support for the notion that feedback may have negative consequences for CMC groups engaging in collaborative decision-making. The results indicate that negative feedback in particular offers enough of a distraction as to become a process loss during group discussion. On the other hand, while it may not serve as a process loss per se, the impact of positive feedback demonstrated in this study raises additional questions. For instance, is positive feedback alone sufficient to produce the phenomenon of groupthink? Does positive feedback alleviate social apprehensions in its recipients, and what performance effects does that have on groups with long-term membership? While the current study provides some initial insight into these important group processes, future research should be conducted in order to gain a more comprehensive look into the influence of feedback.
Positive and Negative Comments Submitted by the Confederate
NOTE: The comments below were generated by subjects participating in the pilot round of the experiment. These comments were submitted verbatim by the confederate in the order displayed below.
1. great what a good idea! YOU ROCK
3. Good job of think outside the box. (sic)
4. I like this, team work
5. You guys were very funny
6. Great ideas guys. Happy to work with you all
7. I think we all thought a lot alike and came up with a good item set
1. I have to say actually that would have been a horrible idea
3. honestly that has to be the worst person at survival i've ever heard of.
4. That was very dumb. You need to start thinking before you speak!!
5. speed it up!!!
6. I am highly thankful this was not real, because I for one believe we would die a horrible death.
7. I think this whole think was a waste of my time. Who would come up with someting like this, (sic)
Adkins, M. and D. Brashers. 1995. "The Power of Language in Computer-Mediated Groups." Management Communication Quarterly 8(3): 289-322.
Alavi, M. 1994. "Computer-Mediated Collaborative Learning: An Empirical Evaluation." MIS Quarterly 18(2): 159-174.
BarNir, A. 1998. "Can Group- and Issue-Related Factors Predict Choice Shift? A Meta-Analysis of Group Decisions on Life Dilemmas." Small Group Research 29(3): 308328.
Baron, R. 2005. ""So Right It's Wrong: Groupthink and the Ubiquitous Nature of Polarized Group Decision-making." Advances in Experimental Social Psychology 37: 219-253.
--. 1986. "Distraction-Conflict Theory: Progress and Problems." In Advances in Experimental Social Psychology. Orlando, FL: Academic Press, Inc.
Basoglu, A., M. Fuller, and T. Sweeney. 2009. "Investigating the Effects of Computer Mediated Interruptions: An Analysis of Task Characteristics and Interruption Frequency on Financial Performance." International Journal of Accounting Information Systems 10(4): 177-189.
Bordia, P. 1997. "Face-to-Face Versus Computer-Mediated Communication: A Synthesis of the Experimental Literature." The Journal of Business Communication 34(1): 99-120.
Bostrom, R., R. Anson, and V. Clawson. 1993. "Group Facilitation and Group Support Systems." In Group Facilitation and Group Support Systems edited by L. Jessup and J. Valacich. New York, NY: Macmillan.
Burgoon, J. and J. Hale. 1983. "Dimensions of Communication Reticence and Their Impact on Verbal Encoding." Communication Quarterly 31: 302-311.
Cecil, E. A., L. L. Cummings, and J. M. Chertkoff. 1976. "Group Composition and Choice Shift: Implications for Administration." Academy of Management Journal 16(3): 412-422.
Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences. Vol. 2. Hillsdale, NJ: Erlbaum.
Connolly, T., L. Jessup, and J. Valacich. 1990. "Effects of Anonymity and Evaluative Tone on Idea Generation in Computer-Mediated Groups." Management Science 36(6): 689-703.
Cooper, D. and J. H. Kagel. 2005. "Are Two Heads Better Than One? Team versus Individual Play in Signaling Games." American Economic Review 95: 477-509.
De Cremer, D. and E. van Dijk. 2002. "Reactions to Group Success and Failure as a Function of Identification Level: A Test of the Goal-Transformation Hypothesis in Social Dilemmas. "Journal of Experimental Social Psychology 38: 435-442.
