-
etap_476 115..140
The Role of MixedEmotions in theRisk Perception ofNovice and
SerialEntrepreneursKsenia PodoynitsynaHans Van der BijMichael
Song
This study examines the role of mixed emotions in the risk
perception of entrepreneurs, animportant determinant of
entrepreneurial decision making. We extend the literature on
mixedemotions by applying the cognitive appraisal tendency approach
and contrasting it withambivalence stemming from the valence-based
approach. We test our hypotheses on a dataset of 253 entrepreneurs
from the United States. We show that mixed and conflictingemotions
are an important predictor of the risk perception of entrepreneurs.
At the sametime, we find that emotional reactions of entrepreneurs
on strategic issues change substan-tially as they found more
ventures and become habitual entrepreneurs.
Introduction
Imagine how much easier life would be if we felt only one
emotionpositive ornegativeat a time. Luckily and unfortunately,
this is often not the case. Entrepreneurs,in particular, are people
who are likely to experience mixed rather than single basicemotions
in their decision making, including their evaluation of
opportunities and assess-ment of risks. In this study, we focus on
the relationship between mixed emotions and therisk perception of
entrepreneurs.
We define entrepreneurs as individuals who recognize and exploit
new businessopportunities by founding new ventures (Baron, 2008;
Shane & Venkataraman, 2000).Because opportunity recognition and
evaluation are key phenomena of entrepreneurship(Alvarez &
Barney, 2007; Shane & Venkataraman), it is important to
understand whatmakes entrepreneurs pursue opportunities. Risk
perception plays an important role in thiscontext. Empirical
research has demonstrated that risk perception strongly influences
riskbehavior in general (Sitkin & Pablo, 1992; Sitkin &
Weingart, 1995), and opportunityevaluation and the decision to
start a venture in particular (Keh, Foo, & Lim, 2002;
Simon,Houghton, & Aquino, 2000). However, recent literature
reviews have called for more
Please send correspondence to: Ksenia Podoynitsyna, tel.: (40)
247-3640; e-mail: [email protected].
PTE &
1042-2587 2011 Baylor University
115January, 2012DOI: 10.1111/j.1540-6520.2011.00476.x
-
research to refine our understanding of the role of
entrepreneurial risk taking (Ireland &Webb, 2007). Until now,
empirical research has emphasized risk-taking propensity(Forlani
& Mullins, 2000; Mullins & Forlani, 2005) and cognitive
biases (Keh et al.;Simon et al.) as determinants of entrepreneurial
risk perception. A recent study by Foo(2011) finds that certain
background emotions can also influence risk perception.
Indeed,because entrepreneurs tend to rely heavily on their
intuition in making decisions (e.g.,Simon et al.) and emotions are
one of the main drivers of intuitive thinking (Epstein,
1994;Kahneman, 2003), emotions are meant to play a key role in
entrepreneurial decisionmaking.
Entrepreneurs likely experience mixed rather than single basic
emotions in theirdecision making. Similar to the executives of
established firms, entrepreneurs often dealwith ambiguous
situations (Eisenhardt & Sull, 2001; Walsh, Meyer, &
Schoonhoven,2006). They operate in environments that are highly
unpredictable and filled with rapidchange (Lichtenstein, Dooley,
& Lumpkin, 2006), while entrepreneurial tasks are highlyvaried
in nature and change significantly as the entrepreneurial process
unfolds (Baron,2006). When performing variable tasks in highly
uncertain and unpredictable environ-ments, an entrepreneurs affect
may play an important role in judgment and decisionmaking (Baron,
2008). Moreover, the greater the ambiguity in their working
environment,the more likely that entrepreneurs will experience
different emotions at the same time(Folkman & Lazarus, 1985;
Larsen, McGraw, Mellers, & Cacioppo, 2004). Thus, study-ing the
relationship between mixed emotions and risk perception can provide
valuableinsights for entrepreneurs.
The extant research on mixed emotions has concentrated on the
simultaneous expe-rience of positive and negative affect (Folkman
& Lazarus, 1985; Fong, 2006; Fong &Tiedens, 2002; Larsen,
McGraw, & Cacioppo, 2001; Larsen et al., 2004; Priester &
Petty,1996, 2001; Schwarz & Weinberger, 1980). We will build on
research examining thecognitive appraisal tendency approach to
emotions and contrast it with the valence-basedapproach. The former
approach focuses on the differences between emotions of the
samevalence, which can be attributed to various cognitive appraisal
patterns associated withthese emotions, such as uncertainty and
controllability, among others (Keltner, Ellsworth,& Edwards,
1993; Lerner & Keltner, 2000, 2001; Smith & Ellsworth,
1985, 1987). Forexample, happiness and hope are two positive
emotions, but happiness tends to beappraised as certain and
controllable while hope tends to be appraised as uncertain
anduncontrollable. Following the logic of these two approaches to
affect, mixed emotionsmay emanate from the simultaneous experience
of positive and negative emotions(ambivalence) or from the
coexistence of emotions with different associated
cognitiveappraisal patterns.
