|Subject Terms:||Organizational behavior|
Multiculturalism & pluralism
Multiculturalism & pluralism
|Classification Codes:||9190: US|
2500: Organizational behavior
9130: Experimental/theoretical treatment
|Copyright Plenum Publishing Corporation Oct 1998|
INTRODUCTION Jackson, May, and Whitney (1995) define diversity as ". . . the presence of differences among members of a social unit" (p. 217). Diversity is an increasingly important factor in organizational life as organizations worldwide become more diverse in terms of the gender, race, ethnicity, age, national origin, and other personal characteristics of their members. For example, demographic and economic changes occurring within the U.S. mean that by the year 2000, the American workforce is likely to be gender-balanced, with only 58% of the workforce comprised of White, nativeborn Americans (Jackson et al., 1995). With the increasing age of the "baby boomers" and the inclusion of "generation X" into organizational life, employees of diverse ages will be more likely to work with one another. In addition, the increasing globalization of business requires employees from different cultures to work together in cross-national teams. Even in purely domestic operations, firms are being forced to form cross-functional, interdepartmental, cross-divisional, and interorganizational alliances in order to make maximum use of scarce resources and thus increase their competitive advantage (Jackson et al., 1995).
Diversity and its impact on how we work together is particularly critical when we look at the current management literature on teamwork and its role in increasing organizational efficiency and effectiveness. Morgan, Glickman, Woodard, Blaiwes, and Salas (1986) define a team simply as ". . . a distinguishable set of two or more individuals who interact independently and adaptively to achieve specified shared, and valued objectives." Guzzo (1995) suggests that, "teams and teaming have become hot topics . . . as organizations have come to rely on team-based arrangements to improve quality, productivity, customer service, and the experience of work for their members." In case after case, we find organizational re-engineers relying heavily on teams and teamwork to build high-performing organizations, as there is a widely held belief that teams are critical tools for solving problems and making decisions in a highly complex, international environment (Tjosvold, 1995). Much of the existing literature on team effectiveness and small group processes has been derived from research on relatively homogeneous groups or where the exact composition of the teams has not been taken into account. It is therefore uncertain as to whether the claims made for the benefits derived from teams and teamwork for improving organizational effectiveness will hold in circumstances involving teams of diverse composition.
To date, research has indicated that diversity among members in small groups can yield both benefits and costs (Campion, Medsker, & Higgs, 1993; Guzzo & Dickson, 1996; Jackson, 1991; Magjuka & Baldwin, 1991). For example, Campion et al. (1993) found that difference in member background and expertise either had no impact or decreased group effectiveness depending on the criterion measures used. Magjuka and Baldwin (1991) found that within-group diversity had positive effects on group performance in a sample of 72 manufacturing teams. Other studies have shown that diverse work groups can be more creative and arrive at better quality decisions than homogeneous groups (McGrath, 1984; McLeod & Lobel, 1992), but they can also be less integrated (O'Reilly, Caldwell, & Barnett, 1989), less satisfied and more prone to turnover (Jackson, Brett, Sessa, Copper, Julin, & Peyronnin, 1991).
Precisely how and why differences among small group members affect group effectiveness is unclear. In their review of research on group effectiveness, Guzzo and Dickson (1996) state that there "is evidence that team effectiveness is well-served by diverse members when teams perform cognitive, creativity-demanding tasks. This is not to say that diverse membership might not pay off in enhanced effectiveness in other task domains; rather, too little is known to draw firm conclusions . . . . In fact there is a real need to develop theory and data on the ways in which dissimilarity among members contributes to task performance" (p. 331).
In this article, we heartily agree with Guzzo and Dickson's call for more precise theories for understanding the effects of group member diversity on group process and performance. We suggest that these types of findings might be explained by using a "group-development" model to examine the impact of diversity on group processes and performance. The model we propose borrows heavily from Jackson et al. (1995) and Milliken and Martins (1996), although it is different in two significant ways. First, we use the terminology of Tuckman (1965) and Tuckman and Jensen (1977), though not their exact stage-development predictions, to identify four key requirements of successful group activity. These are to form a cohesive social unit, to manage conflict that invariably occurs during periods of group storming to establish acceptable norms of behavior and goals for the group, and finally to problem-solve, make decisions, and take actions that allow the group to perform successfully. Our model uses concepts from Jackson et al.'s (1995) and Milliken and Martins' (1996) models, as well as some of our own concepts, to show how diversity affects these four aspects of group development.
