People who send messages (e.g., “I have no fuel”; “I did not receive the attached file”) convey their messages in a way that they think best suits the recipient. Message tuning refers to how senders tailor messages for specific recipients. For example, people give longer and more elaborate street directions and instructions to people whom they presume to be nonnatives or unfamiliar with a city.2 Also, senders capitalize on the knowledge that they believe the recipient already has (e.g., “Turn right when you see that big tree that the city pruned last week”). For this reason, team members send shorter, less complete messages to one another because they believe that they can capitalize on an existing shared knowledge base. However, team members often overestimate the commonality of information they share with others. Consequently, the messages they send become less clear.

Message senders present information that they believe will be favorably received by the recipient and, therefore, distort messages.3 For example, when people present a message to an audience whom they believe has either a pro- or anti-stance on a particular topic, they err in the direction of adopting the audience’s point of view. Because senders who bring bad news are not welcome, they often modify the news. Unfortunately, message distortion can wreak havoc on effective teamwork. Consider how the sexual abuse details in the Jerry Sandusky case were distorted. Former Pennsylvania State University staff members were put on trial in 2011 and were charged with perjury for purposely understating the severity of the child abuse allegations against football coach Jerry Sandusky, brought to their attention by a former assistant coach, Mike McQueary. Mr. McQueary apparently sanitized his version of the witnessed abuse when relaying the details, thereby leading to the staff members’ distorted recount of events.4

The saying-is-believing (SIB) effect occurs when a speaker tunes a message to suit an audience and, in the course of tuning the message, the speaker’s subsequent memories and impressions about the topic change.5 The SIB effect is even more pronounced when the audience validates the communicator’s message.6

Senders are not the only ones who distort messages. Recipients often hear what they want to hear when receiving messages, especially ambiguous ones. For example, when people are given neutral information about a product, they interpret it in a way that is favorable toward their own position. Furthermore, they selectively attend to information that favors their initial point of view and ignore or misinterpret information that contradicts their position.

People are remarkably poor at taking the perspective of others. For example, people who are privy to information and knowledge that they know others are not aware of still act as if others are aware of it, even though it would be impossible for the receiver to have this knowledge.7 This problem is known as the curse of knowledge.8 For example, in one simulation, traders who possessed privileged information that could have been used to their advantage behaved as if their trading partners also had access to the privileged information. Perspective-taking deficiencies also explain why some instructors who understand an idea perfectly are unable to teach students the same idea. Perspective-taking deficiencies explain why teams fail, even though every team member really wants to succeed. Emotional states play a role in effective knowledge transfer. Happy receivers are more likely to absorb and act on new information than angry or frustrated receivers.9 Moreover, knowledge transfer is greater when receivers and senders are in the same high-arousal, affective state, regardless of whether it is positive or negative.

People believe that their thoughts, attitudes, and reasons are much more transparent—that is, obvious to others—than is actually the case.10 Yet, members of teams often have no idea what their leaders are thinking, but the leaders believe they are being perfectly clear. Part of the reason for the illusion of transparency is that people find it impossible to put themselves in the position of the receiver. For example, when people are told to “tap out with their fingers” famous songs such as “Happy Birthday,” they significantly overestimate the likelihood that a listener will understand which song they are tapping—but the listeners hardly ever do!11 Perhaps it is for this reason that most communicators overestimate their effectiveness. In short, people expect others to understand them more often than others actually do.12

Indirect speech acts are the ways in which people ask others to do things—but in indirect ways. For example, consider the various ways of requesting that a person shut a door13:

1. Close the door.

2. Can you close the door?

3. Would you close the door?

4. It might help to close the door.

5. Would you mind awfully if I asked you to close the door?

6. Did you forget the door?

7. How about a little less breeze?

8. It’s getting cold in here.

9. I really don’t want the cats to get out of the house.

Each statement can serve as a request to perform that act, although (except for “close the door”) the sentence forms are not requests but assertions and questions. Thus, statements 2 through 9 are indirect speech acts; a listener’s understanding of the intention behind a communicator’s intention requires an extra cognitive step or two—and can often fail, especially in cases of stress.

