Sie sind auf Seite 1von 19

Academy of Management Journal

2011, Vol. 54, No. 6, 11411158.


http://dx.doi.org/10.5465/amj.2009.0745

EXPLAINING BIAS AGAINST BLACK LEADERS:


INTEGRATING THEORY ON INFORMATION PROCESSING
AND GOAL-BASED STEREOTYPING
ANDREW M. CARTON
The Pennsylvania State University
ASHLEIGH SHELBY ROSETTE
Duke University
Approaches related to inference-based processing (e.g., romance-of-leadership theory)
would suggest that black leaders are evaluated positively after success. In contrast,
approaches related to recognition-based processing (e.g., leader categorization theory)
would suggest that, because of stereotyping, black leaders are evaluated negatively
regardless of their performance. To reconcile this discrepancy, we predicted that
evaluators would engage in goal-based stereotyping by perceiving that black leaders
and not white leadersfail because of negative leader-based attributes and succeed
because of positive nonleader attributes (i.e., compensatory stereotypes). Multilevel
analyses of archival data in the context of college football in the United States supported our predictions.

organizational leadership positions (Alliance


Board for Diversity, 2005; Corporate Board Initiative, 2006; Thomas & Gabarro, 1999). For example, in 2009 only 1 percent of the CEOs of Fortune
500 companies were black (according to blackentrepreneurprofile.com)a rate that is far below
the proportion of blacks in the population of the
United States (12.8 percent, according to census.
gov) and the world (approximately 1520 percent, per nationalpolicyinstitute.org).
In this research, we provide a fresh perspective
on why negative evaluations of black leaders are so
persistent by reinterpreting long-standing theory
related to how leaders are evaluated. Lord and Maher (1993) noted that two dominant streams of research have been especially active in the effort to
understand how perceivers evaluate leaders. The
first stream of research, exemplified by romance-ofleadership theory (Meindl & Ehrlich, 1987), builds
on the notion of inference-based processing. According to this approach, it is presumed that evaluators infer internal attributes of leaders from performance outcomes, regardless of the race of leader.
Specifically, leaders are evaluated positively in the
context of success and negatively in the context of
failure (Hall & Lord, 1995; Meindl & Ehrlich, 1987).
The second stream of research, exemplified by
leadership categorization theory (Lord, Foti, & De
Vader, 1984), builds on the notion of recognitionbased processing. According to this approach, perceivers use preexisting mental categorizations of
prototypical and nonprototypical leaders to deter-

A wealth of evidence suggests that black organizational members are at a disadvantage when they
are evaluated in terms of their leadership ability
(Beatty, 1973; Ford, Kraiger, & Schechtman, 1986;
Greenhaus & Parasuraman, 1993; Greenhaus, Parasuraman, & Wormley, 1990; Knight, Hebl, Foster,
& Mannix, 2003; Orpen, 1981; Powell & Butterfield,
1997; Rosette, Leonardelli, & Phillips, 2008; Yarkin, Town, & Wallston, 1982). The consequences of
negative leader evaluations include poorer prospects for career advancement (Greenhaus & Parasuraman, 1993), a lower likelihood of promotion to
specific positions (Powell & Butterfield, 1997), a
general exacerbation of the barriers to upward
mobility (Morrison & Von Glinow, 1990), and
ultimately the underrepresentation of blacks in

We thank Jason Colquitt and three anonymous reviewers for their guidance and generative feedback throughout the review process. We also thank Bruce Carton,
Gavin Kilduff, Rick Larrick, Leigh Tost, and Thomas
Wicker for their comments on drafts. We thank John
Clithero, Jonathon Cummings, Jeff Edwards, Jim Emery,
Dave Hofmann, Patrick McKay, Sim Sitkin, Jack Soll,
Elena Vidal, and especially Al Mannes for helpful feedback on our methodological approach. We are grateful for
constructive feedback from attendees of seminars at
Georgia Institute of Technology, London Business
School, Penn State University, University of Houston,
and University of Miami.
Editors note: The manuscript for this article was accepted for publication during the term of AMJs previous
editor-in-chief, Duane Ireland.
1141

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holders express
written permission. Users may print, download, or email articles for individual use only.

1142

Academy of Management Journal

mine whether a target has strong leadership ability


and subsequently evaluate prototypical leaders favorably and nonprototypical leaders unfavorably
(Lord et al., 1984). A recognition-based approach
would suggest that perceivers are not inclined to
make positive attributions about black leaders abilities because blacks are usually stereotyped negatively in terms of attributes, such as competence,
that are associated with leaders (Devine, 1989).
Given that (1) evaluations of leaders drawn from
the inference process are based on performance and
(2) evaluations of black leaders drawn from the
recognition process are based on negative stereotypes, a contradiction emerges when one compares
the implications of the two processes. An approach
based on the inference process would suggest that
black leaders are evaluated positively in the context of success, whereas an approach based on the
recognition process would suggest that black leaders are evaluated negatively regardless of performance. Strong evidence in the extant literature indicating that black leaders are evaluated negatively
regardless of performance (Beatty, 1973; Greenhaus
& Parasuraman, 1993; Knight et al., 2003; Powell &
Butterfield, 1997; Yarkin et al., 1982) appears to
favor a recognition-based approach as opposed to
an inference-based approach. Yet organizational researchers have remained silent with respect to how
perceivers negotiate the discrepancy between inference and recognition. This is surprising given that
evidence suggests that perceivers strongly emphasize both processes when evaluating leaders (Lord
& Maher, 1993). Most important, this discrepancy
highlights a gap in existing research that drives the
need for the present research: the mechanisms underlying the persistence of racial disparities in
leadership evaluation remain unclear.
We suggest that an effort to illuminate how perceivers negotiate the discrepancy between the inference and recognition processes will shed light
on both how and why perceivers are persistent in
evaluating black leaders negatively even after
black leaders are observed in the context of performance success. We draw from theory on goal-based
stereotyping (Kunda & Spencer, 2003) to predict
that evaluators adjust the way they use stereotypes
of black leaders according to performance outcomes. Specifically, negative leader-based stereotypes will be applied after performance failure, and
nonleader compensatory stereotypes (i.e., black
leaders succeed because of marginal qualities that
compensate for negative qualities) will be applied after performance success. If blacks are evaluated in terms of leadership ability only in the
context of failure, then negative stereotypes regarding their leadership ability are able to more easily

December

persist. This perspective indicates that goal-based


stereotyping may provide an important missing
link in understanding bias against black leaders
and may serve as an important contributor to barriers that impede the advancement of black leaders
in organizations. In this article, we first more
deeply review how the processes of inference and
recognition serve as the bases of leadership evaluation and then elaborate on how goal-based stereotyping can reconcile instances in which the two
processes are in conflict. We then report tests of our
predictions with archival data.
LEADERSHIP AND INFORMATION
PROCESSING
Perceptions of leaders ultimately drive leader
evaluations. Information processing theories of
leadership relate to the formation of such perceptions. Perceivers filter and organize information related to leaders as well as to the environments in
which leaders are embedded (Lord & Maher, 1993).
The way that perceivers cognitively organize this
information affects the attributions they make
about leaders, which, in turn, determine whether a
leader will be perceived as credible. In this section,
we expand on the two predominant ways that perceivers organize and process information in order
to evaluate organizational leaders: inference and
recognition.
Leadership and Inference
Attribution theory (Kelley & Michela, 1980) suggests that people make sense of their environment
by making inferences (i.e., attributions) about the
attributes of other people, or simple deductions
related to cause and effect: positive outcomes are
caused by positive attributes, and negative outcomes are caused by negative attributes. Unlike
stereotypes, which can be applied prior to observing behavior associated with a target, inferences
occur after such behavior has been observed. Attributions can be internal (an outcome is ascribed to a
personal attribute) or external (an outcome is ascribed to a situational factor).
The inference process is altered in two key ways
in the context of leadership (Agle, Mitchell, & Sonnenfeld, 1999; Calder, 1977; Emrich, 1999; Meindl,
1995; Meindl & Ehrlich, 1987; Pfeffer, 1977; Rush,
Phillips, & Lord, 1981; Shamir, 1991; Weber, Camerer, Rottenstreich, & Knez, 2001). First, perceivers
are motivated to make internal attributions as opposed to external attributions. That is, a fundamental element of the inference process is that responsibility for outcomes rests with leaders themselves

2011

Carton and Rosette

and not with situational factors. Second, perceivers


are more likely to base attributions on performance
rather than other outcomes. In short, perceivers are
likely to increase positive internal attributions of
leaders after performance success and to increase
negative internal attributions of leaders after performance failure.
In explaining the results of two studies that
tested romance-of-leadership theory, Meindl and
Ehrlich (1987: 106) stated that people may see
leadership as the origin or initial force in a causal
chain, ultimately affecting performance by virtue of
its control of other causal forces more contiguous to
outcomes (1987: 106). As another example, Rush
et al. (1981) showed two groups of participants the
same video footage of a leader interacting with
team members. The researchers then gave false performance feedback to participants: they informed
one set of participants that the leaders team was
performing strongly and the other set of participants that the leaders team was performing poorly.
Perceivers who were informed that the leaders
team was performing strongly inferred the existence of stronger leadership attributes.
It follows from the evidence reviewed here that
perceivers will increase positive internal attributions about both black and white leaders after success and increase negative internal attributions
about both black and white leaders after failure.
Hypothesis 1a. More positive internal attributions about leaders are made after performance success than after performance failure.
Hypothesis 1b. More negative internal attributions about leaders are made after performance failure than after performance success.
Leadership and Recognition
Recognition is a process of evaluating leaders
that has been presumed to run in parallel to inference, yet its mechanisms are highly distinct from
inference (Lord & Maher, 1993). Lord and colleagues (1984) suggested that perceivers have prototypes of leaders that encompass conventional
leader characteristics. These prototypes allow perceivers to recognize whether targets are effective
leaders. Perceivers recognize a target as an effective
leader when that target possesses attributes that are
consistent with the leader prototype, such as intelligence and decisiveness.
Stereotypes are likely to influence the prototypes
used in the recognition process. Stereotypes are
generalized beliefs about certain social groups
(Hilton & von Hippel, 1996). A key consideration is
that the effect of stereotypes on recognition differs