The second difference in our model is that we propose factors which moderate the impact of diversity on group performance and have incorporated them specifically into our model. These moderating factors operate at two different points in the process by which diversity affects group development and performance. Our overall model shown in Fig. 1, involves components that are intentionally somewhat general in nature so as to be applicable to a wide range of tasks and groups. In the pages that follow, we will explain each of the components of the model and suggest specific hypotheses generated from the model.
In beginning our description, it seems necessary to "start in the middle" by outlining the group development aspects of our model. There is a vast amount of literature that describes the processes required for effective group functioning and several models of how groups form and develop over time. For example, Pearce and Raylin (1987) outline a number of process activities needed in successful groups. They suggest that work groups require open communication. Groups must be able to coordinate themselves in a flexible manner, and there must be commitment to group goals which both direct and motivate members. Gladstein (1984) suggests that successful groups require open communication, mutual supportiveness, effective conflict management, discussion of strategy, and the appropriate weighing of individual inputs into group decisions.
In our model, we use the terminology of Tuckman (1965) and Tuckman and Jensen (1977) to summarize most of these requirements for good group functioning. We have selected this terminology because it is widely recognized in the group development literature and the concepts of forming, storming, norming, and performing provide a fairly parsimonious set of concepts around which we can build our diversity model. We reiterate that in using Tuckman's terminology, we are not adopting his idea that groups progress through each of these stages of development in a sequential manner. We believe that forming, storming, norming, and performing may occur many times during a group's life. For example, after a significant period of storming, groups may return to forming activities in order to re-establish a sense of social cohesion and attraction to one another. Most individuals can recall group situations where, following a serious episode of storming, group members have behaved in extremely polite and formal ways toward one another, carefully choosing their words and actions so as to re-establish their bonds with other members of the group.[Chart]
As used in our model, the term forming represents a group of activities used to establish a pattern of interaction within a group. Through these behaviors, group members develop a level of attraction and cohesion to one another. Storming on the other hand, occurs when group members have conflicts of interest and significant differences of opinion about group functioning, goals, and tactics. Activities related to managing conflicts and disagreements occur during storming periods. Raven and Rubin (1976) define norms as "standards against which the person can evaluate the appropriateness of behavior . . . providing order and meaning to what otherwise might be seen as an ambiguous, uncertain, perhaps threatening situation" (p. 314). Thus, norming represents a set of activities directed toward the establishment of these standards. In our model, performing represents a class of behaviors which are directed at task accomplishment. These include problem solving, decision making, and various "implementing" activities.
In the discussion of our model, we use the term "group" to refer to small collectives of individuals (ten or less) who have the opportunity for significant, meaningful interaction with one another. These groups, whether social or work-related are "made up of individuals who see themselves and are seen by others as a social entity, who are interdependent because of the tasks they perform as members of the group, who are embedded in one or more larger social systems . . . and who perform tasks that affect others" (Guzzo & Dickson, 1996, pp. 308-309). In addition, these groups operate in an organizational system in which the rules, roles, task definitions, information, and resources needed for the group to perform effectively are not readily and rigidly dictated by the organization, the circumstances, or a group leader (i.e., not an excessively strong organizational context). Our model is based on situations where groups have some discretion or flexibility in their work processes. Thus, the model may be less applicable to groups operating under very tightly controlled conditions.
We focus on the process aspects of performing, in that groups which perform well have been able to establish successful processes for solving problems, combining information to make decisions, as well as procedures for implementing those decisions. We have separated the processes of performing from the cognitive resources that are required for performing. We shall explain this distinction later in the paper, but now will turn our attention to the notion of diversity and describe how different aspects of diversity affect group development processes.
