Group Performance

When important tasks need to be performed quickly or effectively, we frequently create groups to accomplish them. Many people believe that groups are more effective than individuals in performing tasks (Nijstad, Stroebe, & Lodewijkx, 2006), and such a belief seems commonsensical. After all, because groups have many members, they will also have more resources and thus more ability to efficiently perform tasks and make good decisions. However, although groups sometimes do perform better than individuals, this outcome is not guaranteed. Let’s consider some of the many variables that can influence group performance.

Social Facilitation and Social Inhibition

In one of the earliest social psychological studies, Norman Triplett (1898) investigated how bicycle racers were influenced by the social situation in which they raced. Triplett found something very interesting: the racers who were competing with other cyclists on the same track rode significantly faster than those who were racing alone, against the clock. This led Triplett to hypothesize that people perform tasks better when the social context includes other people than when they do the tasks alone. Subsequent findings validated Triplett’s results, and other experiments have shown that the presence of others can increase performance on many types of tasks, including jogging, playing pool, lifting weights, and working on mathematics and computer problems (Geen, 1989; Guerin, 1983; Robinson-Staveley & Cooper, 1990; Strube, Miles, & Finch, 1981). The tendency to perform tasks better or faster in the presence of others is known as social facilitation.

Although people sometimes perform better when they are in groups than they do alone, the situation is not that simple. Perhaps you can remember a time when you found that a task you could perform well alone (e.g., giving a presentation, playing a video game, shooting a basketball free throw, or making a soccer penalty kick) was not performed as well when you tried it with, or in front of, others. Thus it seems that the conclusion that being with others increases performance cannot be entirely true and that sometimes the presence of others can worsen our performance. The tendency to perform tasks more poorly or slower in the presence of others is known as social inhibition. So, as the presence of other people can both improve and worsen individual performance, it is important to explore the different conditions that promote these opposite outcomes.

To study social facilitation and social inhibition, Hazel Markus (1978)  gave research participants both an easy task (putting on and tying their shoes) and an unfamiliar and thus more difficult task (putting on and tying a lab coat that tied in the back). The research participants were asked to perform both tasks in one of three social situations: (a) alone, (b) with a confederate present who was watching them, or (c) with a confederate present who sat in the corner of the room repairing a piece of equipment without watching. As you can see in Figure 10.4, “Group Task Performance,” Markus found first that the difficult task was performed more slowly overall. But she also found an interaction effect, where the participants performed the easy task faster but the more difficult task slower when a confederate was present in the room. Furthermore, it did not matter whether the other person was paying attention to the performance or whether the other person just happened to be in the room working on another task—the mere presence of another person nearby influenced performance.

In this experiment, participants were asked to perform a well-learned task (tying their shoes) and a poorly learned task (putting on a lab coat that tied in the back). There is both a main effect of task difficulty and a task-difficulty-by-performance-condition interaction. Data are from Markus (1978).

These results convincingly demonstrated that working around others could either help or hinder performance. But why would this be? One explanation of the influence of others on task performance was proposed by Robert Zajonc (1965). As shown in Figure 10.5, “Explaining Social Facilitation and Social Inhibition,” Zajonc made use of the affective component of arousal in his explanation. Zajonc argued that when we are with others, we experience more arousal than we do when we are alone, and that this arousal increases the likelihood that we will perform the dominant response—the action that we are most likely to emit in any given situation.

According to the social facilitation model of Robert Zajonc (1965), the mere presence of others produces arousal, which increases the probability that the dominant response will occur. If the dominant response is correct, the task is performed better, whereas if the dominant response is incorrect, the task is performed more poorly.

The important aspect of Zajonc’s theory was that the experience of arousal and the resulting increase in the performance of the dominant response could be used to predict whether the presence of others would produce social facilitation or social inhibition. Zajonc argued that if the task to be performed was relatively easy, or if the individual had learned to perform the task very well (a task such as pedaling a bicycle or tying one’s shoes), the dominant response was likely to be the correct response, and the increase in arousal caused by the presence of others would improve performance. On the other hand, if the task was difficult or not well learned (e.g., solving a complex problem, giving a speech in front of others, tying a lab apron behind one’s back), the dominant response was likely to be the incorrect one; and because the increase in arousal would increase the occurrence of the (incorrect) dominant response, performance would be hindered.

