How Believing in Ourselves Increases

Risk Taking: Perceived Self-Efficacy

and Opportunity Recognition
Norris Krueger, Jr.
Entrepreneurial Strategies, Bozeman, MT 59715
Peter R. Dickson
College of Business, The Ohio State University, Columbus, OH 43210

ABSTRACT
What effect does positive and negative feedback about past risk taking have on the future
risk taking of decision makers? The results of an experimental study show that subjects
who are led to believe they are very competent at decision making see more opportunities
in a risky choice and take more risks. Those who are led to believe they are not very
competent see more threats and take fewer risks. The feelings of self-competence and
self-confidence on one task did not generalize to a similar task. Perception of opportunities
was unexpectedly not related to the perception of threats. As executives bring their personal
perceptual biases to firm decision making, our results identify a serious built-in bias in
SWOT analysis (the analysis of firms’ strengths and weaknesses as related to potential
opportunities and threats). Executives who believe that they and their firm are very competent
will take more risks and vice versa. Our results also provide evidence that the perceived
likelihood of an event depends on whether the event is a loss or a gain. Human decision
making is subject to the general bias that outcome expectations are not independent of
outcome valuations.
Subject Areas: Risk and Uncertainty, Strategy and Policy, and Utility Theory.

INTRODUCTION
In a global marketplace full of resourceful imitators who have a ready supply of
skilled, highly motivated labor and increasing access to capital, American managers
and firms must compete by taking more risks with new, unproven product designs
and manufacturing processes. Despite this brutal economic reality, there has been
extraordinarily litlie research by social scientists, management scientists, or economists
into how managers and owners of firms see opportunities and threats in decisions
and how this affects their product and process risk taking. The current era demands
continuous improvement in products and processes and, hence, continuous experimen-
tation and risk taking [7]. The study of the determinants of such experimentation
and risk taking is therefore particularly timely.
Scholars have known for a long time that risk taking is not based on rational
calculations alone [S], with even the great Lord Keynes, a highly successful speculator
himself, arguing that taking risks requires optimism and what he quaintly, if obscurely,
described as “animal spirits” [4]. Yet, numerous field studies of managers’ and

385
Decision Sciences
Volunte 25 Number 3
Primed in the U.S.A.
386 Self-Eflicacy and Opportunity Recognition

entrepreneurs' personality differences have found that such measures of animal

spirits are weak predictors of risk taking [6] [18].
This paper takes a somewhat different approach to optimism and animal spirits
by studying what might be loosely described as believing in ourselves. Specifically,
we propose and test a model that relates changes in perceived self-efficacy (the
perceived ability to make good decision choices) to changes in risk taking. Bandura
& Wood found that managers with high perceived self-efficacy tend to view
setbacks as learning experiences and thus persevere. Managers with self-doubts
become preoccupied with the risk of failure:
Individuals who believe themselves to be inefficacious are likely to effect
limited change even in environments that provide many opportunities.. . .
They dwell on their personal deficiencies and envision failure scenarios
that beget adverse consequences. [2, p. 8061
This observation suggests that perceptions of opportunity and threat my help
explain the relationship between perceived self-efficacy and risk taking. Furthermore
it raises a serious concern with SWOT, the analysis and fitting of a firms' strengths
and weaknesses to market opportunities and threats. What happens when decision
makers lose confidence in themselves and their firm? Do the weaknesses and
threats dominate, making the manager more risk-averse and, in a world where
competitiveness depends on risk taking, less competitive? In a controlled experimental
study, we explore how changing decision makers' beliefs about their decision
making skills and abilities changes their risk taking and their perceptions of the
presence of threats and opportunities in the decisions they make.

LITERATURE REVIEW
Perceived Self-Efficacy
Managers' behaviors and decisions may have little to do with actual skills and
resources. Instead, how we think and behave is linked far more closely to our own
perceptions of a situation and our perceived competence to control processes and
outcomes in that situation, especially where performance requires persistence.
Measures of perceived self-efficacy show a significant ability to explain and
predict a wide variety of behaviors important to managers, including those featuring
high uncertainty and high stakes such as entrepreneurship [ZO].Perceived self-efficacy
significantly influences aspiration levels, goal commitment, task persistence, and
work attitudes [121. For instance, perceived self-efficacy has proven central to
important practical managerial techniques such as goal-setting and performance
feedback [16]. Managers with high perceived self-efficacy set higher goals and tend
to exceed their goals; managers with low perceived self-efficacy set lower goals
and tend to underachieve [2]. Another important practical aspect is that increased
perceptions of self-efficacy are readily teachable and these increased perceptions
of self-efficacy persist over time [12].
Decision-Making Behavior
Decision making understandably attracts research interest because of its importance
and because of the challenges it presents to researchers [18]. We still have difficulty
explaining (let alone predicting) decisions made under high uncertainty [ 111 [25].
Krueger and Dickson 387