DeNisi, A. and A. Kluger. 2000. "Feedback Effectiveness: Can 360-Degree Appraisals Be Improved?" Academy of Management Executive 14(1): 129-139.
Dennis, A. and M. Garfield. 2003. "The Adoption and Use of GSS in Project Teams: Toward More Participative Processes and Outcomes." MIS Quarterly 27(2): 289-323.
Druskat, V. U. and S. Wolff. 1999. "Effects and Timing of Developmental Peer Appraisals in Self-Managing Work Groups. "Journal of Applied Psychology 84(1): 58-74.
Farh, J.-L., A. Cannella, and A. Bedeian. 1991. "The Impact of Purpose on Rating Quality and User Acceptance." Group & Organization Management 16(4): 367-386.
Geister, S., U. Konradt, and G. Hertel. 2006. "Effects of Process Feedback on Motivation, Satisfaction, and Performance in Virtual Teams." Small Group Research 37(5): 459-489.
Greenberg, J. 1987. "The College Sophomore as Guinea Pig: Setting the Record Straight." Academy of Management Review 12(1): 157-159.
Hebert, B. and J. Vorauer. 2003. "Seeing through the Screen: Is Evaluative Feedback Communicated More Effectively in Face-to-Face or Computer-Mediated Exchanges?" Computers in Human Behavior 19(1): 25-38.
Hiltz, S. R., K. Johnson, and M. Turoff. 1991. "Group Decision Support: The Effects of Designated Human Leaders and Statistical Feedback in Computerized
Conferences. "Journal of Management Information Systems 8(2): 81-108. Hirokawa, R. and D. Gouran. 1989. "Facilitation of Group Communication: A Critique of Prior Research and an Agenda for Future Research." Management Communication Quarterly 3(1): 71-92.
Hollingshead, A., G. Wittenbaum, G. C. Jacobsohn, and S. Fraidin. 2005. "Competitive Members in Cooperative Groups." In IACM 18th Annual Conference. Seville, Spain.
Janis, I. 1972. Victims of Groupthink. 1st ed. Boston, MA: Houghton Mifflin Company. Keyton, J. 1999. "Analyzing Interaction Patterns in Dysfunctional Teams." Small Group Research 30(4): 491-518.
Kiesler, S. and L. Sproull. 1992. "Group Decision Making and Communication Technology." Organizational Behavior and Human Decision Processes 52: 96-123.
Kluger, A. and A. DeNisi. 1996. "The Effects of Feedback Interventions on Performance: A Historical Review, a Meta-Analysis, and a Preliminary Feedback Intervention Theory." Psychological Bulletin 119(2): 254-284.
Kroeck, K. G., B. Avolio, R. L. Small, and W. O. Einstein. 1987. "Shifts in Resource Allocation Preference Following Panel Discussion." Journal of Management 13(4): 713-724.
Kruglanski, A. and D. Webster. 1991. "Group Members' Reactions to Opinion Deviates and Conformists at Varying Degrees of Proximity to Decision Deadline and of Environmental Noise." Journal of Personality and Social Psychology 61: 212-225.
Kuhn, T. and M. S. Poole. 2000. "Do Conflict Management Styles Affect Group Decision Making? Evidence from a Longitudinal Field Study." Human Communication Research 26(4): 558-590.
Lam, S. and J. Schaubroeck. 2009. "Information Sharing and Group Efficacy Influences on Communication and Decision Quality." Asia Pacific Journal of Management.
Laurie, N. and J. Swaminathan. 2009. "Is Timely Information Always Better? The Effect of Feedback Frequency." Organizational Behavior and Human Decision Processes 108(2): 315-329.
Leary, M. 1983. "A Brief Version of the Fear of Negative Evaluation Scale." Personality and Social Psychology Bulletin 9: 371-375.
