Strategic Management Journal, Vol. 19, 39–58 (1998)
COGNITIVE DIVERSITY AMONG UPPER-ECHELON
EXECUTIVES: IMPLICATIONS FOR STRATEGIC
DECISION PROCESSES
C. CHET MILLER
1
*, LINDA M. BURKE
2
AND WILLIAM H. GLICK
3
1
Johnson Graduate School of Management, Cornell University, Ithaca, New York,
U.S.A.
2
Department of Reimbursement Management, Scott and White Memorial Hospital,
Temple, Texas, U.S.A.
3
College of Business Administration, Arizona State University, Tempe, Arizona,
U.S.A.
Diversity among executives is widely assumed to influence a firm’s strategic decision processes,
but empirical research on this linkage has been virtually nonexistent. To partially fill the
void, we drew upon three separate studies to examine the impact of executive diversity on
comprehensiveness of strategic decision-making and extensiveness of strategic planning. Contrary
to common assumptions of researchers and executives, our results suggest that executive
diversity inhibits rather than promotes comprehensive examinations of current opportunities
and threats, and inhibits rather than promotes extensive long-range planning. In light of the
cumulative research showing that firm performance is related to both comprehensiveness and
extensiveness, our results provide evidence for an indirect connection between executive diversity
and firm performance. 1998 John Wiley & Sons, Ltd.
Strat. Mgmt. J., Vol. 19, 39–58 (1998)
INTRODUCTION
Interest in executive diversity has surged in recent
years. Among researchers fueling this surge,
many have argued that higher levels of diversity
lead to executive creativity, more effective execu-
tive decision-making, and more positive organi-
zational outcomes (Bantel and Jackson, 1989).
Other researchers, however, have argued that
higher levels of executive diversity result in less
communication among executives, less effective
executive decision-making, and less positive
organizational outcomes (O’Reilly, Snyder, and
Key words: executive diversity; strategic decision-
making; strategic planning
* Correspondence to: C. Chet Miller, Johnson Graduate School
of Management, Cornell University, Ithaca, New York, U.S.A.
CCC 0143–2095/98/010039–20 $17.50 Received 1 July 1991
1998 John Wiley & Sons, Ltd. Final revision received 3 March 1997
Boothe, 1993). Empirically, research has not pro-
duced consistent support for either of these posi-
tions (cf. Bantel and Jackson, 1989; Glick, Miller,
and Huber, 1993; Jackson et al., 1991; Michel
and Hambrick, 1992; Murray, 1989; O’Reilly et
al., 1993; Smith et al., 1994; Wiersema and
Bantel, 1992, 1993).
One possible explanation for the disappointing
empirical results is that researchers have focused
on demographic diversity rather than cognitive
diversity. Demographic diversity typically is not
hypothesized to have direct effects on processes
or outcomes, but is hypothesized to have indirect
effects through cognitive diversity (Glick et al.,
1993). Thus, it may be that the effects of demo-
graphic diversity are too weak to be detected
consistently. Further, demographic diversity may
not actually affect cognitive diversity, and there-
40 C. C. Miller, L. M. Burke and W. H. Glick
fore may not have any important effects on proc-
esses or outcomes. The linkage between demo-
graphic diversity and cognitive diversity is
assumed to exist by most researchers (e.g., Smith
et al., 1994; Wiersema and Bantel, 1992), but
recent evidence suggests that the linkage may not
exist. Glick et al. (1993), for example, found
that diversity assessed in terms of demographic
features of executives did not correlate with
diversity assessed in terms of cognitive features.
A second possible explanation for the disap-
pointing results applies to studies of organi-
zational outcomes: the mediating effects of proc-
ess variables have not been examined in most
studies of executive diversity and organizational
outcomes. Instead, most researchers (e.g., Bantel
and Jackson, 1989; Wiersema and Bantel, 1992)
have simply related diversity to outcome variables
such as organizational innovation and prof-
itability. It may be that diversity’s effects on
ultimate outcome variables are too weak to be
detected consistently, particularly in cross-
sectional studies.