1143

greatly for white and black leaders. For evaluators


of white leaders, typically no strong stereotypes
guide the recognition process because whites are
considered the normative group, or the standard,
against which others are compared (DiTomaso,
Post, Smith, Farris, & Cordero, 2007). In this way, a
white target can be viewed as somewhat of a blank
slate, and racial stereotypes are unlikely to strongly
bias the way that perceivers evaluate a white
leader.
In contrast, evaluators typically possess very
strong stereotypes of blacks that are likely to influence the recognition process (Hilton & von Hippel,
1996). Common stereotypes of blacks in the United
States include perceptions of being poor, lazy, and
criminal (Devine, 1989). In the context of leadership, the most relevant stereotype is the presumption that blacks are incompetent (Devine & Baker,
1991; Devine & Elliot, 1995; Dixon & Rosenbaum,
2004; Fiske, Cuddy, Glick, & Xu, 2002; Krueger,
1996). The importance of this stereotype becomes
clear when considering that two separate sources
have suggested that competence is the key determinant of perceptions of leadership. First, Lord et al.
(1984) suggested that attributes related to competence (e.g., intelligence and decision making ability) are stronger predictors of the perception of
leadership than other factors, including understanding and dominance. Second, Rudman and
Glick (2001) suggested that competence is the most
important driver of perceptions of leadership in
their two-factor model of agency (in which the
other form of agency is dominance). Specifically,
they described competent leaders as those who are
independent, confident . . . analytical and who
work well under pressure (Rudman & Glick,
2001: 751).
It follows from the evidence reviewed here that
blackand not whiteleaders will be recognized as being incompetent. Accordingly, we
make the following prediction:
Hypothesis 2. Black leaders are perceived as
more incompetent than white leaders.

THE TENSION BETWEEN INFERENCE AND


RECOGNITION
Although research on both inference and recognition has been highly influential in advancing understanding of how perceivers evaluate leaders, little research has examined the relationship between
the two processes. We suggest that the two processes may sometimes conflict when black leaders
are evaluated and that an understanding of how

1144

Academy of Management Journal

perceivers resolve this conflict will shed light on


how bias against black leaders is sustained.
When black leaders fail, the inference process is
likely to reinforce the inclination to view them as
incompetent: perceivers assume that leaders who
fail exhibit weak leadership qualities. Thus, there
is a match in such situations between the negative valence of attributions that are deduced from
the outcome during the inference process and the
negative valence of the attribute that is typically
drawn from black stereotypes during the recognition process (i.e., incompetence). In contrast, there
is a built-in tension between inference and recognition when blacks in leadership positions succeed. An inference-based approach would suggest
that perceivers are motivated to make positive internal attributions in the context of successful performance, whereas a recognition-based approach
would suggest that perceivers are motivated to perceive negative attributes because blacks are assumed to be incompetent (Devine & Elliot, 1995).
There is thus a mismatch in such situations between the positive valence of attributions that perceivers are driven to deduce from the performance
outcome during the inference process and the negative valence of the attribute that is likely to be
drawn from black stereotypes (e.g., incompetence)
during the recognition process. These mismatches
between the recognition and inference processes
are less likely to occur for white leaders because
strong racial stereotypes for whites are absent.
In trying to attain consistent evaluations of targets, perceivers are motivated to have the inference
and recognition processes align (Lord & Maher,
1993). When inference and recognition yield mismatches, perceivers are motivated to reconcile such
discrepancies. Given the centrality of both processes for guiding perceptions of leadership, an
optimal situation exists if perceivers can adhere to
both processes instead of other, less desirable, alternatives, such as abandoning one process in favor
of the other, abandoning stereotypes, or changing
stereotypes. To propose how perceivers can accomplish this, we incorporate theory on goal-based
stereotyping into the dual-process model of inference and recognition.
GOAL-BASED STEREOTYPING: HOW
PERCEIVERS RECONCILE MISMATCHES
BETWEEN INFERENCE AND RECOGNITION
Existing studies that compare black and white
leaders (DiTomaso & Hooijberg, 1996; Knight et al.,
2003; Powell & Butterfield, 1997) rest on the traditional assumption that stereotyping is an unwavering and rigid categorization process based on firmly

December

entrenched beliefs about a target (Bargh, 1999;


Devine, 1989). However, a growing body of evidence that has remained largely untapped by organizational scholarsincluding those who study
leadershipsuggests that predictable occasions exist when the goals of perceivers promote or inhibit
the application of stereotypes. Researchers have
labeled this phenomenon goal-based stereotyping
(Blair, 2002; Kunda, Davies, Hoshino-Browne, &
Jordan, 2003; Kunda & Spencer, 2003; Macrae &
Bodenhausen, 1995; Sinclair & Kunda, 1999). One
set of important goals are those related to comprehension. Kunda and Spencer elaborated on the
meaning of comprehension goals and on how perceivers use them:
Comprehension goals encompass the need to . . .
form coherent impressions. . . . Stereotypes serve
these needs. . . . Increases in the strength of comprehension goals can prompt stereotype activation and
application when stereotyping is assumed to aid
comprehension. . . . Comprehension goals may also
prompt stereotype suppression, if stereotyping is
assumed to disrupt comprehension. People may inhibit the activation and application of stereotypes if
they consider them uninformative. (2003: 524)

We suggest that, to attain more coherent impressions of black leaders (Kunda & Spencer, 2003:
524), perceivers adopt comprehension goals when
they observe black leaders in the context of successa situation that engenders a mismatch between inference and recognition. We further suggest that perceivers achieve these goals via goalbased stereotyping: the application of different
stereotypes depending upon whether leaders are
observed in the context of success or failure.1 We
also contend that perceivers do not typically adopt
comprehension goals for white leaders because the
absence of strong stereotypes for whites suggests
that there are no clear mismatches between inference and recognition.

1
The term application does not imply that the perceiver is conscious of the process. Indeed, stereotypes
can be applied in response to comprehension goals either with or without intention and awareness (Kunda &
Spencer, 2003: 523). This idea is in line with the notion
that there are both automatic and controlled channels in
recognition and inference (Lord & Maher, 1993). The key
issue highlighted here is that controlled and automatic
channels often yield the same outcomes. Therefore, we
remain agnostic with respect to whether the application
of stereotypes occurs with or without conscious intention or awareness.

2011

Carton and Rosette

Application and Inhibition of the Stereotype of


Incompetence
In the context of successful performance, we suggest that perceivers have a goal of eliminating or
reducing the inherent mismatch between the negative attribute of incompetence that is identified
during the recognition process and the positive attributions that perceivers are motivated to make
during the inference process. Specifically, we propose that the stereotype of incompetence will be
applied to black leaders in instances of performance failure and inhibited in instances of performance success. Although we predict (in Hypothesis 1) that perceivers are generally motivated to
make positive attributions after success and negative attributions after failure for both blacks and
whites, the irrelevance of the stereotype of incompetence to whites suggests that perceivers will not
promote or inhibit its application when evaluating
them.2 Given our differing expectations for black
and white leaders, we make the following
prediction:
Hypothesis 3. Race and performance interact
in predicting the application of the stereotype
of incompetence. Black leaders are perceived
to be more incompetent than white leaders in
the context of failure; however, evaluation of
black and white leaders according to the stereotype of incompetence does not differ in the
context of success.
Although inhibiting the application of the incompetence stereotype is likely to partially fulfill
the comprehension goals that perceivers adopt
when evaluating black leaders in the context of
success, this process of inhibition is unlikely to
fully satisfy comprehension goals. Given that it is
perceived that blacks possess a characteristic that is
substantially more negative than those of whites in
the context of failure (incompetence), evaluators
2

Kunda and Spencer (2003) distinguished between


the activation and application of stereotypes. Activation
occurs when stereotypes are accessible in memory. Application occurs when stereotypes affect the judgment of
a target. Application requires activation, whereas activation does not require application. For example, the stereotype of blacks as being aggressive can be made
salient in the environment and subsequently activated in
memory, yet not applied to (i.e., used to judge) a specific
black target. However, if that stereotype were applied to
a black target, then it would have to be activated in
memory first. In this research, we are interested in the
application of stereotypes because the evaluation of leaders will only be affected if stereotypes are applied. Activation is not sufficient in and of itself.