--. 1996. Self-Presentation: Impression Management and Interpersonal Behavior. Boulder, CO: Westview Press.
Lim, L. and I. Benbasat. 1996. "A Framework for Addressing Group Biases with Group Technology." Journal of Management Information Systems 13(3): 7-24.
Lindsley, D., D. Brass, and J. Thomas. 1995. "Efficacy-Performance Spirals: A Multilevel Perspective." Academy of Management Review 20(3): 645-678.
London, M. and V. Sessa. 2006. "Group Feedback for Continuous Learning." Human Resource Development Review 5(3): 303-329.
Lowry, P. B., T. Roberts, N. Romano, P. Cheney, and R. Hightower. 2006. "The Impact of Group Size and Social Presence on Small-Group Communication: Does Computer-Mediated Communication Make a Difference?" Small Group Research 37(6): 631-661.
Martin, C. A. and B. Tulgan. 2001. Managing Generation Y: Global Citizens Born in the Late Seventies and Early Eighties. Amherst, MA: HRD Press.
McGrath, J. 1984. Groups: Interaction and Performance. Englewood Cliffs, NJ: Prentice-Hall.
--and A. Hollingshead. 1994. Groups Interacting with Technology. Thousand Oaks, CA: Sage Publications.
Mejias, R. J. 2007. "The Interaction of Process Losses, Process Gains, and Meeting Satisfaction within Technology-Supported Environments." Small Group Research 38(1): 156-194.
Peterson, R. and K.J. Behfar. 2003. "The Dynamic Relationship between Performance Feedback, Trust, and Conflict in Groups: A Longitudinal Study." Organizational Behavior and Human Decision Processes 92(1): 102-112.
Pinsonneault, A., H. Barki, B. Gallupe, and N. Hoppen. 1999. "Electronic Brainstorming: The Illusion of Productivity." Information Systems Research 10(2): 110-133.
Potter, R. and P. Balthazard. 2004. "The Role of Individual Memory and Attention Processes during Electronic Brainstorming." MIS Quarterly 28(4): 621-643.
Rao, S. and J. Lim. 2000. "The Impact of Involuntary Cues on Media Effects." Paper read at 33rd Hawaii International Conference on System Sciences.
Sarker, S. and S. Sahay. 2003. "Understanding Virtual Team Development: An Interpretive Study. "Journal of the AIS 4: 1-38.
Schiable, T. and A. Jacobs. 1975. "Feedback III: Sequence Effects. Enhancement of Feedback Acceptance and Group Attractiveness by Manipulation of the Sequence and Valence of Feedback." Small Group Research 6(2): 151-171.
Seibold, D., R. Meyers, and Sunwolf. 1996. "Communication and Influence in Group Decision Making." In Communication and Group Decision Making, edited by R. Hirokawa and M. S. Poole. Thousand Oaks, CA: Sage.
Smith, C, R. S. Tindale, and E. Anderson. 2001. "The Impact of Shared Representations on Minority Influence in Freely Interacting Groups." In Group Consensus and Minority Influence edited by C. K. W. D. Drau and N. D. Vries. Maiden, MA: Blackwell Publishers.
Speier, C, J. Valacich, and I. Vessey. 1999. "The Influence of Task Interruption on Individual Decision Making: An Information Overload Perspective." Decision Sciences 30(2): 337-360.
Steiner, I. D. 1972. Group Process and Productivity. New York, NY: Academic Press.
Stone, D. L., H. G. Gueutal, and B. Mcintosh. 1984. "The Effects of Feedback Sequence and Expertise of the Rater on Perceived Feedback Accuracy." Personnel Psychology 37: 487-506.
Sussman, S. and L. Sproull. 1999. "Straight Talk: Delivering Bad News through Electronic Communication." Information Systems Research 10(2): 150-166.
Tabachnick, B. G. and L. S. Fidell. 1996. Using Multivariate Statistics. 3rd ed. New York, NY: HarperCollins.
