The Relationship between Overconfidence and the Introduction of Risky Products: Evidence

from a Field Study

Author(s): Mark Simon and Susan M. Houghton
Source: The Academy of Management Journal, Vol. 46, No. 2 (Apr., 2003), pp. 139-149
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/30040610 .
Accessed: 04/12/2013 00:39
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy
of Management Journal.
http://www.jstor.org
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
c
Academy of Management Journal
2003, Vol. 46, No. 2, 139-149.
THE RELATIONSHIP BETWEEN OVERCONFIDENCE
AND THE INTRODUCTION OF RISKY PRODUCTS:
EVIDENCE FROM A FIELD STUDY
MARK SIMON
Oakland
University
SUSAN M. HOUGHTON
Georgia
State
University
To
date,
no field research has examined the effects of overconfidence on ill-structured
decisions made
by managers,
such as
product
introductions. We
explored
this
gap
in
the literature and
found,
in a
study
of
high-technology firms,
that overconfidence was
positively
related to the
degree
to which
product
introductions were
pioneering (risky).
Further,
managers introducing pioneering products
were more
apt
to
express
extreme
certainty
about
achieving
success,
but these
products
were less
likely
to achieve
success.
Scholars know little about how
managers
select
risky
action,
although
such decisions are the es-
sence of
strategic
choice
(Palmer
&
Wiseman,
1999).
Some research evidence indicates that indi-
viduals'
perceptions
of
risky
situations
may
ex-
plain why they engage
in
risky
behavior
(e.g.,
Sit-
kin &
Weingart, 1995).
Busenitz and
Barney
(1997)
demonstrated that the
perceptual
bias of overcon-
fidence
may
be associated with
entrepreneurial
be-
havior. Camerer and Lovallo
(1999)
greatly
refined
understanding
of the role of overconfidence in
risky
decision
making through
a
rigorous
labora-
tory study.
In this simulated
environment,
they
found that overconfidence was associated with the
risky
action of
entering
a new market. While their
study, along
with Busenitz and
Barney's
work,
highlighted
the
importance
of overconfidence in
risky
decision
making,
Camerer and Lovallo noted
that
"experimental
data are
hardly
conclusive evi-
dence that overconfidence
plays
a role in actual
entry
decisions
by
firms"
(1999: 307).
They
stated
that a critical "scientific
payoff"
would come when
these
relationships
could be observed in the field.
To
date, however,
no studies have examined the
role of overconfidence in actual
entry
decisions,
such as
product
introductions.
Overconfidence occurs when an individual's cer-
tainty
that his or her
predictions
are correct ex-
ceeds the
accuracy
of those
predictions (Klayman,
Soll,
Gonzalez-Vallerjo,
& Barlas, 1999).
Although
dozens of
laboratory
studies have documented the
prevalence
of
overconfidence, few scholars have
explored
overconfidence in the field. The limited
field research on overconfidence
suggests
that over-
confidence is more
likely
to occur when indi-
viduals make
predictions regarding
less
repetitive
decisions. For
instance,
meteorologists
do not over-
estimate the
accuracy
of their
knowledge
when
forecasting
rain
(Murphy
&
Winkler, 1977),
but
they
are overconfident when
predicting
tornadoes
(Murphy
&
Winkler, 1982).
Field research also
sug-
gests
that
receiving
little or
ambiguous
feedback
about
prior
decisions also increases overconfi-
dence. For
example,
doctors are not overconfident
when
predicting patient mortality
in intensive care
units
(Winkler
&
Poses, 1993),
yet they
exhibit
overconfidence when
diagnosing
disease
(Chris-
tensen-Szalanski &
Bushyhead, 1981).
In the former
case, doctors
usually
know the outcomes of their
prior predictions,
but in the latter case
they
often
lack feedback
(Winkler
&
Poses, 1993).
Similarly,
those who receive
delayed
feedback have a much
more difficult time
learning
to calibrate their
judg-
ments than those who receive
rapid
feedback,
such
as
meteorologists forecasting
rain or
bridge players
making
bids
(Keren, 1991).
Although
the above field studies have
implied
that overconfidence is
greater
when a decision en-
vironment is
nonrepetitive
or
ambiguous,
all of
these studies examined
relatively
structured deci-
sion environments.
Strategic
business
decisions,
however,
occur in
extremely
ill-structured decision
We thank Irene
Duhaime,
Harry
Barkema, and three
anonymous
reviewers for their
helpful
comments on
drafts of this work.
This research was funded in
part by
an Oakland Uni-
versity
School of Business Administration Research
Grant, the Center for
Entrepreneurial Leadership,
Inc.,
and the
Ewing
Marion Kauffman Foundation. The con-
tents of this
publication
are
solely
the
responsibility
of
the authors.
139
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
140
Academy of Management Journal April
environments,
lacking
discrete
input
variables or
measurement calibration
(Roll, 1986).
In such ill-
structured decision environments as those sur-
rounding
new
product
introductions, overconfi-
dence
may
be
especially prevalent. Extending
Camerer and Lovallo's
(1999)
line of
inquiry,
we
examined overconfidence in the decision environ-
ment of new
product
introductions in the field.
Importantly,
not all new
product
introduction de-
cisions are the same. Product introductions that are
more
pioneering-that
is,
that create a distinct mar-
ket
category
(Robinson
&
Fornell, 1985)-are
riskier
than incremental
product
introductions,
because
pioneering
introductions are
unique
and decision
makers lack
prior
similar actions to
help
calibrate
judgment
(Golder
&
Tellis, 1993).
These character-
istics
suggest
that
managers may
be even more vul-
nerable to overconfidence when
deliberating
about
pioneering product
introductions than
they
are
when
considering
incremental introductions. Con-
sequently,
we examined whether overconfidence
varied with the extent to which a
product
introduc-
tion was
pioneering.
OVERCONFIDENCE AND THE EXTENT
OF PIONEERING
Overconfidence is excessive
certainty
about
one's
prediction.
One
important prediction
tool
that
managers
use when
attempting
to forecast the
success of their
strategic
actions is called a
diag-
nostic cue. A
diagnostic
cue is an indicator that is
present
most of the time
given
one outcome and
absent most of the time
given
the alternative out-
come
(Soll, 1996). Managers
must
rely
on
previous
examples
of the
diagnostic
cue to make their cur-
rent
prediction.