In this study, we examine 16 emotion adjectives, half of which
have a positive valence(related to the basic emotions of happiness
and hope) and half of which have a negativevalence (related to the
basic emotions of anger and fear). The emotion adjectives
wereselected in such a way that half of them also have a positive
association with certainty andcontrollability cognitive appraisal
dimensions (related to the happiness and anger basicemotions), and
the other half have a negative association with these dimensions
(related tothe hope and fear basic emotions). These dimensions are
shown to mediate the relation-ship between emotions and risk
estimates (Lerner & Keltner, 2001).
This study contributes to the literature by (1) conceptualizing
mixed emotions usingthe cognitive appraisal tendency approach, (2)
contrasting this approach with the mixedemotions conceptualization
based on the valence-based approach on the example of
riskperception of entrepreneurs, and (3) exploring the moderating
effect of experience infounding ventures on these
relationships.
116 ENTREPRENEURSHIP THEORY and PRACTICE
-
Theoretical Framework and Hypotheses
Entrepreneurial Risk Perception and its DriversEntrepreneurial
risk perception is defined as the entrepreneurs assessment of the
risk
inherent in a situation. This definition is consistent with
previous work of Dutton andJackson (1987), Jackson and Dutton
(1988), Sitkin and Pablo (1992), and Sitkin andWeingart (1995). A
situation is considered risky when it is more negative than
positive,there is more potential for loss than for gain, and it is
seen more as a threat than as anopportunity (Highhouse, Paese,
& Leatherberry, 1996; MacGrimmon & Wehrung, 1985;Sitkin
& Weingart).
In the entrepreneurship and management literatures, several
individual and organiza-tional drivers of risk perception have been
proposed and tested. Sitkin and Pablo (1992)suggest that
individuals risk propensity will mediate the relationship between
their riskpreference, inertia, and outcome history and their risk
perception. Moreover, theseresearchers suggest that organizational
characteristics like control systems, problemdomain familiarity,
social influence, top management team homogeneity, and
problemframing can all influence an individuals risk perception.
Sitkin and Weingart (1995) alsofind empirical support for the
mediation effect of risk propensity on the outcome historyrisk
perception relationship and the direct effect of problem framing on
risk perception.Forlani and Mullins (2000) also find an effect of
risk propensity on risk perception. At thesame time, Mullins and
Forlani (2005) find that risk propensity influenced
entrepreneurschoices related to the probability of gain and loss,
while risk perception influencedentrepreneurs choices related to
the magnitude of gains and losses. Simon et al. (2000)and Keh et
al. (2002) examine the role of cognitive biases in entrepreneurial
risk percep-tion. Illusion of control and belief in the law of
small numbers were found to be negativelyassociated with risk
perception.
In considering affect as another antecedent of risk perception,
Johnson and Tversky(1983) find that the experience of negative
affect equally inflated individuals estimatesfor all types of
negative events. Mayer, Gaschke, Braverman, and Evans (1992)
findsimilar results. We refer to the review of Blanchette and
Richards (2010) for an in-depthdiscussion of these two and related
studies. Finally, Foo (2011) finds that the experienceof four
induced emotions (anger, fear, happiness, and hope) unrelated to
the venturedecision had a differential influence on students risk
perceptions of entrepreneurialventure scenarios. Foo also finds a
positive relationship between dispositional happinessand anger
(i.e., happiness and anger as traits) and the risk-taking
propensity of entrepre-neurs. However, despite the likelihood that
entrepreneurs experience mixed emotions dueto the variable and
uncertain nature of their environment (e.g., Lichtenstein et al.,
2006),to date, no study has examined the relationship between mixed
emotions and risk percep-tion of entrepreneurs. In our study we
will focus on this link.
Valence-Based and Cognitive Appraisal Tendency ApproachesMost
studies on entrepreneurship that focus on emotions and affect
follow the
valence-based approach by distinguishing between positive and
negative affect. Forinstance, Baron (1998, 2008) conceptually
clarifies the role of positive and negativeaffect in basic
cognitive processes of entrepreneurs and the potential influences
on keyaspects of the entrepreneurial process. Foo, Uy, and Baron
(2009) examine conse-quences of positive and negative affect and
find that they both impact venture effort.Brundin, Patzelt, and
Shepherd (2008) show that entrepreneurs displays of positiveand
negative emotions influence their employees willingness to act
entrepreneurially.
117January, 2012
-
Shepherd and Cardon (2009) propose that negative emotional
reactions due to projectfailure impact learning from the failure
and the motivation to try again. Cardon,Zietsma, Saparito,
Matherne, and Davis (2005) and Cardon, Wincent, Singh, andDrnovsek
(2009) explain the concept of entrepreneurial passion by using a
parenthoodmetaphor in the entrepreneurial process and by linking
entrepreneurial passion to threerole identities (inventing,
founding, and developing) and conceptually exploring theeffects of
entrepreneurial passion on goal-related cognitions, entrepreneurial
behavior,and entrepreneurial effectiveness.