SOURCES OF DIVERSITY AND THEIR IMPACT ON GROUP DEVELOPMENT
Jackson et al. (1995) differentiate between two sources of diversity-readily detectable attributes and underlying attributes. Readily detectable attributes are those which can be easily recognized in a person such as age, gender, or national/ethnic origin. Underlying attributes represent personal characteristics which are not so easily identifiable, such as cultural beliefs, personality characteristics, or knowledge level. While Jackson et al. (1995) differentiate between readily detectable and underlying attributes and we have adopted this approach in our model, whether this distinction is justified remains an empirical issue.
In our model, we have differentiated between two types of underlying attributes. Some underlying attributes such as cultural values and perspectives, attitudes, values and beliefs, and conflict resolution styles (Underlying Attributes I) are closely correlated with readily detectable attributes. Nationality/ethnic origin is significantly related to cultural values and perspectives. The work of Hofstede (1980) and many others clearly indicates this. Age differences are related to diversity in attitudes and values (e.g., Bantel & Jackson, 1989; Jackson, Brett, Sessa, Cooper, Julin, & Peyronnin, 1991). There is also research to indicate that males and females tend to differ in their attitudes as well as in their approach to disagreements and conflicts. For example, Carli (1988) suggests that women in groups express more agreement and attempt to relieve group tension, while men express more disagreement and focus more on task-oriented behaviors than women. Women tend to use referent power to persuade, while men tend to use expert power. Thus, our first hypothesis is that nationality/ethnic origin, age, and gender will be strongly correlated with the underlying attributes of cultural values and perspectives, personal attitudes, values, and beliefs, and styles of conflict resolution.
Hypothesis 1. Readily detectable attributes will be strongly correlated with Underlying Attributes I.
A second group of attributes (Underlying Attributes II) which includes socioeconomic and personal status, education, functional specialization, human capital assets, past work experiences, and personal expectations, are less strongly connected to the nationality/ethnic origin, age, or gender of work group members. While there are certainly differences among nations in terms of overall economic level, individual members of work groups may come from very economically underdeveloped countries, but, themselves, be of a high socioeconomic level. Male and female group members may bring equivalent or very different human capital assets to a work group. Age is likely to have little relationship to the particular functional specialty which an individual brings to a group. As the dotted line in our model (Fig. 1) indicates, however, we are not suggesting that there are no relationships between these underlying attributes and the detectable attributes of nationality/ethnic origin, age, and gender-only that such relationships are far less clear cut and may vary more greatly from one work group to another.
Hypothesis 2. Readily detectable attributes will be less correlated with Underlying Attributes II than they are with Underlying Attributes I.
How Readily Detectable Attributes Affect Group Development
Readily detectable attributes play a particularly important part in influencing our early social cognitions about a person. Jackson, Stone, and Alvarez (1992) describe social cognitions as ". . . the inferential logic by which people translate easily detected information about demographic attributes into best-guess hypotheses about personal attributes of a stranger" (p. 56). Jackson et al. (1992) summarize a variety of research studies which indicate that as similarity on readily detectable attributes between people decreases a variety of processes occur which affect how those individuals interact with one another. First, differences in readily detectable attributes cause the individuals to become more aware of their own "social identity." The realization that "I am not like you" becomes more salient and draws attention to the characteristics which represent those differences. Second, as a result of the increased salience of nonsimilarity, stereotyping tends to increase which causes misperceptions and bias in how we interpret information about another person. Even if we perceive information somewhat accurately, the retrieval of that information from our memories when needed for later judgments or decisions is more likely to be inaccurate or biased. Third, anxiety about interactions with nonsimilar individuals increases. Finally, the attraction of the other person decreases and negative judgments of dissimilar others increase (Tsui, Egan, & O'Reilly, 1992).
One possible mechanism by which dissimilarity results in increased anxiety and lower interpersonal attraction has been suggested by Jackson, Stone, & Alvarez (1992). They indicate that dissimilarity may affect the costs and benefits of information exchange. This cost-of-interaction concept is consistent with research by Devine (1989) who suggests that heterogeneous groups experience more negative affect because overcoming stereotypes and social categorization processes takes considerable cognitive effort. It is assumed that this extra cognitive work is unpleasant and the negative affect associated with it influences the more general affective response to dissimilar others in the group. While the costs may be higher when dealing with dissimilar others, it may also be more difficult to extract rewards from these experiences. These concepts are readily found in the interpersonal attraction literature. Berscheid (1985) reviewed a variety of models of interpersonal attraction. These models indicate that a particular individual will be attracted to persons who (1) provide rewarding experiences in greater proportion to nonrewarding or punishing ones, and (2) have appealing characteristics and will reciprocate by viewing the focal person as having appealing characteristics as well.