Turner, M. and T. Horvitz. 2001. "The Dilemma of Threat: Group Effectiveness and Ineffectiveness under Adversity." In Groups at Work edited by M. Turner. Mahwah, NJ: Lawrence Erlbaum Associates.
Valacich, J. and C. Schwenk. 1995. "Devil's Advocacy and Dialectical Inquiry Effects on Face-to-Face and Computer-Mediated Group Decision Making." Organizational Behavior and Human Decision Processes 63(2): 158-173.
Waldersee, R. and F. Luthans. 1994. "The Impact of Positive and Corrective Feedback on Customer Service Performance. "Journal of Organizational Behavior 15(1): 83-95.
Watson, D. and R. Friend. 1969. "Measurement of Social-Evaluative Anxiety." Journal of Consulting and Clinical Psychology 33: 448-457.
West, M. A. 1990. "The Social Psychology of Innovation in Groups." In Innovation and Creativity at Work edited by M. A. West and J. L. Fair. Chichester, UK: Wiley and Sons.
Yoo, Y. and M. Alavi. 2001. "Media and Group Cohesion: Relative Influences on Social Presence, Task Participation, and Group Consensus." MIS Quarterly 25(3): 371390.
Zhou, L., J. Burgoon, J. Nunamaker, and D. Twitchell. 2004. "Automated Linguistics-Based Cues for Detecting Deception in Text-Based Asynchronous Computer-Mediated Communication: An Empirical Investigation." Group Decision & Negotiation 13: 81-106.
Laura E. Marler
Assistant Professor of Management
Mississippi State University
Associate Professor of Business Information Systems
Mississippi State University
Table 1 Correlations between Dependent and Control Variables Mean Task Choice Change (SD) Range Perf. Shift Acc. Task 1.95 Performance (0.7) 0 to 5 -- Choice 1.63 Shift (1.0) 0 to 5 0.19 * -- Change in 0.09 Accuracy (0.8) -5 to 5 0.64 ** 0.06 -- Time in 16.89 Discussion (5.2) Unlimited 0.33 ** 0.20 * 0.19 * Fear of Negative 3.46 1 to 7 0.05 0.07 0.13 Evaluation (0.83) Time Fear Neg, Ev Task Performance Choice Shift Change in Accuracy Time in Discussion -- Fear of Negative 0.03 -- Evaluation * p < 0.05, ** p < 0.01, Two-tailed tests. Table 2 Descriptive Statistics and Mean Differences between Treatments Decision Choice Treatment n Accuracy Shift Control 78 2.09 (0.7) 1.73 (1.1) Feedback 89 1.85 (0.7) 1.54 (0.8) p-value 0.01 * 0.23 Partial [[eta].sup.2] 0.03 0.01 Positive FB 52 1.66 (0.7) 1.60 (0.8) Negative FB 37 2.11 (0.6) 1.46 (0.9) p-value 0.005 ** 0.46 Partial [[eta].sup.2] 0.09 0.01 Change in Time in Treatment Accuracy Discussion Control 0.31 (0.7) 16.71 (6.0) Feedback -0.13 (0.8) 17.05 (4.4) p-value 0.00 ** 0.67 Partial [[eta].sup.2] 0.07 0.00 Positive FB -0.28 (0.9) 15.84 (4.6) Negative FB 0.14 (0.8) 18.70 (3.7) p-value 0.02 * 0.003 ** Partial [[eta].sup.2] 0.06 0.10 * p < 0.05 ** p < 0.01
|Printer friendly Cite/link Email Feedback|
|Author:||Marler, Laura E.; Marett, Kent|
|Publication:||Journal of Managerial Issues|
|Date:||Jun 22, 2013|
|Previous Article:||Impression management (IM) behaviors, IM culture, and job outcomes.|
|Next Article:||Employer policies for dependent care: the role of internal demographics.|