How Diversity Hurts Performance

Somehow, those who tout the benefits of "diversity" never present evidence for their claim. It does seem that on the contrary,, diversity worsens performance:

DEMOGRAPHY AND DIVERSITY IN ORGANIZATIONS: A REVIEW OF 40 YEARS OF RESEARCH
Katherine Y. Williams and Charles A. O’Reilly, III

"Most of the research that supports the claim that diversity is beneficial for groups has been conducted in the laboratory or classroom setting, instead of examining intact working groups within an organizational context. In the laboratory the results sometimes show that group diversity can improve the quality of a given decision or the creativity of an idea (e.g., Kent & McGrath, 1969; Priem, Harrison, & Muir, 1995). The research on intact working groups, on the other hand, paints a less optimistic view of the effects of diversity on group functioning. It provides evidence of the possible dysfunctional aspects of heterogeneity in groups, including increased stereotyping, in-group/out group effects, dysfunctional conflict, and turnover (e. g., Linville & Jones, 1980; O’Reilly, Snyder, & Boothe, 1993; Pelled, 1996; Tsui, Egan, & O’Reilly, 1992)...

Much of the literature that supports the claim that diversity is beneficial for groups is often based on variation in individual attributes such as personality, ability, and functional backgrounds, and not on ascriptive attributes such as ethnicity and sex (e.g., Altman & Haythom, 1967; Hoffman, 1959; Levy, 1964; Triandis, Hall, & Ewen, 1965; Zeleny, 1955). With the changes in the demography of the workforce, understanding the effects of visible attributes is even more important than it used to be. Although there is evidence from laboratory research that diversity in ascriptive characteristics can be beneficial to groups (Cox, Lobel, & McLeod, 1991; Watson, Kumar, & Michaelsen, 1993), this is based on only a few studies when considered against the large number of field studies suggesting that heterogeneity in race and gender often have negative effects on group process and performance (e.g., Cummings, Zhou, & Oldham, 1993; Kizilos, Pelled, & Cummings, 1996; Riordan & Shore, 1997; Zenger & Lawrence, 1989)...

Similarity on attributes ranging from attitudes and values to demographic variables increases interpersonal attraction and liking (e.g., Byme, Clore, & Worchel, 1966). Individuals who are similar in background may share common life experiences and values, and may find the experience of interaction with each other easier, positively reinforcing, and more desirable. Similarity provides positive reinforcement for one’s attitudes and beliefs, while dissimilarity is seen as a punishment. For instance, similarity/attraction theory has been embedded in the principle of homophily and the effects it may have on communication in groups (Rogers & Bhowmik, 1971). In a free choice situation, when an individual can interact with any of a number of people, there is a strong tendency for him or her to select a person that is similar (e.g,, Burt & Reagans, 1997; Lincoln & Miller, 1979). Homophily has been observed in friendship and voluntary interactions (Blau, 1977; McPherson & Smith-Lovin, 1987), as well as in organizational settings (e.g., Brass, 1985; Ibarra, 1992; Mehra, Kilduff, & Brass, 1996). Several laboratory studies demonstrate that heterogeneity leads to decreased communication, message distortion, and more errors in communication (e.g., Barnlund & Harland, 1963; Triandis, 1960)...

Information and decision-making theories propose that variance in group composition can have a direct positive impact through the increase in the skills, abilities, information, and knowledge that diversity brings, independent of what happens in the group process (Tziner & Eden, 1985). Demographically diverse individuals are expected to have a broader range of knowledge and experience than homogeneous individuals. For example, the proponents of immigration argue that diversity promotes creativity in the workforce. To accomplish this, Lazear (1997) argues that new immigrants must have information that is different from the existing workforce, have information that is relevant or useful, and must be able to communicate this to others. From this perspective, diversity is valuable when it adds new information. Clearly, this positive impact of diversity can be expected when the task can benefit from multiple perspectives and diverse knowledge, such as innovations, complex problems, or product design. Researchers largely agree that functional or background diversity provides the range of knowledge, skills, and contacts that enhances problem solving (e.g., Ancona & Caldwell, 1992; Bantel & Jackson, 1989; Pelled, Eisenhardt, & Xin, 1997). “Members who have entered the organization at different times know a different set of people and often have both different technical skills and different perspectives on the organization's history" (Ancona & Caldwell, 1992, p. 325)...