Zajonc’s theory explained how the presence of others can increase or decrease performance, depending on the nature of the task, and a great deal of experimental research has now confirmed his predictions. In a meta-analysis, Bond and Titus (1983) looked at the results of over 200 studies using over 20,000 research participants and found that the presence of others did significantly increase the rate of performance on simple tasks and decrease both the rate and the quality of performance on complex tasks.

One interesting aspect of Zajonc’s theory is that because it only requires the concepts of arousal and dominant response to explain task performance, it predicts that the effects of others on performance will not necessarily be confined to humans. Zajonc reviewed evidence that dogs ran faster, chickens ate more feed, ants built bigger nests, and rats had more sex when other dogs, chickens, ants, and rats, respectively, were around (Zajonc, 1965). In fact, in one of the most unusual of all social psychology experiments, Zajonc, Heingartner, and Herman (1969) found that cockroaches ran faster on straight runways when other cockroaches were observing them (from behind a plastic window) but that they ran slower, in the presence of other roaches, on a maze that involved making a difficult turn, presumably because running straight was the dominant response, whereas turning was not.

Although the arousal model proposed by Zajonc is perhaps the most elegant, other explanations have also been proposed to account for social facilitation and social inhibition. One modification argues that we are particularly influenced by others when we perceive that the others are evaluating us or competing with us (Szymanski & Harkins, 1987). This makes sense because in these cases, another important motivator of human behavior—the desire to enhance the self—is involved in addition to arousal. In one study supporting this idea, Strube and his colleagues (Strube, Miles, & Finch, 1981) found that the presence of spectators increased the speed of joggers only when the spectators were facing the joggers and thus could see them and assess their performance.

The presence of others who expect us to do well and who are thus likely to be particularly distracting has been found to have important consequences in some real-world situations. For example, Baumeister and Steinhilber (1984) found that professional athletes frequently performed more poorly than would be expected in crucial games that were played in front of their own fans.

Process Losses and Process Gains

So far in this section, we have been focusing on how being in a group affects individual performance. What about the broader question of whether performance is enhanced when people work in groups, compared with what group members would have achieved if they had been working on their own? Working in groups clearly has some benefits. Because groups consist of many members, group performance is almost always better than the performance of an individual acting alone. Many heads are better than one in terms of knowledge, collective memory, physical strength, and other abilities. The group from the National Aeronautics and Space Administration (NASA) that worked together to land a human on the moon, a music band whose members are writing a new song together, or a surgical team in the middle of a complex operation may coordinate their efforts so well that is clear that the same outcome could never have occurred if the individuals had worked alone, or in another group of less well-suited individuals. In these cases, the knowledge and skills of the individuals seem to work together to be effective, and the outcome of the group appears to be enhanced. When groups work better than we would expect, given the individuals who form them, we call the outcome a process gain.

There are at least some data suggesting that groups may in some cases experience process gains. For instance, Weber and Hertel (2007) found in a recent meta-analysis that individuals can in some cases exert higher motivation when working in a group compared with working individually, resulting in increased group performance. This is particularly true for less capable group members who seem to become inspired to work harder when they are part of a group. On the other hand, there are also costs to working in groups—sometimes being in a group can stifle creativity and increase procrastination, for example. In these cases, the groups experience process losses. A process loss occurs when groups perform more poorly than we would expect, given the characteristics of the members of the group.

One way to think about the benefits of groups is to compare the potential productivity of the group—that is, what the group should be able to do, given its membership—with the actual productivity of the group. For example, on a rope-pulling task, the potential group productivity (the strength with which the group should pull when working together) would be calculated as the sum of all the individual inputs. The difference between the expected productivity of the group and the actual productivity of the group (i.e., the extent to which the group is more or less than the sum of its parts) is determined by the group process, defined as the events that occur while the group is working together on the task. When the outcome of the group performance is better than would be expected on the basis of the members’ characteristics (the group pulls harder than expected), there is a process gain; when the outcome of the group performance is worse than would be expected on the basis of the members’ characteristics, there is a process loss. Mathematically, we can write the following equation to express this relationship:

actual productivity = potential productivity − process loss + process gain.

Group performance is another example of a case in which person and situation variables work together because it depends on both the skills of the people in the group and the way these resources are combined as the group members work together. Let’s now turn to exploring these personal and situational factors in more detail.

Person Variables: Group Member Characteristics

No matter what type of group we are considering, the group will naturally attempt to recruit the best people they can find to help them meet their goals. Member characteristics are the relevant traits, skills, or abilities of the individual group members. On a rope-pulling task, for instance, the member characteristic is the ability of each of group member to pull hard on the rope on his or her own. In addition to having different skills, people differ in personality factors that relate to group performance. Some people are highly motivated to join groups and to make positive contributions to those groups, whereas others are more wary of group membership and prefer to meet their goals working alone. Furthermore, when they are in groups, people may be expected to respond somewhat differently in group interactions, because each is using the group to meet his or her own social and personal goals.