The cognitive appraisal tendency approach represents a viable
and promising alter-native to the valence-based approach. It
focuses on the differences between the emo-tions of the same
valence due to dissimilar cognitive appraisals. Smith and
Ellsworth(1985) summarize the patterns of appraisal for a range of
emotions. Keltner et al.(1993) confirm that not only the valence of
an affect, but also its associated cognitiveappraisal patterns have
an important influence on causal judgments. In particular, theyfind
a dissimilar impact of sadness and anger. Lerner and Keltner (2000,
2001) elabo-rate on these results and present an appraisal tendency
framework, in which eachemotion is defined by a tendency to
perceive new events and objects in ways that areconsistent with the
original cognitive appraisal dimensions of the emotion. They
findthat fearful people made risk-averse choices, while angry
people made risk-seekingchoices. Moreover, risk judgments of angry
people more closely resemble those ofhappy people than those of
fearful people. Lerner and Tiedens (2006) elaborate furtheron the
differences between anger and other emotions and depict the
anticipatory plea-sure of anger that triggers the positive
subjective sense. Foos (2011) research alsosupports the appraisal
tendency approach.
Both the valence-based approach and the cognitive appraisal
tendency approach canbe used to study mixed emotions, which are the
focus of this study. We build further onthese two streams of
research by examining mixed emotions as a potential antecedent
ofentrepreneurial risk perception. In order to build our
hypotheses, we use the affectinfusion model (AIM; Forgas, 1995,
1998) describing the mechanisms of how affect mayinfuse
judgments.
Mechanisms Explaining the Role of Affect in Judgment andDecision
Making
In this study, we define affect as a generic label to refer to
both moods and emotions.Moods are low-intensity, diffuse, and
relatively enduring affective states without a clearantecedent
cause and therefore little cognitive content. Emotions are more
intense, shortlived, and usually have a definite cause and a clear
cognitive content (Forgas, 1995).Researchers on basic (as opposed
to mixed) emotions have embraced the AIM depictinga set of
mechanisms explaining the influence of moods and emotions on
judgment anddecision making (e.g., Baron, 1998, 2008; Forgas, 1995,
1998). The AIM distinguishesbetween judgment processes where a
relatively closed information search is required andprocesses where
a much more open information search is needed. In general,
entrepre-neurs are involved in the latter processes. These
processes have a higher probability ofaffect infusion, that is,
affectively loaded information exerts an influence on and
becomesincorporated into the judgment process, entering into the
judgmental deliberations andeventually coloring the judgment
outcome (Forgas, 1995). The affect infusion worksthrough two
different, but complementary mechanisms: affect as information and
affectpriming.
118 ENTREPRENEURSHIP THEORY and PRACTICE
-
The affect-as-information mechanism describes direct affect
infusion: Instead ofmaking a judgment on the basis of recalling
features of a target, the individual askshimself how he feels about
it. So, in fact the affect itself is the information. In doing
so,individuals may mistake feelings due to a preexisting state for
a reaction to the target(Forgas, 1995). DeSteno, Petty, Wegener,
and Rucker (2000) find empirical support for
theaffect-as-information mechanism in making likelihood estimates
of future events. Loe-wenstein, Weber, Hsee, and Welch (2001) show
that emotional reactions to risky situationsoften diverge from
cognitive assessments of those risks.
The affect-priming mechanism suggests that affect can indirectly
inform social judg-ments by priming the encoding, retrieval, and
selective use of information in the judgmentprocess. Affect priming
can be enacted through four submechanisms. The first one
isselective attention, which means that given the overload of
information in a judgmentprocess and the necessity of making
selections, particular information is selected that iscongruent
with the current affect. Second, in the process of encoding new
information,more attention may be given to the affect-congruent
information. Third, most judgmentprocesses involve the retrieval
and use of prior knowledge structures to interpret
incominginformation. Affect can selectively influence these
retrieval processes, because affect-congruent information may have
a greater likelihood of being retrieved (Forgas, 1995).Forgas also
finds empirical support for the affect-priming mechanism in complex
bar-gaining tasks (Forgas, 1998).
HypothesesWe first focus on mixed emotions with dissimilar
valences. In general, emotion
adjectives that are associated with the basic emotions of
happiness and hope are con-sidered to be positive, while emotion
adjectives associated with the basic emotions ofanger and fear are
considered to be negative (see, e.g., Lerner & Keltner, 2000,
2001).Following the literature on attitudinal ambivalence (Fong,
2006; Fong & Tiedens, 2002;Priester & Petty, 1996, 2001),
we distinguish between dominant emotions and conflict-ing emotions
for each entrepreneur. Dominant emotions are those positive or
negativeemotions that are experienced in greater numbers, while
conflicting emotions are thosethat are experienced in lesser
numbers. For example, when an entrepreneur experiencesfive positive
and two negative emotions, the five positive emotions are dominant
and thetwo negative emotions are conflicting. Another entrepreneur,
experiencing three positiveand four negative emotions, has four
dominant (negative) emotions and three conflicting(positive)
emotions.
A prominent argument in the mixed emotions literature (Priester
& Petty, 1996, 2001)is that the conflicting (rather than
dominant) emotions make the difference in cases ofmixed emotions.
