Does Diversity Pay?

: Race, Gender, and

the Business Case for Diversity
Cedric Herring
University of Illinois at Chicago

The value-in-diversity perspective argues that a diverse workforce, relative to a

homogeneous one, is generally beneficial for business, including but not limited to
corporate profits and earnings. This is in contrast to other accounts that view diversity
as either nonconsequential to business success or actually detrimental by creating
conflict, undermining cohesion, and thus decreasing productivity. Using data from the
1996 to 1997 National Organizations Survey, a national sample of for-profit business
organizations, this article tests eight hypotheses derived from the value-in-diversity
thesis. The results support seven of these hypotheses: racial diversity is associated with
increased sales revenue, more customers, greater market share, and greater relative
profits. Gender diversity is associated with increased sales revenue, more customers, and
greater relative profits. I discuss the implications of these findings relative to alternative
views of diversity in the workplace.

roponents of the value-in-diversity per-

P spective often make the “business case for
diversity” (e.g., Cox 1993). These scholars claim
closer examination of assumptions, a more com-
plex learning environment, and, arguably, bet-
ter solutions to workplace problems (Gurin,
that “diversity pays” and represents a com- Nagda, and Lopez 2004). Because of such puta-
pelling interest—an interest that meets cus- tive competitive advantages, companies increas-
tomers’ needs, enriches one’s understanding of ingly rely on a heterogeneous workforce to
the pulse of the marketplace, and improves the increase their profits and earnings (Florida and
quality of products and services offered (Cox Gates 2001, 2002; Ryan, Hawdon, and Branick
1993; Cox and Beale 1997; Hubbard 2004; 2002; Williams and O’Reilly 1998).
Richard 2000; Smedley, Butler, and Bristow Critics of the diversity model, however, are
2004). Moreover, diversity enriches the work- skeptical about the extent to which these bene-
place by broadening employee perspectives, fits are real (Rothman, Lipset, and Nevitte
strengthening their teams, and offering greater 2003a, 2003b; Skerry 2002; Tsui, Egan, and
resources for problem resolution (Cox 2001). O’Reilly 1992; Whitaker 1996). Scholars who
The creative conflict that may emerge leads to see “diversity as process loss” argue that diver-
sity incurs significant potential costs (Jehn,
Northcraft, and Neale 1999; Pelled 1996; Pelled,
Direct all correspondence to Cedric Herring, Eisenhardt, and Xin 1999). Skerry (2002), for
Department of Sociology (MC 312), University of instance, points out that racial and ethnic diver-
Illinois at Chicago, 1007 W. Harrison, Chicago, IL sity is linked with conflict, especially emotion-
60607 ([email protected]). I would like to thank al conflict among co-workers. Tsui and
William Bielby, John Sibley Butler, Sharon Collins, colleagues (1992) concur, suggesting that diver-
Paula England, Tyrone Forman, Hayward Derrick sity diminishes group cohesiveness and, as a
Horton, Loren Henderson, Robert Resek, Moshe
result, employee absenteeism and turnover
Semyonov, Kevin Stainback, and the editors of ASR
and the anonymous reviewers for their extraordinar-
increase. Greater diversity may also be associ-
ily helpful comments and suggestions. An earlier ated with lower quality because it can lead to
version of this article was presented at the Annual positions being filled with unqualified workers
Conference of the American Sociological Association (Rothman et al. 2003b; see also Williams and
in Montreal, Canada in August 2006. O’Reilly 1998). For these reasons, skeptics

AMERICAN SOCIOLOGICAL REVIEW, 2009, VOL. 74 (April:208–224)

 DOES DIVERSITY PAY?—–209

question the real impact of diversity programs number of customers, relative market share,
on businesses’ bottom line. and relative profitability.
Is it possible that diversity has dual outcomes,
some of which are beneficial to organizations VALUE IN DIVERSITY
and some of which are costly to group func-
tioning? Diversity may be valuable even if The definition of “diversity” is unclear, as
changes in an organization’s composition make reflected in the multiplicity of meanings in the
incumbent members uncomfortable. As literature. For some, the term provokes intense
DiTomaso, Post, and Parks-Yancy (2007:488) emotional reactions, bringing to mind such
point out, “research generally finds that het- politically charged ideas as “affirmative action”
erogeneity on most any salient social category and “quotas.” These reactions stem, in part,
contributes to increased conflict, reduced com- from a narrow focus on protected groups cov-
munication, and lower performance, at the same ered under affirmative action policies, where
time that it can contribute to a broader range of differences such as race and gender are the focal
contacts, information sources, creativity, and point. Some alternative definitions of diversity
innovation.” This suggests that diversity may be extend beyond race and gender to include all
conducive to productivity and counterproduc- types of individual differences, such as ethnic-
tive in work-group processes. Many of the ity, age, religion, disability status, geographic
claims and hypotheses about diversity’s impacts location, personality, sexual preferences, and a
have not been examined empirically, so it is not myriad of other personal, demographic, and
clear what effect, if any, diversity has on the organizational characteristics. Diversity can
overall functioning of organizations, especial- thus be an all-inclusive term that incorporates
ly businesses. people from many different classifications.
In this article, I empirically examine key Generally, “diversity” refers to policies and
questions pertaining to organizational diversi- practices that seek to include people who are
ty and its implications. Does diversity offer the considered, in some way, different from tradi-
many benefits suggested by the value-in-diver- tional members. More centrally, diversity aims
sity thesis? Or, do costs offset potential bene- to create an inclusive culture that values and uses
f its? Perhaps diversity is simultaneously the talents of all would-be members.
associated with the twin outcomes of group- The politics surrounding diversity and inclu-
level conflict and increased performance at the sion have shifted dramatically over the past 50
establishment level. Although no singular years (Berry 2007). Title VII of the 1964 Civil
research design could adjudicate between all Rights Act makes it illegal for organizations to
the claims of proponents and skeptics,1 the ques- engage in employment practices that discrimi-
tions I raise warrant serious examination given nate against employees on the basis of race,
the growing heterogeneity of the U.S. work- color, religion, sex, or national origin. This act
force. Using data from the 1996 to 1997 mandates that employers provide equal employ-
National Organizations Survey, my analyses ment opportunities to people with similar qual-
explore the relationship between racial and gen-
ifications and accomplishments. Since 1965,
der workforce diversity and several indicators
Executive Order 11246 has required govern-
of business performance, such as sales revenue,
ment contractors to take affirmative action to
overcome past patterns of discrimination. These
directives eradicated policies that formally per-
1 It is possible to derive propositions from argu-
mitted discrimination against certain classes of
ments against the business case for diversity thesis, workers. They also increased the costs to organ-
but works in this skeptical vein typically provide cri- izations that fail to implement fair employment
tiques of more optimistic views of diversity, rather practices.
than systematic, alternative theoretical formulations.
By the late 1970s and into the 1980s, there
When skeptics do put forth testable hypotheses, they
tend to focus on intermediary processes and mech- was growing recognition within the private sec-
anisms rather than the diversity–business “bottom tor that these legal mandates, although neces-
line” linkage, per se. I cite such critiques to show that sary, were insufficient to effectively manage
there are reasons to be skeptical about the link organizational diversity. Many companies and
between diversity and business performance. consulting firms soon began offering training
210—–AMERICAN SOCIOLOGICAL REVIEW