In our model, we suggest that readily detectable attributes make individuals aware of their differences from one another, which in turn elicits stereotypes and biases. Overcoming and dealing with these stereotypes and biases requires considerable cognitive effort which increases the costs of interacting with dissimilar others, thereby reducing the benefit-to-reward ratio possible from those encounters. While the cost-of-interaction approach assumes a relatively rational process and the ability to compare relative levels of effort and rewards at some conscious or unconscious level, the extra cognitive effort required to deal with someone who is perceived to be dissimilar may itself be sufficient to elicit negative affect.
As indicated in Fig. 1, readily detectable attributes play a strong role in determining the ability of group members to successfully engage in what is usually the earliest stage of group development-forming. The perceived costs of interacting with others who are dissimilar in nationality/ethnic origin, age, or gender is likely to reduce the overall level of social interaction among group members as well as decrease the interpersonal attractiveness of individual group members and, thus, the overall attractiveness of the group itself. Group cohesion, defined by Johnson and Johnson (1987) as ". . . the extent to which the influences on members to remain in the group are greater than the influences on member to leave the group" will be more difficult to achieve (p. 408). Difficulty in forming a cohesive unit will in turn influence the group's ability to manage conflicts and develop group norms and goals. Thus, the third set of hypotheses derived from our model are:
Hypothesis 3a. Diversity among group members in terms of readily detectable attributes will be strongly and negatively correlated with aspects of the group-forming process. The greater the level of diversity, the less cohesion, social interaction, and attraction experienced by the group.
Hypothesis 3b. Diversity among readily detectable attributes of group members will also be correlated with aspects of storming and norming. However, this relationship operates primarily through the impact this diversity has on forming activities. Thus, when the relationship between readily detectable attributes and forming is partialed out, the correlations between diversity in readily detectable attributes and storming/norming will become either nonsignificant or at least substantially lower than the attribute-forming relationship.
Underlying attributes affect the group development process through their impact on cognitive paradigm dissimilarity. Shaw (1990) presented a model which suggests that cultural differences impact cross-cultural interactions because of the different cognitive structures that each person brings to the situation. Shaw (1990) argued that differences on various cultural dimensions (e.g., such as those described by Hofstede, 1980) affected the types and sizes of schema used to process information as well as the constraint values, and vertical and horizontal structure of those schemata. Similarly, Milliken and Martins (1996) suggest that diversity on a variety of attributes, including values, skills and knowledge, and cohort membership, affect the cognitive paradigm dissimilarity of individuals. Milliken and Martin (1996) suggest that diversity affects the extent to which individuals share similar cognitive schema structures and are likely to attribute the causes of behavior to similar factors. Shaw (1990) and Milliken and Martin (1996) review literature from cognitive and cross-cultural psychology to support their contentions. The impact of culture on cognitive structures and differing attributional processes is widely recognized and supported by this research.
In our model, to the extent that cognitive paradigm dissimilarity exists, the cognitive cost of interpersonal interactions increases. As Devine (1989) suggested, overcoming differences in cognitive categorization processes requires significant cognitive effort and is likely to result in negative feelings about the interaction. Thus, diversity in underlying attributes like cultural values, education, or functional specialization affects group development through its impact on cognitive costs, which in turn affect the desire and ability of group members to form into a cohesive unit.