Models of risk taking originally emphasized individual differences in personality

or preferences as reflected in utility curves. Such models offer explanations of risk
taking and yet are quite limited in their ability to predict. For instance, decision
making depends on ambiguity as well as quantifiable risks 1113, suggesting that
perceptions (e.g., of ambiguity) are critical. More recent models focus on the
situation, in keeping with evidence of large effects arising from the context of the
situation [5] [19] 1221.For instance,Kahneman & Tversky’s prospect theory hypothesizes
that we tend to be risk-averse in situations where the status quo is positive, but
risk-seeking where the status quo is negative [15]. Even if utility functions for
individuals do exist, they can differ widely across individuals and situations: is
gaining $100 through a bequest from a beloved relative as valuable as winning
$100 in poker from an enemy?
Entrepreneurialrisk taking is also very situation-specific [18] [19] [24]. Entrepreneurs
may take sizable risks, even “bet the farm,” yet they do not demonstrate above-average
general propensities to take risk. Entrepreneurs see themselves as above-average
risk takers in their business decision making [19], though they do not exhibit
above-average risk taking in paper-and-pencil decision making exercises [24]. Per-
ceived self-efficacy varies across individuals and situations [121 and risk taking
appears greater in situations where subjects should believe they have greater personal
competence [14]. If we appear to take more risks where we feel more competent,
could risk taking be situation-specific because perceived self-efficacy is situation-
specific?

Attractiveness and Risk: Perceptions of Opportunity and Threat

Decision makers may differ on risk taking because they differ on interpretations
of the same cues about a decision. How managers react to a decision environment
depends on how they interpret cues from that environment. Managers tend to
categorize decision situations into opportunitiesand threats. They tend to see controllable
situations as opportunities and they tend to see uncontrollable situations as threats.
Of course, ‘controllable’ and ‘uncontrollable’ are in the eye of the beholder. Thus,
one manager may view a highly uncertain situation as a potential opportunity while
another may view the same situation as a potential threat [9]. In fact, a given
decision situation may offer cues suggesting that it is both attractive (an opportunity)
and risky (a threat) [17]. This opportunitylthreat framing differs from Kahneman
& Tversky’s [15] sure gain/sure loss framing, but offers some interesting implications
for prospect theory that we will discuss later.
Entrepreneurial managers consistently tend to see more opportunities than do
bureaucrats, typically rating themselves as having above-averageperceived competence,
expressing strong confidence in their abilities and reporting a focus on opportunities
[6] [24]. They tend to believe they can reduce or overcome risk through skill, that
is, perceived self-efficacy [ 181. Those high in perceived self-efficacy should take
greater risks because they do not dwell on failure or uncontrollable threats. Instead,
they frame risky choices as opportunities that they can control and manage through
their skill. This suggests an interesting, testable model of the role of perceptions
of self-efficacy and perceptions of opportunity and threat on risk taking.
388 Self-Eficacy and Opportunity Recognition

Hypotheses
The above literature on perceived self-efficacy, perception of opportunity and
threat,and risk taking is summarized in the model presented in Figure 1. Performance
feedback should significantly influence subjects’ perceptions of self-efficacy and
thus subsequent risk taking. Self-efficacy affects risk taking through changing
situational perceptions of opportunity and threat,which in turn significantly influence
risk taking. The following hypotheses formalize these propositions.
H1:Performance feedback directly influences perceived self-efficacy.
H2: An increase in perceived self-efficacy leads to an increase in subsequent
risk taking.
H3:Perceived self-efficacy influences perceptions of opportunity and threat;
these in t u n influence subsequent risk taking.

RESEARCH DESIGN
To identify whether opportunity and threat perceptions derive from perceived self-
efficacy requires ruling out plausible alternate explanations such as mood, context
and domain effects (such as framing), and individual differences in situational
perceptions (such as familiarity effects). We addressed these in the control provided
by experimental design and through manipulation checks. Controlled experiments
are common practice in decision making research where specific, well-defined
relationships are under investigation.
One of the first concerns we addressed is that success can create good feelings
and failure creates bad feelings. Such mood changes may in turn affect perceptions
of riskiness (and thus risk taking) [l]. To address the argument that we manipulated
mood rather than perceived self-efficacy, we presented subjects with two different
batteries of decision-making tasks and manipulated perceptions of self-efficacy
independently for each. Subjects with increased self-efficacy perceptions on one
task and decreased self-efficacy perceptions on another are unlikely to have much
change in overall mood. Second, to control for possible context and domain effects
each exercise presented subjects with tasks that spanned a number of situations and
decision frames [19][22]. Finally, we collected baseline information on key individual
difference measures [5] [24] to control for any anchoring biases such as those
induced by individual differences.
Experimenters also need to take care that the research design itself does not
introduce spurious effects. For instance, to encourage subjects to take the tasks
seriously, one needs to clearly connect the reward to outcomes in subjects’ minds.
We thus informed them that they would be paid based on their rated performance.