The extent to which member skill influences group performance varies across different group tasks. On a car assembly line, performing the task requires only relatively minimal skills, and there is not a lot of coordination among the individuals involved. In this case, it is primarily the number and skill of the individuals who are working on the task that influences the group outcome. In other cases, such as a surgical team or a work team within a corporation, the group includes individuals with a wide variety of different skills, each working at very different tasks. In cases such as these, communication and coordination among the group members is essential, and thus group process will be very important. As an example of variation in the importance of member skills in different sporting contexts, Jones (1974) found that the skill of individual baseball players accounted for 99% of the team performance on baseball teams (and thus group process accounted for only 1%) but that the skill of individual basketball players accounted for only 35% of the team performance on basketball teams (and thus group process accounted for 65%).

The Importance of the Social Situation: Task Characteristics

Although the characteristics of the group members themselves are critical, they represent only the person part of the equation. To fully understand group performance, we must also consider the particulars of the group’s situation—for instance, the task that the group needs to accomplish. Let’s now consider some of the different types of tasks that might be performed by groups and how they might influence performance (Hackman & Morris, 1975; Straus, 1999).

One basic distinction concerns whether the task can be divided into smaller subtasks or has to be done as a whole. Building a car on an assembly line or painting a house is a divisible task, because each of the group members working on the job can do a separate part of the job at the same time. Groups are likely to be particularly productive on divisible tasks when the division of the work allows the group members to specialize in those tasks that they are best at performing. Writing a group term paper is facilitated if one group member is an expert typist, another is an expert at library research, and so forth. Climbing a mountain or moving a piano, on the other hand, is a unitary task, because it has to be done all at once and cannot be divided up. In this case, specialization among group members is less useful, because each group member has to work on the same task at the same time.

Another way of classifying tasks is by the way the contributions of the group members are combined. On an additive task, the inputs of each group member are added together to create the group performance, and the expected performance of the group is the sum of group members’ individual inputs. A tug of war is a good example of an additive task because the total performance of a team is expected to be the sum of all the team members’ individual efforts.

On a compensatory (averaging) task, however, the group input is combined such that the performance of the individuals is averaged rather than added. Imagine that you wanted to estimate the current temperature in your classroom, but you had no thermometer. One approach to getting an estimate would be to have each of the individuals in your class make his or her estimate of the temperature and then average the estimates together to create a group judgment. On decisions such as this, the average group judgment is likely to be more accurate than that made by most individuals (Armstrong, 2001; Surowiecki, 2004).

Another task classification involves comparing tasks in which the group performance is dependent upon the abilities of the best member or members of the group with tasks in which the group performance is dependent upon the abilities of the worst member or members of the group. When the group’s performance is determined by the best group member, we call it a disjunctive task. Consider what might happen when a group is given a complicated problem to solve, such as this horse-trading problem:

A man buys a horse for $50. He later decides he wants to sell his horse and he gets $60. He then decides to buy it back and pays $70. However, he can no longer keep it, and he sells it for $80. Did he make money, lose money, or break even? Explain why.

The correct answer to the problem is not immediately apparent, and each group member will attempt to solve the problem. With some luck, one or more of the members will discover the correct solution, and when that happens, the other members will be able to see that it is indeed the correct answer. At this point, the group as a whole has correctly solved the problem, and the performance of the group is thus determined by the ability of the best member of the group.

In contrast, on a conjunctive task, the group performance is determined by the ability of the group member who performs most poorly. Imagine an assembly line in which each individual working on the line has to insert one screw into the part being made and that the parts move down the line at a constant speed. If any one individual is substantially slower than the others, the speed of the entire line will need to be slowed down to match the capability of that individual. As another example, hiking up a mountain in a group is also conjunctive because the group must wait for the slowest hiker to catch up.

Still another distinction among tasks concerns the specific product that the group is creating and how that group output is measured. An intellective task involves the ability of the group to make a decision or a judgment and is measured by studying either the processes that the group uses to make the decision (such as how a jury arrives at a verdict) or the quality of the decision (such as whether the group is able to solve a complicated problem). A maximizing task, on the other hand, is one that involves performance that is measured by how rapidly the group works or how much of a product they are able to make (e.g., how many computer chips are manufactured on an assembly line, how many creative ideas are generated by a brainstorming group, how fast a construction crew can build a house).