We suggest that there is yet another point at which underlying attribute diversity can affect the group development process. We argue that it is possible for a group to form effectively (at least initially), i.e., feel that they are a cohesive unit, interact well with one another at a social level, and feel a sense of attraction toward one another, but nevertheless have considerable difficulty in other aspects of group development due to the cognitive paradigm dissimilarity that exists among them. Thus, a person may feel a sense of cohesion and attraction to the people in his or her group because they are of the same nationality, age, and gender. However, this similarity in terms of readily detectable attributes may be accompanied by considerable diversity in underlying attributes. Conflict management during storming may be difficult due to very different perceptions of what conflict means and how it should be addressed. Similarly, though group members of different ethnic origins may like one another, they may have difficulty in developing group norms due to drastically different perceptions of what acceptable behavior in a group means. For example, Huig and Triandis (1986) found that individuals from collective cultures gave high consideration to the implications of their own behavior, shared material and nonmaterial resources with others, and emphasized harmony within the group. In contrast, Bellah, Madsen, Sullivan, Swindler, and Tipton (1985) found that people from individualistic cultures focused on self-reliance and independence, helped their groups only if they received something in return, and felt that competition was normal and good. Thus, our model implies a fourth set of hypotheses:
Hypothesis 4a. Diversity in underlying attributes will be negatively related to the effectiveness of storming and norming aspects of group development.
Hypothesis 4b. Even in groups which have achieved a high level of cohesion and mutual attraction, diversity in underlying attributes will have a negative relationship with the effectiveness of storming and norming activities.
Diversity Management Skills as Moderators
In our model, we suggest that the impact of cognitive paradigm dissimilarity on group storming and norming may be moderated by what we call the diversity management skills that group members bring with them. Working within a diverse group has much in common with the situation that expatriate managers face when they are sent to foreign locations by their organizations. For the expatriate manager, the challenges confronted are likely to focus heavily on differences in national origin and cultural perspectives. In small, diverse work groups, the challenges may stem from differences due to national origin, gender, age, functional specialization, or any of a number of other characteristics. In either case, however, the individuals involved must develop working relationships which allow them to accomplish a variety of complex tasks.
There has been a large amount of research on the characteristics of expatriate managers that aid them in dealing with the cross-cultural differences they face in foreign assignments. To simplify the various findings of the studies, one could say that there have been identified a variety of characteristics which help individuals deal effectively with work situations in which the people involved differ markedly from one another on a variety of characteristics.
One of the most interesting and informative of these studies was by Mendenhall and Oddou (1985). They identified four major dimensions of expatriate acculturation-that is, factors that influenced whether an expatriate could successfully adapt to a foreign assignment. The self-oriented dimension concerns activities that contribute to the expatriate's self-esteem and self-confidence. There are three subfactors in this dimension: (1) reinforcement substitution, (2) stress reduction, and (3) technical competence. People high on reinforcement substitution are able to readily substitute activities in a new culture for those that they liked to do in their home culture. Expatriates also tend to manage cross-cultural situations more readily if they engage in activities that help them reduce stress. In addition, the technical competence of an expatriate also has a strong bearing on their ability to manage cross-cultural situations. Expatriates who are technically competent and feel confident in their ability to handle their international assignments manage cross-cultural challenges more readily than those who are less qualified.
The second major dimension of acculturation, the others oriented dimension, consists of two subfactors: (1) relationship development, and (2) willingness to communicate. Expatriates who are able to develop lasting friendships and close relationships with people different from themselves manage overseas assignments more easily. Additionally, individuals who are willing to communicate in a different language also adapt more readily. It is interesting to note that Mendenhall and Oddou suggest that being willing to communicate may be as important as the actual level of communication skill of the person involved in the cross-cultural situations. For example, someone may have very good foreign language skills, but be unwilling to use those skills in a foreign country. Mendenhall and Oddou (1985) indicate that the a person with lower language skills, but who is willing to use their skills openly, will be better able to manage cross-cultural situations than someone who is an expert in a foreign language but will not speak up. Obviously, the ideal situation would be to have both good communication skills and be willing to use them.
The third dimension of acculturation, the perceptual dimension, concerns the ability of expatriates to understand how their own behavior affects other people as well as why foreign nationals behave the way they do. Individuals who can accurately assess how their behavior is affecting others are more able to manage cross-cultural situations. This self-monitoring ability allows the person to adjust his or her behavior to the circumstances as needed to achieve a more positive interpersonal interaction. Research in social psychology indicates that people are motivated to attribute causes to other people's behavior. Furthermore, this research indicates that we are not very good at accurately assessing the causes of the behavior of people from our own culture and are even more inaccurate in explaining the behavior of persons from cultures very different from our own. Mendenhall and Oddou suggest that expatriates who are able to reserve judgment and gather more facts before assigning causes to behavior are better at cross-cultural interactions. They refer to this ability as evaluation flexibility.