Experimental Protocol
Under controlled laboratory conditions, 153 upper division business majors undertook
a first round of decision making and reported baseline perceptions of the tasks:
perceived self-efficacy and perceptions of opportunity and threat. The experimenter
then appeared to evaluate these ‘results’ while subjects completed a distractor task.
In a 2x2 design, we then randomly assigned subjects to receive noncontingent
Krueger and Dickson 389

Figure 1: Hypothesized model of risk taking with expected signs of relationships.

Change in
Perceived
Opportunity

Performance + Change in Change in

Feedback Perceived Risk Taking
Self-Efficacy Behavior

Change in
Perceived
Threat

feedback, either positive or negative, on the ‘quality’ of their decision making for
each of the two tasks. Subjects then completed a second pair of decision malung
exercises and again reported perceptions of self-efficacy, opportunity, and threat.
Finally, subjects answered diagnostic questions (about the feedback, changes in
mood, etc.), then we thanked, debriefed, and compensated them.
Round 1 of Decision Tasks. Small groups of subjects completed two separate
batteries of ‘dilemma’ decisions and ‘gamble’ decisions. The separate tasks allowed
a test of task specificity and helped control for mood induction effects.
The first battery consisted of eight choice dilemmas based on decision scenarios
used in previous research. Instead of simply choosing between risky and certain
options, subjects stated the probability of the risky option succeeding that made
them indifferent between the options. In keeping with typical practice, these items
assume subjects are advising a close friend who will likely accept the advice (thus
reducing personal biases). For dilemmas, higher probabilities reflect risk aversion
so we subtracted responses from 100, then summed and transformed them into an
eight point scale to match the gamble scores. (See Appendix 1 for sample items.)
The second battery contained eight gambles comparable to decisions used by
Kahneman and Tversky to test framing biases. Each asked subjects to simply
choose between a risky (or riskier) option and a certain (or less risky) option. The
measure of risk taking (0-8) was the number of times the subject chose the risky
(riskier) option.
For each battery, we asked subjects to estimate the likelihood that this type of
decision (‘dilemma’ or ‘gamble’) presented an opportunity and the likelihood that
it presented a threat. Subjects also provided measures of perceived self-efficacy.
The perceived self-efficacy measures asked subjects to provide a specific, quantified
estimate of their ability (e.g., “If you had to make 10 dilemma type questions in
10 minutes, how many good decisions would you make?”) Measures in Round 1
provided a baseline against which we could compare Round 2 measures.
Noncontingent Performance Feedback. While subjects’ first round responses
were ostensibly being evaluated by a computer program “developed by experts to
evaluate decision making skill,” they completed psychological scales that have
been observed to influence risk taking or opportunistic behavior [19] [24].
390 Self-Eficacy and Opportunity Recognition

We randomly assigned subjects to receive performance feedback on the ‘quality’

of their first round decisions. This noncontingent feedback was unrelated to actual
risk taking or performance. We informed subjects that they either had performed
“very well” (6 out of 8 correct) on dilemmas or “very poorly” (2 out of 8) with
no mention of risk or risk taking. Similar feedback indicated they either had done
well or poorly on the gambles (5 of 8 correct or 3 of 8 correct). Neither manipulation
was unrealistically extreme.
Round Two ofDecision Tasks.Following the feedback, subjects then immediately
received two further batteries of eight dilemmas and eight gambles, plus a second set
of items measuring their perceptions of task self-efficacy, opportunity, and threat.
Manipulation Check.. Supporting H1, positive performance feedback did sigmficantly
increase perceived self-efficacy, while negative feedback significantly decreased
perceived self-efficacy for both dilemmas (F= 45.4, p<.OOOl)and gambles (F= 32.8,
p < .OOOl). There was no siginificant difference in individual difference measures
between the four cells. No dilemma-related measure (including perceived self-efficacy)
was significantly associated with feedback on gambles; no gamble-related measure
was associated with feedback on dilemmas. Feedback effects were thus clearly
task-specific, eliminating a mood explanation for our result. A generally positive or
negative affective state created by the feedback would have influenced risk taking
on both tasks.
Debriefing. In answers to debriefing questions, few subjects reported a significant
change in their mood or any suspicion of the experiment’s actual intent. Almost all
subjects recalled their feedback accurately and reported at least grudging belief in the
feedback. After completing the debriefing questions subjects were fully debriefed about
the noncontingent feedback, asked to keep what they had been told confidential,
thanked, and paid $5 each.