programs aimed at “valuing diversity.” In their working together and capitalizing on their indi-
study of private-sector establishments from viduality. The best group decisions and predic-
1971 to 2002, Kalev, Dobbin, and Kelly (2006) tions are those that draw on unique qualities. In
find that programs designed to establish organi- this regard, Bunderson and Sutcliffe (2002)
zational responsibility for diversity are more show that teams composed of individuals with
efficacious in increasing the share of white a breadth of functional experiences are well-suit-
women, black women, and black men in man- ed to overcoming communication barriers: team
agement than are attempts to reduce manageri- members can relate to one another’s functions
al bias through diversity training or reduce social while still realizing the performance benefits of
isolation through mentoring women and racial diverse functional experiences. Williams and
and ethnic minorities. They also find that O’Reilly (1998) and DiTomaso and colleagues
employers who were subject to federal affir- (2007) concur, arguing that diversity increases
mative action edicts are likely to have diversi- the opportunity for creativity and the quality of
ty programs with stronger effects. the product of group work.
During the 1990s, diversity rhetoric shifted The benefits of diversity may extend beyond
to emphasize the “business case” for workforce team and workplace functioning and problem
diversity (Bell and Hartmann 2007; Berry 2007; solving. Sen and Bhattacharya (2001), for
Hubbard 1997, 1999; von Eron 1995). instance, propose that diversity influences con-
Managing diversity became a business neces- sumers’ perceptions and purchasing practices.
sity, not only because of the nature of labor Indeed, Black, Mason, and Cole (1996) find
markets, but because a more diverse workforce that consumers have strongly held in-group
was thought to produce better business results. preferences when a transaction involves signif-
Exploiting the nation’s diversity was viewed as icant customer–worker interaction. Richard
key to future prosperity. Ignoring the fact that (2000) argues that cultural diversity, within the
discrimination limits a society’s potential proper context, provides a competitive advan-
because it leads to underutilization of talent tage through social complexity at the firm level.
pools was no longer practical nor feasible. Irrespective of the specific processes, diversi-
Diversity campaigns became part of the attempt ty may positively influence organizations’ func-
to strengthen the United States and move tioning, net of any internal work-group
beyond its history of discrimination by pro- processes that diversity may impede.
viding previously excluded groups greater The sociological literature on diversity con-
access to educational institutions and work- tinues to grow (e.g., Alon and Tienda 2007;
places (Alon and Tienda 2007). Today, advo- Bell and Hartmann 2007; Ber ry 2007;
cates are asked to find evidence to support the DiTomaso et al. 2007; Embrick forthcoming;
business-case argument that diversity expands Kalev et al. 2006), but, to date, there has been
the talent pool and strengthens U.S. institu- little systematic research on the impact of diver-
tions. Even if the shift from affirmative action sity on businesses’ financial success. Few stud-
to diversity has “tamed” what began as a radi- ies use quantitative data and objective
cal fight for equality (Berry 2007), workforce performance measures from real organizations
diversity has become an essential business con- to assess hypotheses. One exception compares
cern in the twenty-first century. companies with exemplary diversity manage-
ment practices with those that paid legal dam-
THE IMPLICATIONS OF DIVERSITY FOR ages to settle discrimination lawsuits. The results
WORKPLACE DYNAMICS AND BUSINESS show that the exemplary firms perform better,
as measured by their stock prices (Wright et al.
OUTCOMES
1995).
How might diversity affect business outcomes? A second exception is a series of studies
Page (2007) suggests that groups displaying a reported by Kochan and colleagues (2003).
range of perspectives outperform groups of They find no significant direct effects of either
like-minded experts. Diversity yields superior racial or gender diversity on business perform-
outcomes over homogeneity because progress ance. Gender diversity has positive effects on
and innovation depend less on lone thinkers group processes while racial diversity has neg-
with high intelligence than on diverse groups ative effects. The negative relationship between
 DOES DIVERSITY PAY?—–211

racial diversity and group processes, however, The results of research on heterogeneity in
is largely absent in groups with high levels of groups suggest that diversity offers a great
training in career development and diversity opportunity for organizations and an enormous
management. These scholars also find that racial challenge. More diverse groups have the poten-
diversity is positively associated with growth in tial to consider a greater range of perspectives
branches’ business portfolios. Racial diversity and to generate more high-quality solutions
is associated with higher overall performance in than do less diverse groups (Cox, Lobel, and
branches that enact an integration-and-learn- McLeod 1991; Hoffman and Maier 1961;
ing perspective on diversity, but employee par- Watson, Kumar, and Michaelsen 1993). Yet, the
ticipation in diversity education programs has greater the amount of diversity in a group or an
a limited impact on performance. Finally, they organizational subunit, the less integrated the
find no support for the idea that diversity that group is likely to be (O’Reilly, Caldwell, and
matches a firm’s client base increases sales by Barnett 1989) and the higher the level of dis-
satisfying customers’ desire to interact with satisfaction and turnover (Jackson et al. 1991;
Wagner, Pfeffer, and O’Reilly 1984). Diversity’s
those who physically resemble them.2
impact thus appears paradoxical: it is a double-
edged sword, increasing both the opportunity for
DEMOGRAPHIC DIVERSITY AND creativity and the likelihood that group members
ORGANIZATIONAL FUNCTIONING will be dissatisfied and will fail to identify with
Researchers studying demographic diversity the organization. Including influential and
potentially confounding demographic and organi-
typically take one of two approaches in their
zational factors—as I do in the following analy-
treatment of the subject. One approach treats
ses—helps clarify the relationship between
diversity broadly, making statements about het-
diversity and organizational functioning, espe-
erogeneity or homogeneity in general, rather
cially in business organizations.
than about particular groups (e.g., Hambrick
and Mason 1984). The alternative treats each
demographic diversity variable as a distinct the- ALTERNATIVE EXPLANATIONS
oretical construct, based on the argument that For-profit workplaces are appropriate sites for
different types of diversity produce different examining questions about diversity, as they
outcomes (e.g., Hoffman and Maier 1961; Kent are where employment decisions are made and
and McGrath 1969). Instead of assuming that all the settings in which work is performed (Baron
types of diversity produce similar effects, these and Bielby 1980; Reskin, McBrier, and Kmec
researchers build their models around specific 1999). Moreover, it is organizational processes
types of demographic diversity (e.g., Zenger that perpetuate segregation and influence the
and Lawrence 1989). Indeed, Smith and col- character of jobs and workplaces (Tomaskovic-
leagues (2001) argue that scholars should not Devey 1993).
lump women and racial minorities together as In assessing the relationship between diver-
a standard approach to research. The relative sity and business outcomes, the literature sug-
number, power, and status of various groups gests that there are several organizational factors
within organizations can significantly affect that might be influential. According to the insti-
issues such as favoritism and bias, and aggre- tutional perspective in organizational theory,
gating these groups may mask such variations. organizational behavior is a response to pres-
sures from the institutional environment
(Stainback, Robinson, and Tomaskovic-Devey
2 It would be preferable to model an array of inter-
2005). The institutional environment of an
organization embodies regulative, normative,
nal processes in organizations to establish how diver-
and cultural-cognitive institutions affecting the
sity affects business performance (e.g., functionality
of work teams, marketing decisions, or innovation in organization, such as law and social attitudes
production or sales), but this is not possible with (Scott 2003).
data from the National Organizations Survey because According to this formulation of organiza-
such indicators are not available. These topics do tional behavior, adoption of new organization-
offer potentially fruitful paths for future research. al practices is often an attempt to gain legitimacy
212—–AMERICAN SOCIOLOGICAL REVIEW