Priester and Petty (1996) and Priester, Petty, and Park (2007)
suggest thegradual threshold model (GTM) as the most appropriate
formula to calculate mixedemotions (i.e., ambivalence) on the basis
of particular reactions of the respondents. Belowa certain
threshold, ambivalence is a function of both dominant and
conflicting reactions.Above that threshold, ambivalence is a
positive and negatively accelerating function of theconflicting
reactions. That is, although feelings of ambivalence generally
increase asconflicting reactions increase, the initial conflicting
reactions produce a greater increase inambivalence than subsequent
conflicting reactions. In the GTM, the threshold graduallyemerges
as conflicting reactions become more numerous (Priester &
Petty, 1996; Priesteret al.). Furthermore, it does not matter
whether conflicting reactions are positive ornegative (Priester et
al.)thus, for some entrepreneurs the conflicting reactions
(i.e.,emotions) will be positive, for others negative.
119January, 2012
-
We also consider emotions based on different associated
cognitive appraisal pat-terns, in accord with the cognitive
appraisal tendency approach (Keltner et al., 1993;Lerner &
Keltner, 2000, 2001; Smith & Ellsworth, 1985). We assume that
each emotionactivates a cognitive predisposition appraising future
events in line with the centralappraisal dimensions that triggered
the emotion (Lerner & Keltner, 2000). In this view,there is a
strong relation between how a person interprets a situation along
certainemotion-relevant dimensions and what the person feels and
how the person appraisesfuture events (Smith & Ellsworth,
1987). These dimensions include (1) pleasantnessthe extent to which
the situation is perceived as being pleasant or unpleasant; (2)
antici-pated effortthe extent to which the person feels a need to
expend effort (eitherphysical or mental) in the situation; (3)
attentional activitythe extent to which theperson wants to attend
to or shut out the situation; (4) certaintythe extent to which
theperson understands or is sure of what is happening or going to
happen in the situation;(5) responsibilitywho or what the person
perceives as having brought about the situ-ation, either oneself or
someone or something else; (6) controlwho or what the
personperceives as currently being in control of the situation,
either oneself, some other personor persons, or uncontrollable
circumstances; (7) legitimacythe extent to which thesituation is
perceived as fair or unfair; and (8) perceived obstaclethe extent
to whichthe person perceives problems or obstacles in the situation
that are hampering the attain-ment of an desired goal. In line with
Lerner and Keltner (2000, 2001) and Foo (2011),we focus on two of
these dimensions, certainty and control, which mediate the
rela-tionship between emotions and risk perception (Lerner &
Keltner, 2001) and are thus ofgreat importance in the
entrepreneurial setting (Baron, 1998). In general, emotion
adjec-tives that are associated with the basic emotions of
happiness and anger score high onthe control and certainty
dimension, while adjectives associated with hope and fear havea low
score on control and certainty (Smith & Ellsworth). For
instance, fear is triggeredby the perception that negative events
are unpredictable and uncontrollable (Foo; Lerner& Keltner,
2000, 2001).
The ambivalence literature is quite neutral to what constitutes
conflicting and domi-nant reactions to the researched issues
(Priester & Petty, 1996, 2001). Although thevalence-based
approach to emotions can be more easily fit into the ambivalence
concep-tualization (e.g., Fong, 2006; Fong & Tiedens, 2002),
one can also distinguish dominantand conflicting reactions on the
basis of the cognitive appraisal approach. If, for anentrepreneur,
the number of experienced emotional adjectives that score highly on
cer-tainty and controllability exceeds the number of experienced
adjectives that have a lowscore on certainty and controllability,
we call the former emotions dominant and the latterconflicting (and
vice versa). The arguments from the ambivalence literature
regarding theprominence of conflicting emotions influence should
also hold in case of cognitiveappraisals approach (Priester &
Petty, 1996, 2001).
Following the AIM, the effect of conflicting emotions on
judgments in general andrisk perception in particular works through
two complementary mechanisms. First, withdirect
affect-as-information mechanism, where feelings are used as if it
was informationabout the issue in question, as the level of
conflicting emotions regarding a certain issueincreases,
entrepreneurs will experience this odd mixture of emotions more
strongly andtherefore recognize it more easily. As a result,
entrepreneurs will more likely experiencethe focal issue as unusual
and feel uncomfortable about it (Fong, 2006). Feeling
uncom-fortable will lead to a cautious, conservative judgment on
the issue, resulting in a higherperception of risk.
With the indirect affect-priming mechanism, where feelings
determine the typeof information used to assess a certain issue,
the more entrepreneurs experience and
120 ENTREPRENEURSHIP THEORY and PRACTICE
-
recognize conflicting emotions, the more they will be sensitive
to unusual associations(Fong, 2006). Facing an overload of
information in the judgment process, entrepreneursimplicitly or
explicitly have to make choices regarding what information to
retrieve,encode, and use. This may lead to overly simplistic
analyses of complex situations,directing attention to a limited set
of variables and exclusion of other important variables(Simon &
Houghton, 2002). Moreover, it often leads to the retrieval of
success stories andpositive information, neglecting failures and
negative information (Golder & Tellis, 1993;Simon et al.,
2000). Normally, such selection will result in too-optimistic
judgments anda lower perception of risk. However, when
entrepreneurs become more sensitive tounusual associations, they
are more likely to take more variables into account and accessthe
less-obvious knowledge structures, thus enriching the information
search process(Fong). This will lead to a richer set of
associations and interpretations including bothpositive/success
stories and negative/failure stories. Therefore, mixed emotions are
likelyto be associated with a higher perception of risk.