Analysis
Conceptually, using change scores (rather than post-test scores) as the dependent
measures enables the study of how decision making evolves. Statistically, using
change scores offers the advantage of reducing the effects of initial individual
differences in perceived self-efficacy. This increases the sensitivity of the experiment,
but only if assignment to treatments was random. Manipulation checks found no
significant assignment biases.

RESULTS
Change Scores. Changes in risk taking scores were correlated significantly with
self-efficacy and perceptions of opportunity and threat. Correlational analysis
showed that changes in perceived self-efficacy are significantly associated with
subsequent changes in perceived opportunity (r= .60, p<.OOOl) and with changes
in perceived threat (r=-.63, p<.OOOl) for dilemmas. For gambles, we obtained
similar results for perceived opportunity (r=.54, p < .OOOl) and perceived threat
(r=-.28, p < . 0 0 2 ) . Interestingly, opportunity perceptions were not correlated with
threat perceptions (see Appendix 2). We might reasonably expect that changes in
opportunity and threat perceptions would be in opposite directions. Our findings
suggest decision makers consider them somewhat independent constructs [8] [ 171.
Krueger and Dickson 391

ANOVA results on cell means reported in Table 1 and illustrated in Figures 2

and 3 support both H1 and H2. Gamble feedback influenced changes in gamble
risk taking (F-4.4, p < .02) but not dilemma self-efficacy perceptions or risk taking
on dilemmas. Dilemma feedback influenced changes in dilemma risk taking
(F=10.2,p < .OOOl) but not gamble self-efficacy perceptions or risk taking on gambles.
The reported tests are whether the average change score in each cell deviated
significantly from zero.
Path Analyses. Figure 4 presents two path models derived from the second and
third hypotheses and the path coefficients (betas). Change in opportunity perception
was positively correlated with change in perceived self-efficacy while change in
threat perception was negatively correlated with change in perceived self-efficacy.
Opportunity and threat perceptions were both associated with risk taking. After
controlling for opportunity and threat perceptions, the direct path between change
in perceived self-efficacy and change in risk taking was no longer significant. That
is, the intervening changes in opportunity and threat perceptions fully mediated the
relationship between the change in self-efficacy perceptions and the change in risk
taking and did so for both gambles and dilemmas.
Task Specificity. If we look at just those subjects who received positive feedback
on one task and negative feedback on the other (see Table l), we see that risk taking
increased on the former and decreased on the latter, thus ruling out global expla-
nations such as mood and individual differences in risk taking. This supports prior
findings that risk taking is domain or context specific [19] [22].
Limitations
Managers need not react the same way as students in their risk taking, though they are
often comparable in studies using performance feedback [3]; still, future research
should include experienced managers as subjects. The decision tasks used are
well-known and accepted in their respective research domains, but it was clear to
subjects that final outcomes depended on transparently uncontrollable chance they
could not “manage.” This lack of mundane realism could raise problems but here
it happens to make our test more conservative and thus our findings more robust.
That is, if we observe the effects of perceived self-efficacy when the choice
outcome depends on chance, such effects should increase as choice outcomes
depend increasingly on skill. In fact, our subjects did behave as if slall was involved.
The model proved stronger for the dilemmas which offered a greater illusion of
managerial controllability.
Multiple measures of perceived self-efficacy, opportunity, and threat might
allay concerns over construct validity as well as measurement reliability. Again,
however, the lessened sensitivity from using single-item measures made this study
a more conservative test of the hypotheses and the path models.
DISCUSSION
Overall, the results support our hypotheses and model. We found that the influence
of perceived self-efficacy on risk taking was significant and fully mediated by
perceptions of opportunities and threats. That is, self-efficacy perceptions appear
to influence opportunity and threat perceptions, which then influence risk taking.
These results held for both tasks.
392 Self-Eficacy and Opportuniry Recognition

Table 1: Risk taking means and difference scores.

Mean Risk Taking Scores
(and Standard Deviations)
Mean Difference Scores
Gambles Dilemmas between Rounds
Gamble Dilemtna
Feedback Feedback R1 R2 R1 R2 Gambles Dilemmas n
Positive Positive 3.1 3.5 4.5 4.8 +.4a +.3 31
1.4 1.4 .8 1.0 1.9 1.o
Positive Negative 3.3 3.9 4.5 4.0 +.6a -.fia 31
1.3 1.3 .7 .6 1.5 .9
Negative Positive 3.7 3.1 4.7 4.9 -.6’ + .2 31
1.3 2.2 1.0 1.0 1.5 .9
Negative Negative 3.3 3.2 4.6 4.0 -. 1 -.6a 30
1.2 1.4 .8 .8 1.9 1.1
Overall Means 3.4 3.4 4.6 4.5 .o -.1
Bsignificantly different from zero in the hypothesized direction @< .05)

While we are confident that perceived self-efficacy influences decision behavior

through perceptions of opportunities and threats, an interesting question is how
self-efficacy influences prospects of success and failure. One obvious vehicle for
addressing this question is prospect theory.