Although Mendenhall and Oddou (1985) were concerned with identifying characteristics that would assist in the management of cross-cultural encounters, the same characteristics or skills might be equally useful in the management of any situation or environment where diversity is a significant factor. In our model, we suggest that communication skill, willingness to communicate, relationship development, self-monitoring, and evaluation flexibility might be more broadly termed diversity management skills which allow individuals to "work through" the problems caused by cognitive paradigm dissimilarity. Diversity management skills allow group members to communicate more effectively with one another about their differences, to evaluate more accurately how their own behavior is affecting group processes, and to react in a more realistic and less judgmental way to the different attitudes and behaviors expressed by other group members. This leads to our fifth hypothesis from the model.
Hypothesis 5. Diversity management skills will moderate the relationship between cognitive paradigm dissimilarity and a group's ability to storm and norm effectively. When diversity management skills are low, the impact of cognitive paradigm dissimilarity on the development of effective storming and norming processes will be highly negative. When diversity management skills are high, cognitive paradigm dissimilarity will have little impact on storming and norming processes.
Group Behavioral Integration and Performing
In our model, groups which effectively manage the processes of forming, storming, and norming are able to achieve a level of what Hambrick (1994) refers to as behavioral integration. Behavioral integration is "the degree to which the group engages in mutual and collective interaction" (p. 188). Hambrick (1994) suggests that behavioral integration is different from other concepts such as social integration and cohesiveness. He argues that social integration focuses primarily on the affective aspects of the group, while cohesiveness relates primarily to the attraction of members to one another. In contrast, behavioral integration relates to the quantity and quality of information exchanged, the amount of collaborative behavior found in the group, and the ability of the group to engage in joint decision making. Our model suggests that diversity affects the ability of groups to achieve behavioral integration through its impact on the group's processes of forming, storming, and norming.
One might argue that behavioral integration and the performing stage of group development are essentially identical. However, in Hambrick's (1994) definition of behavioral integration, the emphasis is on the ability of a group to make decisions and engage in collaborative action. In terms of the performing stage of group development, the problem solving, decision making, and implementation activities are specifically oriented toward task achievement. In our model, we have identified behavioral integration and performing as two distinct aspects of group activity with behavioral integration being a somewhat more general state of collective activity while performing is activity specifically oriented toward task achievement.
While the distinction between behavioral integration and performing may not be absolutely critical in our model's development, we believe that maintaining this distinction adds increased flexibility in our ability to understand behavior in diverse groups. For example, we have found this distinction useful in understanding the behavior of student project groups in one of our university subjects. Many of these student groups have formed well, manage conflict effectively, and have developed agreed upon standards of behavior. They can make joint decisions and work collaboratively on various group activities. However, much of this behavior is directed at nonproductive activities relative to the actual task at hand. One of our student groups, for example, was excellent at arriving at meetings on time, enjoying one another's company, ordering pizza and beer, and playing football in the park. Their behavior as a group was very integrated, but as far as their "performing" went-well! Such examples suggest that maintaining a distinction between behavioral integration and the performing stage of group development is warranted.
As indicated by our discussion above, the relationship between behavioral integration and the performing aspects of group development is complex. While being behaviorally integrated may not insure that a group will be able to engage in task-oriented problem solving, decision making, and implementation activities, the absence of integration will almost certainly prevent the group from developing effective performance procedures.
Hypothesis 6. Low levels of behavioral integration will be associated with poor task-oriented "performing" aspects of group development. High levels of integration may result in either good or poor task-oriented "performing" activities.
Diversity, the Availability of Group Resources and Task Type
As noted earlier, cognitive dissimilarity will affect the level of behavioral integration in a group through its influence on a group's ability to form, storm, and norm. If cognitive dissimilarity results in a low level of behavioral integration, it may be difficult to arrive at an agreed upon and effective method of problem solving and making decisions or implementing particular actions needed for task accomplishment. This indirect impact of cognitive dissimilarity on performing through its effect on behavioral integration is stated above in Hypothesis 6.