Both mechanisms indicate that an increase in the level of
conflicting emotions willlead to a higher perception of
entrepreneurial risk. Moreover, experiencing conflictingemotions is
associated with a feeling of indecisiveness (Priester & Petty,
1996, 2001). Thehigher the level of indecisiveness, the greater the
level of risk entrepreneurs are likely toperceive. Therefore, we
hypothesize:
Hypothesis 1a: Conflicting emotions in terms of valences (i.e.,
positive vs. negativeemotions) will have a stronger association
with the risk perception of entrepreneursthan dominant emotions in
terms of valences will.Hypothesis 1b: Conflicting emotions in terms
of cognitive appraisal tendencies (i.e.,certain and controllable
vs. uncertain and uncontrollable) will have a stronger asso-ciation
with the risk perception of entrepreneurs than dominant emotions in
terms ofcognitive appraisals will.Although the concepts of dominant
and conflicting emotions can be applied to both
valence-based and cognitive appraisal tendency approaches, there
are numerous argu-ments suggesting the superiority of the latter
approach in explaining the connectionbetween emotions and risk
perception. First, researchers note that the mere ability offeeling
distinct emotions should result in their differential influence on
many cognitiveand motivational processes (cf. DeSteno, Petty,
Rucker, Wegener, & Braverman, 2004).Second, because the
occurrence of a certain type of event roughly predicts the
occurrenceof other events of the same type and because particular
specific emotions mark suchcategories of events, the impact of
affect is likely to be emotion specific rather thanvalence specific
(DeSteno et al., 2000; Fessler, Pillsworth, & Flamson, 2004).
It isunlikely that emotions sharing the same valence would
necessarily have the same effect onrisk taking because valence is
not uniformly associated with different classes of functionalgoals
(Fessler et al.). Studies show that different emotions of the same
valence can havecompletely opposite effects in a variety of
decision-making situations and that emotionswith the same cognitive
appraisal tendency are more alike than emotions of the samevalence
(e.g., DeSteno et al.; Fessler et al.; Lerner & Keltner, 2000,
2001; Raghunathan &Pham, 1999). Likewise, dissimilarity in
terms of cognitive appraisal tendency should bemore influential
than dissimilarity in terms of valence. For risk judgments,
certainty andcontrollability appraisals are shown to be most
essential (Lerner & Keltner, 2000, 2001).Because these two
appraisal tendencies produce opposing effects on risk judgments
(e.g.,Foo, 2011), conflicting emotions with contrasting certainty
and controllability should alsoproduce a greater feeling of
discomfort and irregularity than conflicting emotions basedon
contrasting valences. Once again, feelinguncomfortable will lead to
a cautious,
121January, 2012
-
conservative judgment on the issue, resulting in a higher
perception of risk. Therefore, wehypothesize:
Hypothesis 2: Conflicting emotions in terms of cognitive
appraisal tendencies (i.e.,certain and controllable vs. uncertain
and uncontrollable) will have a stronger asso-ciation with the risk
perception of entrepreneurs than conflicting emotions in terms
ofvalences (i.e., positive vs. negative emotions) will.Promoting
entrepreneurship as the driver of economic growth aims to increase
the
number of entrepreneurs founding new ventures. Having past
experience as an entrepre-neur can increase the probability of
actually transitioning from paid employment toentrepreneurship by
12%, while serial entrepreneurs can account for as many as 30%
ofsuch transitions (Hyytinen and Ilmakunnas, 2007). Serial (also
referred to as habitual)entrepreneurs therefore represent a very
valuable population from a policy perspective(Wiklund &
Shepherd, 2008). Moreover, serial entrepreneurs can play an
important rolein promoting economic growth as they can potentially
build upon their experience andstart more successful ventures. One
of the executive forums of the Journal of BusinessVenturing (1986)
was called To really learn about entrepreneurship, lets study
habitualentrepreneurs, which accurately reflects the relevance of
the study of serial entrepreneursto practitioners and the
popularity of success stories of serial entrepreneurs.
As entrepreneurs found more new ventures and gain more
experience in creating andrunning new businesses, their intuition
develops as well (Mitchell, Friga, & Mitchell,2005). Although
entrepreneurs newly accumulated insights and expert knowledge
struc-tures are pillars supporting intuition (Mitchell et al.;
Shane, 2000), emotions are also oneof the characteristic features
of intuition (Epstein, 1994; Kahneman, 2003). Moreover,intuition is
often depicted as being primarily nonverbal and intimately
associated withaffect (Denes-Raj & Epstein, 1994).
Consequently, as serial entrepreneurs tend to relymore on intuition
in their judgments, they are also more likely to engage emotions in
theirjudgments. Accordingly, both valence-based and cognitive
appraisal tendency perspec-tives on emotions should play greater
roles in the judgment processes of more experi-enced, serial
entrepreneurs. Thus, we hypothesize:
Hypothesis 3a: Conflicting emotions in terms of valences (i.e.,
positive vs. negativeemotions) will play a greater role in risk
judgments of serial entrepreneurs than in riskjudgments of novice
entrepreneurs.Hypothesis 3b: Conflicting emotions in terms of
cognitive appraisal tendencies (i.e.,certain and controllable vs.
uncertain and uncontrollable) will play a greater role inrisk
judgments of serial entrepreneurs than in risk judgments of novice
entrepreneurs.
