BIASES AND HEURISTICS IN STRATEGIC DECISION MAKING: DIFFERENCES

BETWEEN ENTREPRENEURS AND MANAGERS IN LARGE ORGANIZATIONS

LoweU W. Busenitz, Univosity of Houston, CBA - Management Dept., Houston, TX 77204-6283
Jay B. Baniey, Texas A & M University

ABSTRACT obsavati(Hi that entrepioiirairs really are different from

i large caganizations in some important ways.
The lesearch question "How are entFq>reiuurs different from
managers in largeOTganizations?*was addressed. This study
hypothesized that entrepreneurs utilize biases and heuristics in THEORY
strat^c decisitm making more extensively than managers in
large raganizations. The results suggest a txsw and potentially The Mure of research <HI individual differences betweoi
more useful way of thinlfing about the aitrepi«ieurial entrejx-eneuis and manago^ in organizations presents a
diffooice. challenge to researchers. On the one hand, more rigmous
research using traditional psyclK>logical and
personal/demographic differences does not seem to generate
INTRODUCTION more powerful results. On the other hand, eff<»ts to focus on
extemal &u:txs do not c(Hiq>letely abandon individual
Casual observations suggests that individuals \^u) start their diffoeiKxs models. Thus, to c(mtiniK necessary resrarch on
own organizati(ms are somehow "differoit" frwa individuals iiKlividual diffomces between entrepreneurs and managers in
that work in large organizations. Entreixeneurs have been large organizations, the task is not simply to study traditional
described as r i ^ takers and rugged individualists, as engagir^ differences better, but to study new individual differences.
in deviate social behavior, and as being a "breed apart" (e.g.,
Bc^ey & Boyd, 1987; Ginsberg & Buchholtz, 1989). In The maJOT thesis of this paper is that the uses of biases and
ccmtrast, managers in large organizations have been described heuristics in ^cision making are critical in understanding
as being risk averse adhering to tn^oadly accepted mmns of differraces between entrepreneurs and managers in large
behavior (Pettigrew, 1973), and more professiratal and orgariizaticHis. In this ccmtext, biases are subjective and/or
predictable in their decision making (Bamard, 1968). Implicit predisposed opinions that influence the decision process.
in much of this discussion is the assumption that entreproieurs, Heuristics are also subjective in nature, but are specific
or would-be entrepreneurs, need to have the right composition. "informal rules-of-thumb" or intuitive guidelines that yield
quick and usually acceptable decisions to problems (Tversky &
These casual observations have not gone untested in the Kahneman, 1974; Nisbett & Ross, 1980). The term "biases and
research literature. Unfortunately, most effcnts to onpirically heuristics" will be used here to refer to these general decision
describe individual differences between entreproieurs and making phenomena.
managers in lax;ge organizations have met with liniited success
(Low & MacMillan, 1988). These results have led many In ^ite of the obvious impcntance of strategic decision making
researchers to abandon the search for individual differences in emerging ventures (Hamlnick & Crozier, 1985), only a few
between entrepreneurs and managers in large organizations, stu<Ues have specifically examined ckcision making among
and look for explanations of entrejK'eneurial behavicn*. entre(»-eneurs (Gilmrae & Kazanjian, 1989; Smith, Gannon,
However, much phenomena surrounding the establishment of Giimm, & Mitchell, 1988). Moreover, virtually none of this
new firms remains unexplained. Reasons for the high failure research uses biases and heuristics in the study of strategic
rate of new start-ups remains largely unexplained. The use of decisicm making among entrepreneurs. This is the case even
various business and economic principles ttiat help explain the though it is widely recogni^ that both the cranplexity of
maneuvers of large firms are often of little help in decisi<His &cing entrepreneurs (e.g.. Smith et al., 1988) artd the
understanding the successes and failures of new ventures. speed with \ ^ c h entre(Hieneurs must make decisions
(Eisenhardt, 1989) are the ccmdititms under which bias« and
The purpose of this paper is to refocus attention on individual heuristics are most likely to opoBte (Stevensmi et al., 1990;
differences explanations of entrepreneurial behavior. However, Nisbett & RMS, 1980). The one exception to this lack of
rather than focusing on the individual diffoences previously inquiry into the use of cognitive biases and heuristics in
studied, this paper examines differences in the decision making entrqneneurial decisicm making is the work by Manimala
pnicesses used by oitrepreneurs and manago^ in large (1992), w*o found 77 heuristics to be used more by
organizations. Building on non-rational decision making entrejHeneurs in highly innovative ventures.
models from behavioral decision thecny (Stevenson,
Busem^er, & Naylor, 1990), we argue that entrepreneurs are
more subject to decision making biases than managers in lar^e
organizations. Two decision making biases and heuristics are The decision was nuule to exam the reptresentativen^s
examined: overconfidence and representativeness (Fischhoff, heuristic and the overc<mfidence bias because they are two of
Slovic, & Lichtenstein, 1977; H<«arth, 1987; Tversky & the most widely discussed biases and heuristics (e.g., Hc^arth,
Kahneman, 1974). By examining these cognitive bias 1987; Kahnanan et al., 1982; Tverdcy & Kahnanan, 1974) and
individual differences, this pajior reafilrms the casual these two were cranpatible with the research design used for