1145

are also likely to be motivated to perceive that black


leaders possess a characteristic that is substantially
more positive than those of white leaders in the
context of success. These perceptions allow the
attainment of more consistent and coherent impressions. That is, for the sake of consistency, evaluators are motivated to believe that blacks have an
extremely good characteristic that balances out, or
compensates for, an extremely bad characteristic
(incompetence). Perceivers will endow successful
black leaders with attributes that match the target
of evaluation (i.e., blacks) and can be interpreted to
predict a positive performance outcome. We suggest that such attributes are compensatory stereotypes.
Application and Inhibition of Compensatory
Stereotypes
A number of researchers have suggested that perceivers can use compensatory stereotypes to offset
negative stereotypes when evaluating a variety of
targets in different contexts (Biernat, Sesko, & Amo,
2009; Kay, Gaucher, Napier, Callan, & Laurin, 2008;
Stone, Perry, & Darley, 1997; Yzerbyt, Provost, &
Corneille, 2005). We introduce this concept to the
study of race and leadership. Whereas many stereotypes of blacks (e.g., lazy, ignorant, and hostile) typically lead to negative consequences
(Devine & Elliot, 1995), others (e.g., controlling,
relational, humorous, calculating) may be
perceived as compensatory stereotypes that are related to positive outcomes because they can literally be interpreted as compensating for incompetence in the context of leadership. Instead of being
congruent with the sense of competence perceived
to be exhibited by effective leaders (Lord et al.,
1984), these attributes are typically viewed as relatively primitive adaptations (Ogbu, 1985) and are
therefore not considered necessary attributes of effective leaders.
Given the wide variety of organizational contexts
(Pfeffer, 1997), the nature of compensatory stereotypes is likely to vary significantly across different
organizations and industries. In organizations that
are very hierarchical and use a command-and-control model of organizing, black leaders may be perceived as being successful because they are controlling (Ogbu, 1985), but not because they are
competent leaders. In organizations with a flat
structure in which leaders have less formal authority, perceivers may ascribe the success of black
leaders to teddy bear attributes indicative of
warmth and friendliness (Livingston & Pearce,
2009). Given that research on leadership evaluation
suggests that evaluators strongly emphasize per-

1146

Academy of Management Journal

sonal competence over all other traits when evaluating leadership ability (Lord et al., 1984), attributes such as warmth and friendliness can be
viewed as compensatory. As support for this claim,
existing research outside of the context of leadership has suggested that blacks are perceived to be
warm as a way to compensate for lack of intellectual ability and discipline, key attributes of competence (Biernat et al., 2009).
In addition to depending on differing organizational structures, the range of possible compensatory stereotypes is likely to depend upon the idiosyncratic characteristics of organizations. In
organizations that have a history of questionable
ethical practices, evaluators may perceive that
black leaders have attained success because they
used manipulation or corruption to compensate for
incompetence (Ogbu, 1985). In organizations with a
socially oriented mission (e.g., political organizations, churches, fund-raising organizations, and advocacy organizations), evaluators may believe that
black leaders have attained success because they
accrue resources and attention by making noise
and drawing attention to themselves as opposed to
wielding influence via a refined ability to mobilize
and coordinate a set of followers (Dixon, 2008). In
entrepreneurial start-ups that lack resources and
are at a high risk of failing, evaluators may focus on
stereotypes related to the survival instincts of
black leaders (Ogbu, 1985). In organizations specializing in the arts (record companies and dance
studios), evaluators may perceive that black leaders
are successful because they have a primitive natural ability, such as natural rhythm (Kawakami &
Dovidio, 2001), instead of the ability to form a
compelling vision and outsmart competitors. In organizations in which jobs require physical strength,
such as fire and police departments, evaluators
may assume that better physical prowess allows
black leaders to succeed (Miller, 2004). Similarly,
in athletic contexts, athleticism is likely to be a
key compensatory stereotype for athletes in leadership positions (Stone, Lynch, Sjomeling, & Darley, 1999).
In short, perceivers view compensatory stereotypes as attributes that blacks possess to compensate for incompetence. Our key insight is thus twofold: compensatory stereotypes (1) can be used in
the inference process to explain performance success and (2) will not be perceived as contradicting
the stereotype of incompetence and the general belief that blacks are unlikely to be effective leaders.
Although we suggest that compensatory stereotypes will be applied to explain performance success (a positive stereotype matches a positive
outcome), their application could possibly surface

December

another conundrum: an attribute that causes positive outcomes cannot be sensibly linked to a negative outcome (failure) via the inference process. To
address this mismatch, we suggest that compensatory stereotypes will be inhibited when black leaders are evaluated in the context of performance
failure. Like the stereotype of incompetence, compensatory stereotypes are not typically used to label white leaders because white leaders are not
presumed to have to compensate for incompetence.
We describe shortly how the present research
focuses on athletes in leadership positions in college football organizations. In accordance with the
previous literature review, an important compensatory stereotype for black athletes in leadership
positions in the sports context is athleticism. Given
our differing expectations for black versus white
leaders and the specific context of sports in the
present research, we make the following
prediction:
Hypothesis 4. Race and performance interact
in predicting the application of the compensatory stereotype of athleticism. Black leaders
are perceived to be more athletic than white
leaders in the context of success; however,
there is no difference between how black and
white leaders are evaluated in the context of
failure.
METHODS
Research Context: College Football in the United
States
In the present research, we focus on how black
and white leaders in sports organizations are evaluated. Sports organizations comprise one of the
largest business industries, generating well over
US$100 billion annually (Slack, 1997), and they are
useful contexts for research because they are populated by highly motivated actors operating in real,
but semicontrolled, environments (Keidel, 1987;
see also Pfeffer & Davis-Blake, 1986; Staw & Hoang,
1995). Sports contests can be viewed as extreme
situations in which certain factors are unambiguous and allow clear conclusions to be made (Eisenhardt, 1989: 537). For example, performance is typically unambiguous in terms of success and failure
(winning vs. losing), and there exists a prominent
compensatory stereotype associated with blacks:
athleticism. Additionally, sports leaders are some
of the most influential organizational leaders. For
example, they typically demand higher rates as
speakers than leaders from any other industry, according to allamericanspeakers.com (accessed in
2010). In this article, we focus on quarterbacks in

2011

Carton and Rosette

the sport of American football, as the quarterback is


one of the most important leadership positions in
sports organizations.3
Methodological approach: Archival study. In
this research we employ an archival study.
Whereas goal-based stereotyping has been tested
almost exclusively in the laboratory (as exceptions,
see Dasgupta and Asgari [2004] and Rudman, Ashmore, and Gary [2001]), an archival study permits
unobtrusive measures that do not rely on artificial
manipulations and an externally valid way to assess how perceivers use stereotypes. Specifically,
we collected information related to how quarterbacks were perceived in published news media
documents because these documents house a wide
array of evaluations of quarterbacks from different
sources. This approach mandated that we first
place perceptions of evaluators into categories. We
therefore conducted a content analysis.
Terms of data selection. The first step of a content analysis involves dictating the terms of data
selection (Smith, 2000). We investigated the electronic versions of articles on local news websites,
sports-specific websites, and general news websites
published during four periods of the 2007 Football
Bowl Subdivision (formerly known as Division
1-A) college football season, the highest level of
competition in college football in the United States.
These four periods were in early September, midSeptember, mid-October, and late October.4 Both

The current research rests on the assumption that


quarterbacks not only play important roles as leaders in
their respective sports organizations but also that observers perceive their position as important leaders. Writing on sports organizations supports this assumption,
suggesting that quarterbacks are highly revered for their
leadership ability and possess symbolic value as leaders
(Bass, 2009). Research on leadership perception also supports this assumption, suggesting that, like political and
military leaders, sports leaders symbolize leadership
(Lord & Maher, 1993). Consider popular metaphors for
quarterbacks, such as leader of the gridiron and general on the field, conjuring the hierarchical schemes of
military coordination and placing quarterbacks in important leadership roles. Further, search engine results
(google.com, accessed in 2010) indicated that the word
leader is paired more with quarterback than with any
other sports position.
4
We collected data for several periods to obtain a more
representative sample over timein this case, over the
course of a season. Given the possibility that evaluations
of leaders may change during a season, a longitudinal
design was likely to be more accurate and representative
than a cross-sectional design. The specific number of
four periods was determined because it accounted for the
nuances of college football. Specifically, the first period

1147

black and white quarterbacks were well represented in the Football Bowl Subdivision: in the
2007 college football season there were 328 white
quarterbacks and 124 black quarterbacks.
We sampled newspaper reports for each of the
119 teams in the Football Championship Subdivision in 2007. For each university, we sampled articles from one local paper, using the following
criteria to select the one local paper per team: (1)
the paper belonged to a Top 100 media market,
according to data from the Digital Syndicate Network; (2) it was the most geographically proximal
newspaper relative to the universitys central campus; (3) it was within the same state; (4) it had at
least one article about the universitys football team
in each data collection period; and (5) it was not
one of the Top 15 national newspapers (i.e., a paper
that had one of the top 15 volumes of circulation in
the country). For a university in a state that did not
house a Top 100 media market, we chose the most
widely circulated newspaper in the state. To consider a complete range of media sources, we also
collected data from 25 other sources: the 15 newspapers with the highest national circulations, according to 2007 data from the Audit Bureau of
Circulations, 2007; the top five sports-specific websites; and the top five general news websites. We
selected the sports-specific and general news sites
using the 2007 Alexa Web Search Top 500 website
statistics.
Data were sampled from one article per week for
each newspaper. If there were several articles, then
one article was randomly selected. We collected
data on the quarterback on each team who received
the most media scrutiny during the 2007 season;
this was usually the starting quarterback or the
quarterback who received the most playing time.
This process resulted our identifying 119 quarterbacks (1 from each team) for inclusion in our sample. The race of the selected quarterback was
coded. Ultimately, we found no evaluative phrases
for 6 quarterbacks, so our final sample contained
113, of whom 82 were white and 31 were black.
Two research assistants collected the articles, typically from the Sunday through the Wednesday
subsequent to the previous weekends football game.