Control VariablesIn this study, we use the basic emotions of
anger, fear, happiness, and hope as control
variables. On the basis of past research, we expect these
emotions to be associated withentrepreneurial risk perception.
However, the direction of the associations is not clear.
Thevalence-based approach would predict a positive association of
anger and fear withentrepreneurial risk perception and a negative
association of happiness and hopewith entrepreneurial risk
perception (Johnson & Tversky, 1983). The cognitive
appraisaltendency approach would generally predict a positive
association of fear and hope withentrepreneurial risk perception
and a negative association of anger and happinesswith
entrepreneurial risk perception (Lerner & Keltner, 2000, 2001).
At the same time, not
122 ENTREPRENEURSHIP THEORY and PRACTICE
-
all of the cognitive appraisal dimensions show a stable link
between emotions and theirappraisal tendencies, making these
predictions preliminary (Smith & Ellsworth, 1987;Tong et al.,
2009). Another factor making these predictions more difficult is
the nature ofour sample. We also control for entrepreneurs who are
familiar with the case we presentin the study and their
entrepreneurial experience.
Method
Sample and Data CollectionThe participants for this study were
drawn from the list of 6,359 founders of venture-
backed firms provided by VentureOne, a leading venture capital
(VC) research companybased in San Francisco. VentureOne began
tracking equity investment in 1992. It collectsdata by surveying VC
firms for recent funding activities and portfolio updates,
gatheringinformation through direct contacts at venture-backed
companies, and investigatingvarious secondary resources such as
company press releases and initial public offeringprospectuses from
VentureOne 2001. We randomly selected 1,100 entrepreneurs
withcomplete contact information for the study.
In administering the scenario and the related questionnaire, we
followed the totaldesign method for survey research (Dillman,
1978). The first mailing packet included apersonalized letter, a
project fact sheet, the survey, a priority postage-paid envelope
withan individually typed return-address label, and a list of
research reports available toparticipants. The package was sent by
priority mail to each randomly selected entrepre-neur. As a result
of some packages being undeliverable for name or address reasons,
theadjusted sample comprised 776 entrepreneurs.
To increase the response rate, we sent four follow-up mailings
to the companies. Oneweek after the mailing, we sent a follow-up
letter. Two weeks after the first follow-up, wesent a second
package with the same content as the first package to all
nonrespondingcompanies. After two additional follow-up letters, we
received completed questionnairesfrom 255 entrepreneurs,
representing a response rate of 32.9%. Two respondents
indicatedthat they had no active experience as venture founders and
were omitted from the analysis.In our final sample, 181 were male
and 72 were female. The highest degree obtained for46 of the
entrepreneurs was high school; for 124, bachelor degree; for 74,
masters degree;and for 9, doctoral degree. The mean age of the
respondents was 42.4 years. On average,they founded 4.4 ventures,
were currently involved in 2.1 ventures, and had 14.5 years
ofentrepreneurial experience.
MeasurementsIn our study, we used existing cases and scales from
the literature. We conducted a
pretest by extensively interviewing 12 entrepreneurs. At the
beginning of each interview,entrepreneurs told us about the
background of their ventures, how they started, how theydiscovered
the opportunity, and how the business idea developed over time.
This allowedus to get the conversation started and better interpret
their answers on the questionnaire.In the last part of the
interview, we used the protocol method and asked the
entrepreneursto think aloud as they filled out the questionnaire
(Hunt, Sparkman, & Wilcox, 1982).The interviews were recorded
and two researchers made careful notes of the verbaliza-tions and
the thinking process of the entrepreneurs. The analysis of
interviews led to minorchanges in the wording of the instructions
and cases description.
123January, 2012
-
For the sake of ecological validity of our results, we chose to
center measures of allthe variables in our study around a strategic
issue, signifying an important environmentalevent, trend, or
development for which future resolution will be sought (Dutton
&Jackson, 1987). The selected issue was extensively pretested
in earlier studies to be surethat it represents an ambiguous event
that could be interpreted both as a positive and as anegative
change in the competitive environment (Highhouse et al., 1996). As
a result, themeaning of the strategic issue is not inherent in the
external environment and it is up tofirms decision makers to
interpret the event (Dutton & Jackson). Meanings attached
tostrategic issues color the subsequent affective and cognitive
decision processes of thedecision makers. This ambiguity allowed us
to maximize the chance that entrepreneurs inour sample will
experience mixed emotions, the focus of our study (Folkman &
Lazarus,1985; Larsen et al., 2004).
The strategic issue we chose was originally used by Dutton and
Jackson (1987) as ananecdotal example that everybody can easily
relate to. Considering the potential differ-ences in the
backgrounds of entrepreneurs, this seems to be the most appropriate
choice.Highhouse et al. (1996) adapted the issue in their study of
strategic-issue framing. In ourstudy, we used the version found in
Highhouse et al. It reads as follows:
Imagine that you are the owner of a retail clothing store. You
have historicallyprospered by occupying a choice downtown location.