in the eyes of important constituents, and not Stinchcombe (1965) offers a rationale for why
necessarily an attempt to gain greater efficien- establishment age may also be meaningful. He
cy (DiMaggio and Powell 2003). Based on insti- suggests a “liability of newness,” whereby organi-
tutional theory, for-profit businesses that are zational mortality rates decrease with organiza-
accountable to a larger public may be more sen- tional age. Younger organizations are more prone
sitive to public opinion on what constitutes to mortality than are older organizations, so they
legitimate organizational behavior. Publicly- approach threats to their existence differently. It
held, for-profit businesses’ employment prac- is possible, therefore, that organizations of dif-
tices should thus be more subject to public ferent ages vary in their responses to racial and
scrutiny. These businesses should employ rela- gender diversity concerns.
tively more minorities than private-sector The labor pools from which establishments
employers, to the degree that they are under hire may be important relative to the effects of
greater pressure to achieve racial and ethnic diversity on business outcomes. The sex and
diversity, as public sentiment views such poli- racial composition of regional labor markets may
cies as a necessary element of legitimate organi- influence an establishment’s composition, as well
zational governance (Edelman 1990). as its relative success (Blalock 1957). Regional
Organizational size is also important since it differences in residential segregation, however,
is positively related to sophisticated personnel may obscure regional effects of demographic
systems (Pfeffer 1977) that may contribute to composition (Jones and Rosenfeld 1989).
diversity in the workplace. Large organizations Finally, industrial-sector variations may relate
concerned about due process and employment simultaneously to levels of diversity and busi-
practices will institute specific offices and pro- ness performance. In particular, organizations
cedures for handling employee complaints in the service sector will be more proactive
(Gwartney-Gibbs and Lach 1993; Welsh, with regard to racial and gender diversity than
Dawson, and Nierobisz 2002). At the same time, will those that produce tangible goods, as their
if some organizations, when left to their own performance depends more on public goodwill.
devices, prefer hiring whites over racial minori- There are also reasons, however, to believe that
ties, their larger size and slack resources give the economic sector in which businesses oper-
them more ability to indulge preferences for ate can matter. Compared with manufacturing
white workers (Cohn 1985; Tolbert and and public service, service-sector establish-
Oberfield 1991). ments are more likely to exclude blacks, espe-
Large establishments also tend to make greater cially black men, by using personality traits and
efforts at prevention and redress because they appearance as job qualifications (Moss and
have direct legal obligations. Antidiscrimination Tilly 1996).
laws make discrimination against minorities and
women potentially costly, but not all establish- DIVERSITY, BUSINESS
ments are subject to these laws. Federal law ban- PERFORMANCE, AND
ning sex discrimination in employment exempts EXPECTATIONS
firms with fewer than 15 workers, and enforce-
ment efforts often target large firms (Reskin et Evidence that directly assesses the business
al. 1999). Moreover, affirmative action regula- case for diversity has, at best, proven elusive.
Building on the present discussion, I offer sev-
tions apply only to firms that do at least $50,000
eral predictions. In its most basic form, the
worth of business with the federal government
business case for diversity perspective predicts
and have at least 50 employees (Reskin 1998).
a return on investment for diversity (Hubbard
Establishment size may thus be related to vul-
2004). It is thus fairly easy to derive some
nerability to equal employment opportunity and
straightforward expectations:
affirmative action regulations, which in turn
should be related to increased racial and ethnic Hypothesis 1a: As racial workforce diversity
diversity. Indeed, Holzer (1996) shows that affir- increases, a business organization’s sales
mative action implementation has led to gains in revenues will increase.
the representation of African Americans and Hypothesis 1b: As gender workforce diversity
white women in firms required to practice affir- increases, a business organization’s sales
mative action. revenues will increase.
 DOES DIVERSITY PAY?—–213

Hypothesis 2a: As racial workforce diversity The resulting sample is representative of U.S.
increases, a business organization’s number profit-making work organizations. For each
of customers will increase. organization sampled, Dun and Bradstreet’s
Hypothesis 2b: As gender workforce diversity Market Identifiers Plus service provides sever-
increases, a business organization’s number al important pieces of information: company
of customers will increase. name, address, and telephone number; size (in
Hypothesis 3a: As racial workforce diversity terms of number of employees and sales vol-
increases, a business organization’s market ume); year started; and business trends for the
share will increase. past three years. In addition, historical infor-
Hypothesis 3b: As gender workforce diversity mation on the sampled organizations is available
increases, a business organization’s market from the Dun’s Historical Files. The unit of
analysis is the workplace.
share will increase.
Hypothesis 4a: As racial workforce diversity
increases, a business organization’s profits DIVERSITY
relative to its competitors will increase. There are two basic approaches to measuring
Hypothesis 4b: As gender workforce diversity diversity, either globally or as distinct indicators.
increases, a business organization’s profits Following Skaggs and DiTomaso (2004), I opt
relative to its competitors will increase. for separate but parallel indicators of racial and
It is possible that racial and gender diversity gender diversity. There are several reasons for
in the workforce are related to some outcomes this. Previous research shows that race and gen-
but not others. My analyses thus incorporate an der as bases for diversity are extremely impor-
array of tangible outcomes and benefits sur- tant in understanding human transactions. For
rounding sales revenue, customer base, market most people, these group identities are not eas-
share, and relative profitability. Following the ily changeable. In addition, the base of knowl-
literature, I also examine relations between edge in the social sciences is more fully
diversity and business outcomes net of other fac- developed for these identities than for others that
tors (e.g., legal form of organization, establish- may be relevant. On a pragmatic level, indica-
ment size, company size, organization age, tors for these two dimensions are readily avail-
industrial sector, and region). able in the data source, while others are not
(e.g., sexual preference or age).
The specific indicator draws from the Racial
DATA AND METHODS
Index of Diversity (RID) (Bratter and Zuberi
The data come from the 1996 to 1997 National 2001; Zuberi 2001), which provides an unbiased
Organizations Survey (NOS) (Kalleberg, estimator of the probability that two individu-
Knoke, and Marsden 2001), which contains als chosen at random and independently from
information from 1,002 U.S. work establish- the population will belong to different racial
ments, drawn from a stratified random sample groups. The index ranges from 0 to 1. A score
of approximately 15 million work establish- of 0 indicates a racially homogenous population;
ments in Dun and Bradstreet’s Information 1 indicates a population where, given how race
Services data file. I use data from the 506 for- is distributed, every randomly selected pair is
profit business organizations that provided composed of persons from two different racial
information about the racial composition of groups.
their full-time workforces, their sales revenue,
(⌺ ni (ni – 1))
their number of customers, their market share, RID = 1 –
and their profitability. The NOS concentrates on N (N – 1)
U.S. work establishments’ employment con-
1 – (1/i)
tracts, staffing methods, work organization, job
training programs, and employee benefits and In this formula, N is the total population, ni
incentives. The data include additional infor- denotes that population is separate racial group
mation about each organization’s formal struc- (i), and i refers to number of racial groups. The
ture, social demography, environmental RID is a measure of the concentration of racial
situation, and productivity and performance. classification in a population if the population
214—–AMERICAN SOCIOLOGICAL REVIEW