For
example,
when
predicting
a
product
introduction's success,
a
manager may
use
"positive
customer feedback
prior
to an introduc-
tion" as a
diagnostic
cue that has been
frequently
associated with the outcome of
"achieving
suffi-
cient demand." A
diagnostic
cue
may
be a
stronger
predictor
of success in common decision contexts
because the current decision is similar to
previous
situations. In uncommon decision contexts, how-
ever, the
diagnostic
cue
may
be a weaker
predictor
of success. We
argue
that the decision context for
considering
a
pioneering product
introduction dif-
fers
substantially
from the decision context of more
incremental product
introductions. Pioneering
product
introductions
represent
a decision context
in which the cue is
likely
to have
poor predictive
validity
because
pioneering
introductions, by
defi-
nition, lack historical
precedent.
For
example,
early
customer feedback is more weakly
associated
with the success of
pioneering products
than it is
with the success of incremental
products
because
customers are not familiar with the
product
cate-
gory
in the case of the former
(Atuahene-Gima,
1995).
Overconfidence occurs when
managers
overesti-
mate the
predictive validity
of a cue.
Managers may
accurately
assess the
predictive validity
of a
diag-
nostic cue in common decision
contexts,
such as
incremental
product
introduction decisions. Man-
agers
can recall a broad set of similar situations
when the
diagnostic
cue was and was not associ-
ated with the outcome of interest.
Therefore,
the
diagnostic
cue in an incremental context does not
lull the
managers
into overconfidence in their
pre-
dictions. The usefulness of a
diagnostic
cue breaks
down, however,
in
atypical
decision contexts,
such
as
pioneering product
introductions,
because the
managers may
not
appreciate
the
poor predictive
ability
of the cue.
Instead,
the
managers
become
overconfident because
they disproportionately
ob-
serve instances of the cue's association with
posi-
tive outcomes.
Managers
in a
pioneering
decision
context
may
notice
examples
in which the
diagnos-
tic cue
predicted
success because successful
pio-
neering
efforts often
yield
dramatic results,
such as
reinventing
whole industries. In these instances,
the cue
may
be
especially
salient and available to
the
manager (Tversky
&
Kahneman, 1974).
In con-
trast,
pioneering products
that fail are shorter lived
and more
easily
overlooked than less
pioneering
products
(Golder
& Tellis, 1993).
In addition,
peo-
ple's
natural reluctance to discuss and
publicize
stories of instances when
they
did not achieve their
goals (Levinthal
&
March, 1993) may dispropor-
tionately
affect the
availability
of accounts of times
when the
diagnostic
cue was not associated with
success in
pioneering products.
Therefore,
manag-
ers in a
pioneering
decision context
may
overesti-
mate the
predictive validity
of a cue because
they
have salient information about the instances when
the cue was associated with a
positive
outcome and
limited information about instances when the cue
was associated with a
negative
outcome,
even if the
negative
outcome was more likely.
Consequently,
we maintain that overconfidence
is
likely
to occur in
pioneering
decision contexts
more often than in incremental decision contexts.
We further
suggest
that the
presence
of overconfi-
dence, in turn, encourages managers
to
pursue
ac-
tions that are riskier than those
they might
have
pursued
without a biased
perception
of risk.
Many
studies have demonstrated the
importance
of risk
perception
in
explaining risky
action
(e.g., Hough-
ton, Simon, Aquino,
&
Goldberg, 2000; Weber &
Hsee, 1998). Scholars
studying
first movers (Lieber-
man &
Montgomery, 1988), plant expansion (Nutt,
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
2003 Simon and
Houghton
141
1993),
and innovation
(Staw, 1991)
have
proposed
that
managers proceed
with
risky
action because
they
do not
perceive
the action's riskiness
(Kahne-
man &
Lovallo, 1993).
Similarly,
we
argue
that
overconfidence is a
specific
cause of underestimat-
ing
the riskiness of a
proposed
action.
Therefore,
managers may
be more
likely
to take the
risky
ac-
tion of
introducing
a
pioneering product
introduc-
tion when
they
are overconfident. This leads us to
hypothesize:
Hypothesis
1. Greater
overconfidence
is asso-
ciated with
introducing products
that are more
pioneering
than incremental.
It is
important
to note that overconfidence
may
occur at
any given
level of
expressed certainty.
For
example,
a
manager
who does not
express
much
confidence in his or her
predictions may
still be
overconfident if the
predictions
are
consistently
wrong,
while a
manager
who is
extremely
certain of
his or her
predictions
is not overconfident if the
predictions
are
always
correct. In this
research,
we
identify
the
special
case of extreme
certainty.
Ex-
treme
certainty
occurs when an individual
per-
ceives
virtually
no
possibility
of failure.
Being
wrong
when one
expresses
extreme
certainty may
be
especially devastating
for
organizations.
Thus,
consistent with
Fischhoff, Slovic,
and Lichten-
stein's research
(1977),
our
study
focuses on the
most
dramatic,
and
potentially damaging,
case of
overconfidence,
extreme
certainty.
Although managers
in a
pioneering
decision con-
text
may
be overconfident more often than
manag-
ers in less
pioneering
decision
contexts,
they
are
not
automatically
more
likely
to
express
extreme
certainty
about their
predictions.
The
question
arises: Do
managers introducing pioneering prod-
ucts
express
extreme
certainty
more or less fre-
quently
than
managers introducing
less
pioneering
products?
On the one
hand,
managers
in
pioneering
decision contexts
might
be less
likely
to
express
extreme
certainty
than
managers
in other decision
contexts, because
diagnostic
cues are weaker
pre-
dictors of success in
pioneering
contexts-an
argu-
ment that on its face has a certain intuitive
appeal.
On the other hand, managers
in
pioneering
deci-
sion contexts
may
be
relying
on a few salient ex-
amples
of
past
situations in which a
given diagnos-
tic cue was associated with success in order to
determine their level of
certainty
about their own
prediction. They may
lack a
larger,
more
represen-
tative, set of
examples
that includes instances in
which
diagnostic cues were not associated with
success. In contrast, managers
in an incremental
product
introduction decision context
may
have a
broader set of
examples
of
diagnostic
cues and out-
comes. In this broader
set, it seems reasonable to
assume that for at least some
examples,
a
diagnostic
cue would not be associated with success
(Soll,
1996).
These
countervailing examples may temper
expectations
and
may
be less
likely
to
generate
extreme
certainty.
Empirical
studies
examining
other
types
of
risky
action
provide
some
evidence,
albeit tentative, of a
positive relationship
between extreme
certainty
and
pioneering.