Finally, we can differentiate intellective task problems for which there is an objectively correct decision from those in which there is not a clear best decision. On a criterion task, the group can see that there is a clearly correct answer to the problem that is being posed. Some examples would be finding solutions to mathematics or logic problems, such as the horse-trading problem.

On some criterion tasks, the correct answer is immediately seen as the correct one once it is found. For instance, what is the next letter in each of the following two patterns of letters?

J F M A M _

O T T F F _

In criterion problems such as this one, as soon as one of the group members finds the correct answer, the problem is solved because all the group members can see that it is correct. Criterion tasks in which the correct answer is obvious once it is found are known as “Eureka!” or “Aha!” tasks (Lorge, Fox, Davitz, & Brenner, 1958), named for the response that we have when we see the correct solution.

In other types of criterion-based tasks, there is an objectively correct answer, although that answer is not immediately obvious. For instance, consider again the horse-trading problem. In this case, there is a correct answer, but it may not be apparent to the group members even when it is proposed by one or more of them (for this reason, we might call this a “non-Eureka” task). In fact, in one study using the horse-trading problem, only 80% of the groups in which the correct answer was considered actually decided upon that answer as the correct one after the members had discussed it together.

In still other criterion-based tasks, experts must be used to assess the quality or creativity of the group’s performance. Einhorn, Hogarth, and Klempner (1977) asked groups of individuals to imagine themselves as a group of astronauts who are exploring the moon but who have become stranded from their base. The problem is to determine which of the available pieces of equipment (e.g., oxygen bottles, a rope, a knife) they should take with them as they attempt to reach the base. To assess group performance, experts on the difficulties of living in space made judgments about the quality of the group decisions. Non-Eureka tasks represent an interesting challenge for groups because even when they have found what they think is a good answer, they may still need to continue their discussion to convince themselves that their answer is the best they can do and that they can therefore stop their deliberation.

In contrast to a criterion task, in a judgmental task there is no clearly correct answer to the problem. Judgmental tasks involve such decisions as determining the innocence or guilt of an accused person in a jury or making an appropriate business decision. Because there is no objectively correct answer on judgmental tasks, the research approach usually involves studying the processes that the group uses to make the decision rather than measuring the outcome of the decision itself. Thus the question of interest on judgmental tasks is not “Did the group get the right answer?” but rather “How did the group reach its decision?” Evaluating the quality of how the decision was reached, compared with the decision itself, can be particularly challenging (Johnson & Johnson, 2012).

So, clearly the nature of the task will influence group performance and whether people perform better together, as opposed to alone. One phenomenon that dramatically illustrates these points is that of social loafing.

Social Loafing

In a seminal study of group effects on individual performance, Ringelmann (1913; reported in Kravitz & Martin, 1986) investigated the ability of individuals to reach their full potential when working together on tasks. Ringelmann had individual men and groups of various numbers of men pull as hard as they could on ropes while he measured the maximum amount that they were able to pull. Because rope pulling is an additive task, the total amount that could be pulled by the group should be the sum of the contributions of the individuals. However, as shown in Figure 10.7, “The Ringelmann Effect,” although Ringelmann did find that adding individuals to the group increased the overall amount of pulling on the rope (the groups were better than any one individual), he also found a substantial process loss. In fact, the loss was so large that groups of three men pulled at only 85% of their expected capability, whereas groups of eight pulled at only 37% of their expected capability.

Ringelmann found that although more men pulled harder on a rope than fewer men did, there was a substantial process loss in comparison with what would have been expected on the basis of their individual performances.

This type of process loss, in which group productivity decreases as the size of the group increases, has been found to occur on a wide variety of tasks, including maximizing tasks such as clapping and cheering and swimming (Latané, Williams, & Harkins, 1979; Williams, Nida, Baca, & Latané, 1989), and judgmental tasks such as evaluating a poem (Petty, Harkins, Williams, & Latané, 1977). Furthermore, these process losses have been observed in different cultures, including India, Japan, and Taiwan (Gabrenya, Wang, & Latané, 1985; Karau & Williams, 1993).

Process losses in groups occur in part simply because it is difficult for people to work together. The maximum group performance can only occur if all the participants put forth their greatest effort at exactly the same time. Since, despite the best efforts of the group, it is difficult to perfectly coordinate the input of the group members, the likely result is a process loss such that the group performance is less than would be expected, as calculated as the sum of the individual inputs. Thus actual productivity in the group is reduced in part by these coordination losses.