to which each individual belongs is considered R and AID-G). Both models work well, but the
to be one racial group, and n1 .|.|. ni is the num- AID-R and AID-G models with the parity
ber of individuals in the various groups (so that assumption give a better fit. Moreover, they
3n i = N for a population with three racial provide a theoretical rationale for why diversi-
groups). If all the individuals in a racial group ty is related to business outcomes. Given these
are concentrated in one category, the measure results, I use both the AID-R and AID-G (par-
would be .0. To constrain the RID between .0 ity) operationalizations (for both race and gen-
and 1.0, I calculate the actual level of diversity der).
as a proportion of the maximum level possible In this study, the superordinate racial group,
with the specified number of races, that is, 1 – whites, make up 75 percent of the population in
(1/i). In a population with an even racial distri- the 1996 to 1997 NOS. Scores on the asym-
bution, the RID would equal 1.00. metrical index of diversity for race thus range
One limitation of the RID (as with most from a low of 0 (homogenous white) to a high
indexes of dissimilarity) is that it treats over- of 25 (racial parity). Males, the superordinate
representation and underrepresentation of sub- gender, make up 54 percent of the population
groups as mathematically equivalent for in the 1996 to 1997 NOS. Scores on the asym-
computational purposes. Arguably, an organi- metrical index of diversity for gender thus range
zation with a 20 percent underrepresentation of from a low of 0 (homogenous male) to a high
minorities (relative to the population) would be of 46 (gender parity).
very different from an organization with a 20
percent overrepresentation of minorities (rela-
tive to the population). BUSINESS PERFORMANCE
To correct for this limitation, I modify the To measure average annual sales revenue,
RID to incorporate the idea that power rela- respondents were asked, “What was your organi-
tions between superordinate and subordinate zation’s average annual sales revenue for the past
groups are asymmetrical. The key idea is that two years?” Responses range from 0 to 60 bil-
of parity. The Asymmetrical Index of Diversity lion dollars. For multivariate analysis, I add a
(AID) is small number (.01) to each observation before
AID = (1 – S) if S > P dividing it by 1 million and taking the log of
these values.
AID = (1 – P) if P ≥ S To measure the number of customers, respon-
where S is the proportion of the organization dents were asked, “About how many customers
composed of the superordinate group, and P is purchased the organization’s products or ser-
the proportion of the population composed of vices over the past two years?” Responses range
the superordinate group. from 0 to 25 million customers.
Scores can range from 0 (completely homo- To measure perceived market share, respon-
geneous organization composed of the super- dents were asked, “Compared to other organi-
ordinate group) to “parity” (i.e., 1 – P) (where zations that do the same kind of work, how
the superordinate and subordinate groups have would you compare your organization’s per-
reached proportional representation in the formance over the past two years in terms of
organization). For example, if the superordi- market share? .|.|. Would you say that it was (1)
nate group constitutes 75 percent of the popu- much worse, (2) somewhat worse, (3) about the
lation, AID scores can range from 0 to .25 (or same, (4) somewhat better, or (5) much bet-
25 when multiplied by 100). If the superordinate ter?”
group constitutes 50 percent of the population, To measure relative profitability, respondents
AID scores can range from 0 to .50 (or 50 when were asked, “Compared to other organizations
multiplied by 100). that do the same kind of work, how would you
The racial diversity index and the gender compare your organization’s performance over
diversity index are both operationalized using the past two years in terms of profitability? .|.|.
two alternative premises: (1) without the under- Would you say that it was (1) much worse, (2)
lying notion of parity (i.e., RID) and (2) with an somewhat worse, (3) about the same, (4) some-
underlying notion of parity mattering (i.e., AID- what better, or (5) much better?”
 DOES DIVERSITY PAY?—–215

CONTROLS an establishment and its sales revenue, number

of customers, market share, or profitability?
Respondents were asked about their establish- Figure 1 illustrates the distribution of estab-
ments’ legal form of organization (if for prof- lishments with low, medium, and high levels of
it). Specifically, “What is the legal form of racial and gender diversity. Thirty percent of
organization? Is it a sole proprietorship, part- businesses have low levels of racial diversity, 27
nership or limited partnership, franchise, cor- percent have medium levels, and 43 percent
poration with publicly held stock, corporation have high levels. Regarding gender diversity, 28
with privately held stock, or something else?” percent of businesses have low levels, 28 per-
Responses are dummy coded for proprietor- cent have medium levels, and 44 percent have
ship, partnership, franchise, and corporation. I high levels.
exclude not-for-profit organizations because Table 1 presents means and percentage dis-
the focus of this study is on the “business case tributions of various business outcomes of estab-
for diversity.” lishments by their levels of diversity. Average
To determine organization size, respondents sales revenues are associated with higher levels
were asked about the number of full-time and of racial diversity: the mean revenues of organ-
part-time employees who work at all locations izations with low levels of racial diversity are
of the business. Responses range from 15 to roughly $51.9 million, compared with $383.8
300,000 employees. Respondents were also million for those with medium levels and $761.3
asked about the establishment size (i.e., the million for those with high levels of diversity.
number of full-time and part-time employees on The same pattern holds true for sales revenue
the payroll at their particular address). by gender diversity: the mean revenues of organ-
Responses are coded from 1 to 42,000. izations with low levels of gender diversity are
To determine an organization’s age, respon- roughly $45.2 million, compared with $299.4
dents were asked, “In what year did the organ- million for those with medium levels and $644.3
ization start operations?” The difference million for those with high levels of diversity.
between the year of the survey and the year of Higher levels of racial and gender diversity
establishment yields the organization age, rang- are also associated with greater numbers of cus-
ing from 1 to 361 years. tomers: the average number of customers for
To measure industry, respondents were asked organizations with low levels of racial diversi-
about the main product or service provided at ty is 22,700. This compares with 30,000 for
their establishments. Responses are dummy those with medium levels of racial diversity
coded into the following 12 categories: agri- and 35,000 for those with high levels. The mean
culture; mining; construction; transportation number of customers for organizations with
and communications; wholesale; retail; finan- low levels of gender diversity is 20,500. This
cial services, insurance, and real estate; business compares with 27,100 for those with medium
services; personal services; entertainment; pro- levels of gender diversity and 36,100 for those
fessional services; and manufacturing. with high levels.
The region in which the establishment is Table 1 also shows that businesses with high
located is dummy coded into four categories: levels of racial (60 percent) and gender (62 per-
Northeast (CT, DE, MA, ME, NH, NJ, NY, PA, cent) diversity are more likely to report higher
RI, and VT), Midwest (IA, IL, IN, KS, MI, MN, than average percentages of market share than
MO, ND, NE, OH, OK, SD, and WI), South are those with low (45 percent) or medium lev-
(AL, AR, DC, FL, GA, KY, LA, MD, MS, NC, els of racial (59 percent) and gender (58 percent)
SC, TN, TX, VA, and WV), and West and oth- diversity. A similar pattern emerges for organ-
ers (AK, AZ, CA, CO, HI, ID, MT, NM, NV, izations reporting higher than average prof-
OR, UT, WA, and WY). itability. Less than half (47 percent) of
organizations with low levels of racial diversi-
ty report higher than average profitability. In
RESULTS
contrast, about two thirds of those with medi-
How is diversity related to organizations’ busi- um and high levels of racial diversity report
ness performance? Does a relationship exist higher than average profitability. Also, organi-
between the racial and gender composition of zations with high levels of gender diversity (62
216—–AMERICAN SOCIOLOGICAL REVIEW

Figure 1. Percentage Distribution of Racial and Gender Diversity Levels in Establishments

Table 1. Means and Percentage Distributions for Business Outcomes of Establishments by Levels
of Racial and Gender Diversity

Racial Diversity Level Gender Diversity Level

Low Medium High Low Medium High
Characteristics (<10%) (10–24%) (25%+) (<20%) (20–44%) (45%+) Overall
Percent in racial or gender diversity category 30 27 43 28 28 44 100
Mean sales revenue (in millions) 51.9 383.8 761.3 45.2 299.4 644.3 456.3
Mean number of customers (in thousands) 22.7 30.0 35.0 20.5 27.1 36.1 31.9
Percent with higher than average market share 45 59 60 45 58 62 56
Percent with higher than average profitability 47 63 61 45 58 62 56

percent) are more likely to report higher than This research establishes the correlation and plausi-
average profitability than are those with low ble explanations about why it matters. Unfortunately,
(45 percent) or medium levels (58 percent) of the limitations of cross-sectional survey data pre-
gender diversity. vent a full exploration of mechanisms. First, in using
Although interesting, the descriptive statistics only survey data, it is difficult to rule out the possi-
do not provide much information about the net bility of unmeasured differences between organiza-
relationship between diversity and organiza- tions with low and high diversity. Second, I can
specify plausible models for a direct causal link from
tions’ business performance.3 To address this
diversity at Time 1 to business outcomes at Time 2.
But, I could also specify equally plausible models for
business outcomes at Time 1 that generate (or at
3 Before engaging in disputes about why diversi- least permit) various levels of diversity at Time 2.
ty matters to business outcomes, it is essential to Another limitation of cross-sectional survey research
first establish conclusively that there is a correlation. is its inability to formally identify mechanisms.
 DOES DIVERSITY PAY?—–217

Table 2. Regression Equations Predicting Sales, Number of Customers, Market Share, and Relative
Profitability with Racial and Gender Diversity