85
this study. In this study, aitrqjroieiiis are tkse ^»ito are asames tiiat a large sanQiIe will be repesoitf^ve of a givoi
organizatianal founders v ^ are cmrently involved in tte populaticii. Ifowewer, Kn1in«nOT et al. (19^) noted that
f(»inding px)cess. As already noted, managers are individuals decisi<Hi makers tend to overestinute the representativaiess of
with middle to upper level res^xHisibilities with »ibstantial small saiiq}ies, feeling as if small samples may be very
oversight in large organizatirais. rqH«seatative of an entire populaticm.

While much of the research on biases ai^ heuristics toids to DecisiiHi making in the entrepreneurship area may actually be
link the nature of the t a ^ with the use (^ ^>ecific heuristics, oihanced by using biases and heuri^cs because small and non-
tme emetpng generalizaticm is that overconfidence in representative samples are goienilly all that exist. In bringing
judgments is widespread (Fischhoff et al., 1977). Decision an untested product to mailcet, reliable market research is
makers tend to be overly optimistic in their estiinatiim abilities generally very limited, since pnxhicts so oftrai deal with
iqx)n receiving initial infonnati<m (FischhofT et al., 1977), previously ncmexistent markets. Thus, an entrepreneur is likely
particularly when they are relatively unfamiliar with the to get some limited feedback from biased sanq)!^ such as a
problem and/or substantial lmcertainty exis^ (Lichtenstein & few p(rtential customers,friemls,and/OT a colleague or two. On
FischlK>fr, 1977). Furthomore, decision makers tend to be fbe basis of these biased samples and qiecific rules-of-thumb,
ovenxHifident in their initial assessments and slow to revise entrepreneurs think they have an adequate "feel" for the market
their initial judgmoits a{)propdately as additicsial infonnation and move fOTward. On the other hand, manages in large
beccMies available. (xrganizaticms are more likely to be xocex ratiraial in their
decisicm making for several reascms. First, cause-effect
Consistent with the Sequent obserraticms that eaitre{M7eneurs relaticmsUps can be better established because they deal with a
tend to be overly enthusiastic regarding the pc^ential of their large number of similar situations. Second, they have access to
venture's siKxess, Coqier, Dunkelbog, and Woo (1988) found more and better informaticm, and use of this information is
that entrepreneurs assigmd a higher {Hobability of success to oihanced by tteir linkages with numoDus colleagues. Finally,
their own voitures vdiile [sescribing noticeably lower odds of they are vatxc femiliar with c(Hnprdtensive ctecision making
success to other ventures like theirs'. As new infonnation patterns doe to their training and the en^rgence of more
becomes available suggesting arevisionin the original ^cision rational decisi<m making within their oi^anizations. This leads
or that the success of the i^w venture is highly unlikely, to the following hypothesis:
necessary adjustnmnts are likely to be ign<»^ or minimized.
While Bazerman (1986) and others have made suggestions as Hypothesis 2: Entrepreneurs will manifest representative
to how to avoid the pitfalb of overconfideiKe, this nmy be heuristics more extensively than managers in large