Indirect speech acts are a function of the magnitude of the request being made (i.e., trivial requests, such as asking someone for the time of day, are easy to accommodate; asking someone if you can have a job is much more difficult to accommodate), the power the recipient has over the sender, and the social distance in the culture.14 Thus, as the magnitude of requests increases, the power distance increases, and the social distance increases, requests made by team members will become more indirect.

The uneven communication problem refers to the fact that in virtually any group a handful of people do the majority of the talking. For example, in a typical five-person group, two people do over 70 percent of the talking; in a six-person group, three people do over 70 percent of the talking; and in a group of eight, three people do 67 percent of the talking.15 Unfortunately, people who do the majority of the talking may not be the people who are the most informed about the problem. Exhibit 6-2 plots the percentage of communication attributed to each member in groups of four, six, and eight members; in all cases, communication is uneven and skewed.

Consider a typical group decision-making task in the ABC Company. Suppose a three-member top-executive committee, Allen, Booz, and Catz, is charged with the task of hiring a new manager for an important division within ABC. The company has determined that six pieces of information are critical to evaluate for this position:

• Previous experience (A)

• Academic grades (B)

• Standardized test scores (C)

• Performance in round 1 interview (D)

• Cultural and international experience (E)

• Letters of recommendation (F)

Allen, Booz, and Catz have narrowed the competition down to three candidates: Kate, Ken, and Kerry. As is standard practice in the company, members of the hiring committee specialize in obtaining partial information about each candidate. Stated another way, each member of the hiring committee has some of the facts about each candidate, but not all of the facts. Thus, Allen, Booz, and Catz are information dependent on one another.

What will happen when they discuss various candidates for the job? Consider three possible distributions of information (see Exhibit 6-4):

• Nonoverlapping case Each partner, Allen, Booz, and Catz, has unique information about each candidate.

• Distributed, partial overlap Each partner not only knows something about each candidate that others also know (common information) but also knows some unique information.

• Fully shared case Each partner knows full information about each candidate. In this sense, the partners are informational clones of one another.

The only difference among these three cases is the information redundancy, or how equally the information is distributed among decision makers. The collective intelligence of the partners is identical in all three cases. Does the distribution of information affect the way the partners make decisions? In a rational world, it should not, but in real teams, it does. The impact of information on the aggregate decision of the team is directly related to the number of members of the team who know the information prior to making a group decision. Stated simply, information that is held by more members before team discussion has more influence on team judgments than does information held by fewer members, independent of the validity of the information.

Simply put, the main determinant of how much a given fact influences a group decision is not the fact itself, but rather, how many people happen to be aware of this fact prior to group discussion. This team fallacy is known as the common information effect.23

This means that even though (in an objective sense) the six pieces of information are really equally important, the top management group will tend to overemphasize information (such as A and C in the distributed case) more than is warranted.

The common information effect has several important consequences. First, team members are more likely to discuss information that everyone knows, as opposed to unique information that each may have. This often means that technical information (which is often not fully shared) is not given the weight that experts believe it should have. Information that people have in common is not only more likely to be discussed, but it also gets discussed for a longer period of time, and this too can exert a significant bias on the integrity of decision making. People are better at remembering information they have read or described more than information they hear from others.24 The result is that teams often fail to make the decision that would be supported if all the team members had full information about the choices.

A hidden profile is a superior decision alternative, but its superiority is hidden from group members because each member has only a portion of the information that supports this superior alternative.25 Stated another way, the information held in common by group members favors a particular choice, whereas the unshared information contradicts the choice.

For the sake of example, let’s consider an executive meeting, in which three candidates (Alva, Jane, and Bill) are under consideration for promotion to partner in the organization—obviously, an extremely important decision. Each of the three candidates has been with the company for some number of years; each has made a different number and type of accomplishments. The executive group can promote only one person for the position at this time.