You have just found out thatplans are underway to construct a large
suburban shopping mall ten miles from yourcurrent place of
business. (p. 97)Highhouse et al. (1996) found that the description
above is a neutral description of the
strategic issue; it is neither representing a positive situation
nor a negative one. Thus,positive or negative assessments of this
situation are interpretations of the decisionmakers.
Dependent Variable. To measure risk perception, we asked the
respondents to character-ize on 7-point scales whether they saw the
strategic issue as an opportunity or as a threatand as having a
potential for loss or a potential for gain. We also asked whether
theycharacterized the issue as a positive or a negative situation
(Jackson & Dutton, 1988;MacGrimmon & Wehrung, 1985; Sitkin
& Weingart, 1995). Cronbachs a for the riskperception construct
is 0.81.
Independent Variables. To objectively assess mixed emotions
(Priester & Petty, 1996),respondents were asked to rate on
7-point scales the extent to which each of 16 emotionadjectives
described how they were feeling about the strategic issue, allowing
for naturallyoccurring emotions. This approach differs from the
priming technique to make the par-ticipants feel a certain emotion
that is unrelated to the specific issue involved (e.g., Foo,2011;
Lerner & Keltner, 2001). Sixteen emotion descriptors were
selected from a list of25 adjectives presented by Smith and
Ellsworth (1987); eight positive and eight negativefeelings were
selected to allow for contrasting the valence-based and cognitive
appraisaltendency approaches. Also, eight of the 16 adjectives were
associated with high certaintyand controllability appraisals and
eight were associated with relatively low certainty
andcontrollability appraisals (Lerner & Keltner, 2000, 2001;
Smith & Ellsworth, 1985, 1987).Table 1 summarizes the valences
and dimensions for each emotion adjective.
By employing a mean split, we determined whether a particular
emotion adjective wasexperienced (1) or not (0). This served as the
basis for calculating the mixed emotionsmeasures for both the
valence-based and cognitive appraisal tendency approaches.Because
the ambivalence literature shows that conflicting emotions drive
ambivalence, we
124 ENTREPRENEURSHIP THEORY and PRACTICE
-
first computed the separate dominant and conflicting emotions
scores in order to test theirrelative importance for entrepreneurs.
For the valence-based approach, we calculated thedominant emotions
for each respondent on the basis of positive and negative valence
bytaking the greater of either the sum of experienced positive
adjectives or the sum ofexperienced negative adjectives (Priester
& Petty, 1996, 2001). For the cognitive appraisaltendency
approach, we calculated dominant emotions associated with
dissimilar cognitiveappraisals by taking the greater of either the
sum of experienced emotions with highcertainty and controllability
appraisals or the sum of experienced emotions with lowcertainty and
controllability appraisals. For conflicting emotions, we followed
the sameprocedure, but took the lesser instead of the greater of
the sums of experienced emotions(Priester & Petty, 1996,
2001).
In order to calculate mixed emotions for both approaches, we
used the GTM formulasuggested by Priester and Petty (1996) as the
most appropriate measure of ambivalence:
5C Dp 1/C
where C is equal to the magnitude of conflicting reactions, D is
equal to the magnitude ofdominant reactions, p is less than 1 (0.5
in the present instance), and a constant of 1 isadded to C and D
(Priester & Petty, 2001).
Control Variables. The findings of Smith and Ellsworth (1987)
demonstrate that the 16emotion adjectives correspond to four basic
emotions: anger, fear, happiness, and hope.Anger was depicted by
such adjectives as resentful, angry, surprised, and scornful;
fearby nervous, afraid, restless, and anxious; happiness by elated,
happy, amused, and proud;and hope by expectant, hopeful,
challenged, and interested. After the exploratory factoranalysis we
dropped the adjective anxious from the fear construct and the
adjectiveinterested from the hope construct due to their cross
loadings. On the basis of the
Table 1
Characteristics of 16 Emotional Adjectives
Emotionaladjective Valence
Certaintyappraisal
Controllabilityappraisal
Afraid Negative Low LowAmused Positive High HighAngry Negative
High HighAnxious Negative Low LowChallenged Positive Low LowElated
Positive High HighExpectant Positive Low LowHappy Positive High
HighHopeful Positive Low LowInterested Positive Low LowNervous
Negative Low LowProud Positive High HighResentful Negative High
HighRestless Negative Low LowScornful Negative High HighSurprised
Negative High High
125January, 2012
-
remaining adjectives, Cronbachs a for anger, fear, happiness,
and hope were 0.83, 0.83,0.79, and 0.78, respectively, suggesting
good reliabilities (Nunnally, 1978). Finally,although emotions are
one of the characteristic features of intuition (Epstein,
1994;Kahneman, 2003), entrepreneurial intuition can also be based
on prior knowledge andexpert knowledge structures (Mitchell et al.,
2005; Shane, 2000). We thus added theentrepreneurs familiarity with
the strategic issue and number of ventures founded ascontrols to
our model in order to distinguish knowledge-based effects from our
findings.We measured familiarity by the extent to which
entrepreneurs could relate the strategicissue to one of the
decisions they had made in the past. Number of ventures founded
wasmeasured by the number of ventures in which the entrepreneurs
were actively involved asa founder. Finally, we added the common
control variables of age, gender, type, and levelof education.