Busenitz and
Barney (1997)
found
that even
though company
founders are in
very
ambiguous
environments,
they
more
frequently
ex-
press
extreme
certainty regarding
their
knowledge
than other
managers
do.
Cooper,
Woo, and Dunkel-
berg
(1988)
found that 35
percent
of
entrepreneurs
were 100
percent
confident that their new ventures
would
succeed,
even
though
new ventures fre-
quently
fail.
Similarly,
Corman, Perles, and Van-
cini
(1988)
determined that
despite
the inherent
uncertainty
that
high-technology companies
face,
two-thirds of their
managers
believed their own
firms' actions contained zero risk.
Therefore,
we
hypothesize
the
following:
Hypothesis
2. Extreme
certainty
is associated
with
introducing products
that are more
pio-
neering
than incremental.
In addition to extreme
certainty,
another variable
in our
study
is
achieving
success. This led us to
address the
relationship
between
achieving
success
and
pioneering.
Product introductions are an
evolving process,
and the root of a failure
may
be in
any part
of the
process.
For
example, managers
often encounter
technological problems
well before
an introduction is
complete.
In
fact, researchers
investigating
overconfidence have often measured
judgment
errors that occur at the start of a
process
by observing
results after the fact
(e.g.,
Winkler &
Poses, 1993). Thus,
one could even view failure to
succeed as
capturing
an initial
prediction
error as-
sociated with a decision's context. As we
argued
earlier, the actual
predictive validity
of a
diagnostic
cue
may
be weaker for
managers
in
pioneering
de-
cision contexts than it is for those in incremental
decision contexts. Yet these
managers may
still
proceed
with an introduction, not
recognizing they
are
acting
on unreliable cues. Because the cues
they
are
acting
on are in
actuality
less associated with
success, we would
expect
them to encounter failure
more
frequently. Therefore, like other researchers
(e.g.,
Golder & Tellis, 1993), we
expected
to find a
negative relationship
between
achieving
success
and
pioneering:
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
142
Academy of Management Journal April
Hypothesis
3.
Achieving
success is
negatively
associated with
introducing products
that are
more
pioneering
than incremental.
METHODS
Sample
Selection
To test the
hypothesized relationships,
we exam-
ined the
product
introduction decisions of
manag-
ers of small
companies
in the
computer industry.
We studied a
single industry
to
help
establish the
boundary
conditions needed for
theory develop-
ment
(Wiseman
&
Catanach, 1997).
We chose the
computer industry
because its
product
introduc-
tions
vary greatly
in
risk,
given
the
industry's
short
product
life
cycles
and
rapidly changing
customer
needs. Smaller firms
provide
an ideal
setting
in
which to
explore
ill-structured decision
making
be-
cause
they
have fewer formal
procedures, greater
centralization,
and lower
organizational
inertia;
these conditions maximize the effects of
manage-
rial
cognition
on
organizational
outcomes.
Using
a
comprehensive
list of
Georgia-based computer
companies,
the
Georgia Technology
Sourcebook,
we identified 135 small
computer companies
(each
with fewer than 100
employees)
that
anticipated
launching
a new
product
within 30
days
or had
just
launched a
product
within the
past
three months.
Executives in 65
(48%)
of these firms
agreed
to
participate,
and 55
(41%)
of these firms
completed
all
phases
of the
project.
The
participating
firms
had an
average age
of 13
years
and an
average
size
of 48
employees
and did not differ
from nonpartic-
ipants
in
age (p
=
.27)
or size
(p
=
.42).
In each
sampled company,
the
responding
executive
(58
percent
were at the CEO or
president
level,
and 42
percent reported directly
to their
CEOs)
was the
person judged by company managers
to be the most
influential in the focal
product
introduction decision.
Research
Design
McNamara and
Bromiley (1999) argued
that re-
searchers
may
make drastic errors in the
design
and
interpretation
of
risk-taking
models in field studies
if
they
fail to examine the issues
managers actually
attend to. Therefore, we did not ask
managers
about
their confidence in the overall success of their
product introductions, because we believed that
question
was too broad for most
managers
to an-
swer
critically. Instead, we
disaggregated
the
prod-
uct introduction decision into more discrete com-
ponents
that were salient and
important
to the
managers.
We asked
managers
to describe what
factors
they
believed were
important
in order for
their
product
introductions to succeed.
By
examin-
ing managers'
beliefs about
specific
success factors,
rather than their beliefs about the overall success of
a
product
introduction,
we
attempted
to focus on
the
finer-grained aspects
of the
managers'
decision-
making perceptions.
This
approach
is consistent
with other scholars' assertions that beliefs about
success factors are often a
prerequisite
to determin-
ing
whether an action is feasible,
and that
they may
thus
play
a crucial role in the
relationship
between
risk
perception
and action
(Mason
& Mitroff, 1981).
We first collected data from each
participating
firm around the time the
company
launched its
product
to assess the extent to which the
product
was
pioneering,
the success factors the
manager
was
focusing
on,
and the
manager's
level of cer-
tainty
about
achieving
each success factor.
Eigh-
teen months
later,
we collected data to determine
whether the new
product
introduction had
achieved the
specified
success factors.
Instrument
development
and data collection.
We
pretested
the interview
protocols
with
eight
individuals who were either
product
introduction
or
computer industry experts,
or
experts
in both
areas. We refined our
interviewing protocol
in view
of their feedback so that the
questions
were
clear,
relevant,
not
leading,
and not value-laden. For ex-
ample,
the
pretest experts uniformly
told us that
managers
of small
companies
do not
conceptualize
success in terms of
achieving
a
specific
number
(for
instance,
a 30
percent
market
share). Accordingly,
we did not
require respondents
to
provide quanti-
tative
responses
to
express
their level of
certainty
about
achieving
each success factor. Once the in-
terview
protocols
were established,
we
spent
four
hours
training
three MBA
students,
who conducted
most of the
audiotaped
interviews. The
training
focused on how to start the
interview,
clarify
inter-
view
questions, distinguish
between
complete
and
incomplete
answers,
and
probe
for more detail
without
leading
the interviewee. To further stan-
dardize the
process,
we
accompanied
each inter-
viewer on his or her first five interviews. The
interviews were conducted in the
responding
man-
agers'
own offices; each interview lasted about 30
minutes. At the time of the interview, we also col-
lected other documentation about the new
product
to assist in
determining
the extent to which
the introduction was
pioneering. Specifically,
we
talked with other
managers
in the firm, interviewed
competitors,
examined the
physical products
and
promotional materials, and
gathered externally
available market information.