difBcult {(s entrepreneurs. Many of the solutions to their organizatitms.
jKoblems are simply unknown in advance, and little
infonnati<Hi is generally available that would ccmtradict their
optimisn. Praliaps more inqwrtantly, if entrefneneurs had a METHODS
lower level of conifidence in their decision Tnakfng ability, nK>st
new ventures would iKver get launched. Their overctmfidence Sam/Oes
keeps them from being overwhelmed with the multiple hurdles
they face. On the other haiul, mana^rs in large OTganizations As implied by these hypotheses, samples from two populaticms
generally deal with mere establi^ied markets, more were drawn: a sample of entreprenrairs and a sample of
information, and relatively established (H^anizational routines. managers in large organizations. The oitrepreneurs were
These ob^rvations lead to the following hypothesis: (n:ganizationeil founders who are currently involved in the start-
up process. The industries included the manufacturing of
Hypothesis 1: Entrepreneurs will manifest a stronger plastics and electr<Hucs. A pricai, it was thought that these
overconfidence bias than managers in large categories would represent a higher percentage of newly
organizations. enKiging firms because they represent mcare dynamic
industries. From a sample of 573 firms, most of \ ^ c h were in
Inferences in decision making are made by using current the plastics or electronic manufacturing industries, 176
infonnation to derive &rther propositions regarding a p'oblem. responses were received for a 31% response rate with a usable
This inference fs^ocess fieqiKntly relies on the concept of response rate of 22%. To test fOT a biased resp(Hise based on
"represoitativeness," a heuristic Tverslcy & Kahneman (i974) industry classificaticm, non^espondents were compared to
defined as a relation between a hypothetical process and some resp(»idiraits based on the two-digit SIC cat^ories identified
event associated with the process. This heuristic suggests that above. The results from the chi-sguare test suggested that the
decisicm makas tend to oversimplify alternative outcranes by usable respraise was not biased (X^^ (5) = 1.782; p = .878).
as^Kiating them with current knowledge and personal rules
that have emerged frtrni prior experiences. People predict In this study, managers in large arganizati(His were defined as
outcomes on the bases of salient features of events, and they individuals wiio have reqxHisibility for at least two functicmal
ignore other important characteristics of the evidence, including areas (such as marketing, finance, perscnmel, R&D, and
the validity and reliability of the data as well as the actual manufacturing) and wrak for a publicly owned OTganization
probabilities of events occurring (Bar-Hillel, 1979). While with more than 10,000 employees. Th^e managers are often
previous research has identified various dimoisicms of referred to as divisicKial managers or general managers because
representativeness, this study focuses on insensitivity to sample they oversee multiple areas (sample average was 4.S5
size. The law of large numbers (or insensitivity to sample size) fiuu^onal areas). A usable r^fponse rate of 54% was received.