(very early in the season) consisted of games outside of


each teams conference. This period may have been qualitatively distinct from the conference schedule because
competition is typically weaker. We then collected data
for three periods within the conference schedule (one in
the beginning, one in the middle, and one at the end),
which accounted for the majority of the season.

1148

Academy of Management Journal

Coding procedures. The next stage of the content


analysis involved categorizing the comments made
about the quarterbacks. We trained three blind coders (none of whom were the two assistants who
collected the articles). The coders were undergraduate students who were recruited through an advertisement and received monetary compensation
for their work. The advertisement noted that potential coders were not required to have previous coding experience but should be familiar with the rules
of college football as well as football terminology
that members of the media are likely to use. Although the coders were trained to ensure that they
were coding for the correct quarterback (and not
other players who might be mentioned in an article), they were not aware of other information, such
as article titles and names of the media reporters.
Upon completion of the coding, one of the authors
double-checked the data to ensure that the coded
evaluative phrases corresponded to the correct
quarterbacks in the coding spreadsheet.
According to Smiths (2000) discussion of content analyses in archival research, a text unit is the
actual unit of text that is coded, whereas the context unit is the unit that a researcher references to
derive the meaning of the text unit and to subsequently place it into the appropriate category. As
the text unit, we selected adjectives and adjective
phrases, defined as any words or phrases that modified a noun or pronoun (McFadyen, 2007). We
were also interested in adverbs and adverb phrases,
which modified verbs. Hereafter we label all four of
these as evaluative phrases. Evaluative phrases
were the terms most likely to capture descriptions
of identifiable attributes of quarterbacks (e.g., he is
intelligent or fleet-footed). Our noun of interest
was typically quarterback, although it was also
sometimes a word or phrase referencing his performance outcomes. The verb of interest was the word
or phrase identifying the quarterbacks actions. Excluded from the analyses were narratives and descriptions of events, such as he completed a pass.
That is, we only coded evaluative words and
phrasesthose terms that indicated some type of
evaluation of a quarterbacks ability.
Coding categories. For the dependent variable
incompetence, we instructed the coders to code
any phrases that indicated that quarterbacks lacked
the attributes of competence outlined by Rudman
and Glick (2001), such as determination and analytical ability (one attribute, computer skill,
was not relevant to this context), as well as two
attributes related to competence noted by Lord and
colleagues (1984): intelligence and decisiveness.
The dependent variable athleticism related to the
most prominent compensatory attribute for black

December

leaders in sports (Stone et al., 1999). This coding


category included attributes that reflected physical
ability, such as strength, speed, and agility (Mercurio & Filak, 2010; Stone et al., 1999). The presumption that black athletes are highly athletic yet lack
the attributes of competence described by Rudman
and Glick (2001) and Lord and colleagues (1984)
has been supported by research from a number of
social science disciplines, including communication studies (Buffington & Fraley, 2008; Rada &
Wulfemeyer, 2005), social psychology (Devine &
Baker, 1991; Stone et al., 1997), and sports management (Murrell & Curtis, 1994; Rasmussen, Esgate, &
Turner, 2005).
Although not directly related to our hypotheses,
we also coded for competence (quarterbacks described as possessing, as opposed to lacking, the
attributes listed above for incompetence) and lack
of athleticism (quarterbacks who lacked, instead of
possessed, the attributes listed above for athleticism). We expected that these two attributes
would not be applied or inhibited in different frequencies based on black leaders performance because they are not stereotypes of blacks. Additionally, they would not be applied for whites because
whites are perceived as having a balance between
competence and athleticism, and thus as neither
extremely weak nor extremely strong on either attribute. For our final two coding categories, we
coded positive evaluative phrases that were unrelated to athleticism and competence and negative
phrases that were unrelated to incompetence and
lack of athleticism as positive phrases and negative
phrases, respectively.5
Ambiguous evaluative phrases. When the category to which a codable evaluative phrase belonged
was not obvious on its own merit, the coders used
the context unit to determine the meaning of the
evaluative phrase (Smith, 2000). The context unit
involved three total sentences for each evaluative
phrase: the sentences (1) before, (2) after, and (3)
containing the evaluative phrase. For example, the
meaning of a phrase describing how a quarterbacks
performance was littered with errors was clarified by examples of how the quarterback made poor
decisions, suggesting that the phrase related to incompetence.
Interrater reliability. Because the coding involved a two-step sequence of (1) independently
identifying units of an article that were codable as
evaluative phrases and then (2) independently cat5
Analyses conducted with positive evaluative phrases
including competence and athleticism and with negative
evaluative phrases including incompetence and lack of
athleticism revealed the same outcomes.

2011

Carton and Rosette

egorizing the codable text units, two separate measures of interrater reliability were needed. We used
Guetzkows U (Folger, Hewes, & Poole, 1984) to
measure how reliably two coders broke a given
body of data into the same number of units. Values
of Guetzkows U below 0.1 are considered to indicate very high agreement. Coders 1 and 2 (Guetzkow U 0.03), coders 1 and 3 (Guetzkow U
0.03), and coders 2 and 3 (Guetzkow U 0.07) all
attained measures indicating very high agreement.
Second, once a text was broken into codable units,
the coders could classify each unit into one of the
four categories. At this stage, we calculated interrater reliability using Cohens kappa, a reliability
statistic that controls for chance agreements in calculations of agreement rates between two independent coders (Howell, 1992). Values above 0.75 are
considered to indicate very high agreement (Fleiss,
1981). Coders 1 and 2 ( 0.90), coders 1 and 3
( 1.00), and coders 2 and 3 ( .83) attained
very high levels of agreement on this statistic. Once
these levels of interrater reliability were demonstrated, the three coders coded the content independently.
Measures
Dependent variables: Evaluative phrases. The
primary dependent variables were the number of
evaluative phrases recorded in the week after the
previous weekends game for each category: incompetence, athleticism, competence, lack of athleticism, positive attributes, and negative attributes.
For example, if seven unique evaluative phrases
related to incompetence were published in reference to a particular quarterback in a given week,
then the value for the category of incompetence
was 7.
Independent variables: Race and performance.
There were two independent variables. The first
was the race of a quarterback: white quarterbacks
were coded 0 and black quarterbacks were coded 1.
The second was performance. College football
games are usually played on Saturdays. Typically,
the majority of observations of quarterbacks published during the first few days of a given week
(when the data were collected) relate to the outcome of the previous Saturdays game. Therefore,
we coded performance according to whether a
quarterbacks team won or lost during the weekend
before the data on evaluative phrases were collected. Winning served as a proxy for performance
success, and losing served as a proxy for performance failure.
Control variables. First, we controlled for the
number of words written per week per quarterback

1149

because the number of evaluative phrases could be


inflated if a large volume was written about a quarterback, or deflated if little was written about him.
Second, we included proxies to control for the possibility that accepted indicators of competence/incompetence were driving differences in (1) performance outcomes and/or (2) evaluative phrases.
Proxies for competence/incompetence are especially hard to attain; nonetheless we wanted to
control for accepted indicators of competence/incompetence to rule out alternative explanations of
our findings. We first searched for college or high
school grade point average (GPA) for the quarterbacks. College GPA was available in very few cases,
and high school GPA was available for only 35.4
percent (n 40) of the quarterbacks, according to
rivals.com for 2010. We therefore searched for a
more widely available proxy. After an exhaustive
search, we decided to use the peer assessment score
of the school in which a quarterback was enrolled;
this rating of a schools academic reputation by
administrators at other institutions is available at
usnews.com and ranges from 1 (marginal) to 5
(distinguished). Admittedly, this proxy is not a
direct measure of the quarterbacks competence/
incompetence levels, but of college administrators
perceptions of the academic rigor of universities.
Nonetheless, more selective schools are likely to be
more selective with respect to which football players are admitted (Ferris, Finster, & McDonald,
2004). Given that this measure was a less direct
proxy than GPA, we analyzed a subsample with
GPA as the proxy for competence/incompetence
while analyzing the full sample with peer assessment score as the proxy for competence/incompetence.
Third, we included a proxy to control for the
possibility that actual levels of athleticism were
driving (1) performance outcomes and/or (2) evaluative phrases. We used rushing yards by each
quarterback in all of the games that were included
in our analysis. This measure was useful because it
assessed a quarterbacks physical performance in
the exact same game as the one used to code perceivers evaluations. Rushing yards assesses how
far a quarterback advanced into the opposing
teams territory without being tackled. Unlike
many other physical tasks in football, this ability
requires speed, agility, and superior body control,
all of which are central to the stereotype of blacks
as being athletic and possessing outstanding physical prowess (Entine & Smith, 1999; Miller, 2004;
Stone et al., 1999).
Fourth, we accounted for the source of the evaluative phrases for each quarterback in each data
collection period. Phrases were either recorded by

1150

Academy of Management Journal

third parties (i.e., members of the media) or members of the same sports organization as the quarterback (in quotes from media reporters). For this control variable, greater values indicated that a larger
proportion of the evaluative phrases in an article
had quotes from members of the same sports organization as the quarterback and a smaller proportion
of the evaluative phrases in an article were reported
by members of the media. It was important here to
account for source because variation among perceivers might depend on whether they desired a
quarterback to perform well, as would members of
the same organization as the leader, or instead had
no direct stake in the outcome and therefore were
less invested in the performance of the quarterback,
as members of the media typically are.
Fifth, data were collected on the race of the
source (whether the evaluators were black, white,
or another race) in a subsample of quarterbacks (n
53). However, it was discovered that there was
extremely low variance for this factor, with 94.8
percent of the sources being white. Therefore, this
factor was not included as a control in the analyses.
In sum, in the full sample we controlled for words
published per quarterback, peer assessment of a
quarterbacks academic institution, rushing yards,
and source of the evaluations. In the subsample, we
controlled for words published per quarterback,
GPA, peer assessment, and source of the evaluations.