Analysis and Results
Prior to testing the hypotheses, we examined the correlation
matrix, shown inTable 2. The measures of mixed, conflicting, and
dominant emotions appear to behighly correlated (0.740.96), so we
do not include any two of these variables in thesame regression
model. The correlations between these types of emotions and
basicemotions are also quite large, although lower (i.e., ranging
between 0.47 and 0.70 withthe majority around 0.60). In models
combining basic and mixed emotions, we mean-centered all emotion
variables in order to eliminate the nonessential
multicollinearitydue to the computation formula of mixed emotions
components (Aiken & West, 1991;Cohen, Cohen, West, & Aiken,
2003; Kenny & Judd, 1984). An application of Belsley,Kuh, and
Welsch (1980) diagnostic tests indicated no serious
multicollinearity problemsin the mean-centered regression models
(maximum variance inflation factor was5.5 < 0; maximum condition
index was 22.3 < 30, while the proportion of varianceinflated
for emotions did not exceed 0.05 < 0.50). However, because the
maximum vari-ance inflation factor does not exceed 1.83 in other
models, caution is necessary wheninterpreting the results when both
basic emotions and mixed emotions are entered intothe same
models.
To test the hypothesized relationships, we performed
hierarchical regression analysis(Cohen & Cohen, 1983; Cohen et
al., 2003). This method examines the effects of addi-tional
variables above and beyond the effects of the variables in the
previous model. InTable 3, we present a summary of the results of
eight regression models.
All the regression models were highly significant, with F-values
ranging between 4.29and 8.71 (p < .001). In the first baseline
model, we only examined the effects of the controlvariables on
entrepreneurial risk perception. In model 2, we added the basic
emotionsbased on positive and negative valence. In model 3, we
added the mixed emotions basedon positive and negative valence. In
models 4 and 5, we split the mixed emotions frommodel 2 into
dominant and conflicting emotions in order to determine their
relative effecton mixed emotions. In model 6, we added the mixed
emotions associated with differentcognitive appraisals. Finally, in
models 7 and 8, we split the mixed emotions from model6 into
dominant and conflicting emotions in order to determine their
relative effect onmixed emotions. The results from the hierarchical
regressions suggest that all modelsrepresent significant
improvements over model 1.
We also conducted regressions by adding the basic emotions to
models 3 to 8 (notreported in Table 3). Only mixed and conflicting
emotions in terms of cognitive appraisaltendencies remained
significant with a coefficient (standard error) of 0.065
(0.035),
126 ENTREPRENEURSHIP THEORY and PRACTICE
-
Tabl
e2
Des
crip
tive
Stat
istic
s(N=
253)
Cons
truct
sM
ean
SD1
23
45
67
89
1011
1213
1415
16
1.R
iskpe
rcep
tion
4.22
1.33
2.M
ixed
(Val
.)7.
873.
600.
263.
Confl
ictin
g(V
al.)
3.11
2.32
0.33
0.96
4.D
omin
ant(
Val.)
4.89
2.30
0.35
0.74
0.82
5.M
ixed
(Cog
n.A
ppr.)
7.93
3.53
0.37
0.78
0.84
0.82
6.Co
nflic
ting
(Cog
n.A
ppr.)
3.13
2.33
0.41
0.83
0.91
0.88
0.96
7.D
omin
ant(
Cogn
.App
r.)4.
872.
350.
260.
850.
890.
910.
680.
778.
Ang
er4.
331.
250.
220.
650.
670.
640.
600.
650.
649.
Fear
4.53
1.39
0.26
0.69
0.70
0.64
0.62
0.65
0.67
0.49
10.
Hap
py3.
681.
480.
430.
470.
560.
570.
560.
610.
500.
320.
3211
.H
ope
4.93
1.14
0.07
0.61
0.65
0.61
0.51
0.59
0.66
0.43
0.41
0.21
12.
Age
42.3
812
.76
0.02
0.02
0.04
0.01
0.01
0.04
0.02
0.05
0.02
0.07
0.03
13.
Bus
ines
sde
gree
0.60
0.49
0.06
0.24
0.29
0.30
0.23
0.27
0.31
0.13
0.16
0.08
0.35
0.01
14.
Deg
ree
level
2.18
0.77
-0.0
20.
120.
150.
140.
090.
120.
160.
090.
120.
000.
200.
090.
2015
.Fa
mili
arity
with
issue
5.22
1.57
0.07
0.14
0.18
0.21
0.15
0.19
0.19
0.08
0.15
-0.0
70.
20-0
.06
0.27
0.01
16.
Fem
ale
entr
epre
neur
1.28
0.45
-0.1
10.
090.
100.
030.
020.
060.
070.
000.
020.
03-0
.02
0.05
0.02
0.01
0.10
17.
Num
bero
fven
ture
s4.
402.
820.
260.
470.
540.
510.
450.
520.
510.
380.
460.
240.
470.
040.
480.
240.
290.
07
Note
:Co
rrela
tions
of|0
.13|an
dab
ove
are
signi
fican
tatp