Eighteen
months after we
gathered
the first
phase
of data, we mailed the
original respondents
a
ques-
tionnaire about whether or not the
product
had
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
2003 Simon and
Houghton
143
achieved certain success factors.
Fifty-five (85%)
of
the 65
original
firms
completed
this final
segment
of data collection. The
high
overall
response
rate,
as well as the fact that
nonrespondents
had had
potentially positive (having
been
acquired)
and
negative (having
failed)
outcomes minimized con-
cerns about
response
bias
(McDougall
&
Oviatt,
1996).
Additionally,
the 10
nonresponding
firms
did not differ from the rest of the
sample
in number
of
employees (p
=
.81),
firm
age (p
=
.83),
extent to
which the
product
was
pioneering (p
=
.95),
or
initial level of
certainty (p
=
.82).
Content
analysis.
The
audiotaped
interviews
from the first
phase
of data collection were tran-
scribed and
content-analyzed.
Content
analysis
is a
valid
way
to uncover and measure
underlying
de-
cision
processes
and
cognition
because the lan-
guage
an individual uses reflects his or her
cogni-
tion
(Winograd, 1983).
Researchers have used
content
analysis
to
identify
the heuristics used
by
the
managers
of innovative firms
(Manimala, 1992)
and to determine how
cognitive
biases affect deci-
sion
processes (Haley
&
Stumpf,
1989).
We fol-
lowed the
four-step process
recommended
by
(Winograd, 1983)
to ensure reliable and valid cod-
ing
of the success factors that the
managers
consid-
ered. We
repeated
this
coding process
to determine
each
manager's
level of
certainty
about
achieving
each of the success factors and the extent to which
the
product
was
pioneering.
To code
pioneering,
however,
we considered all other
product
docu-
mentation,
in addition to the focal
manager's
inter-
view statements.
In the first
step,
we decided on the size of the text
units
(for instance, word,
phrase,
sentence,
para-
graph,
and so
forth)
to
analyze. Generally,
the
smaller the text
unit,
the more reliable the
coding
but the
greater
the
potential
to miss the
phenome-
non of interest. We determined that individual sen-
tences, or
occasionally strings
of
sentences,
con-
tained discrete ideas.
Second,
on the basis of the
new
product
introduction
literature,
we
developed
a
preliminary
list of success factors
(for instance,
being
able to
develop
a
quality product). Using
a
"hold-out"
sample
of
responses,
we matched inter-
view text units to the
preliminary
list of success
factors to
clarify
classification decisions. This
pro-
cess resulted in a list of success factor
categories,
along
with extensive definition
write-ups.
The cat-
egory
definitions became the basis for the
coding
rules. In the third
step,
we used these
coding
rules
to
practice coding
with a different hold-out
sample
of interviews. We each coded the hold-out
sample
independently
and obtained
nearly identical re-
sults. We then discussed each coded text unit to
eliminate
possible ambiguity
in the
coding
rules.
The
coding taxonomy proved
to be both
compre-
hensive and reliable.
In the fourth
step,
the final
coding,
we
indepen-
dently
coded each interview and then recoded the
same
transcripts
in a different
order,
to establish
test-retest
reliability by ensuring
that there were no
coding
biases due to text unit order or time frame.
For the success factors
categories,
we attained test-
retest
reliability
of 100
percent
and interrater
agree-
ment for 95
percent
of the cases. We then followed
the
four-step procedure
to code both a
manager's
level of
certainty
with
respect
to each success factor
(test-retest
reliability
of 100
percent;
interrater
agreement
of 96
percent)
and the extent of
pioneer-
ing
(test-retest
reliability
of 100
percent;
interrater
reliability
of 96
percent).
All
coding discrepancies
were resolved
through
discussion to mutual
consensus.
Variable Measurement
Extreme
certainty.
When individuals are mak-
ing relatively
few
predictions,
or
predicting unique
events,
such as whether
they
will achieve a success
factor,
overconfidence can
only
be determined to
exist when
they express
extreme
certainty
and turn
out to be
wrong (Fischhoff
et
al., 1977).
Conse-
quently,
we used a conservative measure of cer-
tainty
in this
study by differentiating
between ex-
treme
certainty
and all other levels of
certainty.
The
response
for each success factor was coded either
"extreme
certainty" (1)
or
"not" (0).
Extreme cer-
tainty
included statements like "I have no
doubt,"
"definitely," "completely
sure,"
and
"absolutely
positive." Responses
that indicated some window
of
doubt,
such as
"maybe"
were coded 0. The final
extreme
certainty
score was the
percentage
of times
a
manager
was coded 1 for each success factor he or
she had identified.
Achieved success
factors.
To determine whether
a new
product
introduction achieved the success
factors that the
manager
associated with it had
identified,
we matched the initial success factors to
the data in the
second-phase survey
about the
prod-
uct introduction's success. We
adapted
the second-
phase survey
from
past product
introduction re-
search
(Cooper
& Kleinschmidt, 1996) to
capture
whether the
product
achieved the success factors
according
to six multi-item scales. The
average
rat-
ing
on each of five two-item scales measured
whether a firm
successfully coped
with external
technological change (a
=
.80), maintained
quality
standards (a =
.77), achieved sufficient demand
(a = .74), controlled
product-related expenses (a
=
.81), and overcame
competitive challenges (a
=
.86). The
average
of a three-item scale measured
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
144
Academy of Management Journal April
whether the firm's
product
introduction
proceeded
at an
acceptable speed
(a
=
.70).
Managers
re-
sponded
to each item on a
seven-point Likert-type
scale where 1
represented "strongly agree"
and 7
represented "strongly disagree."
For
example,
an
item on the
quality
standards scale read "The
prod-
uct was
bug-free"
(reverse-coded).
Any
score lower
than
4,
the
midpoint
on the
scale,
indicated that to
some
degree
the
respondent perceived
that the firm
had failed to achieve success in that factor. The
final measure was the
percentage
of the success
factors
originally
mentioned
by
the
respondent
that
were
subsequently
achieved. Table 1
provides
descriptive
information about the success factor
categories.
Overconfidence.