December

phrases (e.g., competence and incompetence) occurred strictly because articles with more volume
(and, subsequently, more words written about a
particular quarterback) contained more instances of
each type of phrase. We controlled for the volume
of information written about each quarterback in
subsequent analyses.
Demonstrating Internal Validity
Our study presents an opportunity to examine
whether the evaluation of quarterbacks performance was driven by goal-based stereotyping.
Given that we predicted that stereotypical beliefs
about incompetence and athleticism would be promoted and inhibited as a consequence of performance success, a necessary first step was to rule out
the possibility that quarterbacks (1) race and/or (2)
recognized indicators of incompetence and athleticism determined weekly performance. Ruling out
these alternative explanations would largely dispel
the possibility of reverse causality. First, race
was not a significant predictor of performance
( 0.08, s.e. 0.07, p .25). Second, peer
assessment of universities was also not a significant
predictor of performance, ( 0.02, s.e. 0.05,
p .71). Third, the proxy measure of actual athleticism, rushing yards, was also not a significant
predictor of performance ( 0.01, s.e. 0.01,
p .11).

RESULTS
Hierarchical Linear Modeling

Table 1 provides correlations. Separate statistical


analyses suggested that the positive correlations
between the positive and negative evaluative

We were interested in the frequency with which


perceivers used each of the six types of evaluative

TABLE 1
Means, Standard Deviations, and Correlationsa
Variableb, c
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.

Performance
QB race
Peer assessment of QBs school
Rushing yards (during game)
Words published per QB per interval
Source of phrase
Evaluative phrases: Incompetence
Evaluative phrases: Athleticism
Evaluative phrases: Competence
Evaluative phrases: Lack of athleticism
Evaluative phrases: Positived
Evaluative phrases: Negatived

Mean

s.d.

10

11

0.55
0.27
3.09
15.57
663.93
0.14
0.37
0.97
1.56
0.17
0.92
0.21

0.50
0.45
0.65
35.29
779.55
0.29
1.52
2.59
3.78
0.71
1.94
0.75

.07
.02
.05
.14
.08
.09
.15
.07
.06
.18
.07

.11
.08
.03
.02
.09
.18
.02
.01
.03
.05

.11
.12
.02
.02
.08
.05
.02
.01
.10

.02
.15
.07
.00
.03
.06
.06
.05

.07
.29
.55
.45
.22
.56
.27

.03
.01
.06
.00
.08
.06

.14
.11
.31
.14
.24

.38
.18
.45
.05

.20
.44
.06

.10
.19

.16

Correlations greater than .09 are significant at p .05. Correlations greater than .13 are significant at p .01.
QB quarterback.
c
All variables except for QB race and peer assessment of QBs school are level 1.
d
Positive evaluative phrases include all positively valenced phrases other than athletic and competence; negative evaluative
phrases include all negatively valenced phrases except incompetence and unathletic.
a

2011

Carton and Rosette

phrases to evaluate quarterbacks. Given that we


measured how perceivers viewed quarterbacks in
four different periods during a single football season, the observations for each of the four periods
were nested within quarterbacks. There was thus a
level 1 (observations) by level 2 (quarterbacks) hierarchical data structure in which performance (the
key level 1 variable) could change week-to-week for
all quarterbacks. To analyze these nested data, we
employed a multilevel analysis, hierarchical linear
modeling (Hofmann, Griffin, & Gavin, 2000).
Hypothesis 1
The preconditions discussed by Hofmann and
colleagues (2000) for using hierarchical linear modeling were not needed to test Hypothesis 1 since it
involved the prediction of whether the level 1 variable of performance predicted the frequency of positive and negative attributions. We did, however,
control for race (a level 2 variable). Results supported Hypothesis 1a, which predicts that positive
attributions of leaders will increase after performance success ( 0.50, s.e. 0.22, p .05), and
Hypothesis 1b, which predicts that negative attributions of leaders will increase after performance
failure ( 0.25, s.e. 0.10, p .05).
Hypotheses 24
Main effects models. Following Hofmann and
colleagues (2000), we completed all necessary preconditions for testing Hypotheses 2 4.6 To test Hypothesis 2, we conducted a main effects model for
the incompetence dependent variable (evaluative
phrases connoting incompetence). Results supported Hypothesis 2 by showing that black quarterbacks were significantly more likely than white
6

All of the necessary preconditions were met. Oneway analyses of variance (ANOVAs) demonstrated significant variance across level 2 factors in the dependent
variables of interest, incompetence and athleticism, with
ICC1s indicating that 62.9 percent of the variance in
incompetence evaluative phrases and 65.9 percent of
variance in athleticism evaluative phrases existed between quarterbacks. A random-coefficients regression
model including level 1 variables (i.e., control variables
as well as performance, which varied within each quarterback across the four intervals) showed significant between-quarterbacks variance in intercepts. The necessary
preconditions for testing Hypothesis 2 were therefore
met. Finally, a random-coefficients regression model
demonstrated that there was significant between-quarterbacks variance in slopes for incompetence and athleticism evaluative phrases. The necessary preconditions for
testing Hypotheses 3 and 4 were therefore met.

1151

quarterbacks to be perceived as incompetent


( 0.94, s.e. 0.45, p .05). A chi-square difference test revealed that this main effects model had
a significant improvement in model fit over a
model with the control variables (2[2] 313.31,
p .01).
Although we did not hypothesize a main effect
for athletic evaluative phrases, such a test was necessary to create a main effects model against which
a model with the interaction terms could be compared. This model indicated that black quarterbacks were significantly more likely than white
quarterbacks to be perceived as athletic ( 1.64,
s.e. 0.35, p .01). A chi-square difference test
revealed that this main effects model had a significant improvement in model fit over a model with
the control variables (2[2] 274.48, p .01).
Models with interaction terms. In Hypothesis 3,
we predict an interaction between race and performance in predicting frequency of evaluative
phrases connoting incompetence. Results supported this prediction ( 0.95, s.e. 0.33, p
.01). A chi-square difference test revealed a significant improvement in model fit over the main effects model (2[2] 7.84, p .01). In Hypothesis 4, we predict an interaction between race and
performance in predicting frequency of evaluative
phrases connoting athleticism. Results supported
this prediction ( 2.35, s.e., 0.67, p .01). A
chi-square difference test revealed that this model
had a significant improvement in model fit over the
main effects model (2[2] 12.96, p .01). See
Figure 1 for plots of the interactions.
We needed to conduct further analyses to determine if the specific interaction patterns we predicted in Hypotheses 3 and 4 were supported. In
line with previous research treating interaction
terms in hierarchical linear models identically to
those in ordinary least squares regression models
(Hofmann, Morgeson, & Gerras, 2003), we used
simple slopes analyses to probe the two significant
interactions and determine if the two predicted
patterns would hold for each dependent variable.
We first probed the results of Hypothesis 3. As
predicted, black quarterbacks were perceived to be
significantly more incompetent than whites when
their respective teams lost (i.e., performance failure) (t[359] 9.23, p .01), but this difference
was not found when their respective teams won
(i.e., performance success). We then performed
similar analyses to probe the results of Hypothesis 4. As predicted, black quarterbacks were perceived to be significantly more athletic than whites
when their respective teams won (i.e., performance
success) (t[359] 4.81, p .01), whereas this dif-

1152

Academy of Management Journal

FIGURE 1
Evaluative Phrases Published per Week Referring
to Incompetence and Athleticisma
(1A) Quarterback Is Incompetent
3

2.5
2

Evaluative Phrases
Published
1.5
per Week
1

0.5
0

White
Black
0
Loss

(1B) Quarterback Is Athletic


3

2.5

Evaluative Phrases 2
Published
1.5
per Week
1
White
Black

0
0
Loss

disadvantage when they are evaluated and that this


bias in the evaluation process has worked against
the advancement of black leaders in organizations
(Greenhaus & Parasuraman, 1993; Knight et al.,
2003; Orpen, 1981; Powell & Butterfield, 1997; Yarkin et al., 1982). Yet the mechanisms that undergird
this bias had remained largely untapped by researchers. Our results shed light on these mechanisms by illuminating both how and why such
strong bias has been sustained against black leaders. We now discuss the specific theoretical, empirical, and managerial contributions gleaned from
our findings.
Theoretical Contributions

1
Win

0.5

December

1
Win

a
Simple slopes tests revealed that the difference between
blacks and whites was significant for losses but not for wins for
the stereotype of incompetence; simple slopes tests also revealed that the difference between blacks and whites is significant for wins but not for losses for the compensatory stereotype
of athleticism.

ference was not found when their respective teams


lost (i.e., performance failure).
Analyses of additional dependent variables
and subsample. We also tested whether there was
an interaction between race and performance in
predicting two other coding categories: competence
and lack of athleticism. These interactions were not
significant. Finally, we replicated the analyses for
Hypotheses 2 and 3 in the subsample using GPA as
a control variable. Blacks were significantly more
likely than whites to be considered incompetent
( 2.52, s.e. 0.94, p .01), and the interaction
of race and performance predicting the frequency
of incompetence evaluative phrases was significant
( 1.49, s.e. 0.57, p .01).
DISCUSSION
There exists ample evidence from prior research
to support the claim that black leaders are at a