The purpose of the current research is to theo-
retically and empirically examine the linkage
between cognitive diversity and two strategic
process variables. By focusing on cognitive as
opposed to demographic diversity, we address the
first possible explanation for the disappointing
previous results, and we move towards the cogni-
tive construct of paradigm heterogeneity high-
lighted by Hambrick (1994). By focusing on
process variables that may mediate between diver-
sity and organizational outcomes, we address the
second possible explanation. To the extent that
cognitive diversity is found to influence the two
strategic process variables, evidence will have
been provided for an indirect diversity–
profitability linkage because the two process vari-
ables we investigate have repeatedly been found
to influence firm profitability.
COGNITIVE DIVERSITY,
COMPREHENSIVENESS, AND
EXTENSIVENESS
Definitions
Comprehensiveness of strategic decision proc-
esses and extensiveness of strategic planning are
important strategic process variables. Comprehen-
siveness, the process variable hypothesized most
얀 1998 John Wiley & Sons, Ltd. Strat. Mgmt. J., Vol 19, 39–58 (1998)
often as an intervening variable in discussions of
executive diversity and firm performance, is
defined as the extent to which an upper-echelon
executive group utilizes an extensive decision
process when dealing with immediate opportuni-
ties and threats (Fredrickson and Mitchell, 1984).
Behavioral indicators of the level of comprehen-
siveness include the extent to which brainstorming
sessions occur, the number of alternative solutions
that are seriously considered, and the extent to
which quantitative analyses are conducted. The
amount of investigatory work carried out to
handle an immediate situation is the key. It must
be emphasized that comprehensiveness pertains
to the absolute amount of investigatory activity
rather than investigatory completeness per se.
1
Extensiveness of strategic planning is defined
as the extent to which an upper-echelon executive
group utilizes a substantial planning process to
formulate long-term goals and strategies for the
firm. The same behavioral indicators relevant for
comprehensiveness are relevant for extensiveness,
but rather than examining those indicators in the
context of current problem-solving for immediate
opportunities and threats, they are examined in
the context of long-term planning (e.g., brain-
storming over a response to an immediate threat
pertains to comprehensiveness whereas brain-
storming over where the firm as a whole should
be in 10 years pertains to long-term planning).
As with comprehensiveness, it must be empha-
sized that extensiveness of planning pertains to
the absolute amount of activity rather than com-
pleteness per se.
As pointed out above, extensiveness and com-
prehensiveness differ in terms of their foci: adapt-
ing to and shaping the long-term future as
opposed to solving today’s problems. For
example, a firm might not engage in 3-, 5-, or
10-year planning that encompasses all or most of
1
Completeness differs from amount of activity in that com-
pleteness is amount of activity divided by total possible
activity. If the amount of investigatory activity found in
strategic decision-making is the same for a small firm in a
simple industry and a larger firm in a more complex industry,
then completeness is higher for the small firm (where, in
theory, the total possible investigatory activity is smaller due
to the simpler context). Since total possible activity for a given
situation is very difficult to determine, and since theoretical
arguments relating diversity and decision-making apply most
clearly to amount of activity rather than true completeness
(see arguments in the text), our focus on amount of activity
seems appropriate.
Cognitive Diversity among Upper-echelon Executives 41
the firm (low extensiveness), but it may attack
comprehensively a strategic problem that just
arose regarding availability of raw materials for
one of its product lines (high comprehensiveness).
Miller and Toulouse (1986), Priem, Rasheed and
Kotulic (1995), and others have made the same
basic distinction we are making between exten-
siveness and comprehensiveness.