We measured overconfidence
by
first
comparing
the extreme
certainty
measure to
the achieved success factor measure for each suc-
cess factor the
respondent
identified. An individ-
ual was overconfident on
any given
success factor
if he or she received an extreme
certainty
score of 1
but the
subsequent
achieved success factor score
was below 4
(indicating
lack of
success). Next,
the
composite
overconfidence measure was calculated
as the
percentage
of success factors for which each
respondent
was overconfident.
Extent
of pioneering.
To assess the riskiness of a
firm's action,
we measured the extent to which the
product
introduction was
pioneering.
Most risk
measures in field studies use
corporate-level
stock
market and
accounting
data.
By aggregating
the risk
entailed in
multiple
actions, however,
those stud-
ies mask the
relationship
between a
manager's
de-
cision
process
and the
taking
of actions that
vary
in
riskiness
(McNamara
&
Bromiley,
1997).
We used a
finer-grained
measure of risk that isolates the rela-
tionship
between
components
of a
manager's
deci-
sion
process
(that is,
the different success
factors)
and
taking
a
strategic
action that varies in riskiness
(that is,
introducing
a new
product).
We measured
the extent to which each new
product
was market-
pioneering using
the
coding procedure
described
above. Our
pioneering coding categories,
which
were based on other studies of
pioneering (e.g.,
Golder & Tellis, 1993), included the
following:
how
the
product
differed from
competitive offerings,
what distribution channels the
manager planned
to
use,
which customers the
product targeted,
how
those customers
currently
were
meeting
their
needs,
and when other
companies
introduced
products
within the same market
category.
We oc-
casionally
encountered
apparently conflicting
evi-
dence,
but we were able to resolve the contradic-
tions
simply by asking
a few
follow-up questions.
The final
coding ranged
from
1,
"not at all
pioneer-
ing,"
to
4,
"very pioneering."
For
example,
one
company
in our
sample,
which
we call MotorView,
introduced a
pioneering
new
product
that was coded 4. The
product
was a
digital
video
imaging system
for
monitoring heavily trav-
eled street intersections in cities with
high
accident
rates that can
electronically
transmit vehicle iden-
tification information for
ticketing
motor vehicle
violations. At the time of the
introduction,
the firm
had no U.S.
competitors,
and
only
two
European
companies provided
a service that was
remotely
similar;
the
European systems,
however,
relied on
camera film
technology requiring
remote film
pro-
cessing
and
handling.
In
contrast,
another
company,
which we call
Bricklayer,
had a
nonpioneering
new
product
in-
troduction that was coded 1. It introduced a com-
bination software and hardware
product
that acts
as a firewall between an untrusted network of com-
puters
and a trusted network of
computers.
This
product
monitors the data traffic
going
into and
coming
out of the customer's trusted
network,
keeping
unauthorized users out of the network,
not
allowing any computers
in the network to interact
with untrusted
computers.
There are 28 other com-
panies
that have firewall
products
and 11 that have
products
in the class of the
Bricklayer product.
Thus,
the
product
is not
pioneering
because the
firewall
concept
is well known in that
marketplace.
Control variable. We controlled for risk
propen-
sity,
which refers to one's
general tendency
to take
or avoid risk
(Sitkin
&
Weingart,
1995),
because it
may
be associated with
taking
a
specific risky
ac-
tion. The
respondents completed
a risk
propensity
questionnaire approximately
one week before the
initial interview to reduce
any potential
influence
of the risk
questions
on the
responses
to other
seg-
TABLE 1
Descriptive
Data
Regarding
Success Factors
Achieved Maintained
Coped
with Overcame Controlled Proceeded at
Satisfactory Quality Technological Competitive
Product Acceptable
Success Factors Demand Standards
Change Challenges Expenses Speed
Percent mentioned 53% 18% 12% 7% 6% 4%
Percent achieved 60% 38% 59% 60% 80% 63%
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
2003 Simon and
Houghton
145
ments of the research. The instrument was a six-
item
(a
=
.74),
five-point
Likert scale
adapted
from
Busenitz and
Barney
(1997).
Analysis
This
study
tested the
hypotheses using
hierarchi-
cal
multiple regression.
To examine whether over-
confidence was associated with
products
that are
more
pioneering
than incremental
(Hypothesis 1),
in
step
1 we
regressed
the control variable,
risk
propensity, against
the criterion
variable,
extent of
pioneering. Step
2 then added overconfidence to
the
equation.
To test
Hypotheses
2 and
3, we en-
tered risk
propensity
and,
in
step
2,
added extreme
certainty
and the achieved success factors.
RESULTS
Table 2
displays
the
descriptive
statistics of the
variables. As
expected,
overconfidence is
highly
correlated with both extreme confidence and the
achieved success
factors,
suggesting
that we were
correct to include overconfidence in a
separate
equation.
Table 3
presents
the result of the hierar-
chical
regression equations.
The results of model 2
indicate that
greater
overconfidence is associated
with
products
that are more
pioneering
than incre-
mental. The results of model 3 confirm that ex-
treme confidence
(Hypothesis
2)
and the achieved
success factors
(Hypothesis
3)
are both associated
with
pioneering
introductions in the directions
predicted.
To ensure that the
significance
of
Hy-
pothesis
3 was not an artifact of our use of a dichot-
omous measure of success
(that is,
any rating
lower
than 44 indicated failure on the
seven-point
scales
measuring
the achievement of
success),
we also
tested the
relationship using
the
average
of the
participants' responses
to these six scales. Results
obtained
using
the
averaged
metric
(3
=
-.32, p
<
TABLE 2
Means,
Standard
Deviations,
and Correlationsa
Mean s.d. 1 2 3 4
1.
Pioneering
2.04 1.15
2. Overconfidence 0.22 0.29 .24*
3. Extreme
certainty
0.49 0.39 .30** .48***
4. Achieved success factors 0.56 0.35
-.17'
-.53*** .13
5. Risk
propensity
3.70 0.59 .21t -.07 .07 .13
a
n
=
55; one-tailed test results are
reported.
p
< .10
*
p
< .05
**
p
< .01
***
p
< .001
TABLE 3
Results of Hierarchical
Regression Analysisa
Pioneering
Predictor Variable Model 1 Model 2 Model 3
Step
1: Control
Risk
propensity 0.21'
0.23* 0.22*
Step
2: Predictors
Overconfidence 0.26*
Extreme
certainty
0.31 **
Achieved success factors -0.24*
AR2,
change
from model 1 .07 .14
AF,
change
from model 1 3.84* 4.16**
R2 .04 .11 .18
Adjusted R2 .03 .08 .13
F 2.34' 3.15* 3.65**
a
n =
55; standardized coefficients and one-tailed test results
are
reported.
tp < .10
*
p
< .05
**
p
< .01
.01)
indicated the
relationship
was consistent
across different
aggregation
measures.