Perceived Self-Efficacy and Prospect Theory

Kahneman and Tversky’s prospect theory [15] explains why decision makers will
take more risks when faced with a certain loss than when faced with a certain gain.
Such “desperate” risk seeking is observed in actual gambling, investment (such as
the S&L crisis), and strategic decisions. A decision’s prospect involves estimating
its expected value by weighing the gains/losses of all of the possible outcomes by
their probability of occurrence. Prospect theory explained the risk avoiding and risk
seeking that they observed by proposing that decision makers transform an objective
gain/loss into a subjective gaidloss. The value function used by a decision maker
to transform an objective gain into a perceived gain is different from the second
value function used to convert an objective loss into a perceived loss, that is, the
value function is domain specific. In general, the transformational value function
for losses is steeper for losses than for gains.
Critical to our discussion, the objective probability of an outcome is also
transformed into a subjective likelihood by a general probability function that is
not domain specific. The same general probability function that a decision maker
uses to transform the objective probability of a loss into a perceived probability is
used to transform the objective probability of a gain into a perceived probability.
We have shown the influence of perceived self-efficacy and perceptions of
opportunity and threat on choosing the risky option. Thus, changing perceived
self-efficacy either changes the general probability function or the domain-specific
Krueger and Dicbon 393

Figure 2: Effect of feedback on dilemma and gamble self-efficacy.

Changes in Dilemma
Self-Efficacy

+LO 4- .., +1.17

ongambles ,*

Feedback

-1.0

Positive feedback

+1,30 - - - - - - - - - -- - - - +1.30
on gambles

+LO --
Negative Positive
Feedback Feedback Direction
I I t ofDilemma
Feedback

-1.0 --
\
Negative feedback
-1.13
on gambles

value functions. It is our contention that an increase in perceived self-efficacy

influences prospects of success and failure by amplifying the perceived likelihood
of gains and deflating the perceived likelihood of losses. Given that perceived
self-efficacy increases expectations of success (and decreases expectation of failure),
the probability function for gains must differ from the probability function for
losses, requiring appropriate respecification of prospect theory’s expected utility
394 SelfEficacy and Opportunity Recognition

Figure 3: Effect of feedback on dilemma and gamble risk taking.

Changes in Dilemma
Risk Taking

t
+.5
t Posilwefwdbadc
ongambles *.
.. +.3
Direction
of Dilemma
Feedback

-.5

+1.0 --
Positive feedback

+.6 .--...---\.
=.
on gambles

+.4
Negative
Feedback
I
Positive
Feedback
-- Direction
ofDilemma
Feedback

on gambles
-.6

-1.0 --
I

model. The possibility that prospect probabilities are losslgain specific and, hence,
not independent of the prospect dates back at least as far as the early work of Ward
Edwards in the 1950s.
Intuitively, a professional tennis player in a confidence crisis about playing on
grass withdraws from Wimbledon because she thinks her chances of winning are
low, not because she thinks less of the prize money or the prestige. Gamblers on
Krueger and Dickson 395

Figure 4: Path diagrams of dilemma and gamble regression results. All paths are
significant @c .O1).

Change in
Perceived
Opportunity
for Dilemmas

Change in Change in
Perceived Dilemma
Self-Efficacy Risk Taking
for Dilemmas Behavior

Change in
Perceived
Threat
for Dilemmas

Change in
Perceived
+.338 Opportunity
for Gambles

Change in
Perceived
Self-Efficacy
/ Change in
Gambke
Risk Taking
for Gambles
\
-.256 Change in
Perceived
Threat
Behavior

for Gambles

losing streaks stop because they think their chances of winning are less and not
because they think any less of the winnings. In fact, given their losses they probably
think more of any possible winnings.
A perceived loss of self-efficacy can lead to a mindset that dwells on negative
outcomes, that is, the threat of losses. Such dwelling is likely to increase the
perceived availability (in Tversky & Kahneman's terms) and hence the subjective
likelihood of the loss. Thus, subjective probabilities of a loss will be inflated and,
conversely, subjective probabilities of a gain will be deflated. If perceived self-efficacy
increases expectations of success and decreases expectations of failure, the probability
function for gains must, inevitably, be different than for losses.
The only other way to explain our results is that higher perceived self-efficacy
increases the slope of the value function in the gain domain and/or decreases the
slope of the value function in the loss domain. Conventional economic theory
assumes that the shape of the value function determines the extent of risk aversion.
Perceived self-efficacy could have an observed, but spurious effect on the value
function because prior success increases both one's self-efficacy and one's cumulative
396 Self-Eficacy and Opportunity Recognition