The group can benefit the organization by pooling individual members’ information so as to gain a complete picture of the qualifications of each candidate. This is particularly important when individual members of the decision-making team are biased by virtue of their own agendas.

Information that is known to only one or a few members will often be omitted from discussion.26 Team members are more likely not only to mention information if it was known to all before discussion but also to bring it up repeatedly and dwell on it throughout the discussion. Thus, the team decision will often reflect the common knowledge shared by members before discussion rather than the diverse knowledge emanating from their unique perspectives and experiences.

Consider the scenario in Exhibit 6-5. In this situation, the initial bias favors Bill. At the outset of the meeting, each team member has more information about him (five pieces of information). The information the team has about Bill is fully shared, meaning that all team members are apprised of this candidate’s qualifications prior to the meeting. Obviously, Bill has done an excellent job of marketing his own achievements within the organization!

However, consider Alva, who has a combined total of eight pieces of favorable information supporting his candidacy for the partnership. However, each member of the executive team is privy only to three pieces of information about this candidate, and the information is not redundant. In an objective sense, Alva is by far the most qualified; yet his accomplishments are not fully shared among the top management team—a factor that will not be corrected with discussion (at least unstructured discussion).

If this team was immune to the common information effect, and the members optimally combined and pooled their unique information, a hidden profile would emerge. A hidden profile is a conclusion that is apparent only after team members have fully shared information. In this case, Alva would prevail.

Common information also affects people’s memory for team discussions. People recall fewer unshared arguments from team discussion.27 Moreover, analysis of recorded discussions reveals that unshared arguments are less likely to be expressed.28

Unstructured, free-style discussion, even among trained professionals who have every motivation to make an accurate diagnosis, is insufficient for ensuring the quality of outcomes. We want to point out some plausible-appearing solutions that do not work, either because they actually reinforce the problem or do not address the problem adequately.

Mental models are mental representations of the world that allow people to understand, predict, and solve problems in a given situation.68 Mental models can represent a simple physical system, such as the trajectory of a thrown object; mental models also can represent a complex social system, such as an organization or financial system.

A team mental model is a common understanding that members of a group or team share about how something works.69 From this, members form expectations about what others will do in a given situation. Team members have mental models not only about the work they do but also about the operation of their team. Mental models do not develop naturally; rather, team mental models develop through the process of role identification behaviors (through which team members share information regarding their specialized knowledge, skills, and abilities).70 Teams with a cognitive foundation, like a mental model, perform better than teams that lack a cognitive foundation.71

There are two key considerations in terms of the mental models that members have about their actual work: the accuracy of the model and the degree of correspondence (or noncorrespondence) between members’ models.

Many people supplement their own memories, which are known to be highly limited and unreliable, with various external aids. For example, PDAs and GPSs allow us to store and retrieve important information externally rather than in our own long-term memory. Similarly, other people (e.g., friends, family, coworkers, and teammates) also function as external memory aids.

A transactive memory system (TMS) is a shared system for attending to, encoding, storing, processing, and retrieving information.78 Think of TMS as a division of mental labor. When each person learns in some general way what the other people on the team may know in detail, team members can share detailed memories. In essence, each team member cultivates the other members as external memory, and in doing so, they become part of a larger system. A TMS develops implicitly in many teams to ensure that important information is not forgotten. A TMS is a combination of two things: knowledge possessed by particular team members and awareness of who knows what. In this way, a TMS serves as an external storage device, such as a library or computer that can be visited to retrieve otherwise unavailable information. Teams that have a TMS have access to more and better information than any single group member does alone. A TMS is more beneficial to small groups that use quality as a performance measure, but more beneficial to large groups, groups in dynamic task environments, and groups in volatile knowledge environments that involve time as a critical performance measure.79

Team members instinctively expect that the “experts” on the team will remember the details most closely associated with their area of expertise. Even when the experts are not so clearly defined, people specialize in remembering certain kinds of information, and it is generally understood by all members of the team (although often implicitly) which person is to remember what. This way of processing information provides an advantage to teams because they collectively remember and use more information than individuals acting on their own—even the same number of individuals considered separately.