DISCUSSION
Hundreds of
laboratory
studies on
cognitive
bi-
ases,
including
overconfidence,
have
greatly
added
to
knowledge
about the
pervasiveness
of these de-
cision
processes (Hogarth,
1980).
These studies
have demonstrated that the extent of biases can be
extremely
sensitive to decision
context,
leading
scholars to
investigate
the
predominance
of
system-
atic
judgment
errors outside of the
lab,
in natural
decision
settings. Accordingly,
researchers have
examined the
predictions
of TV
game
show
guests
(Berk,
Hughson,
&
Vandezande, 1996),
real estate
agents
(Northcraft
&
Neale, 1987),
commercial
bankers
(McNamara
&
Bromiley, 1999),
auditors
(Ashton
&
Hubbard-Ashton, 1990),
and
profes-
sional
bridge players (Keren, 1991),
and found
mixed results across these
applied
situations. In
each of these field
studies, however,
the research-
ers have studied
relatively
structured decisions.
Although
these
investigations
of structured situa-
tions have broadened
understanding
of the choice
behavior,
cognitive
bias scholars have
argued
that
ill-structured decisions
represent
the most interest-
ing
and
important judgments
of choice
(e.g.,
Soll,
1996).
Consistent with
this, McNamara and Bromi-
ley (1997)
suggested
that ill-structured decision sit-
uations
may
be the most common arena for the bias
of
overconfidence,
yet
no research
published
to
date has examined this
relationship
in the field.
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
146
Academy of Management Journal April
The current
study
addressed this
gap
with a field
study
of the
prevalence
of overconfidence in one
type
of ill-structured decision
situation,
introduc-
ing
a new
product.
In
addition,
we examined over-
confidence across a
range
of
product
introduction
decisions,
varying
from lower-risk, incremental,
product
introductions to
higher-risk, pioneering,
introductions. We found that overconfidence was
associated with
introducing products
that were
more
pioneering
than incremental. We also found
that
managers
who were
extremely
certain that
they
would achieve certain success factors were more
likely
to introduce
pioneering
rather than incre-
mental
products. Finally,
as
predicted, achieving
the measured
aspects
of success was less associated
with
pioneering product
introductions than with
incremental introductions.
This
study
makes a
significant
contribution to an
ongoing methodological
debate in the
cognitive
bias literature. Some scholars have
argued
that ex-
perimental designs may
overstate the
presence
of
bias because the studies lack
meaningful monetary
incentives,
which decreases
subjects'
motivation to
act
rationally (Schwarz, 1994). Furthermore,
labo-
ratory settings may provide subjects
with
espe-
cially
salient
signals
that would not occur in more
natural environments
(Schwarz, 1994). Others,
however,
have countered that the
uncertainty,
com-
plexity,
and stress associated with
making
actual
strategic
decisions
might
increase the
presence
of
cognitive
biases
(e.g.,
Duhaime &
Schwenk, 1985).
These scholars have asserted that
many
real-life
situations are like the
laboratory settings, forcing
people
to make decisions without the benefit of
experience,
and
therefore,
the bias
(for
example,
the
primacy
effect)
that is found in the lab
may
also
occur in actual
organizations.
Our
study provides
some
insight
into this debate
by confirming
that
overconfidence occurs in actual
strategic
decision
situations that are ill-structured. Further,
our
study
highlights
a
pattern
of an increase in overconfi-
dence as the riskiness of the decision context in-
creases. In more
general
terms,
the research illus-
trates the benefits of iterative alternation between
the
laboratory
and field
by using
well-established
laboratory findings
to examine
phenomena
in nat-
ural environments. We also believe that
returning
to a
laboratory design may
be the best next
step
for
teasing
out the causal directions of the relation-
ships
that we found.
Similarly,
an
experimental
design may
be the best
approach
for
isolating
the
underlying dynamics
of actual
diagnostic
cues and
their
relationship
to levels of confidence (Soll,
1996).
In addition to
contributing
these
methodological
findings,
this
study
also sheds some
light
on risk-
taking
behavior in the field. Some researchers have
concluded that individuals decide to take certain
high-risk
actions because
they perceive
them to be
less
risky
than others do
(e.g.,
Sitkin &
Weingart,
1995).
With a few
exceptions (e.g.,
McNamara &
Bromiley,
1997), however,
field research has not
measured the
accuracy
of
perceptions
of risk. Our
findings suggest
that
managers taking
riskier ac-
tions are too certain
they
will achieve success and
thereby
underestimate risk. We also extend
previ-
ous studies
by comparing
the same
phenomenon
across
varying
levels of risk. To
date,
studies have
examined biases at a
single
level of risk
(e.g.,
Si-
mon,
Houghton,
&
Aquino,
2000).
The current
study
overcomes this
limitation,
indicating
that
overconfidence and extreme
certainty play
a
greater
role in riskier actions than in less
risky
ones.
Future research should build on the
findings
of
this
study
to address its limitations. First,
as is
typical
of
many organizational
studies of risk tak-
ing
(Palmer
&
Wiseman, 1999),
our research did not
address the studied
managers' perceptions
of re-
turn. Because the link between risk
taking
and re-
turn is
complex (Jia, Dyer,
&
Butler, 1999;
Weber &
Milliman, 1997)
and often
depends
on
context,
ex-
amining
this
relationship
was
beyond
the
scope
of
this
study,
but it should be
incorporated
into future
research.
Second,
we examined smaller firms in
one
industry,
a
sampling pattern
that limits the
study's generalizability.
Third,
the
project's sample
was
relatively
small,
primarily
because of the
heavy
demands of
studying
overconfidence in the
field,
which
required respondents
to
participate
in
interviews,
provide
archival information,
and com-
plete
instruments 18 months later.
Despite
the
small
sample,
the
study
detected the
hypothesized
associations,
suggesting
that our
findings may
be
robust.
We also
guarded against retrospective
bias
(Hu-
ber &
Power, 1985)
by interviewing
the
managers
close to the time of the
product
introduction,
inter-
viewing
the most
knowledgeable person
involved
in each decision, guaranteeing confidentiality,
and
making
the
study's
results useful to the
respon-
dents. In our
study, any retrospective
bias would
have reduced the
presence
of overconfidence in
two
ways. First, managers
who had
already
en-
countered
problems
with the
emerging product
in-
troduction of concern
might
be reluctant to state
they
had been certain, at the time of the decision,
that the
product
could achieve the success factors.