gains. The perceived self-efficacy associated with having accumulated assets (one’s
‘endowment’) might lessen the slope of the value function in the loss domain
because the loss is seen as relatively smaller than one’s endowment. But if true,
this would also decrease the slope of the value function for gains. Moreover, such
effects cannot explain our results as our feedback effects were independent of
baseline measures of perceived self-efficacy.
For those who find the above logic uncompelling, we have direct empirical
evidence that perceived self-efficacy influences the subjective probabilities rather
than the subjective values. Second round gambles included a version of the Ellsberg
gamble [lo] that offered subjects the choice of drawing a ball from Box A containing
50 red balls and 50 black balls, or from Box B containing 100 red or black balls
(number of red and black unspecified). Given this choice, subjects tend to choose
Box A more often as it provides more certain information about the outcome. We
expected that a higher percentage of our subjects in the low perceived self-efficacy
condition would choose Box A compared to the group in the high perceived self-efficacy
condition because A is the more certain choice. The difference was significant in
the expected direction (67 percent versus 44 percent, pC.05). The interesting point
is that we cannot explain this result through an effect on the value function because
the payoffs were the same for both options. Self-efficacy perceptions must have
influenced the probability function, and influenced it more when the odds were
more uncertain (i.e., unspecified).

CONCLUSIONS
A great deal of decision making involves choosing between different alternatives
that vary in their perceived returns and in the perceived probabilities of achieving
these returns. The question we have explored is fundamental to all such decision
making: Does perceived decision making self-efficacy increase a decision maker’s
likelihood of choosing a riskier course of action, rather than a safer option?
We found support for our hypotheses that an increase in self-efficacy increases
perceptions of opportunity and decreases perceptions of threat and that changing
opportunity and threat perceptions changes risk taking. This support was found even
though our subjects’ ability to exercise control over the outcomes was transparently
minimal. Thus, it is reasonable to expect that the observed relationships will be
even stronger in situations where actors perceived that skill (efficacy) influences
outcomes. This proposition is worthy of further research.
Heath & Tversky [14] present evidence that actual efficacy increased risk
taking; we find that self-perceptions of efficacy also increase risk taking. March &
Shapira [18] suggest that successful risk takers will become more optimistic; we
find that optimistic decision makers, in turn, take more risks. That is, perceptions
of opportunity and threat fully explained the relationship between perceived self-
efficacy and risk taking.
At the same time we find that perceptions of opportunity and perceptions of
threat were unrelated. This suggests that they may not represent the poles of a
single construct underlying risk taking but rather different dimensions of a decision
that decision makers may trade off [8].
SWOT analysis, a well known strategic decision tool, recommends that a firm’s
management should match its strengths and weaknesses to appropriate opportunities
Krueger and Dickson 397

and threats in the marketplace. Our results suggest a major problem with SWOT
analysis, a bias that decreases the competitive rationality of managerial decision
making [ 6 ] . Managers who perceive that their firm has many strengths and few
weaknesses (and are consequently more confident in their decision making) will
“see” more opportunities and fewer threats in a SWOT analysis. They will take
more risks, such as choosing the harder to achieve but higher payoff prospect. The
reverse is also true. Managers who perceive that their firm has many weaknesses
and few strengths (and are consequently less confident in their decision making)
will “see” more threats and fewer opportunities in a SWOT analysis. They will
take fewer risks, such as choosing the status quo over a riskier strategy. One way
of reducing this perceptual trap is to have executives undertake analysis of external
opportunities and threats before undertaking analysis of internal strengths and
weaknesses that may affect their self-confidence and self-competence.
A more general managerial implication of our findings is the potential for a
demoralizing self-fulfilling prophecy that some large organizations seem to have
experienced. Poor results lead to a loss in self-efficacy and fatalistic resignation.
Managers focus excessively on the threats rather than the opportunities in SWOT
analyses. The perceived less risky status quo is maintained and, in a changing
market place, the situation gets even worse. The solution is to build organizational
and decision making self-confidence through leadership and, if necessary, changing
personnel. Leaders who lose their confidence, freeze up, and, by default, choose the
status quo must be replaced. New managers with higher perceived self-confidence
in needed areas are more prone to seek and accept opportunities in new directions.
Their own perceived self-confidence can also boost perceived self-efficacy
throughout the organization, empowering members and facilitating initiation of
risky organizational change and of new marketplace strategies. On the other hand,
very high perceived self-efficacy might amplify escalation of commitment to a
highly risky strategy and further explain the folly of generals and chief executives
and Miller’s “Icarus” paradox [20],which details how perceptions of competence
lead managers and their organizations to extend a risky strategy to competitively
irrational extremes.
With the current emphasis being given to continuous improvement and organizational
change, personal and organizational self-efficacy is a vital resource that should be
nurtured and carefully managed to effect such change. Building perceived self-efficacy
in organization members permits managers and organizations to be more receptive
to opportunities and to be more persistent in pursuing them.
Our suggested respecification of the expected utility model and prospect theory
is also worthy of further research. Brady and Lee [4] propose that decision weights in
a choice model should not be objective probabilities but perceptions of information
adequacy such as Keynes’ degree of completeness of information coefficient, w .
Just as we believe that perceived self-efficacy influences subjective probabilities
and this effect is gain/loss specific, perceived self-efficacy may also influence
perceptions of information adequacy and this effect is also gainlloss specific. A
proposition that information processing scholars might explore is whether managers
with low decision making self-efficacy will give more weight and credibility to
threat-related information than they do to opportunity-related infomation. This is
similar to individuals with low self-esteem accepting negative feedback as being
398 Self-Efficacy and Opportltniry Recognition