However, diversity and its resulting cognitive paradigm dissimilarity may also have a positive impact on the quality of final group outcomes by providing valuable resources (e.g., alternative viewpoints, a broader range of expertise, and other human capital assets) to be used in problem solving, decision making, and implementing activities. For example, Adler (1990) suggests that group diversity has many advantages such as increased creativity, increased flexibility, increased problem-solving skills, and less "groupthink." Bottger and Yetton (1988) report a study of midlevel managers and graduate management students working in groups on a decision task. The results of the study support their contention that the performance of a group is a function of both group resources (defined as group member task knowledge) and the efficacy of group process. Bottger and Yetton (1988) define group processes in terms of the group's ability to manage conflict and the appropriateness of the group's decision scheme (i.e., the rational use of the group's knowledge resources). Bottger and Yetton's model is quite similar to ours in that, to use our terminology, the performance of a group is a function of the group's available performance related cognitive resources (similar to task member knowledge), behavioral integration (which includes an aspect of conflict management), and performing activities (which represents aspects of the schemata used by the group to make decisions). Both we and Bottger and Yetton (1988) find it useful to distinguish among these different aspects of group activity and performance. Bottger and Yetton's (1988) research indicated a unique contribution of each of their variables to the prediction of group performance.
The impact of diversity on the performance of a group and the individual members within the group will be moderated by the nature of the group task. By group task, we mean tasks that are either directly or indirectly dependent on group processes for their accomplishment. Tasks that can be performed completely independently by individual members, and that have minimal impact on other group members and their tasks, are obviously less likely to influence or be influenced by the impact of diversity on group processes. Bantel and Jackson (1989) found that when a task is complex and nonroutine, cognitive diversity is an advantage in that the need to reconcile different views of the problems faced results in more group discussion, less likelihood of groupthink, and thus better quality decisions. Jackson, Stone, and Alvarez (1992) identified three types of tasks. Performance tasks engage primarily perceptual and motor skills (e.g., sorting objects or assembling component parts). For such tasks, demographic diversity decreases performance. Intellective tasks are cognitively-based tasks which require collective problem-solving activities where there is a correct answer (e.g., solving a puzzle or a mathematical problem). There is no conclusion about diversity's impact on intellective tasks. Idea generation and decisionmaking tasks are also cognitively based but require consensus to arrive at a preferred solution. In these types of tasks, Jackson et al. (1992) indicate that heterogeneous teams perform better.
Adopting another perspective, Larkey (1996) identified two types of group task situations. In the first, the group confronts a task which is likely to cause conflict and threat to individuals within the group (e.g., the selection of a new supervisor from within the group, or a competition for the award for the most successful saleperson). In this case, the nature of the task may strengthen individual cultural group identity. Cultural identification would tend to lead to categorization processes (as discussed earlier in the model) which would then result in negative evaluations, exclusion, adherence to cultural communication style, suppression of different views, and distortion of the content and intent of communications. Thus, the ability of the group to achieve an adequate level of behavioral integration would be diminished. A second type of situation that may face a group is one in which the nature of the task threatens, challenges or rewards the workgroup in its entirety (e.g., mounting a defense against the threatened loss of group project funding, or a competition for the award of most successful sales team). Larkey (1996) argues that this would create workgroup identity and allow members to consider each others' individual characteristics. This might then enhance the recognition of individual cognitive resources that group members bring to the task and thus potentially enhance group performance.
Hypothesis 7. For some tasks which require high levels of idea generation, decision making, and can be accomplished only through collective group effort, diversity will increase the ability of the group to perform well by increasing the available cognitive resources of the group. This positive effect will occur only for groups that have achieved at least a moderate level of "performing" effectiveness which will allow the group to make adequate use of the resources available among its members. For other tasks (such as those which require only the use of simple motor skills or which may be divided into distinct components that can be accomplished singly by individual members of the group), diversity will have little effect on group performance.