Comprehensiveness and extensiveness have
both been found to impact firm profitability. For
comprehensiveness, some research has suggested
that a large amount of investigatory activity in
handling immediate opportunities and threats is
harmful for firms in turbulent industries (i.e.,
Fredrickson and Mitchell, 1984), but most empiri-
cal research has suggested positive effects for
firms in turbulent industries and null effects for
firms in stable industries (see Bourgeois and
Eisenhardt, 1988; Glick et al., 1993; Miller and
Toulouse, 1986; and Priem et al., 1995). Simi-
larly, research findings related to long-term plan-
ning and performance have been mixed over the
years, but two recent meta-analyses (Boyd, 1991;
Miller and Cardinal, 1994) provide strong evi-
dence that extensiveness of strategic planning
positively influences firm performance, especially
in turbulent industries.
2
Cognitive diversity is defined in terms of dif-
ferences in beliefs and preferences held by upper-
echelon executives within a firm. More speci-
fically, cognitive diversity refers to variation in
beliefs concerning cause–effect relationships and
variation in preferences concerning various goals
for the organization (Miller, 1990). Such variation
underlies differences in perspectives that tend to
endure through time. Because variation in endur-
ing beliefs and preferences tends to create dis-
agreements when specific strategic issues are
being considered (see, for example, Lant, Milli-
2
In published research relevant to the linkage between com-
prehensiveness and firm performance, operational definitions
(i.e., measures) have consistently been focused on absolute
amount of investigatory activity rather than completeness per
se (see, for example, Fredrickson, 1984; Miller and Toulouse,
1986; and Priem et al., 1995). Constitutive definitions, how-
ever, have sometimes been focused more on true completeness
(e.g., Fredrickson, 1984). In published research relevant to
the linkage between extensiveness of planning and firm per-
formance, operational definitions have typically been focused
on amount of activity (or amount of formal activity) rather
than completeness per se (see Miller and Cardinal, 1994).
Differences in terminology in the planning–performance litera-
ture, however, have obscured this consistency. Extensiveness
is perhaps a reasonable unifying term.
얀 1998 John Wiley & Sons, Ltd. Strat. Mgmt. J., Vol 19, 39–58 (1998)
ken and Batra, 1992), cognitive diversity probably
influences both comprehensiveness and exten-
siveness. The direction of the effects of cognitive
diversity, however, is unclear, with some argu-
ments suggesting positive effects while others
suggest negative effects. Arguments suggesting
positive effects are more prevalent and are
presented first.
Arguments suggesting positive effects
At least three arguments suggest that cognitive
diversity positively influences comprehensiveness
and extensiveness. The first of these concerns
disagreements as a basic resource. When there are
many disagreements surrounding an immediate
opportunity or threat, or a long-range plan, upper-
echelon executives as a group and as individuals
are aware of more issues, more ways of viewing
each issue, and more alternative courses of action
(Bantel and Jackson, 1989; Lant et al., 1992;
Wiersema and Bantel, 1992). Once aware of the
range of issues and options, the upper-echelon
group can discuss them, commission relevant
analyses, hire consultants for help in areas of
weak knowledge, and so on. If there are few
or no disagreements at the outset, upper-echelon
executives are less likely to consider a wide range
of issues and options because they simply would
not think of many of them. As Lant et al. point
out, disagreements can ‘result in more extensive
discussion of strategic options, more learning
opportunities, and, thereby, reduce the likelihood
of a groupthink-type phenomenon occurring’
(1992: 591).
The second argument is a simple argument put
forth by Fredrickson and Mitchell (1984), Glick
et al. (1993), and others. This argument concerns
costs. When there are many disagreements in
strategic decision-making, upper-echelon execu-
tives are more likely to expend the resources
necessary for more analyses, more consultants,
and more discussions. In other words, the need
to resolve disagreements or at least partially rec-
oncile divergent positions in order to move for-
ward leads to a greater willingness to expend the
resources necessary for high comprehensiveness
and extensiveness. When there are few disagree-
ments, executives will not feel a need to expend
such resources.