The effects of diversity can be moderated by the situation. Since some of the potentially negative effects of increased diversity result from cognitive processes (e.g., stereotyping), it is reasonable that the same cognitive processes may offer a means for reducing the negative effects. For instance, creating a common identity or goal may, as Sherif (1936) demonstrated over 50 years ago, reduce in-group/out-group biases and promote solidarity. More recently, research has shown that strong, collectivistic cultures may reduce invidious social categorization effects (e.g., Chatman, Polzer, Barsade, & Neale, 1997; O’Reilly & Chatman, 1996). Similarly, other actions that call attention to similarities or differences may accentuate or diminish social categorization and similarity processes. Some types of diversity training may unwittingly highlight differences and create exclusive rather than inclusive categorizations (Nemeth & Christensen, 1996). Finally, other contextual influences such as technology or task design may also increase or decrease normative and informational influences on interpretations. For example, jobs may be designed that signal to people that they are interdependent or independent (Jehn, Northcraft, & Neale, 1997). These contextual influences can act to focus attention in ways that can subtly focus interpretations on similarities or differences. As such, they may unwittingly help or hinder group process and performance...

The overall effect of increasing diversity is likely to have a u-shaped form with some increments of diversity having large positive increases in group problem-solving capability with comparatively small negative effects on group functioning. Large amounts of diversity in groups may offer little in the way of added value from unique information and make group cohesion and functioning difficult...

In general, there is strong evidence that diversity in tenure is associated with lower levels of social integration, poorer communication, and higher turnover in groups. Although under some circumstances turnover may have positive effects (Staw, 1980), the effects of tenure diversity found in the research reviewed here are considered negative; that is, those who are most different in terms of tenure are most likely to exit...

Functional background may serve as a proxy for the information, knowledge, skills, and expertise that individuals bring to a group. The research suggests that the diversity of information functionally dissimilar individuals bring to the group improves performance in terms of creativity, but not necessarily implementation. For example, functionally diverse groups are slower (Hambrick, Cho, & Chen, 1996) and have lower cohesion than homogeneous groups (Ancona & Caldwell, 1992, p. 338). However, the overall evidence strongly suggests that functional diversity is likely to stimulate task conflict and improve performance (Jehn, Northcraft, & Neale, l997; Pelled, Eisenhardt, & Xin, 1997)...

Overall, the research on age diversity suggests that groups with higher variations in their age composition may have slightly lower levels of effective group process than more homogeneous groups. The expectation, drawn from an information and decisionmiaking theory, that age differences within a group may index differences in perspective and values that are useful for cognitive performance is not supported by the literature Instead, the literature suggests that age diversity is associated with increased turnover and withdrawal, especially of those individuals who are most different...

The results of research on gender diversity suggest that the proportion of men and women present in the sample may be an important predictor of the results. In general, gender diversity has negative effects on groups, especially on males. It is associated with higher tumover rates, especially among those who are most different. The studies also reveal that women and men respond differently, and may have different experiences as a minority. Men display lower levels of satisfaction and commitment when they are in the minority, while women appear less likely to have a negative psychological reaction. This is despite the fact that men in female-dominated groups are more likely to be accepted, less likely to be treated with hostility, and less likely to be stereotyped. Given the asymmetrical findings, future research on the effects of gender diversity needs to pay close attention to the proportions of men and women in the sample if results are to be interpretable (e,g., Ely, 1994)...