Theory on goal-based stereotyping. Our results


provide strong evidence that goal-based stereotyping helps sustain bias against black leaders. Specifically, our analyses of archival data demonstrated
that perceivers applied different stereotypes depending upon whether black leaders but not
white leaderssucceeded or failed. As opposed to
the traditional assumption on the part of organizational scholars that stereotypes are rigidly applied
when perceivers observe leaders (DiTomaso &
Hooijberg, 1996; Knight et al., 2003; Powell & Butterfield, 1997), evaluators of leaders in our data set
were flexible in the way that they applied their
beliefs so that they could comprehend situations in
which they were presented with information (i.e.,
performance) that contradicted prevailing stereotypes of black leaders. More specifically, our test of
Hypothesis 3 indicated a leader-based attribute was
applied after performance failure but inhibited after performance success, and our test of Hypothesis 4 indicated that a nonleader attribute was applied after performance success and inhibited after
performance failure. For example, one black quarterback in our data set was perceived to be successful because he had the speed to get away but
not because he was competent. Another black quarterback was perceived to be unsuccessful because
he was inferior in making decisions under pressure but not because he was unathletic. Goalbased stereotyping allowed perceivers to maintain
(1) the inference process by attributing positive attributes to black leaders after success and negative
attributes after failure (as shown in the test of Hypothesis 1) and (2) to maintain the recognition process by sustaining their belief that black leaders are
incompetent across situations (as shown in the test
of Hypothesis 2). In an extension of prior evidence
that suggests that perceivers possess dueling schemata when they evaluate targets in organizations
(Foldy, 2006), our findings suggest that the perfor-

2011

Carton and Rosette

mance context triggers which schema is emphasized in a given instance when the target of evaluation is a black leader.
An additional contribution to research on goalbased stereotyping is our incorporation of the concept of compensatory stereotypes (Biernat et al.,
2009; Kay et al., 2008; Stone et al., 1997; Yzerbyt et
al., 2005) to help clarify how perceivers can explain
positive outcomes without undermining the stereotype that black leaders are incompetent (and, therefore, weak leaders). Whereas a compensatory stereotype in the context of sports is athletic, the
nature of compensatory stereotypes will be different in other organizational contexts (Devine & Elliot, 1995; Dixon, 2008; Kawakami & Dovidio, 2001;
Miller, 2004; Ogbu, 1985). For example, a black
CEO in a Fortune 500 firm may be viewed as successful because he or she is nonthreatening and
likable (Livingston & Pearce, 2009).
Theory on information processing. Our research
also independently contributes to information processing theories of leadership. Information processing is often described as the foundation of leadership because it relates to how evaluators perceive
that certain leaders have more credibility than others (Lord & Maher, 1993: 55). At the most fundamental level, we illuminated what can occur at the
intersection between the two dominant processes
that perceivers use to evaluate leaders: recognition
and inference (Lord & Maher, 1993). Although theory on information processing is widely employed
in research on leadership (Hall & Lord, 1995;
Hanges, Lord, & Dickson, 2000; Lord, 1985; Lord et
al., 1984; Lord & Maher, 1993; Meindl & Ehrlich,
1987; Phillips & Lord, 1982), existing research
has not frequently investigated recognition and inference in tandem.
Our findings indicate that the two processes are
likely to conflict when black leaders are evaluated,
but not when white leaders are evaluated. According to inference-based approaches (e.g., theory on
romance of leadership), leaders, regardless of race,
should be evaluated positively when they are observed in the context of success (Meindl, 1995). Yet
according to recognition-based approaches (e.g.,
leader categorization theory), the stereotype of incompetence leads blacks to be recognized as weak
leaders regardless of performance (Lord et al.,
1984). One of our principle contributions was to
offer a potential reconciliation for this discrepancy
by making two propositions. First, when recognition-based attributes match attributions derived
from inference-based processes (i.e., blacks are incompetent when they fail and possess strong compensatory stereotypes when they succeed), perceivers sustain full comprehension. Second, when

1153

there are mismatches between inference and recognition (i.e., it is difficult to label black leaders as
incompetent when they succeed and as possessing
strong compensatory stereotypes when they fail),
perceivers are motivated to better comprehend
what appears to be two incompatible impressions,
and they subsequently adopt comprehension goals.
In contrast to our results for black leaders, the
current results support the theoretical proposition
that white leaders are not strongly stereotyped (DiTomaso et al., 2007), and therefore goal-based stereotyping is not needed to reconcile specific attributes of whites with performance outcomes.
Although perceivers made more positive evaluations of white leaders after success than after failure, we suggest that perceivers were not motivated
to apply and inhibit specific attributes in the same
way that they were for black leaders.
Empirical Contributions
Our research used several methodological approaches that provided value-added contributions
to research on goal-based stereotyping, a construct
that has typically been examined in the context of
laboratory experiments (Blair, 2002). Our analysis
provided support for goal-based stereotype application in an organizational settingthat of college
football. Even though we attained this external validity, we also retained internal validity in a way
that is familiar to experimenters because we collected data on perceiver evaluations immediately
after performance outcomes (i.e., the results of football games). This timing allowed for a degree of
causal inference because the evaluations of quarterbacks occurred after performance was publically
known (Winship & Morgan, 1999). Coupled with
our juxtaposition of black and white leaders, this
procedure rendered our archival study similar to
that of an experiment.
Managerial Implications
Our findings have direct implications for the
leadership evaluation instruments that are frequently used to determine hiring, promotion, and
retention decisions. We suggest that the phenomenon of goal-based stereotyping may systematically
bias leader evaluations against black leaders, in
part explaining the glass ceiling faced by black
leaders in organizations (Morrison & Von Glinow,
1990). Perhaps the most meaningful finding for
practitioners is that success may not be credited to
the leadership ability of blacks, but instead to attributes that are perceived to compensate for incompetence. In response to potential systematic bias

1154

Academy of Management Journal

derived from goal-based stereotyping, we present a


two-step recommendation that managers can use to
identify and amend situations in which goal-based
stereotyping may affect the evaluation of black
leaders. The first step involves detecting whether
goal-based stereotyping is occurring, and the second step involves addressing goal-based stereotyping when it does occur.
Identify if compensatory stereotyping is occurring. Management should determine if goal-based
stereotyping is occurring in their organization. For
example, consider two leaders of different but comparable departments or areas, one black and one
white, who attain the same success on a performance metric (e.g., percentage increase in quarterly
sales revenue over the previous year). If different
descriptions are consistently used to evaluate different leaders with comparable performance outcomes, then goal-based stereotyping is likely to be
occurring.
We now consider in more detail specific steps
needed to detect if such patterns are actually occurring. First, managers may need to alter leadership instruments, which typically consist of text
that describes the behavior and performance of a
leader, to account for more attributes on leadership
evaluation surveys. Beyond characteristics that are
thought to be especially representative of leaders
(e.g., competence), managers can incorporate a set
of factors that are not deemed to be traditional
leadership characteristics or behaviors and instead
may be perceived as compensatory but nonetheless pivotal for organizational success. In line with
our earlier arguments, managers should take note
that different stereotypes are likely to be deemed as
compensatory in different organizational contexts
and adjust their instruments accordingly.
Second, managers need clear performance referents against which to compare attributions of leaders to determine if the perception of certain qualities increases after success in contrast to failure.
Managers should focus on the performance outcomes that are most relevant to an organization,
such as productivity and market share, and outcomes that can most directly be attributed to leader
influence, such as commissions, external ratings of
team performance, and turnover. The most important performance metrics will depend upon the
industry, stakeholders, and other situational factors. Given that in many organizations few blacks
are in leadership positions (per blackentrepreneur
profile.com, accessed in 2010), it may be more realistic to track individual black and white leaders
over time than to compare a sample of black and
white leaders at a single point in time. Several
evaluations could then be assessed over time rela-

December

tive to the same key performance metrics (e.g., productivity or market share).
Once these two factors (expanded leadership instruments and clear performance metrics) are in
place, then management is in position to detect
whether goal-based stereotyping is occurring by assessing whether evaluators use stereotypes of incompetence in the context of poor performance and
compensatory stereotypes in the context of strong
performance to evaluate black leaders and not
white leaders.
Enact perception-based reform. If evidence of
goal-based stereotyping exists, then we suggest that
managers enact perception-based reform. We propose perception-based reform in addition to or as
an alternative toposition-based reform, which includes affirmative action and diversity initiatives,
because the assumption underlying perceptionbased reform is that the fundamental source of discrimination arises from the minds of evaluators as
opposed to institutional routines, standardized
procedures, and historical inertia (Coate & Loury,
1993). Although position-based reform may be necessary to provide the opportunity for black leaders
to be successful, it may not be sufficient to curtail
bias against them because our findings suggest that
evaluators may not attribute success to the leadership ability of blacks.
There are several possible ways to achieve perception-based reform. First, management should
provide individual information about its leaders,
especially black leaders (e.g., educational background, back stories, advertisement of personal accomplishments). Individuating information has
been shown to minimize the application of stereotypes (Kunda, Davies, Adams, & Spencer, 2002;
Kunda & Sherman-Williams, 1993). As an additional measure, black leaders themselves can make
their colleagues and subordinates more aware of
their qualifications, aptitude, and experience. Although it is an additional burden that black leaders
will likely have to bear, they can nonetheless take
action themselves to address the bias caused by
goal-based stereotyping. Second, managers should
ensure that prevailing information in their organizational context does not align with existing stereotypes. This can help uproot stereotypes over time
(Hilton & von Hippel, 1996). That is, information
that runs counter to negative black stereotypes
should be easily accessible (e.g., workers are exposed to pictures of successful black leaders in
hallways and regular advertising about the creative
solutions that black leaders have devised). Finally,
management can provide incentives for accurate
judgment. Although incentives do not entirely
eliminate biases, they have been shown to reduce