Second, managers might
not admit
they
failed to
achieve a success factor, if 18 months earlier
they
had stated that
they
were certain
they
would
achieve the success factor. We minimized the
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
2003 Simon and
Houghton
147
chance of a
spurious relationship
between
certainty
and
achieving
the success factors
by measuring
them
differently, collecting
data on each
variable
at
times 18 months
apart,
and not
reminding
the man-
agers
of their
responses
to earlier
segments
of the
research.
Limitations
notwithstanding,
the
study
has some
important managerial implications.
While overcon-
fidence
may
facilitate a
society's
economic
progress
by spurring experimentation
(Levinthal
& March,
1993),
it can lead
many
individual firms down
"pathways
to disaster" and to ultimate failure
(Barnes, 1984).
Given the
importance
of the various
aspects
of success examined here to
product
intro-
ductions,
managers
who are overconfident of
achieving
them
may
be
especially likely
to hurt
overall
performance. Although
overconfidence is
difficult to
control,
researchers have
suggested
some
decision-making
tools to
help managers
guard against
overconfidence.
Schweiger,
Sand-
berg,
and
Ragan
(1986)
highlighted
several
group
decision-making techniques,
such as devil's advo-
cacy
and dialectical
inquiry,
that
may help
mini-
mize individual-level biases
during
decision mak-
ing.
Also,
Winkler and Poses
(1993)
suggested
that
individuals
may
limit their own overconfidence
by
writing
down all the reasons
supporting
their
pre-
dictions and all the reasons
disconfirming
them.
It is
important
to note that we are not
necessarily
recommending
that
managers
decrease risk
taking,
but instead are
suggesting
that
managers
should not
rely
on
inappropriately
extreme
certainty
to facili-
tate the decision
process,
because such reliance can
lead to
overcommitting up-front
resources,
hinder-
ing willingness
to monitor
assumptions,
and
failing
to make needed
adjustments (Sykes
&
Dunham,
1995). Instead,
organizations
should build
appro-
priate
cultures and rewards
systems
that facilitate
informed risk
taking
and
learning.
Taken as a
whole,
the current research indicates
that
considering
certain
types
of actions
may place
managers
in a decision environment that increases
their levels of
certainty,
which,
in
turn,
encourages
action.
Ironically,
it
may
be more difficult to
achieve success in these environments. It is our
hope
that these
findings
will
spur
additional field
and
laboratory studies, thereby generating
a
greater
accumulation of
knowledge
about the
relationship
between overconfidence and
risky
action in differ-
ent decision contexts.
REFERENCES
Ashton, R.,
&
Hubbard-Ashton, A. 1990. Evidence-
responsiveness
in
professional judgment:
Effects of
positive
versus
negative
evidence and
presentation
mode.
Organizational
Behavior and Human Deci-
sion
Processes,
46: 1-19.
Atuahene-Gima, K. 1995. An
exploratory analysis
of the
impact
of market orientation on new
product perfor-
mance: A
contingency approach. Journal of
Product
Innovation
Management,
12: 275-293.
Barnes, J.
1984.
Cognitive
biases and their
impact
on
strategic planning. Strategic Management Journal,
5: 129-137.
Berk, J., Hughson,
E., & Vandezande,
K. 1996. The
price
is
right,
but are the bids? An
investigation
of rational
decision
theory.
American Economic
Review,
86:
954-970.
Busenitz,
L. W., &
Barney, J.
1997. Biases and heuristics
in
strategic
decision
making:
Differences between
entrepreneurs
and
managers
in
large organizations.
Journal of
Business
Venturing,
12: 9-30.
Camerer, C., & Lovallo, D. 1999. Overconfidence and
excess
entry:
An
experimental approach.
American
Economic
Review,
89: 306-318.
Christensen-Szalanski, J.,
&
Bushyhead, J.
1981.
Physi-
cians' use of
probabilistic
information in a real clin-
ical
setting. Journal of Experimental Psychology:
Human
Perception
and
Performance,
7: 928-935.
Cooper,
A.
C., Woo, C. Y., &
Dunkelberg,
W. C. 1988.
Entrepreneurs' perceived
chances for success.
Jour-
nal
of
Business
Venturing,
3: 97-108.
Cooper,
R.
G.,
&
Kleinschmidt,
E.
J.
1996.
Winning
busi-
nesses and
product development:
Critical success
factors. Research
Technology Management, 39(4):
18-32.
Corman, J., Perles, B.,
&
Vancini, P. 1988. Motivational
factors
influencing high-technology entrepreneur-
ship. Journal of
Small Business
Management,
26(1): 36-42.
Duhaime, I.,
&
Schwenk, C. R. 1985.
Conjectures
on
cog-
nitive
simplification
in
acquisition
and divestment
decisions
making. Academy of Management
Re-
view,
10: 287-295.
Fischhoff, B., Slovic, P., &
Lichtenstein, S. 1977. Know-
ing
with
certainty:
The
appropriateness
of extreme
confidence.
Journal of Experimental Psychology:
Human
Perception
and
Performance, 3: 552-564.
Golder, P. N., & Tellis, G. J. 1993. Pioneer
advantage:
Marketing logic
or
marketing legend. Journal of
Marketing Research, 20: 158-170.
Haley, U., &
Stumpf,
S. A. 1989.
Cognitive
trails in stra-
tegic decision-making: Linking
theories of
personal-
ities and
cognitions. Journal of Management Stud-
ies, 26: 477-497.
Hogarth,
R. M. 1980.
Judgment
and choice: The
psychol-
ogy of
decision. New York:
Wiley.
Houghton, S., Simon, M., Aquino, K., &
Goldberg,
C.
2000. No
safety
in numbers: Persistence of biases
and their effects on team risk
perception
and team
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
148
Academy of Management Journal April
decision
making. Group
and
Organization
Man-
agement:
An International
Journal.
25: 325-353.
Huber, G.,
&
Power,
D. 1985.
Retrospective reports
of
strategic-level managers:
Guidelines for
increasing
their
accuracy. Strategic Management Journal,
6:
171-180.
Jia, J. M.,
Dyer, J. S.,
&
Butler, J.
C. 1999. Measures of
perceived
risk.
Management
Science,
45: 519-532.