The third argument suggesting positive effects
is more complex than the preceding two argu-
42 C. C. Miller, L. M. Burke and W. H. Glick
ments. This third argument suggests that disagree-
ments affect upper-echelon cohesion. Cohesion,
in turn, is expected to affect comprehensiveness
and extensiveness. Thus, cohesion is expected to
mediate partially the effects of cognitive diversity
on comprehensiveness and extensiveness.
Cohesion is defined as the extent to which
upper-echelon executives like one another and
stick up for each other (see O’Reilly, Caldwell,
and Barnett, 1989). It is related to Hambrick’s
(1994) concept of behavioral integration. Social
psychological arguments related to interpersonal
attraction and inferred evaluations (see Condon
and Crano, 1988) suggest that cognitive diversity
negatively influences cohesion. Through a process
of inferred evaluation, individuals assume that a
person who agrees with them also likes them.
This process combined with the frequent obser-
vation that individuals tend to like those who like
them (Aronson and Worchel, 1966; Condon and
Crano, 1988) yields the expectation of a negative
relationship between cognitive diversity and
upper-echelon cohesion. Stated more simply:
‘individuals will feel closer to and identify with
persons who share similar beliefs and values’
(Wagner, Pfeffer, and O’Reilly, 1984: 77).
The second part of this diversity–process argu-
ment relates cohesion and strategic processes.
Specifically, the second part of the argument
suggests that cohesion negatively affects compre-
hensiveness and extensiveness. The principal rea-
son for expecting cohesion to negatively affect
comprehensiveness and extensiveness concerns a
desire for amicable relations among cohesive
executive teams. Amicable relations can be dis-
rupted by many of the tactics that promote com-
prehensive decision-making and extensive plan-
ning processes, such as playing devil’s advocate
and insisting on consulting outsiders to confirm
or disconfirm beliefs held by other team members.
Within cohesive executive teams, ideas put on
the table early are more likely to go unchallenged
and uninvestigated. In contrast, executives in
teams that are not cohesive are more likely to
challenge opinions put forth by their colleagues.
These executives are more likely to encourage
debate and initiate investigations designed to
uncover flaws in their colleagues’ reasoning.
Consistent with this reasoning, Janis has argued
that extremely high levels of cohesion can lead to
groupthink—‘a deterioration of mental efficiency,
reality testing, and moral judgment that results from
얀 1998 John Wiley & Sons, Ltd. Strat. Mgmt. J., Vol 19, 39–58 (1998)
in-group pressures’ (1972: 9). Group members are
thought to value group membership to the point
where fear of ostracism and fear of membership
loss result in conformity and unquestioned accept-
ance of ideas from an early decision contributor or
from a group leader. Particularly problematic is the
fact that the group as a whole often refuses to seek
or accept input from outsiders.
In summary, cognitive diversity is believed by
many to negatively affect cohesion and cohesion
is believed to negatively affect comprehensiveness
and extensiveness, resulting in an overall positive
linkage between diversity and comprehensiveness
and an overall positive linkage between diversity
and extensiveness. This cohesion argument com-
bined with the resource and cost arguments
presented earlier provides strong support for the
popular expectation that cognitive diversity posi-
tively affects comprehensiveness and exten-
siveness. For further theoretical support of the
positive effects position, see Finkelstein and
Hambrick (1996: 146–147).
Arguments suggesting negative effects
Although the most popular perspective suggests
positive effects, there are two arguments suggest-
ing cognitive diversity negatively affects compre-
hensiveness and extensiveness. First, diversity
often implies disagreement over strongly held
preferences and beliefs that will not be compro-
mised. Thus, extensive decision-making may lead
to head-butting rather than to issue resolution
(Glick et al., 1993). If so, one or a few executives
may quietly address strategic issues behind the
scenes while not opening up the process to others.
In such situations, existing executive diversity
would not have a chance to cause further analyses
or debates. Second, cognitive diversity often
implies that different people will use their own
specialized languages, images, and stories to com-
municate with each other. As numerous
researchers (e.g., Daft and Lengel, 1986) have
suggested, such differentiation can lead to com-
munication failures. To the extent that communi-
cation failures occur, one or a few executives may
quietly address strategic issues behind the scenes.