86
The SIC far the managers i n d u e d in this sample came bom FIGURE 1
ihe 1300, 3400, 3500, 3600, and 3800. llie remits of the chi- Calibrati<m Curves ttx Bitiiqxeneurs and Managers
square te^ bc^weoi useable r«qxmses and non-resp(»dents
again suggest that therespcmsewas nal biased (X^ (4) = 3.973;
p = .59). 1.0

Mi

Ovavmtfitknce. To measure overcraifideiux, we replicated UNDERCONFIDENCE

the widely cited ^udies conducted by Fischlmff et al. (1977) DOMAIN
and Lichtoistein and Fischhoff (1977). The resulting data / / /

d-J
facilitated Ilie ccnnputatioa of a statistical score for each
observati(Hi and calibratitm curves for our eirtrqireneurs and
managers. This curve reveals how closely the d»ignated
probability stxse for each category (.SO, .60, etc.) and for each
ffoap (in this case entrepreneurs versus managers in large
OTganizations) qrproximates the perfectly calilsBted curve.

R^resaitath a. Following the lead of Fraig, Krantz, and

Nisbett (1986) mid Fong and Nisbett (1991) decisicm scenarios
rqiresenting various types of real-to-life strategic decisicsis was Entrepreneurs
used. Both problems used fen- this study pratrayed a strategic
decision pitting two alternatives against (me another. One
alternative was baaed <m quantitative/statistical infcmnaticm
v ^ l e the other was based on heuristic reasoning. Subjects /
were told to decide between the two alternatives for each .5 OVERCONFIDENCE
problem and then to desoibe their reasoning fw reaching the DOMAIN
designated decision. Following the coding schema develqied
by Ffflig ai^ colleagues, co^rs then analyzed these responses
to detomine whether heuristic type reasoning was used by the .4
respondents. There was exact agreement betweoi coders 84% n,= .5 .6 .7 .8 .9 1.0
of the time. These results were tlwn s u m n ^ across the two Probability Respcmse
problems to create a single 3-category variable (0-2). A "2"
indicated that an individual used statistical reasoning across
both poblems vMle a "0" indicated that only heuristic
reasoning was used.
Since the dependent variable in this model was dichotomous,
ic^istic ri^ressicm vras enq)loyed as tiie primary test of
Control variables. While research on differences between Hypotheses 1 and 2 (coded "1" frar entreimieurs and "2"
entrepreneurs and managers in large organizatitms has managers). Given this coding. Hypothesis 1 suggests that the
generated mixed results, we chose to include several of these coefficioit of the ovenxmfidence variable will be positive and
variables: risk-taking, need for conformity, age, educaticm, significant, while Hypothesis 2 suggests that the
networking intensity, and ec(m<miic factors. re[xesentativaKss coefiScient will be negative and significant.
As shown in Table 1, both variables are significant and in the
expected directi<m. Of tte c<mtrol variables, education,
RESULTS ccmformity, and alertness remain statistically significant. The
risk-taking, age, and networidng variables were non-significant
First, bivariate analysis was conducted with the overconfidence Overall, these results support the emergir^ consensus that
variable. The calibraticm curve shown in Figure 1 represents psychological, perscmal/ctemographic, and broader social and
the agp^^ated results of the number of correct responses in exxaxxsac factms have a limited ability to distinguish
each cat^ory (50%, 60%, ... 100%) divided by the total entreproieurs fivm managers in large organizatirais. More
number of responses in each category. For exan^le, if aU the impcHtantly for this study, consistent stqiport was found for
responses in the .70% category givrai by managers were correct Hypotheses 1 and 2. Even after ccmtroUing for previously
7 out of 10 times, then one would conclude that as a group they examined factors, the ovenxmfidence and representativeness
tend to be perfectly calibrated (at least in that category). The measures remain statistically significant, and help distinguish
charting of these two curves in Figure 1 indicates that between entrq)reneurs and managers in large organizations.
entrepreneurs are overconfident in their choices in five out of An industry analysis was also conducted and the representative
the six probability categories wliile managers were heuristic and ovorconfldence bias still remained significant.
overconfldmt ordy three out of the six cat^ories. Additionally,
entrepreneurs were more overconfident than manago^ in lai^e
organizations in each of the categories except in the .8 range
probability where they were very nearly identical.