Model 1 Model 2 Model 3 Model 4

Independent Variables Sales Customers Market Share Profitability
Constant 4.847*** 12111.880*** 3.364*** 3.361***
Racial diversity .087*** 509.398*** .009*** .008**
Gender diversity .042*** 278.971*** .004* .006**
R2 .058*** .038*** .021*** .025***
N 506 506 506 470
Notes: Coefficients are unstandardized. For the dummy (binary) variable coefficients, significance levels refer to
the difference between the omitted dummy variable category and the coefficient for the given category.
* p < .1; ** p < .05; *** p < .01.

issue more rigorously, Tables 2 and 3 present diversity increases, estimates of relative prof-
results from multivariate analysis. Table 2 itability also increase. The results are statistically
shows the relationship between racial and gen- significant and consistent with Hypotheses 4a
der diversity in establishments and logged sales and 4b.
revenue, number of customers, estimates of But do these results hold up once alternative
relative market shares, and estimates of relative explanations are taken into account? Table 3
profitability. Model 1 shows that diversity and presents the same relationships as Table 2, but
sales revenues are positively related. Diversity it includes controls for alternative explanations.
accounts for roughly 6 percent of the variance Model 1 of Table 3 presents the relationship
in sales revenue. These results are fully con- between racial and gender diversity in estab-
sistent with Hypotheses 1a and 1b. Model 2 lishments and logged sales revenue. In Model
shows that racial and gender diversity are also 1, the relationships between racial and gender
significantly related to the number of cus- diversity and sales revenues remain significant
tomers. As the racial and gender diversity in (p < .05), net of controls for legal form of organ-
establishments increase, their number of cus- ization, company size, establishment size, organ-
tomers also increases. ization age, industrial sector, and region. Model
Diversity accounts for less than 4 percent of 2 shows that a one unit increase in racial diver-
the variance in number of customers, but the sity increases sales revenues by approximately
relationship is statistically significant for both 9 percent; a one unit increase in gender diver-
racial and gender diversity. These results are sity increases sales revenues by approximately
consistent with Hypotheses 2a and 2b. Model 3 percent. Combined, these factors account for
3 in Table 2 shows a positive relationship 16.5 percent of the variance in sales revenue.
between racial and gender diversity and esti- The results in Table 3 provide support for
mates of relative market shares. As diversity in Hypotheses 1a and 1b (i.e., as a business organi-
establishments increases, estimates of relative zation’s racial and gender diversity increase, its
market share also increase significantly. The sales revenue will also increase).
relationship is statistically significant for racial Model 1 of Table 3 examines alternate expla-
diversity and marginally significant for gender nations of sales revenue. Net of all other factors,
diversity. These results are consistent with privately-held corporations have significantly
Hypotheses 3a and 3b. Finally, Model 4 displays lower sales revenues than do other legal forms
the relationship between racial and gender diver- of business. No other legal forms depart sig-
sity and estimates of relative profitability. As nificantly from the omitted category. Model 1
also shows that sales revenues increase mar-
ginally as company size increases, and signifi-
Different researchers can offer several hypotheses cantly as establishment size increases and
about the mechanisms that produce the observed organizations age.
relationship between variables (in this case, diversi- The standardized coefficients for this model
ty and business outcomes). (not reported in Table 3) show that a one stan-
218—–AMERICAN SOCIOLOGICAL REVIEW

Table 3. Regression Equations Predicting Sales, Number of Customers, Market Share, and Relative
Profitability with Racial and Gender Diversity and Other Characteristics of
Establishments

Model 1 Model 2 Model 3 Model 4

Independent Variables Sales Customers Market Share Profitability
Constant 4.998*** 61545.4 3.403*** 3.363***
Racial diversity .093*** 433.86*** .007** .006*
Gender diversity .028** 195.642** .001 .005**
Proprietorship –.821 –370.78 –.232* –.161
Partnership .663 –6454.6 –.017 .256
Public corporation –.109 7376.29 .214* .202
Private corporation –1.484** –8748.7* .008 .019
Company size .00001* .352* .000** .000**
Establishment size .00001** .119 .000 .000
Organization age .013** 44.813 .001 .001
Agriculture –1.942 4188.66 –.206 –.033
Mining .739 –28856* –.168 –.264
Construction –.967 –875.7 –.152 –.036
Transportation/communications –.052 1498.75 .119 .226
Wholesale trade .008 –16383** .136 –.064
Retail trade –1.183** 7209.83* .08 –.087
F. I. R. E. –.683 –8335.4 –.212 .085
Business services –1.49* 1552.28 .204* .112
Personal services –1.566* 1480.03** .423** –.001
Entertainment –4.708** –1504.5 –.191 –.076
Professional services –.615 –13539** .138 –.023
North 2.196*** 23143.9*** –.06 –.039
Midwest 2.616*** 14968.3*** –.023 –.073
South 1.82*** 21152.8*** –.055 .059
R2 .165*** .155*** .075** .064**
N 506 506 469 484
Notes: Coefficients are unstandardized. For the dummy (binary) variable coefficients, significance levels refer to
the difference between the omitted dummy variable category and the coefficient for the given category.
* p < .1; ** p < .05; *** p < .01.

dard deviation increase in racial diversity pro- .05), controlling for legal form of organization,
duces a .188 standard deviation increase in sales company size, industrial sector, and region. A
revenue. Furthermore, the relationship between one unit increase in racial diversity increases the
racial diversity and sales revenue (Beta = .188) number of customers by more than 400; a one
is stronger than the impact of company size unit increase in gender diversity increases the
(Beta = .067), establishment size (Beta = .087), number of customers by nearly 200. The over-
and organization age (Beta = .077). Gender all model accounts for 15.5 percent of the vari-
diversity is also important to sales revenue. It ance in number of customers. These results
maintains a statistically significant relationship fully support Hypotheses 2a and 2b (i.e., as a
to sales revenue (Beta = .082), net of controls. business organization’s racial and gender diver-
Therefore, not only are racial and gender diver- sity increase, its number of customers will also
sity significantly related to sales revenue, but increase).
they are among its most important predictors. Model 2 also examines alternate explana-
Model 2 in Table 3 presents the relationship tions of sales revenue. Again, the relationship
between racial and gender diversity in estab- between racial diversity and number of cus-
lishments and number of customers. This rela- tomers (Beta = .120) is stronger than the impact
tionship remains statistically significant (p < of company size (Beta = .056), establishment
 DOES DIVERSITY PAY?—–219