Coordination losses become more problematic as the size of the group increases because it becomes correspondingly more difficult to coordinate the group members. Kelley, Condry, Dahlke, and Hill (1965) put individuals into separate booths and threatened them with electrical shock. Each person could avoid the shock, however, by pressing a button in the booth for three seconds. But the situation was arranged so that only one person in the group could press the button at one time, and therefore the group members needed to coordinate their actions. Kelley and colleagues found that larger groups had significantly more difficulty coordinating their actions to escape the shocks than did smaller groups.

However, coordination loss at the level of the group is not the only explanation of reduced performance. In addition to being influenced by the coordination of activities, group performance is influenced by self-concern on the part of the individual group members. Since each group member is motivated at least in part by individual self-concerns, each member may desire, at least in part, to gain from the group effort without having to contribute very much. You may have been in a work or study group that had this problem—each group member was interested in doing well but also was hoping that the other group members would do most of the work for them. A group process loss that occurs when people do not work as hard in a group as they do when they are alone is known as social loafing (Karau & Williams, 1993).

Research Focus

Differentiating Coordination Losses from Social Loafing

Latané, Williams, and Harkins (1979) conducted an experiment that allowed them to measure the extent to which process losses in groups were caused by coordination losses and by social loafing. Research participants were placed in a room with a microphone and were instructed to shout as loudly as they could when a signal was given. Furthermore, the participants were blindfolded and wore headsets that prevented them from either seeing or hearing the performance of the other group members. On some trials, the participants were told (via the headsets) that they would be shouting alone, and on other trials, they were told that they would be shouting with other participants. However, although the individuals sometimes did shout in groups, in other cases (although they still thought that they were shouting in groups) they actually shouted alone. Thus Latané and his colleagues were able to measure the contribution of the individuals, both when they thought they were shouting alone and when they thought they were shouting in a group.

The results of the experiment are presented in Figure 10.8, which shows the amount of sound produced per person. The top line represents the potential productivity of the group, which was calculated as the sum of the sound produced by the individuals as they performed alone. The middle line represents the performance of hypothetical groups, computed by summing the sound in the conditions in which the participants thought that they were shouting in a group of either two or six individuals, but where they were actually performing alone. Finally, the bottom line represents the performance of real two-person and six-person groups who were actually shouting together.

Individuals who were asked to shout as loudly as they could shouted much less so when they were in larger groups, and this process loss was the result of both motivation and coordination losses. Data from Latané, Williams, and Harkins (1979).

The results of the study are very clear. First, as the number of people in the group increased (from one to two to six), each person’s individual input got smaller, demonstrating the process loss that the groups created. Furthermore, the decrease for real groups (the lower line) is greater than the decrease for the groups created by summing the contributions of the individuals. Because performance in the summed groups is a function of motivation but not coordination, and the performance in real groups is a function of both motivation and coordination, Latané and his colleagues effectively showed how much of the process loss was due to each.

Social loafing is something that everyone both engages in and is on the receiving end of from time to time. It has negative effects on a wide range of group endeavors, including class projects (Ferrari & Pychyl, 2012), occupational performance (Ülke, & Bilgiç, 2011), and team sports participation (Høigaard, Säfvenbom, & Tønnessen, 2006). Given its many social costs, what can be done to reduce social loafing? In a meta-analytic review, Karau and Williams (1993) concluded that loafing is more likely when groups are working on additive than non-additive tasks. They also found that it was reduced when the task was meaningful and important to group members, when each person was assigned identifiable areas of responsibility, and was recognized and praised for the contributions that he or she made. These are some important lessons for all us to take forward here, for the next time we have to complete a group project, for instance! 

As well as being less likely to occur in certain tasks under certain conditions, there are also some personal factors that affect rates of social loafing. On average, women loaf less than men (Karau & Williams, 1993). Men are also more likely to react to social rejection by loafing, whereas women tend to work harder following rejection (Williams & Sommer, 1997). These findings could well help to shed some light on our chapter case study, where we noted that mixed-gender corporate boards outperformed their all-male counterparts. Simply put, we would predict that groups that included women would engage in less loafing, and would therefore show higher performance.

Culture, as well as gender, has been shown to affect rates of the social loafing. On average, people in individualistic cultures loaf more than those in collectivistic cultures, where the greater emphasis on interdependence can sometimes make people work harder in groups than on their own (Karau & Williams, 1993).