Group and Individual Level Outcomes
Jackson et al. (1995) distinguish between individual and team-related consequences of diversity. They further divide these consequences into those related primarily to task accomplishment or relationships among group members. Jackson et al. (1995) discuss a wide variety of possible consequences of diversity. In terms of individual consequences, they identify several potential dependent variables including personal performance, feelings of satisfaction with personal performance, acquisition of skills and knowledge, and feelings of satisfaction with group members and social relationships. At the team level, Jackson et al. (1995) note that little research has been conducted on consequences other than team performance and membership stability (turnover). Turnover, however, can also be a source of diversity to the extent that subsequent group members introduce new attributes and resources to the group, as well as create group cohort effects. As with individual consequences, at the team level, group satisfaction with performance, satisfaction with aspects of the group process, and team learning and skill acquisition are suggested by Jackson et al. as important dependent variables worthy of research.
In Fig. 1, we have listed only four major group and individual level outcomes predicted by our model-performance, satisfaction with performance, satisfaction with group members, and satisfaction with group process. This is intended to be an illustrative rather than exhaustive list of important outcomes. Any of the individual or team consequences identified by Jackson et al. (1995) or other outcomes discussed in the Milliken and Martins (1996) model could be included in our list of outcomes.
In the preceding pages, we have described a group-development approach to the study of the impact of diversity on groups. While the model borrows heavily from the work of Jackson et al. (1995) and Milliken and Martins (1996), we believe our model provides a more detailed and precise mapping of the effects of diversity on group processes and performance and yields readily testable hypotheses for future research.
We have suggested that differences in readily detectable attributes affect group processes primarily through their impact on the perceived cognitive costs of interpersonal interactions. Diversity in underlying attributes affects group processes both through the mechanism of cognitive costs as well as more directly through the impact of cognitive paradigm dissimilarity and its effect on the ability of groups to manage conflict and develop group norms. Furthermore, we suggest that the impact of cognitive paradigm dissimilarity on group processes is moderated by the level of diversity management skills group members bring to group situations.
We have combined the concepts of Tuckman (1965) with that of Hambrick (1994) to suggest that diversity, through its impact on the forming, storming, and norming activities of the group, will determine the overall level of behavioral integration in group activities. Behavioral integration, in turn, influences the ability of the group to develop systems of problem solving, decision making, and implementation. Depending on the task confronted by the group, diversity may enhance group performance and attitudes by increasing the available task-related cognitive resources in the group. However, for diversity to have this positive effect, the group must have achieved a sufficient level of "performing" to make full use of these available resources.
The scope of this article does not allow for the discussion of a number of other issues that may have a bearing on the effects of diversity on group processes and performance. While we have mentioned several different sources of diversity (i.e., readily detectable attributes and two categories of underlying attributes), future research could identify alternative dimensions of diversity that might affect group development and performance. In addition to types of diversity, the configuration of diversity (e.g., the degree of heterogeneity in the group, and the presence of minority factions, coalitions, or tokens) should also be considered, along with the attributes of group leaders in relation to the rest of the membership.
Another area that deserves research attention is the role of the societal and organizational context in which groups are embedded. For example, the attitudes prevalent in the society at large should affect the likelihood that readily detectable attributes will elicit negative stereotypes and increase the cognitive cost of interpersonal interactions among diverse group members. Similarly, organizational policies and procedures would be expected to shape work group practices. It would be interesting to evaluate how closely organizational claims of valuing diversity correspond with the level of diversity management skills among its members.
Although our model is intended primarily as a guide for future research, it may also assist practitioners in their work with diverse work groups and teams. It highlights the importance of considering the impact of both apparent and less visible aspects of diversity, and suggests the need to foster in organizational members specific diversity management skills rather than just a broad appreciation of diversity.
While some of the factors discussed in our model may be difficult to measure, such as the evaluation of cognitive costs/benefits of interactions, or the level of cognitive paradigm dissimilarity, we believe that our model adds significantly to the overall picture of how group composition affects group performance and attitudes. Identifying reliable measures of our model's concepts as well as testing specific hypotheses derivable from the model should provide significant direction for future research on the impact of diversity on small groups for some time to come.[Reference]