87
 TABLSl These obsenwticQs have a pc^entially inqxxtant in^jact in
Results of Logis^c R^ressi(» Analysis analyzing tius iielsti<Miidup between decisi(xi «m«l^ii|g and
peifoiinaix:e. If different individuals and ixganizatirais are
cognitivdy teased in diffoest ways, then they may make
Mdqjendent Variables Parameter Wald strat^c choices in finuiamentally difftaent ways. If these
Estimate Chi-Square cognitive biases are difBcult to change, t!;^ may represent
somces of sustained differaices amcmg individuals ami fiims.
SiK^ differences, in the field of strategic mam^^ement, have
Intercept -6.14»»* 14.77 been ^xma to be sources of sustained cc»npetitive advantage
Risk-taking .05 .22 aiKl sRistaiiwd cooqj^tive disadvantage (Barney, 1991). By
Confonnity .33** 7.35 recognizing thiU strategic ^cisirais can be ncarrati(»ial in
Educaticm 1.06*'* 25.13 differmt w i ^ , this research points to a possible ccmnecti«i
Age .02 1.07 b^ween c(^tive dieories of dedsitm making, strat^ic
NetwMking -.01 2.04 mana^ment, and firm perfmnance. Of course, additi(»ial
Alertness -.27+ 3.60 resemch will need to be conducted to exmnine tliese potential
Rqiresentativeness 1.79**» 22.22 linkages.
Overccmfideoce -2.87* 5.50
In additi(m to extending this researdi to examine otho' kinds of
biases and lwuristics, future wtrak will need to eicanune
Pseudo-R2 37.1 wtKther noD-rational decision making remains stable over time.
Model Chi-Square %.6*** Some have argiKd that biases and heuristics are often aj^lied
df 155 in an uncraiscious manner (Kahneman et al., 19^), and thus,
Hit Ratio (%) 78 are relatively immune fruxn change fx modification.
Altenu^vely, otl»rs have noted that decision biases can be
COTrectedthrou^training(Fraig&Nisbett, 1991). Iftheuseof
p<.10 • p<.05 p<.01 p<.001 biases and heuristics are stable over time, then it would follow
that those \vixo are unaanfcntable with heuristics based
decisitm making, on average, will remain associated with larger
DBCUSSION firms «4iere this type of decisi<m making is less frequently
required. On the aOier band, those \«4io are comfcntable with
Common experience suggests that entrepreneurs and managers creating and relying on tl^se decisi<Ri making shortcuts are
in large organizati<His are, somehow, different from each a^ixer. likely to be attracted to entrefn^neurial sitings wtere these
It has been soniewiiat diaxincerting that most academic efforts (tecision skills are best utilized (Schneider, 1987). The
at discovering the sources of these experienced differences have assumption future research could address is wiiether
met with limited success. This paper presents empirical individuals with diffoent decision preferences will naturally
evidence suggesting that entrepreneurs and managers in large and efficiently select into organizational contexts wiiere those
organizations aie different from one another, and that these preferences are valued and accepted, such as into
differraices are substantial in size. By applying the theory of entrejneneurial firms.
biases and heuristics, this pqier has sliown that entrepreneurs
and managers in large organizaticms think differently. Of
course, future research will need to exanmie other biases and There are several practical implications that emerge frrmi this
heuristics, to more completely explain the diffraences between study as well. Fra* example, most research on biases and
entrepreneurs and managers in large oiganiza!i<ms. heuristics focuses on (tecisicm errcHS caused by these deviations
frxHn rational decisi<xi making (Lopes, 1991). We speculate,
This research on two biases and heuristics helps resolve some without these bias^ and heuristics, many decisions wcmld
counter-intuitive conclusions frtmi previous woik. For nevo^ be made. In entrepreneurial ventures, in particular, the
example, many psycholc^cal based studies have shown that the wincbw of (^jpcntunity would often be gone by the time all the
risk taking propensity of entrepreneurs is not greater than that necessaiy infonnati(m becanK available fen' more raticnal
of managers in large organizations (Low & McMillan, 1988). decision making. Additionally, successfully starting a new
This is tnie, even though, objectively, aitrepreiKurs %em to business usually involves overcoming multiple hurdles. Using
accept higher levels of risk in their career choices and business biases and heiuistics as simplifying mechanians fw
decisions than managers in large organizaticms. This apparent with these multiple problems may be cnicial. To &ce
contradicti<m may be resolved by recognizing that the mote hurdlesfrYsna rational perspective would not only
extensive use of the overconfidence and representativeness by decisions, but would in all likelihood beccmie ov
entreprrateurs is likely to lead them to perceive less risk in a
given decision situatirai than managers. By beii^ more willing As an illustratim of die possible benefits of biased decisirai
to generalize frtnn limited experience (represoitativeness), and making, c(Hisider the following. One entrejxeneur from our
by feeling more confident tlmt they will be able to ovenxHne sample wrote us because he todc issue with Hie quantitative
any unforeseen challenges, entiefH-eneurs may conclude that a emi^iasis of our study. Ovorall, he rejected this systenurtic
situation is simply less risky. Thus, it is not differences in ri^ a{qxY>ach to decisimi making. Ife stated:
pnqiensity that distinguishes entrefn-eneurs from managers in You see, pe(^le wfto are engaged in businesses such as
large organizaticms, but the ways they perceive and think about mine, are rarely influencai by surveys became Aey don't
place any stock in them. Survey reports, in general, are
risk.