Although the issue of race-ethnic diversity is important for society and organizations, research on its impact remains inconclusive. The optimists argue, based on information/decision theory, that ethnic diversity can promote creativity and improve decision making. Paradoxically, there is some evidence from field studies that supports this conclusion, but, contrary to information/decision theory predictions, these results occur independent of group-process variables; that is, diversity improves performance controlling for group process. The pessimists, using similarity/attraction and social categorization theories, note that ethnic diversity can, unless successfully managed, have negative effects on group process. Consistent findings show that individuals who are different from the majority race in an organization are more likely to leave, to be less satisfied and psychologically committed to the organization, and to receive lower performance evaluations. Interestingly, several studies find that these effects are more pronounced for Whites than minorities (Riordan & Shore, 1997; Tsui, Egan, & O’Reilly, 1992). Overall, the evidence for ethnic diversity seems more consistent with predictions of similarity/attraction and social categorization theories than with information and decision making.

A limitation to the existing research is that most studies have only examined blacks and whites, or whites and “others.” Yet, as decades of research in cross-cultural psychology has shown, there are important differences within and across ethnic groups (e.g., Kim, Park, & Suzuki, 1990; Phinney, 1996) that may be relevant within organizations (e.g, O’Reil1y, Williams, & Barsade, 1997; Riordan & Shore, 1997), It is unclear that Asian Americans, for instance, Will have the same experience as African Americans in majority Anglo-American organizations. Further, the effects of proportions may also have important effects on race/ethnic diversity just as it does on sex composition (e.g., Espinoza & Garza, 1985; Garza & Santos, 1991; Tsui, Egan, & O’Reil1y, 1992).

Variations in other individual differences have also been studied and found to have important effects on the process and performance of groups. Some of the earliest laboratory studies conceptualized diversity in terms of variations in personality, attitudes, and values and found positive effects for these (e.g., Torrance, 1957). In 1959 Hoffman examined 30 small groups in the laboratory and found that groups that were diverse in terms of their personality characteristics produced higher quality outputs and tended to produce more inventive solutions. Hoffman and Maier (1961) found further support for the positive effects of diversity in terms of personality. Triandis, Hall, and Ewen (1965) also found that heterogeneity in attitudes, but not abilities, was associated with increased creativity.

Bochner and Hesketh (1994), using an Australian sample, found that people who were different from others in their work groups on power distance and collectivism perceived that they were discriminated against more frequently, but valued their cultural differences more highly. This finding suggests that heterogeneity in cultural values may have important effects on individual outcomes. For example, Dutch researchers found that individuals who were not Dutch tended to be less satisfied with their jobs than their Dutch co-workers (Verkuyten, de long, & Masson, 1993). Although the evidence for cultural diversity is intriguing, organizational demographers have seldom focused on this type of diversity as it affects group process and performance.

Another stream of research of potential relevance for understanding the impact of diversity are studies of the effects of “minority influence” on decision making. Charlan Nemeth and her colleagues (e.g., Nerneth, 1986; Nemeth & Kwan, 1987) have shown that when people hold strong, consistent views different from the majority, they can often have an effect on decisions beyond what their proportion Would suggest. Research suggests informational social influence may dampen some of the conformity pressures of normative social influence (Moscovici, 1985). Thus, insofar as demographic differences also index differences in information, a minority in a group may raise issues that can affect the g1'oup’s decision making...

Based on the studies reviewed here, two major findings from the research on demography and diversity appear to be well supported. First, there is substantial evidence from both laboratory and field studies conducted over the past four decades that variations in group composition can have important eflects on group functioning. These studies show that increased diversity, especially in terms of age, tenure, and ethnicity, typically has negative effects on social integration, communication, and conflict (e.g., Chatman et 211., 1997; Ibarra, 1992; Iehn, Northcraft, & Neale, 1997; Lott & Lott, 1965; O’Rei1ly et al., 1989; Pelled, Eisenhardt, & Xin, 1997; Smith et al., 1994). Diverse groups are more likely to be less integrated, have less communication, and more conflict. Interestingly, the one exception to this pattern is with regard to functional diversity or educational background. For this variable, increased diversity has been shown under some circumstances to increase communication (e.g., Aneona & Caldwell, 1992; Glick, Miller, & Huber, 1993; Jehn, Northcraft, & Neale, 1997). In addition to the effects on social integration, communication, and conflict, research has also linked group diversity to several other process variables such as increased in-group/out-group biases, stereotyping, and other cognitive biases that can negatively affect group functioning (e.g., Crocker & Major, 1989; Lorber & Farrell, 1991; Messick & Massie, 1989; Moreland, 1985).