more credible. Conversely, managers with high perceived self-efficacy may place
more credence and weight on opportunity-related information.
The interactive effects of self-efficacy and framing effects such as those induced
by gain/loss framing or by endowment effects from accumulated assetslwealth also
need to be explored. Finally, such relationships need to be studied using decisions
where the choice outcomes are not always irreversible and can be changed by
continuous managerial learning.
Our results may also have implications for buyer as well as seller behavior.
Buyers with higher self-efficacy may be less brand loyal and take greater risks in
adopting new products. This effect may only occur with complex products, involving
new technology, where efficacy at using the product is an important determinant
of usage success. The implications for professional and personal financial management
are also interesting. The illusion that one has become very skilled at picking stocks
may lead to ever greater risk taking, which may ultimately be disastrous. Note also
that there are multiple dimensions or types of risk that one might perceive (financial,
personal, social, etc.) and increasing perceived self-efficacy at one type of risk need
not carry over to other dimensions [ 131.
Finally, our findings complement Schumpeter’s proposition that if managers
are more optimistic and take more risks, then society will enjoy more innovations
[23]. High perceived self-efficacy by business decision makers thus benefits both
firms and society. It may also help explain Schumpeter’s observed business cycles.
A successful dose of risk taking in an economy, such as a dramatic new technological
innovation like the automobile, creates new jobs. A second-order effect of such
successful risk taking and the ensuing economic growth is that it leads to an
increase in general business confidence and the rate of risky innovation in the
economy that further expands growth. On the downside of a cycle, an economy
declines when declining confidence leads to less risky innovative activity, which
leads to less efficiency-promoting creative destruction and to less job creation,
which in turn leads to still less confidence and still less risky innovation. [Received:
January 3, 1993. Accepted: May 16, 1994.1

REFERENCES
[ I ] Ajzen, I. Attitudes, traits and actions: Dispositional prediction of behavior in social psychology.
Advunces in Experirirenial Social Psychology, 1987, 20, 1-63.
[2] Bandura, A., & Wood, R. The effect of perceived controllability & performance standards on
self-regulation of complex decision making. Journal of Personaliv & Social Psychology, 1989,
56, 805-814.
[3] Bateman, T., & Zeithaml, C. The psychological context of strategic decisions: A model and
convergent experimental findings. Strategic Managerrrent Journal, 1989, 10, 59-74.
[4] Brady, M., & Lee, H . Dynamics of choice behavior: The logical relation between linear objective
probability and nonlinear subjective probability. Psychological Reports, 1989, 64, 9 1-97.
[5] Bromiley, P., & Curley, S . Individual differences in risk takmg. In J. Yates (Ed.),Risk raking
behavior. New York: Wiley, 1992, 87-132.
[6] Cooper, A., Dunkelberg, W., & Woo, C. Entrepreneurs’ perceived chances for success. Journal
of Business Venruring, 1988, 3, 97-108.
171 Dickson, P. Toward a general theory of competitive rationality. Journal of Marketing, 1992, 56,
69-83.
[8] Dickson, P., & Giglierano, J. Missing the boat and sinking the boat: A conceptual model of
entrepreneurial risk. Journal of Marketing, 1986, 50, 58-70.
Krueger and Dickson 399