88
 most highly prized by those individuals who lack sufftdent Ginsberg, A., & Buchholtz, A. 1989. Are entrepreneurs a
knowledge ^a matter in which they are required to make a toeed apart? A look at the evidence. Joumal of Genaal
decision. It is my considered opinion that those individuals Managanent. 15(2): 32-40.
are not going to be found successfully engaging in
entrepreneurial businesses. Hambrick, D , & Crozier, L. 1985. Stumblers and stars in the
managanent of rs^d growth. Joumal of Biisiness
Venturing. 1(1): 31-45.
The use of biases and heuristics may also offer some help in
explaining why oitreproieurs sometinies make bad manago^. Hogarth, R. M. 1987. Judgement and choice: The psychology
While the use of cognitive biases may be beneficial in some ofdecisicms. New York: John wiley & Sons.
circumstances, it can lead to major errxH^ in others (TvCTsky &
Kahneman, 1974). While research has yet to establish
perfranumce in[q>lications, it is possible that the mrae extensive Kahneman, D., Slovic, P., & Tversky, A. 1982. Judgment
use of heuristics in strat^c decision making may be a great under uncertainty: Heuristics and biases. New York:
advantage during the startup years. However, it may also le»i Cambridge University Press.
to the demise of a tnisiness, particularly as a firm matures.
Lichtoistein, S., & Fischhoff, B. 1977. Do those who know
m<He also know more about how much they know?
Organizational Behavior and Human Performance. 20:
REFERENCES 159-183.
Bar-Hillel, M. 1979. The role of sample size in sample Lopes, L. L. 1991. The rhetoric of irrationality. ThecB:y &
evaluation. Op>ani7ational BehaviOT and Human Psychology. 1(1): 65-82.
Performance. 24: 245-257.
Low, M. B., & MacMillan, I. C. 1988. Entrepreneurship: Past
Barnard, C. 1968. The functions of the executive. Boston: research and future challenges. Joumal of Management.
Harvard UnivCTsity Press. 14(2): 139-161.
Bamey, J.B. 1991. Firm resources and sustained ccnnpetitive Manimala, M. J. 1992. Entrepreneurial Heuristics: A
advantage. Joumal of Management. 17: 99-120. asnparison between high PI (pioneering-innovative) and
low PI ventures. Joumal of Business Venturing. 7: 477-
Bazerman, M. H. 1986. Managerial decision inakinp. New 504.
York: John Wiley & Sons.
Nisbett, R., & Ross, L. 1980. Human Inference: Strategies
Begley, T., & Boyd, D. 1987. Psychological characteristics and ShOTtcomings of Social Judgments. Englewood Cliffs,
associated widi perframance in entrepreneurial firms and NJ: Prentice-HaU.
smaller businesses. Joumal of Business Venturing. 2: 79-
93. Petti^ew, A. M. 1973. The politics of organizational decision
making. London: Tavistock.
Cooper, A. C , Dunkelberg, W.C, & Woo, C.
Entrepreneurs' perceived chances for success. Schneider, B. 1987. The peq)le make the place. Personnel
Business Voituring. 3: 97-108. Psychology. 40: 437-453.
Eisenhardt, K. 1989. Making fast strategic decisions in high- Smith, K., GamKm, M., Grimm, C , & Mitchell, T. 1988.
velocity envinmments. Academy of Management Joumal. Decision making behavior in smaller oitrepreneurial and
32(3): 543-576. larger professicmally managed firms. Joumal of Business
Venturing. 3: 223-232.
Fischhoff, B., Slovic, P., & Lichtenstein, S. 1977. Knowing
with certainty: The appropriateness of extreme confidence. Stevenson, M. K., Busemeyer, J. R., & Naylor, J. C. 1990.
Joumal of Experimental Psychology: Human Perception Judgmoit and decision-making theory, ia M. D. Dunnette
and PerfOTmance. 3: 552-564. & L. M. Hough (Eds.), Handbook of industrial and
organizational psychology (2nd ed): 283-374. Palo Alto:
Fong, G. T., & Nisbett, R. E. 1991. Immediate and delayed Ccmsulting Psycholc^sts Press.
transfer of training effects in statistical reascming. Joumal
of Experimental Psychology: General. 120(1): 34-45. Tversky, A., & Kahneman, D. 1974. Judgment undo-
uncertainty: Heuristics and biases. Science. 185: 1124-
Fong, G. T., Krantz, D. H., & Nisbett, R. E. 1986. The effects 1131.
of statistical training on thinking about eyeryday problems.
Cognitive Psychology. 18: 253-292.

Giimore, T. N., & Kazanjian, R. K. 1989. Clarifying decision

making in high-growth ventures: The use of responsibility
charting. Joumal of Business Venturing. 4: 69-83.

89