The main disadvantage is that team members depend on one another for knowledge and information. Teams that have been working together for years find it nearly impossible to reconstruct interactions with clients and other shared experiences without the other team members present.

What does a TMS do for teams? How does it affect productivity in terms of the key dimensions of performance criteria? More important, how can the manager best capitalize on the strengths of a TMS while minimizing the liabilities?

TMS eliminates much of the coordination loss that can plague team effectiveness.86 Teams that have a transactive memory structure because their members are familiar with one another are less likely to fall prey to the common information effect compared with teams composed of previously unacquainted persons.87 Thus, the key question for the manager is how to ensure that teams develop an accurate TMS.

Probably the most straightforward way is to simply ask members of the team to indicate what knowledge bases the other members of the group possess. If there is high intrateam agreement, the TMS is higher than if there is low agreement about who knows what.

In one investigation, the completion times of teams of doctors performing total joint replacements in hospital surgeries were examined.88 Three types of learning were examined: organizational experience (i.e., the number of times that kind of procedure had been performed); individual experience (i.e., the number of times a given person on a given team had performed the surgery); and team experience (i.e., the number of times any two people on a team had performed the surgery together). If successful surgical procedures were simply a function of accumulated expertise, then “team learning” should not matter. However, it does: The more times people have worked together as a team, the faster (and smoother) is their surgery. For example, holding all other kinds of experience constant, a team whose members have performed ten total knee replacements together takes 5 percent less time to complete the procedure when compared to a team that is just as accomplished but that has no experience in working together.

In an analysis of organizational learning in improvement project teams in hospital neonatal intensive care units, two distinct factors emerged as key: learn-what (activities that identify current best practices) and learn-how (activities that operationalize practices in a given setting).106 The hospital teams had greater success when they implemented practices that were supported by extensive evidence and when project team members engaged in learning activities designed to promote engagement by the unit. Thus, both learn-what (content) and learn-how (means) are important.

Many groups are “porous” in the sense that newcomers join groups and people rotate in and out of the group. Groups are more likely to adopt the routine of a rotator when they share a superordinate identity with that member.107 A group is also more likely to adopt a routine from a rotator when it is superior to their own. When groups do not share a superordinate identity, they fail to adopt the rotator’s ideas and knowledge, even when it is superior to their own and would have improved their performance.

How does direct experience with a task versus experience acquired vicariously from others affect team performance? Direct task experience leads to higher levels of team creativity and more divergent products than does indirect task experience (vicarious experience).108 Teams who learn through direct task experience develop better transactive memory systems than do teams who learn vicariously.

It is important to consider the feedback offered via experience; feedback that is too detailed and specific might do more harm than good. A study of 48 trainees in a furniture factory performance simulation revealed that increasing feedback specificity hindered knowledge transfer. Those who received less specific feedback relied more heavily on explicit information processing and had more exposure to the challenging aspects of the task than did those who received more specific feedback.109

It is inevitable that teams will fail or that the environment will challenge their ability to perform in unexpected and dramatic ways. Studies of teams’ ability to adapt to sudden and dramatic changes in workload, for example, reveal that learning and performance orientation affect their ability to adapt.110 And studies of team project failure reveal that teams who believe that their organization normalizes failure are able to recover more quickly.111

Teams whose members work together for longer periods are more likely to develop a TMS and will, therefore, be more productive. However, there is a countervailing force at work in teams that have been together for long periods of time, namely, routinization. That is, because a TMS is basically a set of expectations, certain working relationships may become entrenched over time. For example, groups that experience partial membership change tend to rely on the TMS structure that their old-timers developed in their original group; this is ultimately detrimental to performance.112 However, when old-timers are instructed to reflect upon their collective knowledge, these negative effects are minimized. When delegation is optional (which a TMS does not ensure), and in a world in which the team’s expectations about what is needed (e.g., consumer demand) are accurate, then more TMS should basically lead to more routinization of the task and, hence, to a more efficient channeling of efforts by team members (because there will be less coordination loss involved in understanding each team member’s role). Such expectations would seem to be best when there is little need for innovation. Thus, there is a precarious trade-off of sorts between routinization and innovation.