It is also clear that not all of the group process variables investigated operate independently of one another. For example, Smith and colleagues (1994) found that social integration and informal communication were related. Chatrnan and colleagues (1997), using a business simulation with MBA students as subjects, found that organizational culture moderated the effects of diversity such that as dissimilarity increased conflict was seen as more beneficial for those groups with a collectivistic culture. Other studies have shown that under different conditions conflict may have positive or negative effects (e.g., Jehn, Northcraft, & Neale, 1997; O’Reilly, Williams, & Barsade, 1997; Pelled, Eisenhardt, & Xin, 1997; Stephan, 1985). Studies such as these underscore the importance of contextual variables as moderators of diversity effects. Thus, consistent with the model outlined in Figure 1, there is also evidence that demographic effects may be moderated by variables such as culture, technology, and task design.

A second supportable conclusion is that at the micro level, increased diversity typically has negative effects on the ability of the group to meet its members' needs and to function effectively over time. The literature shows clearly that individuals are affected by the demographic composition of their work groups. The preponderance of evidence shows that increased diversity within a group can be associated with lower levels of satisfaction and commitment (Riordan & Shore, 1997; Tsui, Egan, & O‘Rei1ly, 1992 ), lower performance evaluations for those who are different (Greenhaus, Parasuraman, & Wonnley, 1990; Holahan, 1979; Judge & Ferris, 1993; Sackett, DuBois, & Noe, 1991; Tsui & O’Rei1ly, 1989), and higher levels of absenteeism and turnover (Cummings, Zhou, & Oldham, 1993; Jackson et al., 1991; McCain, O’Reilly, & Pfeffer, 1933; O’Reilly, Caldwell, and Barnett, 1989; O’Reil1y, Snyder, & Boothe, 1993; Pfeffer & O’Reilly, 1987; Tsui, Eisenhardt, & Xin, 1992; Wagner, Pfeffer, & O’Reilly, 1984; Wiersema & Bird, 1993). In general, more visible demographic characteristics such as sex and ethnicity have larger negative effects than variables that are less visible like age (Cummings, Zhou, & Oldham, 1993; Pelled, 1993).

What is less clear from this research is precisely how and when differences in minority status lead to negative outcomes (e.g., Espinoza & Garza, 1985; Garza & Santos, 1991; Riordan & Shore, 1997). Research on gender, for example, suggests that men in the minority may react more negatively than women (Riordan & Shore, 1997; Wharton & Baron, 1987). To resolve this ambiguity, future research might productively examine not just overall measures of variation in group composition as captured by the coefficient of variation, but also consider proportional measures within groups. Using proportional measures may allow us to discover differential effects for specific minority samples.

At a more macro level, the evidence for performance effects is less clear. There is some indication that at the organizational level top management heterogeneity can be positively related to organizational performance (e.g., Bantel & Jackson, 1989; Eisenhardt & Schoonhoven, 1990; Murray, 1989). However, other studies that look directly at process variables find similar effects for top team homogeneity (e.g., Michel & Harnbrick, 1992; O’Reilly & Flatt, 1989; O‘Reilly, Snyder, & Boothe, 1993; Zajac, Golden, & Shortell, 1991). Triandis, Hall, and Ewen (1965) suggested that dissimilarity in groups, while offering the potential for more creative solutions, also led to more difficulties in group functioning. They concluded that in order to capture the benefits of diversity, groups needed to be able to resolve their differences. Over 30 years later, this conclusion remains valid. A critical determinant of the outcome is whether the diversity is having constructive or destructive effects on the process (e.g., Flatt, 1996; Iaquinto & Frederickson, 1996). Part of the reason for these differences lies in the complexity of the diversity-process-performance chain and the competing effects of the added value of increased information versus the increased difficulties of communicating and solving problems in highly diverse groups. At this macro level, any causal linkages between the composition of the senior team and organizational outcomes has to be a function of a large number of other unmeasured variables (Jehn, Northcraft, & Neale, 1997)...