2011

Carton and Rosette

them (Camerer & Hogarth, 1999). Recall that our


findings indicate that perceivers make attributions
that are not in line with reality: differences between
black and white leaders abilities were perceived
even when we controlled for accepted indicators of
ability. Thus, incentives to make accurate judgments may decrease the distinctions perceivers
make between black and white leaders.
Limitations and Directions for Future Research
We cannot glean from the current research
whether goal-based stereotyping in the context of
leader evaluation remains as potent in other business settings, especially those in which compensatory stereotypes may not be as strong. Future research should investigate other business contexts,
particularly those in which performance outcomes
are more ambiguous. Such investigations may draw
important boundary conditions in relation to the
influence of goal-based stereotyping on leader evaluations. A second limitation relates to the internal
validity of the archival study. Although we can
draw clear inferences related to causality because
evaluations took place after the previous weekends
performance, perceivers were not randomly sorted
into different conditions. Instead, all perceivers of a
quarterback in our sample (e.g., a media reporter
and players at an institution) had the same opportunity to comment on that quarterback.
Scholars looking to leverage our findings to advance future research might follow a number of
fruitful paths. One possibility is that researchers
can aim to illuminate the complementarity between two separate findings: white leaders (and not
black leaders) are categorized as prototypes (i.e.,
representative targets) of leaders during the evaluation process (Rosette et al., 2008); and black leaders (and not white leaders) are stereotyped (i.e.,
assumed to have highly distinctive traits) during
the evaluation process, as demonstrated by the
findings in our study. The former finding may explain why white leaders are in a position to be
evaluated favorably, whereas the latter finding may
explain why black leaders are in a position to be
evaluated unfavorably. An integration of these two
general findings may provide a more complete picture of race and the evaluation of leadership.
Conclusions
A wealth of existing research suggests that black
leaders are not evaluated comparably to white leaders, resulting in severe consequences for black organizational members. One hope is that the continued success of black leaders will eventually uproot

1155

existing stereotypes. Our findings cast doubt on


this optimism by demonstrating that bias against
black leaders is sustained because evaluators are
flexible in the way that they use stereotypes. According to evaluators of black leaders in our study,
failure was due to negative leadership attributes,
whereas success was due to nonleadership attributes. This research indicates that the flexibility of
the human mind presents yet another obstacle in
the push for a world in which prejudice does not
obstruct the advancement of minorities in organizations.

REFERENCES
Agle, B. R., Mitchell, R. K., & Sonnenfeld, J. A. 1999. Who
matters to CEOs? An investigation of stakeholder
attributes and salience, corporate performance, and
CEO values. Academy of Management Journal, 42:
507525.
Alliance Board for Diversity. 2005. Women and minorities on Fortune 100 boards. New York: Catalyst.
Bargh, J. A. 1999. The cognitive monster: The case
against the controllability of automatic stereotype
effects. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology: 361382. New
York: Guilford.
Bass, T. 2009. The basic fundamentals of a quarterback.
Youth football directory. http://wvrsn.com/index.
cfm?foraction-Home-DisplayStory&Item|D665.
Beatty, R. W. 1973. Blacks as supervisors: A study of
training, job performance, and employers expectations. Academy of Management Journal, 16: 196
206.
Biernat, M., Sesko, A. K., & Amo, R. B. 2009. Compensatory stereotyping in interracial encounters. Group
Processes and Intergroup Relations, 12: 551563.
Blair, I. V. 2002. The malleability of automatic stereotypes and prejudice. Personality & Social Psychology Review, 6: 242261.
Buffington, D., & Fraley, T. 2008. Skill in Black and
White: Negotiating Media Images of Race in a Sporting Context. Journal of Communication Inquiry, 32:
292310.
Calder, R. J. 1977. An attribution theory of leadership. In
B. M. Staw & G. R. Salancik (Eds.), New directions in
organizational behavior: 179 204. Chicago: St.
Clair.
Camerer, C. F., & Hogarth, R. M. 1999. The effects of
financial incentives in experiments: A review and
capital-labor-production framework. Journal of Risk
and Uncertainty, 19: 7 42.
Coate, S., and Loury, G. C. 1993. Will affirmative-action

1156

Academy of Management Journal

December

policies eliminate negative stereotypes? American


Economic Review, 83: 1220 1240.

ceived status and competition. Journal of Personality and Social Psychology, 82: 878 902.

Corporate Board Initiative. 2006. 2005 update on the


Committee of 100s Asian & Asian Pacific American
(APA) corporate board report card.

Fleiss, J. L. 1981. Statistical methods for rates and proportions (2nd ed.). New York: Wiley.

Dasgupta, N., & Asgari, S. 2004. Seeing is believing: Exposure to counterstereotypic women leaders and its
effect on the malleability of automatic gender stereotyping. Journal of Experimental Social Psychology,
40: 642 658.
Devine, P. G. 1989. Stereotypes and prejudice: Their
automatic and controlled components. Journal of
Personality and Social Psychology, 56: 680 690.
Devine, P. G., & Baker, S. M. 1991. Measurement of racial
stereotype subtyping. Personality and Social Psychology Bulletin, 17: 44 50.
Devine, P. G., & Elliot, A. J. 1995. Are racial stereotypes
really fading? The Princeton trilogy revisited. Personality and Social Psychology Bulletin, 21: 1139
1150.
DiTomaso, N., & Hooijberg, R. 1996. Diversity and the
demands of leadership. Leadership Quarterly, 7:
163187.
DiTomaso, N., Post, C., Smith, D. R., Farris, G. F., &
Cordero, R. 2007. Effects of structural position on
allocation and evaluation decisions for scientists
and engineers in industrial R&D. Administrative
Science Quarterly, 52: 175207.

Foldy, E. G. 2006. Dueling schemata: Dialectical sensemaking about gender. Journal of Applied Behavioral Science, 42: 350 372.
Folger, J., Hewes, D., & Poole, M. 1984. Coding social
interaction. In B. Dervin & M. Voight (Eds.), Progress
in communication sciences: 115161. Norwood, NJ:
Ablex.
Ford, J. K., Kraiger, K., & Schechtman, S. L. 1986. Study
of race effects in objective indices and subjective
evaluations of performance: A meta-analysis of performance criteria. Psychological Bulletin, 99: 330
337.
Greenhaus, J. H., & Parasuraman, S. 1993. Job performance attributions and career advancement prospects: An examination of gender and race effects.
Organizational Behavior and Human Decision
Processes, 55: 273297.
Greenhaus, J. H., Parasuraman, S., & Wormley, W. M.
1990. Effects of race on organizational experiences,
job performance evaluations, and career outcomes.
Academy of Management Journal, 33: 64 86.
Hall, R. J., & Lord, R. G. 1995. Multi-level informationprocessing explanations of followers leadership perceptions. Leadership Quarterly, 6: 265287.

Dixon, J. C., & Rosenbaum, M. S. 2004. Nice to know you?


Testing contact, cultural, and group threat theories
of anti-Black and anti-Hispanic stereotypes. Social
Science Quarterly, 85: 257280.

Hanges, P., Lord, R., & Dickson, M. 2000. An information


processing perspective on leadership and culture: A
case for connectionist architecture. Applied Psychology, 49: 133161.

Dixon, T. L. 2008. Network news and racial beliefs: Exploring the connection between national television
news exposure and stereotypical perceptions of African Americans. Journal of Communication, 58:
321337.

Hilton, J. L., & von Hippel, W. 1996. Stereotypes. In J. T.


Spence, J. M. Darley, & D. J. Foss (Eds.), Annual
review of psychology, vol. 47: 237271. Palo Alto,
CA: Annual Reviews.