size (Beta = .004), and organization age (Beta gender diversity is a strong predictor in three of
= .027). Gender diversity is also more highly the four indicators.
related to number of customers, maintaining a
statistically significant relationship (Beta = CONCLUSIONS
.079) net of controls. These results again sug-
gest that racial and gender diversity are among This article examines the impact of racial and
the most important predictors of number of cus- gender diversity on business performance from
tomers. two competing perspectives. The value-in-diver-
Model 3 in Table 3 shows that racial diversi- sity perspective makes the business case for
ty (Beta = .088) maintains its significant rela- diversity, arguing that a diverse workforce, rel-
tionship to market share, controlling for legal ative to a homogeneous one, produces better
form of organization, company size, and indus- business results. Diversity is thus good for busi-
trial sector. The relationship between gender ness because it offers a direct return on invest-
diversity (Beta = .025) and relative market share, ment, promising greater corporate profits and
however, becomes nonsignificant. Model 3 earnings. In contrast, the diversity-as-process-
accounts for 7.5 percent of the variance in esti- loss perspective is skeptical of the benefits of
mates of relative market shares. These findings diversity and argues that diversity can be coun-
are consistent with Hypothesis 3a (i.e., as a terproductive. This view emphasizes that, in
business organization’s racial diversity increas- addition to dividing the nation, diversity intro-
es, its market share will also increase), but they duces conflict and other problems that detract
do not support Hypothesis 3b. Still, racial diver- from an organization’s efficacy and profitabil-
sity is among the most important predictors of ity. In short, this view suggests that diversity
relative market share. impedes group functioning and will have neg-
Model 4 in Table 3 reports the net relation- ative effects on business performance.
ship between racial and gender diversity and rel- A third, paradoxical view suggests that greater
ative profitability. In this case, the relationship diversity is associated with more group con-
between racial diversity and relative profitabil- flict and better business performance. This is
ity remains marginally significant (p < .1), net possible because diverse groups are more prone
of other factors. The relationship between gen- to conflict, but conflict forces them to go beyond
der diversity and relative profitability remains the easy solutions common in like-minded
statistically significant. Model 4 explains less groups. Diversity leads to contestation of dif-
than 7 percent of the variance in estimates of rel- ferent ideas, more creativity, and superior solu-
ative profitability, but the results are consistent tions to problems. In contrast, homogeneity
with Hypotheses 4a and 4b (i.e., as a business may lead to greater group cohesion but less
organization’s racial and gender diversity adaptability and innovation.
increase, its profits relative to competitors will Drawing on data from a national sample of for-
also increase). Both racial diversity (Beta = profit business organizations (the 1996 to 1997
.076) and gender diversity (Beta = .089) are National Organizations Survey), my analyses
among the most important predictors of relative center on eight hypotheses consistent with the
profitability. value-in-diversity (business case for diversity)
Overall, the multivariate analyses strongly perspective and use both objective and percep-
support the business case for diversity per- tual indicators. Although some may see percep-
spective. The results are consistent with all but tual indicators as problematic, the combined use
one of the hypotheses suggesting that racial and of indicators in these analyses is a strength of the
gender diversity are related to business out- research design. It is highly unlikely that the dis-
comes. The significant relationships between tinct indicators have the same set of unobserved
diversity and various dimensions of business processes or sources of measurement error that
performance remain, even controlling for other might produce spurious results.
important factors, such as legal form of organ- The results support seven of the eight
ization, company and establishment size, organ- hypotheses: diversity is associated with
ization age, industrial sector, and region. Indeed, increased sales revenue, more customers, greater
racial diversity is consistently among the most market share, and greater relative profits. Such
important predictors of business outcomes, and results clearly counter the expectations of skep-
220—–AMERICAN SOCIOLOGICAL REVIEW

tics who believe that diversity (and any effort to sible that these relationships exist because of
achieve it) is harmful to business organizations. some other dynamic that the models do not
Moreover, these results are consistent with argu- consider. Nevertheless, the results presented
ments that a diverse workforce is good for busi- here, based on a nationally representative sam-
ness, offering a direct return on investment and ple of business organizations, are currently the
promising greater corporate profits and earn- best available.
ings. The statistical models help rule out alter- The concept of diversity was originally cre-
native and potentially spurious explanations. ated to justify more inclusion of people who
So, how is diversity related to the bottom-line were traditionally excluded from schools, uni-
performance of organizations? Critics assert versities, corporations, and other kinds of organ-
that diversity is linked with conflict, lower group izations. This research suggests that
cohesiveness, increased employee absenteeism diversity—when tethered to concerns about par-
and turnover, and lower quality and perform- ity—is linked to positive outcomes, at least in
ance. Nevertheless, results show a positive rela- business organizations. The findings presented
tionship between the racial and gender diversity here are consistent with arguments that diver-
of establishments and their business function- sity is related to business success because it
ing. It is likely that diversity produces positive allows companies to “think outside the box”
outcomes over homogeneity because growth by bringing previously excluded groups inside
and innovation depend on people from various the box. This process enhances an organiza-
backgrounds working together and capitaliz- tion’s creativity, problem-solving, and per-
ing on their differences. Although such differ- formance. To better understand how diversity
ences may lead to communication barriers and improves business performance, future research
group conflict, diversity increases the opportu- will need to uncover the mechanisms and
nities for creativity and the quality of the prod- processes involved in the diversity–business-
uct of group work. Within the proper context, performance nexus.
diversity provides a competitive advantage
Cedric Herring is Interim Department Head of the
through social complexity at the firm level. In
Sociology Department at the University of Illinois at
addition, linking diversity to the idea of parity Chicago. In addition, he is Professor of Sociology and
helps illustrate that diversity pays because busi- Public Policy in the Institute of Government and
nesses that draw on more inclusive talent pools Public Affairs at the University of Illinois. Dr. Herring
are more successful. Despite the potentially is former President of the Association of Black
negative impact of diversity on internal group Sociologists. He publishes on topics such as race
processes, diversity has a net positive impact on and public policy, stratification and inequality, and
organizational functioning. employment policy. He has published five books and
Alternatively, it is possible that the associa- more than 50 scholarly articles in outlets such as the
American Sociological Review, the American Journal
tions reported between diversity and business
of Sociology, and Social Problems. He has received
outcomes exist because more successful busi- support for his research from the National Science
ness organizations can devote more attention Foundation, the Ford Foundation, the MacArthur
and resources to diversity issues.4 It is also pos- Foundation, and others. He has shared his research
in community forums, in newspapers and magazines,
on radio and television, before government officials,
and at the United Nations.
4 Some researchers try to work around survey data

limitations by estimating fixed- and random-effects

models and using instrumental variables approach- APPENDIX
es to correct for unmeasured heterogeneity. While How do the organizations in the sample com-
panel data or even replicated cross-sectional data
pare on their characteristics? Table A1 shows
might allow one to stipulate temporal order, cross-sec-
that establishments with low levels of racial
tional survey data can offer little in adjudicating such
issues. Indeed, cross-sectional survey research is diversity are more likely to be sole proprietor-
poorly equipped to offer a definitive answer on such ships (21 percent) than are those with medium
issues. Without direct, cross-time measures of the cen- (5 percent) or high levels (11 percent) of racial
tral variables, it is difficult to discern which causal diversity. The same general pattern holds true
explanations, if any, may be at work. for gender diversity and the tendency for estab-
 DOES DIVERSITY PAY?—–221

Table A1. Means and Percentage Distributions for Characteristics of Establishments by Levels of
Racial Diversity and Gender Diversity

Racial Diversity Level Gender Diversity Level

Low Medium High Low Medium High
Characteristics (<10%) (10–24%) (25%+) (<20%) (20–44%) (45%+) Overall
Percent proprietorship 21 5 11 19 8 10 15
Percent partnerships 9 4 8 3 6 11 7
Percent corporations (and others) 70 91 81 78 84 79 78
Mean percent female 37 41 48 8 32 64 46
Mean organization size 129.8 1,265.8 546.0 343.2 799.0 391.0 581.8
Mean organization age 29.7 35.9 33.1 33.7 34.6 30.2 32.8
Percent in agriculture 1 <1 3 3 1 1 17
Percent in mining <1 <1 1 2 1 <1 1
Percent in construction 7 6 4 17 2 1 6
Percent in transportation/comm. 5 6 8 11 8 2 7
Percent in wholesale 5 5 5 4 5 5 5
Percent in retail 19 20 25 11 16 28 21
Percent in F.I.R.E. 8 8 5 3 2 12 6
Percent in business services 10 9 7 12 12 5 9
Percent in personal services 6 3 5 2 2 8 4
Percent in entertainment 1 <1 1 2 1 <1 1
Percent in professional services 19 14 11 6 16 20 14
Percent in manufacturing 19 30 25 28 33 17 27
Percent Northeast 20 19 9 14 12 20 15
Percent Midwest 34 30 14 22 31 22 27
Percent South 22 23 37 28 28 28 28
Percent West 24 28 39 36 29 30 31

lishments to be proprietorships. Establishments female. Establishments with low levels of racial