Dutton, J. The making of organizational opportunities: Interpretive pathway to organizational

change. In B. Staw & L. Cummings (Eds.), Research in organizational behavior (15). Green-
wich: JAI, 1993, 195-226.
Ellsberg, D. Risk, ambiguity and the Savage axioms. Quarterly Journal of Economics, 1961, 75,
643-669.
Ghosh, D., & Ray, M.R.Risk attitude, ambiguity tolerance and decision making: An exploratory
investigation. Decision Sciences, 1992, 23, 43 1-444.
Gist, M. Self-efficacy: Implications for organizational behavior and human resource manage-
ment. Acaderriy of Managenrent Review, 1987, 12, 472-485.
Halvena, W.J., & DeSarbo, W.S. On the measurement of perceived consumer risk. Decision
Sciences, 1991, 22(4), 927-939.
Heath, C., & Tversky, A. Performance and belief-Ambiguity and competence in choice under
uncertainty. Journal of Risk and Uncertainty, 1991, 4 , 5-28.
Kahneman, D., & Tversky, A. Prospect theory: An analysis of decisions under risk. Econoinet-
rica, 1979, 47, 263-291.
L a k e , E., Frederick, E. Lee, C., & Bobko, P. Effects of self-efficacy, goal-setting and performance.
Journal of Applied Psychology. 1984, 69,241-25 I .
Lopes, L. Between hope and fear: The psychology of risk. Advances in Experimental Social
Psychology, 1987, 20, 255-295.
March, J., & Shapira, Z. Variable risk preferences and the focus of attention. Psychological
Review, 1992, 99(1), 172-183.
MacCrimmon, K., & Wehrung, T. Taking risks. New York: Free Press, 1986.
Miller, D. The lcarus paradox. New York Harper Business, 1990.
Scherer, R., Adams, J., Carley, S., & Wiebe, F. Role model performance effects on development
of entrepreneurial career preferences. Entrepreneurship Theory and Practice, 1989, 13, 53-81.
Schoemaker, P. Are risk-attitudes related across domains and response mcdcs? Managenrent
Science, 1990, 34: 145 1- 1463.
Schumpeter, J. Capitafisrn, socialisni and dernocracy (3rd ed.). New York: Harper Row, 1950.
Shapero, A. Some social dimensions of entrepreneurship. In C. Kent et al. (Eds.), The encyclo-
pedia of entrepreneurship. Englewood Cliffs, NJ: Prentice-Hall, 72-90, 1982.
Thaler, R. Quasi-rational ecoriornics. New York: Sage, 1991.

APPENDIX A
Sample Items

“Gamble” example:
You are considering a price increase for your product.

Possible Actions Expected Outcomes

A. Do NOT 100 percent certainty that the competitors will not raise prices and
raise price. that you will not lose any customers. Profit = $400,000.
B. DO raise price 50 percent chance that competitors will also raise prices and you
won’t lose any customers. Profit = $600,000.
50 percent chance that competitors will NOT raise prices and that you
will lose customers. Profit = $220.000.

“Dilemma” Example.
A neighbor, with whom you have had a friendly relationship for years is bored with
his law practice and wants to start his own business around his hobby. Before he
can know whether his business will succeed, he will have given up practicing law,
his partnership in his law firm, and will have invested most of his net worth and
400 Self-Efficacy and Opportuniry Recognition

seven years of his life. Should h e make a go of it, he will become a wealthy man,
as you define and as h e defines it. Should he fail, h e would have to start over from
scratch, borrowing money from friends and family, older and wiser. He has total
belief in his ability to make it. What odds for success would you require before
you would recommend that he go for it? %
(Remember, higher probabilities reflect more caution.)

APPENDIX B
Correlation Matrices (N=153)
Variables 1 2 3 4
Dilemmas
1. Perceived Self-Efficacy 1.000 .602 -.633 ,434
2. Perceived Opportunity - 1.OOo -.212* .311
3. Perceived Threat 1.ooo -.352
4.Risk Taking - 1.000
Gambles
1. Perceived Self-efficacy 1.000 ,532 -.276 ,322
2. Perceived Opportunity - 1.ooo -.022* ,279
3. Perceived Threat - - 1.ooo -. 136
4.Risk Taking - 1.ooo

Note: All correlations are significant atp=.001 except those marked with an asterisk (*)
which ate not significant at p=.IO.

Nortis Krueger, Jr., received his Ph.D from The Ohio State University in entrepreneur-
ship and strategic management. His current interest is in entrepreneurship, especially the
cognitive aspects of strategic decision making. He researches and consults with technol-
ogy-based startups in Bozeman, Montana.
Peter R. Dickson is the Crane Professor of Strategic Marketing at The Ohio State
University College of Business. His current research focus is on the competitive rationality
of firms in oligopoly markets. This includes studies of risk taking in new product develop-
ment and pricing decisions. Professor Dickson has articles recently published and forth-
coming in the Journal of Marketing, Journal of Retailing, Journal of Consirnier Research,
Journal of Advertising Research, Journal of Econoniic Psychology, and Marketing Letters.