Muminghan and Conlon (1991), in a Clever study of British string quartets, found that the most successful groups were those that were able to contain conflict so that it did not become disruptive. Quartets that confronted conflict were less successful. Wall and Nolan (1986) found that lower levels of conflict were associated With greater satisfaction and equity among group members. But what is the theoretical basis of these effects? If social categorization and similarity/attraction biases are the driving force behind the negative effects of diversity, what actions may be taken to address these directly? Evidence from Mullen and Copper (1994), O’Reilly, Williams, and Barsade (1997), and Gaertner and colleagues (1990) suggest that rather than emphasizing demographic differences and increasing task conflict, social categorization processes that emphasize common goals and identities and inhibit dysfunctional conflict may enhance group processes and performance.

For example, research has shown that increased cooperation can reduce categorization biases and in-group distinctions (Gaertner et al., 1990). Further, when attempts are made to deliberately promote identification with the larger group and minimize subgroup identification, intergroup bias can be reduced even more (Gaertner, Mann, Murrell & Dovidio, 1989). This may be done by identifying salient out- groups or through an emphasis on collectivistic cultures to help override the natural inclinations people have to make invidious social categoiizations (O‘Reilly & Chatrnan, 1996). These processes are consistent with the findings of Chatman and colleagues (1997) who find that collectivistic cultures mitigate the negative effects of diversity. Tushman and O’Reilly (1996) contend that for long-term success organizations and groups need to be “ambidextrous”; that is, able to tolerate diverse structures and cultures. They argue that this is done with strong, inclusive cultures that promote a common identity even in the face of differing perspectives. The use of this cognitive approach may also be seen in the Chatman and Barsade (1995) study showing how organizational cultures may be created that foster commitment and cooperation.

Another possible cause of reduced conflict and increased cooperation may stem from Gaertner and Dovidio‘s (1977) notion of “aversive racism”; that is, faced with strong normative pressures to override invidious social categorizations, group members may enhance their ability to perform by consciously overriding the propensity to differentiate in-groups and out-groups. This may improve teamwork because of the awareness of the social stigma attached to socially inappropriate social categorization. As O’Reilly, Williams, and Barsade (1997) have suggested, an important function of management may be to use the psychology of self-categorization to help employees to identify with a culture that is inclusive and not with in-groups based on characteristics unrelated to job performance like sex and race...

There is an impressive amount of high-quality laboratory and field research on diversity and demography in organizations. Overall, this research offers convincing support for the argument that variations in group demography can have both direct and indirect effects on group process and performance. Under ideal conditions increased diversity may have the positive effects predicted by information and decision theories. However, consistent with social categorization and similarity/attraction theories, the preponderance of empirical evidence suggests that diversity is most likely to impede group functioning. Unless steps are taken to actively counteract these effects, the evidence suggests that, by itself, diversity is more likely to have negative than positive effects on group performance. Simply having more diversity in a group is no guarantee that the group will make better decisions or function effectively. In our view, these conclusions suggest that diversity is a mixed blessing and requires careful and sustained attention to be a positive force in enhancing performance...

Ignoring the negative consequences of diversity is not the answer. Ironically, understanding these negative effects may provide a solution for its more pernicious effects. This is the good news from this review."


So diversity hurts group performance, and to reduce the negative effects of diversity, it is best to get people to think of themselves as the same - in other words, emphasising your diversity increases the negative impact of diversity.

One corollary of this is that identity politics is harmful when the identity group has to work with others in a larger context.


Addendum: See also How Diversity Hurts Performance (Revisited)