with low levels of racial diversity are at least as diversity also tend to be smaller (130 employ-
likely to be partnerships (9 percent) as are those ees) than those with medium (1,266 employees)
with high (8 percent) or medium levels (4 per- or high (545 employees) levels of racial diver-
cent) of racial diversity. Businesses with low lev- sity. The same pattern is true for gender diver-
els of racial diversity are less likely to be sity. The establishments with low levels of racial
corporations (70 percent) than are those with diversity also tend to be slightly newer (30 years
medium (91 percent) or high levels (81 per- old) than those with medium (36 years old) or
cent) of diversity. Establishments with low lev- high (33 years old) levels of racial diversity.
els of gender diversity are slightly less likely to Businesses with the highest levels of gender
be corporations (78 percent) than are those with diversity tend to be the newest (30 years old),
medium (84 percent) or high levels (79 per- compared with those with low (34 years old) or
cent) of gender diversity. medium (35 years old) levels of gender diver-
Although the differences are not large, busi- sity.
nesses with low levels of racial diversity have Table A1 shows that industrial sectors do not
slightly lower percentages of female employees differ greatly by levels of racial or gender diver-
(37 percent) than do those with medium (41 per- sity. The two apparent exceptions are in the
cent) or high levels (48 percent) of racial diver- retail and manufacturing sectors: 19 percent of
sity. These patterns differ when broken down by organizations with low levels of racial diversi-
gender diversity: among businesses with low ty, 20 percent of those with medium levels, and
gender diversity, 8 percent of the employees 25 percent of those with high levels are in the
are female; among those with medium gender retail sector. The same pattern holds true for gen-
diversity, 32 percent of their employees are der diversity: 11 percent of organizations with
female; and among those with high gender low levels of gender diversity, 16 percent with
diversity, 64 percent of their employees are medium levels, and 28 percent with high levels
222—–AMERICAN SOCIOLOGICAL REVIEW

are in the retail sector. Establishments in the Black, Genie, Kevin Mason, and Gene Cole. 1996.
manufacturing sector are more likely to have “Consumer Preferences and Employment
medium levels of racial (30 percent) and gen- Discrimination.” International Advances in
der (33 percent) diversity than they are to have Economic Research 2:137–45.
Blalock, Herbert M. 1957. “Percent Nonwhite and
low racial (19 percent) or gender (28 percent)
Discrimination in the South.” American
diversity or high racial (25 percent) or gender
Sociological Review 22:677–282.
(17 percent) diversity. Bratter, Jenifer and Tukufu Zuberi. 2001. “The
The table also indicates that there are region- Demography of Difference: Shifting Trends of
al differences in regard to various levels of racial Racial Diversity and Interracial Marriage,
diversity: 20 percent of organizations with low 1960–1990.” Race and Society 4:133–48.
racial diversity are located in the Northeast, Bunderson, J. Stuart and Kathleen M. Sutcliffe. 2002.
compared with 19 percent of those with medi- “Comparing Alternative Conceptualizations of
um levels and 9 percent with high levels. For Functional Diversity in Management Teams:
gender diversity, 14 percent of organizations Process and Performance Effects.” Academy of
with low diversity, 12 percent of those with Management Journal 45:875–93.
Cohn, Samuel. 1985. The Process Of Occupational
medium diversity, and 20 percent of those with
Sex-Typing: The Feminization Of Clerical Labor
high diversity are in the Northeast. More than
in Great Britain. Philadelphia, PA: Temple
one third (34 percent) of businesses with low University Press.
levels of racial diversity are located in the Cox, Taylor. 1993. Cultural Diversity in
Midwest, compared with 30 percent of those Organizations: Theory, Research, and Practice.
with medium levels and 14 percent of those San Francisco, CA: Berrett-Koehler.
with high levels. Twenty-two percent of busi- ———. 2001. Creating the Multicultural
nesses with low levels of racial diversity are Organization: A Strategy for Capturing the Power
located in the South, compared with 23 per- of Diversity. San Francisco, CA: Jossey-Bass.
cent of those with medium levels and 37 percent Cox, Taylor and Ruby L. Beale. 1997. Developing
of those with high levels. Nearly one quarter (24 Competency to Manage Diversity. San Francisco,
percent) of businesses with low levels of racial CA: Berrett-Koehler.
Cox, Taylor H., Sharon A. Lobel, and Poppy Lauretta
diversity are located in the West, compared with
McLeod. 1991. “Effects of Ethnic Group Cultural
28 percent of those with medium levels and 39 Differences on Cooperative and Competitive
percent of those with high levels. The percent- Behavior on a Group Task.” Academy of
age of establishments located in the South (28 Management Journal 34:827–47.
percent) does not vary by level of gender diver- DiMaggio, Paul and Walter Powell. 2003. “The Iron
sity, and the percentage of establishments locat- Cage Revisited: Institutional Isomorphism and
ed in the West does not appear to be related Collective Rationality in Organizational Fields.”
systematically to level of gender diversity. Pp. 243–53 in The Sociology of Organizations:
Classic, Contemporary, and Critical Readings,
edited by M. J. Handel. London, UK: Sage
REFERENCES Publications.
Alon, Sigal and Marta Tienda. 2007. “Diversity, DiTomaso, Nancy, Corinne Post, and Rochelle Parks-
Opportunity, and the Shifting Meritocracy in Yancy. 2007. “Workforce Diversity and Inequality:
Higher Education.” American Sociological Review Power, Status, and Numbers.” Annual Review of
72:487–511. Sociology 33:473–501.
Baron, James N. and William T. Bielby. 1980. Edelman, Lauren B. 1990. “Legal Environments and
“Bringing the Firm Back In: Stratif ication, Organizational Governance: The Expansion of
Segmentation, and the Organization of Work.” Due Process in the American Workplace.”
American Sociological Review 45:737–65. American Journal of Sociology 95:1401–40.
Bell, Joyce M. and Douglas Hartmann. 2007. Embrick, David G. Forthcoming. “What is Diversity?
“Diversity in Everyday Discourse: The Cultural Re-examining Multiculturalism, Affirmative
Ambiguities and Consequences of ‘Happy Talk.’” Action, and the Diversity Ideology in Post-Civil
American Sociological Review 72(6):895–914. Rights America.” Sociology Compass.
Berry, Ellen C. 2007. “The Ideology of Diversity Florida, Richard and Gary Gates. 2001. Technology
and the Distribution of Organizational Resources: and Tolerance: The Importance of Diversity to
Evidence from Three U.S. Field Sites.” Paper pre- High-Technology Growth. Washington, DC: The
sented at the Annual Meeting of the American Brookings Institution.
Sociological Association, New York, New York. ———. 2002. “Technology and Tolerance:
 DOES DIVERSITY PAY?—–223

Diversity and High-Tech Growth.” Brookings Performance.” Journal of Experimental Social