For much of the work that organizations do, routinization is a good thing; however, for a large part of what organizations do, innovation is desirable and necessary to meet the competitive challenges we outlined in Chapter 1. Thus, a well-defined TMS could hinder the team’s ability to be adaptive.

For these reasons, there may be significant problems associated with extended team longevity. As a case in point, consider an R&D facility of a large American corporation.113 The division, which included 345 engineers and scientific professionals, was geographically isolated from the rest of the organization. Katz examined 50 project groups in this division that varied greatly in terms of their longevity—that is, how long members of one group had worked with one another. For a period of 15 weeks, professionals kept records of their work-related communication; any time that a group member consulted or spoke with others, whether at the water cooler or in the parking lot, this was recorded. Later, the department managers (a total of seven) and the two lab directors evaluated the performance of each project produced by each group with which they were technically familiar. The performance of these groups increased as they gained longevity, but only up to a point. After 5 years of working together, team project performance declined steeply, as did intraproject communication, organizational communication, and external professional communication—basically all the types of communication that serve to bring fresh ideas to the group.

Four behavioral changes took place in groups that worked together for over 5 years:

• Behavioral stability Project members interacting over a long time develop standard work patterns that are familiar and comfortable. This can happen very rapidly—for instance, the way people in a group tend to sit in the same places in meeting after meeting, even when there may be no logical reason for doing so. Over time, this behavioral stability leads to isolation from the outside. The group can grow increasingly complacent, ceasing to question the practices that shape their behavior.

• Selective exposure There is a tendency for group members to communicate only with people whose ideas agree with their own. It is related to the homogeneity bias—the tendency to select new members who are like members in the existing group. Over time, project members learn to interact selectively to avoid messages and information that conflict with their established practices and disposition.

• Group homogeneity Groups that are separated from the influence of others in the organization develop a homogeneous set of understandings about the group and its environment.

• Role differentiation Groups that work and train together become increasingly specialized in project competencies and roles. This results in greater role differentiation, which in turn results in less interaction among group members because the roles and expectations held by each are so well entrenched. Consequently, they lose access to much of the internal talent, and their ability to learn new ideas from one another is diminished.

Thus, in terms of actual task performance, as team longevity increases, certain social processes conspire to lower levels of project communication, which in turn decrease project performance. Project groups become increasingly isolated from key information sources both within and outside their organizations with increasing stability in their membership. Reductions in project communication adversely affect the technical performance of project groups. Variations in communication activities are more associated with the tenure composition of the project group than with the project tenures of the individual engineers. Stated another way, it is not the age of the employee that is of critical importance, but the age of the team. Furthermore, individual competence does not account for differences in performance. Thus, it is not the case that the older, less skilled members were working in teams that were of greater longevity. Furthermore, the longevity of project groups does not appear to be part of the mental models of the managers—virtually no one was aware of the tenure demographics of their project groups.

What does all of this mean for team longevity? A certain amount of familiarity is necessary for teams to work together in a productive fashion. The effect of working together tends to make team members grow more familiar with each other’s relevant knowledge base, and hence, a TMS can develop. A TMS can be helpful in tasks where coordination losses need to be reduced and tactical precision is key. Although a certain amount of routinization is desirable in any team, the overly routinized team hinders communication and obstructs innovation. Looking at this question from a team design standpoint might offer some insights. It may be desirable, for example to design some teams whose primary objective is to act as innovation experts for the creation and transfer of the organization’s best practices. Groups are more likely to benefit from the knowledge and ideas brought by a newcomer when they share a superordinate identity with one another.114 Moreover, when members sit in an integrated fashion (versus sitting on opposite sides of a table), a superordinate identity is more likely to be established.