Kahneman, D.,
&
Lovallo,
D. 1993. Timid choices and
bold forecasts: A
cognitive perspective
on risk tak-
ing. Management
Science, 39(1):
17-32.
Keren, G. 1991. Calibration and
probability judgments:
Conceptual
and
methodological
issues. Acta
Psy-
chologica,
77: 217-273.
Klayman, J.,
Soill,
J., Gonzalez-Vallejo, J.,
& Barlas,
S.
1999. Overconfidence: It
depends
on how, what, and
whom
you
ask.
Organizational
Behavior and Hu-
man Decision Processes,
79: 216-247.
Levinthal, D.,
& March, J.
1993. The
myopia
of
learning.
Strategic Management Journal,
14: 95-112.
Lieberman, M., &
Montgomery,
D. 1988. First-mover ad-
vantages. Strategic Management Journal,
9: 41-58.
Manimala,
M. 1992.
Entrepreneurial
heuristics: A com-
parison
between
high PI (pioneering-innovative)
and
low
PI
ventures.
Journal of
Business
Venturing,
7:
477-504.
Mason, R.,
& Mitroff, I.
1981.
Challenging strategic
planning assumptions.
New York:
Wiley.
McDougall,
P.
P., & Oviatt,
B. M. 1996. New venture
internationalization: A
follow-up study. Journal of
Business
Venturing,
11: 23-41.
McNamara, G., &
Bromiley,
P. 1997. Decision
making
in
an
organizational setting: Cognitive
and
organiza-
tional influences on risk assessment in commercial
lending. Academy of Management Journal,
40:
1063-1088.
McNamara, G.,
&
Bromiley,
P. 1999. Risk and return in
organization
decision
making. Academy of
Man-
agement Journal,
42: 330-339.
Murphy,
A.
H.,
& Winkler,
R. L. 1977.
Reliability
of
subjective probability
forecasts of
precipitation
and
temperature. Applied Statistics, 26: 41-47.
Murphy,
A. H., & Winkler, R. L. 1982. Subjective proba-
bilistic tornado forecasts: Some
experimental
re-
sults. Monthly
Weather Review, 110: 1288-1297.
Northcraft, G., & Neale, M. 1987.
Experts, amateurs, and
real estate: An anchoring-and-adjustment perspec-
tive on
property pricing
decisions. Organizational
Behavior and Human Decision Processes, 39:
84-97.
Nutt, P. C. 1993. Flexible decision styles
and the choices
of
top
executives. Journal of Management Studies,
30: 695-721.
Palmer, T., & Wiseman, R. 1999. Decoupling
risk taking
from income stream
uncertainty:
A holistic model of
risk.
Strategic Management Journal,
20: 1037-1062.
Robinson, W. T.,
& Fornell,
C. 1985. Sources of market
pioneer advantages
in consumer
goods
industries.
Journal of Marketing
Research,
22: 305-317.
Roll,
R. 1986. The hubris
hypothesis
of
corporate
take-
overs.
Journal of Business,
59: 197-217.
Schwarz,
N. 1994.
Judgment
in a social context: Biases,
shortcomings,
and the
logic
of conversation. In M. P.
Zanna
(Ed.),
Advances in
experimental
social
psy-
chology,
26: 123-162.
Schweiger,
D.,
Sandberg,
W., &
Ragan, J.
1986.
Group
approaches
for
improving strategic
decision
making:
A
comparative analysis
of dialectic
inquiry,
devil's
advocacy,
and consensus.
Academy of Manage-
ment
Journal,
5: 111-128.
Simon, M.,
Houghton,
S.
M.,
&
Aquino,
K. 2000.
Cogni-
tive biases, risk
perception,
and venture formation:
How individuals decide to start
companies. Journal
of
Business
Venturing,
15: 113-134.
Sitkin,
S.
B.,
&
Weingart,
L. R. 1995. Determinants of
risky decision-making
behavior: A test of the medi-
ating
role of risk
perceptions
and
propensity.
Acad-
emy of Management Journal,
38: 1573-1592.
Soll, J.
1996. Determinants of overconfidence and mis-
calibration: The roles of random error and
ecological
structure.
Organizational
Behavior & Human De-
cision Processes,
65: 117-137.
Staw, B. M. 1991.
Dressing up
like an
organization:
When
psychological
theories can
explain organizational
action.
Journal of Management,
17: 805-819.
Sykes,
H., & Dunham,
D. 1995. Critical
assumption plan-
ning:
A
practical
tool for
managing
business devel-
opment
risk.
Journal of
Business
Venturing,
10:
413-424.
Tversky,
A.,
& Kahneman,
D. 1974.
Judgment
under un-
certainty:
Heuristics and biases. Science,
185: 1124-
1131.
Weber, E. U.,
& Hsee, C. 1998. Cross-cultural differences
in risk
perception,
but cross-cultural similarities in
attitudes towards
perceived
risk.
Management
Sci-
ence,
44: 1205-1217.
Weber, E. U., & Milliman, R. 1997. Perceived risk atti-
tudes: Relating
risk
perception
to risky
choice. Man-
agement Science, 43: 122-143.
Winkler, R., & Poses, R. 1993. Evaluating
and combining
physicians' probabilities
of survival in an intensive
care unit.
Management Science, 39: 1526-1543.
Winograd,
T. 1983. Language
as a cognitive process.
Reading,
MA: Addison-Wesley.
Wiseman, R. M., & Catanach, A. H., Jr. 1997. A
longitu-
dinal disaggregation
of
operational
risk under chang-
ing regulations:
Evidence from the
savings
and loan
industry. Academy of Management Journal, 40:
799-830.
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions
2003 Simon and
Houghton
149
Mark Simon
([email protected])
is an associate
pro-
fessor of
strategic management
at Oakland
University.
He
received his Ph.D. from
Georgia
State
University.
His
research interests include risk
perception
and
cognitive
biases as
they
relate to
proactive
firm behaviors.
Susan M.
Houghton ([email protected])
is an as-
sociate
professor
of
strategic management
at the
J.
Mack
Robinson
College
of Business,
Georgia
State
University.
She received her Ph.D. from the
University
of North
Carolina at
Chapel
Hill. Her research focuses on the
cognitive aspects
of
strategic
decision
making
and exec-
utive team
cognitive processes.
This content downloaded from 111.68.100.252 on Wed, 4 Dec 2013 00:39:21 AM
All use subject to JSTOR Terms and Conditions