Review 20:32–36. Psychology 5:429–40.
Gurin, Patricia, Biren (Ratnesh) A. Nagda, and Kochan, Thomas, Katerina Bezrukova, Robin Ely,
Gretchen E. Lopez. 2004. “The Benef its of Susan Jackson, Aparna Joshi, Karen Jehn, Jonathan
Diversity in Education for Democratic Leonard, David Levine, and David Thomas. 2003.
Citizenship.” Journal of Social Issues 60:17–34. “The Effects of Diversity on Business
Gwartney-Gibbs, Patricia A. and D. H. Lach. 1993. Performance: Report of the Diversity Research
“Sociological Explanations for Failure to Seek Network.” Human Resource Management 42:3–21.
Sexual Harassment Remedies.” Mediation Moss, Philip and Chris Tilly. 1996. “Soft Skills and
Quarterly 9:365–74. Race: An Investigation Of Black Men’s
Hambrick, Donald C. and Phyllis A. Mason. 1984. Employment Problems.” Work and Occupations
“Upper Echelons: The Organization as a Reflection 23:252–76.
of its Top Managers.” Academy of Management O’Reilly, Charles A., D. E. Caldwell, and W. P.
Review 9:193–206. Barnett. 1989. “Work Group Demography, Social
Hoffman, L. Richard and Norman R. F. Maier. 1961. Integration, and Turnover.” Administrative Science
“Quality and Acceptance of Problem Solutions by Quarterly 34:21–37.
Members of Homogeneous and Heterogeneous Page, Scott E. 2007. The Difference: How the Power
Groups.” Journal of Abnormal and Social of Diversity Creates Better Groups, Firms, Schools,
Psychology 62:401–07. and Societies. Princeton, NJ: Princeton University
Holzer, Harry J. 1996. What Employers Want. New Press.
York: Russell Sage Foundation Pelled, Lisa Hope. 1996. “Demographic Diversity,
Hubbard, Edward E. 1997. Measuring Diversity Conflict, and Work Group Outcomes: An
Results. Petaluma, CA: Global Insights. Intervening Process Theory.” Organization Science
———. 1999. How to Calculate Diversity Return 7:615–31.
on Investment. Petaluma, CA: Global Insights. Pelled, Lisa Hope, Kathleen M. Eisenhardt, and
Katherine R. Xin. 1999. “Exploring the Black
———. 2004. The Diversity Scorecard: Evaluating
Box: An Analysis of Work Group Diversity,
the Impact of Diversity on Organizational
Conflict and Performance.” Administrative Science
Performance. Burlington, MA: Elsevier
Quarterly 44:1–28.
Butterworth-Heinemann.
Pfeffer, Jeffery. 1977. “Toward an Examination of
Jackson, Susan E., Joan F. Brett, Valerie I. Sessa,
Stratification in Organizations.” Administrative
Dawn M. Cooper, Johan A. Julin, and Karen
Science Quarterly 22:553–67.
Peyronnin. 1991. “Some Differences Make a
Reskin, Barbara F. 1998. The Realities of Affirmative
Difference: Individual Dissimilarity and Group
Action in Employment. Washington, DC: American
Heterogeneity as Correlates of Recruitment, Sociological Association.
Promotions, and Turnover.” Journal of Applied Reskin, Barbara F., Debra B. McBrier, and Julie A.
Psychology 76:675–89. Kmec. 1999. “The Determinants and
Jehn, K. A., G. B. Northcraft, and M. A. Neale. 1999. Consequences of Workplace Sex and Race
“Why Differences Make a Difference: A Field Composition.” Annual Review of Sociology
Study of Diversity, Conflict and Performance in 25:335–61.
Workgroups.” Administrative Science Quarterly Richard, Orlando C. 2000. “Racial Diversity, Business
44:741–63. Strategy, and Firm Performance: A Resource-
Jones, Jo Ann and Rachel A. Rosenfeld. 1989. Based View.” The Academy of Management
“Women’s Occupations and Local Labor Markets: Journal 43:164–77.
1950 to 1980.” Social Forces 67:666–92. Rothman, Stanley, Seymour Martin Lipset, and Neil
Kalev, Alexandria, Frank Dobbin, and Erin Kelly. Nevitte. 2003a. “Does Enrollment Diversity
2006. “Best Practices or Best Guesses? Assessing Improve University Education?” International
the Efficacy of Corporate Affirmative Action and Journal of Public Opinion Research 15:8–26.
Diversity Policies.” American Sociological Review ———. 2003b. “Racial Diversity Reconsidered.”
71:589–617. The Public Interest 151:25–38.
Kalleberg, Arne L., David Knoke, and Peter Marsden. Ryan, John, James Hawdon, and Allison Branick.
2001. National Organizations Survey (NOS), 2002. “The Political Economy of Diversity:
1996–1997 [Computer file]. ICPSR version. Diversity Programs in Fortune 500 Companies.”
Minneapolis, MN: University of Minnesota Center Sociological Research Online (May).
for Survey Research [producer], 1997. Ann Arbor, Scott, W. Richard. 2003. Organizations: Rational,
MI: Inter-university Consortium for Political and Natural, and Open Systems. Upper Saddle River,
Social Research [distributor], 2001. NJ: Prentice Hall.
Kent, R. N. and J. E. McGrath. 1969. “Task and Sen, Sankar C. and B. Bhattacharya. 2001. “Does
Group Characteristics as Factors Influencing Group Doing Good Always Lead to Doing Better?
224—–AMERICAN SOCIOLOGICAL REVIEW

Consumer Reactions to Cor porate Social von Eron, Ann M. 1995. “Ways to Assess Diversity
Responsibility.” Journal of Marketing Research Success.” HR Magazine (August):51–60.
38:225–43. Wagner, Gary W., Jeffrey Pfeffer, and Charles A.
Skaggs, Sheryl L. and Nancy DiTomaso. 2004. O’Reilly. 1984. “Organizational Demography and
“Understanding the Effects of Workforce Diversity Turnover in Top-Management Groups.”
on Employment Outcomes: A Multidisciplinary Administrative Science Quarterly 29:74–92.
and Comprehensive Framework.” Pp. 279–306 in Watson, Warren E., Kannales Kumar, and Larry K.
Research in the Sociology of Work: Diversity in the Michaelsen. 1993. “Cultural Diversity’s Impact
Workforce, edited by N. DiTomaso and C. Post. on Interaction Process and Performance:
Oxford, UK: Elsevier. Comparing Homogeneous and Diverse Task
Skerry, Peter. 2002. “Beyond Sushiology: Does Groups.” Academy of Management Journal
Diversity Work?” Brookings Review 20:20–23. 36:590–602.
Smedley, Brian D., Adrienne Stith Butler, and Lonnie Welsh, Sandy, Myrna Dawson, and Annette
R. Bristow, eds. 2004. In the Nation’s Compelling Nierobisz. 2002. “Legal Factors, Extra-Legal
Interest: Ensuring Diversity in the Health-Care Factors, or Changes in the Law? Using Criminal
Workforce. Washington, DC: The National Justice Research to Understand the Resolution of
Academies Press. Sexual Harassment Complaints.” Social Problems
Smith, D. Randall, Nancy DiTomaso, George F. 49:605–23.
Whitaker, William A. 1996. White Male Applicant:
Farris, and Rene Cordero. 2001. “Favoritism, Bias
An Affirmative Action Expose. Smyrna, DE:
and Error in Performance Ratings of Scientists
Apropos Press.
and Engineers: The Effects of Power, Status, and
Williams, Katherine and Charles A. O’Reilly. 1998.
Numbers.” Sex Roles 45:337–58.
“Demography and Diversity: A Review of 40 Years
Stainback, Kevin, Corre L. Robinson, and Donald
of Research.” Pp. 77–140 in Research in
Tomaskovic-Devey. 2005. “Race and Workplace Organizational Behavior, edited by B. Staw and R.
Integration: A Politically Mediated Process?” Sutton. Greenwich, CT: JAI Press.
American Behavioral Scientist 48:1200–1228. Wright, Peter, Stephen P. Ferris, Janine S. Hiller, and
Stinchcombe, Arthur L. 1965. “Social Structure and Mark Kroll. 1995. “Competitiveness through
Organizations.” Pp. 142–93 in Handbook of Management of Diversity: Effects on Stock Price
Organizations, edited by J. G. March. Chicago, IL: Valuation.” Academy of Management Journal
Rand-McNally. 38:272–87.
Tolbert, Pamela S. and Alice Oberfield. 1991. Zenger, Todd R. and Barbara S. Lawrence. 1989.
“Sources of Organizational Demography: Faculty “Organizational Demography: The Differential
Sex Ratios in Colleges and Universities.” Sociology Effects of Age and Tenure Distributions on
of Education 64:305–15. Technical Communication.” Academy of
Tomaskovic-Devey, Donald. 1993. “The Gender and Management Journal 32:353–76.
Race Composition of Jobs and the Male/Female, Zuberi, Tukufu. 2001. “The Population Dynamics
White/Black Pay Gaps.” Social Forces 92:45–76. of the Changing Color Line in the USA.” Pp.
Tsui, Ann S., T. D. Egan, and Charles A. O’Reilly. 145–67 in Problem of the Century: Racial
1992. “Being Different: Relational Demography Stratification in the United States at Century’s
and Organizational Attachment.” Administrative End, edited by E. Anderson and D. Massey. New
Science Quarterly 37:549–79. York: Russell Sage Foundation.