Hypotheses
To summarize, several arguments suggest that
cognitive diversity has positive effects and several
Cognitive Diversity among Upper-echelon Executives 43
arguments suggest that cognitive diversity has
negative effects on comprehensiveness and exten-
siveness. Despite the counter arguments, we
initially adopted the most popular perspective and
hypothesized that:
Hypothesis 1: Cognitive diversity positively
influences comprehensiveness of strategic
decision-making.
Hypothesis 2: Cognitive diversity positively
influences extensiveness of strategic planning.
Control variables
Our purpose is to examine the impact of cognitive
diversity on comprehensiveness and exten-
siveness. It is not our goal to develop a complete
model of causal factors related to strategic
decision-making. Nonetheless, to control for fac-
tors that may play a major role in determining
comprehensiveness and extensiveness, we
included variables other than cognitive diversity
in our empirical work. Specifically, a priori,we
included environmental turbulence and firm size.
These variables have appeared with great fre-
quency in previous discussions of factors that
may influence strategic decision-making (e.g.,
Capon, Farley, and Hulbert, 1987; Fredrickson
and Iaquinto, 1989; Fredrickson and Mitchell,
1984; Grinyer, Al-Bazzaz and Yasai-Ardekani,
1986; Kukalis, 1989; Lindsay and Rue, 1980;
Mintzberg, 1973; Odom and Boxx, 1988). With
respect to the impact of size, it may be that
larger organizations, because of greater com-
plexity, have more comprehensive, extensive stra-
tegic processes. Further, large size and its attend-
ant complexity also may lead to more diverse
upper-echelon management groups, thereby cre-
ating a spurious connection between diversity and
strategic processes. With respect to turbulence, it
may be that organizations facing higher turbu-
lence have more comprehensive, extensive stra-
tegic processes because of the higher levels of
change and uncertainty that must be handled.
Further, turbulent environments may lead to more
diverse upper-echelon management groups within
organizations, thereby creating a spurious connec-
tion between diversity and strategic processes.
These theoretical positions certainly are not the
only positions that could be put forth, but they
얀 1998 John Wiley & Sons, Ltd. Strat. Mgmt. J., Vol 19, 39–58 (1998)
do indicate the importance of controlling for size
and turbulence.
METHODS AND RESULTS
The results of a single study are affected by the
particular research methods used in that study,
and they may also be affected by sampling error
(for an excellent discussion of these issues, see
Hunter and Schmidt, 1990). Thus, the results of
a single study must always be viewed cautiously
as validity and generalizability are not assured.
One method of partially overcoming these prob-
lems involves incorporating multiple studies in a
single research effort. Researchers (e.g., Brockner
et al., 1993; Simons, 1993) adopting such repli-
cated designs (Hunter and Schmidt, 1990) allow
for richer assessments of the validity and general-
izability of their research findings. Accordingly,
three different studies were used to test our
hypotheses.
Study 1
Sample
Chief executive officers of 315 firms were asked
to include their firms in this study. Thirty-eight
agreed to do so. The 38 chief executives led
firms in a wide variety of industries, including
aircraft engine manufacturing, oil well drilling,
and air transportation. The participation rate of
12 percent is fairly low, but not inconsistent with
many other studies of this type. The 315 firms
were selected by randomly sampling firms that
(1) were nondiversified and that (2) were listed
in Moody’s published materials (i.e., Moody’s
Industrial Manual, OTC Manual, and Transpor-
tation Manual). Nondiversified firms were se-
lected for reasons unrelated to the present work.
Data and measures
Data were collected from chief executive officers
through a four-page questionnaire. Two dimen-
sions of cognitive diversity were assessed: diver-
sity among executives concerning preferred goals
for the firm (i.e., preference diversity) and diver-
sity among executives concerning the nature of
cause–effect relationships (i.e., belief diversity).
Overall cognitive diversity was an aggregation of