Eisenhardt, K. M. 1989. Building theories from case


study research. Academy of Management Review,
14: 532550.
Emrich, C. G. 1999. Context effects in leadership perception. Personality and Social Psychology Bulletin,
25: 9911006.
Entine, J., & Smith, E. 1999. Why black athletes dominate sports and why were afraid to talk about it.
New York: Public Affairs.
Ferris, E., Finster, M., & McDonald, D. 2004. Academic fit
of student-athletes: An analysis of NCAA division
I-A graduation rates. Research in Higher Education,
45: 555575.
Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. 2002. A
model of (often mixed) stereotype content: Competence and warmth respectively follow from per-

Hofmann, D. A., Griffin, M. A., & Gavin, M. B. 2000. The


application of hierarchical linear modeling to organizational research. In K. J. Klein & S. W. J. Kozlowski
(Eds.), Multilevel theory, research, and methods in
organizations: 467511. San Francisco: Jossey-Bass.
Hofmann, D. A., Morgeson, F. P., & Gerras, S. J. 2003.
Climate as a moderator of the relationship between
leader-member exchange and content specific citizenship: Safety climate as an exemplar. Journal of
Applied Psychology, 88: 170 178.
Howell, D. C. 1992. Statistical methods for psychology
(3rd ed.). Boston: Duxbury.
Kawakami, K., & Dovidio, J. F. 2001. The reliability of
implicit stereotyping. Personality and Social Psychology Bulletin, 27: 212225.
Kay, A. C., Gaucher, D., Napier, J. L., Callan, M. J., &
Laurin, K. 2008. God and the government: Testing a

2011

Carton and Rosette

compensatory control mechanism for the support of


external systems. Journal of Personality and Social
Psychology, 95: 18 35.
Keidel, R. W. 1987. Team sport models as a generic
organizational framework. Human Relations, 40:
591 612.
Kelley, H. H., & Michela, J. L. 1980. Attribution theory
and research. In M. R. Rosenzweig & L. W. Porter
(Eds.), Annual review of psychology, vol. 31: 457
501. Palo Alto, CA: Annual Reviews.
Knight, J. L., Hebl, M. R., Foster, J. B., & Mannix, L. M.
2003. Out of role? Out of luck: The influence of race
and leadership status on performance appraisals.
Journal of Leadership and Organizational Studies,
9: 8593.
Krueger, J. 1996. Personal beliefs and cultural stereotypes
about racial characteristics. Journal of Personality
and Social Psychology, 71: 536 548.
Kunda, Z., Davies, P. G., Adams, B. D., & Spencer, S. J.
2002. The dynamic time course of stereotype activation: Activation, dissipation, and resurrection. Journal of Personality and Social Psychology, 82: 283
299.
Kunda, Z., Davies, P., Hoshino-Browne, E., & Jordan, C.
2003. The impact of comprehension goals on the ebb
and flow of stereotype activation during interaction.
In S. J. Spencer, S. Fein, M. P. Zanna & J. M. Olson
(Eds.), Motivated social perception: The Ontario
Symposium, vol. 9: 120. Mahwah, NJ: Erlbaum.

1157

Macrae, C. N., and Bodenhausen, G. V., Milne, A. B.


1995. The dissection of selection in person perception: Inhibitory processes in social stereotyping.
Journal of Personality and Social Psychology, 69:
397 407.
McFadyen, H. 2007. The function of phrases. University
of Ottawa Writing Centre. http://www.writingcentre.
uottawa.ca/hypergrammar/phrfunc.html.
Accessed
March 2010.
Meindl, J. R. 1995. The romance of leadership as a follower-centric theory: A social constructionist approach. Leadership Quarterly, 6: 329 341.
Meindl, J. R., & Ehrlich, S. B. 1987. The romance of
leadership and the evaluation of organizational performance. Academy of Management Journal, 30:
91109.
Mercurio, E., & Filak, V. 2010. Roughing the passer: The
framing of black and white quarterbacks prior to the
NFL draft. Howard Journal of Communications, 21:
56 71.
Miller, P. B. 2004. The anatomy of scientific racism. In
P. B. Miller & D. K. Wiggins (Eds.), Sport and the
color line: Black athletes and race relations in
twentieth-century America: 327344. New York:
Routledge.
Morrison, A. M., & Von Glinow, M. A. 1990. Women and
minorities in management. American Psychologist,
45: 200 208.

Kunda, Z., & Sherman-Williams, B. 1993. Stereotypes


and the construal of individuating information. Personality and Social Psychology Bulletin, 19: 90 99.

Murrell, A. J., & Curtis, E. M. 1994. Causal attributions of


performance for black and white quarterbacks in the
NFL: A look at the sports pages. Journal of Sport
and Social Issues, 18: 224 233.

Kunda, Z., & Spencer, S. J. 2003. When do stereotypes


come to mind and when do they color judgment? A
goal-based theoretical framework for stereotype activation and application. Psychological Bulletin, 129:
522544.

Ogbu, J. U. 1985. A cultural ecology of competence


among inner-city blacks. In M. B. Spencer, G. K.
Brookins, & W. R. Allen (Eds.), Beginnings: The
social and affective development of black children: 45 66. Hillsdale, NJ: Erlbaum.

Livingston, R. W., & Pearce, N. A. 2009. The teddy-bear


effect: Does having a baby face benefit black chief
executive officers? Psychology Science, 20: 1229
1236.

Orpen, C. 1981. Causal attributions for the success and


failure of black and white managers. Journal of Organizational Behavior, 2: 81 87.

Lord, R. G. 1985. An information processing approach to


social perceptions, leadership, and behavioral measurement in organizations. In L. L. Cummings &
B. M. Staw (Eds.), Research in organizational behavior, vol. 7: 87128. Greenwich, CT: JAI.
Lord, R. G., Foti, R. J., & De Vader, C. L. 1984. A test of
leadership categorization theory: Internal structure,
information processing, and leadership perceptions.
Organizational Behavior and Human Performance, 34: 343378.
Lord, R. G., & Maher, K. J. 1993. Leadership and information processing: Linking perceptions and performance. London and New York: Routledge.

Pfeffer, J. 1977. The ambiguity of leadership. Administrative Science Quarterly, 2: 104 112.
Pfeffer, J. 1997. New directions for organization theory.
New York: Oxford University Press.
Pfeffer, J., & Davis-Blake, A. 1986. Administrative succession and organizational performance: How administrator experience mediates the succession effect.
Academy of Management Journal, 29: 72 83.
Phillips, J. S., & Lord, R. G. 1982. Schematic information
processing and perceptions of leadership in problem-solving groups. Journal of Applied Psychology,
67: 486 492.
Powell, G. N., & Butterfield, D. A. 1997. Effect of race on

1158

Academy of Management Journal

December

promotions to top management in a federal department. Academy of Management Journal, 40: 112
128.

Stereotype threat effects on black and white athletic


performance. Journal of Personality and Social
Psychology, 77: 12131227.

Rada, J., & Wulfemeyer, K. T. 2005. Color coded: Racial


descriptors in television coverage of intercollegiate
sports. Journal of Broadcasting and Electronic Media, 49: 65 85.

Stone, J., Perry, Z. W., & Darley, J. M. 1997. White men


cant jump: Evidence for the perceptual confirmation of racial stereotypes following a basketball
game. Basic and Applied Social Psychology, 19:
291306.

Rasmussen, R., Esgate, A., & Turner, D. 2005. On your


marks, get stereotyped, go! Novice coaches and black
stereotypes in sprinting. Journal of Sport and Social Issues, 29: 426 436.
Rosette, A. S., Leonardelli, G. J., & Phillips, K. W. 2008.
The white standard: Racial bias in leader categorization. Journal of Applied Psychology, 93: 758 777.
Rudman, L. A., Ashmore, R. D., & Gary, M. L. 2001.
Unlearning automatic biases: The malleability of
implicit prejudice and stereotypes. Journal of Personality and Social Psychology, 81: 856 868.
Rudman, L. A., & Glick, P. 2001. Prescriptive gender
stereotypes and backlash toward agentic women.
Journal of Social Issues, 57: 743762.
Rush, M. C., Phillips, J. S., & Lord, R. G. 1981. Effects of
a temporal delay in rating on leader behavior descriptions: A laboratory investigation. Journal of
Applied Psychology, 66: 442 450.
Shamir, B. 1991. The charismatic relationship: Alternative explanations and predictions. Leadership
Quarterly, 2: 81104.
Sinclair, L., & Kunda, Z. 1999. Reactions to a black professional: Motivated inhibition and activation of
conflicting stereotypes. Journal of Personality and
Social Psychology, 77: 885904.
Slack, T. 1997. Understanding sport organizations.
Champaign, IL: Human Kinetics.
Smith, C. P. 2000. Content analysis and narrative analysis. In H. T. Reis & C. M Judd (Eds.), Handbook of
research methods in social and personality: 313
335. New York: Cambridge University Press.
Staw, B. M., & Hoang, H. 1995. Sunk costs in the NBA:
Why draft order affects playing time and survival in
professional basketball. Administrative Science
Quarterly, 40: 474 494.
Stone, J., Lynch, C. I., Sjomeling, M., & Darley, J. M. 1999.

Thomas, D. A., & Gabarro, J. J. 1999. Breaking through:


The making of minority executives in corporate
America. Boston: Harvard Business School Press.
Weber, R., Camerer, C., Rottenstreich, Y., & Knez, M.
2001. The illusion of leadership: Misattribution of
cause in coordination games. Organization Science,
12: 582598.
Winship, C., & Morgan, S. L. 1999. The estimation of
causal effects from observational data. In J. Hagan &
K. Cook (Eds.), Annual review of sociology, vol. 25:
659 707. Palo Alto, CA: Annual Reviews.
Yarkin, K. L., Town, J. P., & Wallston, B. S. 1982. Blacks
and women must try harder: Stimulus persons race
and sex attributions of causality. Personality and
Social Psychology Bulletin, 8: 2124.
Yzerbyt, V., Provost, V., & Corneille, O. 2005. Not competent but warm... really? Compensatory stereotypes
in the French-speaking world. Group Processes and
Intergroup Relations, 8: 291308.

Andrew M. Carton (amc30@psu.edu) is an assistant professor of management and organization in the Smeal
College of Business at The Pennsylvania State University. He received his Ph.D. from Duke Universitys Fuqua
School of Business. His research focuses on the relationship between intergroup relations, goal systems, and
leadership in organizations.
Ashleigh Shelby Rosette (arosette@duke.edu) received
her Ph.D. from Northwestern University and is an associate professor of management and organizations at Duke
University. Her research interests include diversity-related perceptions in leadership, workplace discrimination, systems of privilege, and negotiation strategies.

Copyright of Academy of Management Journal is the property of Academy of Management and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.

Das könnte Ihnen auch gefallen