INFORMATION EXPOSURE, OPPORTUNITY EVALUATION, AND ... · herd, 2006; Shane, 2001; Shane &...

INFORMATION EXPOSURE, OPPORTUNITY EVALUATION,AND ENTREPRENEURIAL ACTION: AN INVESTIGATION OF

AN ONLINE USER COMMUNITY

ERKKO AUTIOImperial College Business School

LINUS DAHLANDERESMT European School of Management and Technology

LARS FREDERIKSENAarhus University

We study how an individual’s exposure to external information regulates the evalua-tion of entrepreneurial opportunities and entrepreneurial action. Combining data frominterviews, a survey, and a comprehensive web log of an online user communityspanning eight years, we find that technical information shaped opportunity evalua-tion and that social information about user needs drove individuals to entrepreneurialaction. Our empirical findings suggest that reducing demand uncertainty is a centralfactor regulating entrepreneurial action, an insight that received theories of entrepre-neurial action have so far overlooked.

Opportunity discovery and entrepreneurial ac-tion are at the heart of modern theories of entrepre-neurship (Kirzner, 1973, 1997; McMullen & Shep-herd, 2006; Shane, 2001; Shane & Venkataraman,2000). The recognition of third-person entrepre-neurial opportunities—that is, the conjecture thatan opportunity exists for the willing and able to sellproducts and services at a price greater than thecost of their production—is a necessary but notsufficient prerequisite for entrepreneurial action(Grégoire, Shepherd, & Schurer Lambert, 2010;

Shane & Venkataraman, 2000).1 Because entrepre-neurship is purposive and consequential social ac-tion (Granovetter, 1985), an individual also needsto form a belief that the opportunity offers a feasibleand desirable course of action for him- or herself, asotherwise the individual would not act. To advancethe empirical understanding of entrepreneurial ac-tion, we therefore study how individuals recognizeand evaluate third-person opportunities for entre-preneurial action and what factors influence theconjecture that the recognized opportunities repre-sent feasible and desirable courses of action (Shep-herd, McMullen, & Jennings, 2007).

Although opportunity recognition and evalua-tion are key prerequisites of entrepreneurial action,there is surprisingly little empirical research onhow individuals recognize and evaluate third-per-

All authors contributed equally to this work. We thankthe many people from the Propellerhead online usercommunity and the Propellerhead software staff for con-tributing to this research over a number of years. Wereceived valuable comments from this journal’s pastchief editor, Duane Ireland; from Oliver Alexy, MattBothner, Nicolai Foss, Eric von Hippel, Nicos Nicolaou;from seminar participants at Bologna University, CassBusiness School, ESMT, SKEMA Business School, Til-burg University, DRUID, an Academy of Managementconference, and the workshop “Open and User Innova-tion”; and from three anonymous reviewers. ProfessorAutio’s research was supported by the UK EnterpriseResearch Centre sponsored by the ESRC, BIS, BBA, andTBS (grant ref. ES/K006614/1) and the EPSRC-sponsoredInnovation Studies Centre (grant ref. EP/F036930/1). Wehave no financial interest in Propellerhead.

1 We use the term “opportunity discovery” to refer toinstantaneous reduction in radical uncertainty (Kirzner,1997). “Opportunity recognition” refers to an individu-al’s awareness of third-person opportunities for entrepre-neurial action. “Opportunity evaluation” refers to his/heraction of evaluating whether third-person opportunitiesrepresent feasible and desirable first-person opportuni-ties for entrepreneurial action. We operationalize oppor-tunity evaluation as business planning and entrepreneur-ial action as specific venture organizing and operatingactivities (also see Methods).

1348

� Academy of Management Journal2013, Vol. 56, No. 5, 1348–1371.http://dx.doi.org/10.5465/amj.2010.0328

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

son opportunities, and when opportunity evalua-tion is likely to lead to entrepreneurial action(Haynie, Shepherd, & McMullen, 2009). The argu-ably dominant influence on research on entrepre-neurial opportunity discovery—the “Austrian” tra-dition—has treated it as a virtually instantaneousevent that is almost automatically followed by op-portunity pursuit (Kirzner, 1997). In this tradition,emphasis is attached to factors regulating opportu-nity discovery, while the processes regulating op-portunity evaluation and entrepreneurial action aregiven less attention (Shane, 2000; Shane & Venkat-araman, 2000).

In an important complement to the opportunitydiscovery tradition, McMullen and Shepherd(2006) argued that central to entrepreneurshipis not whether objective and observable opportuni-ties exist independently of individuals, but rather,whether individuals perceive opportunities and actupon them. Their action approach portrays indi-viduals as reacting to external stimuli and formingthird- and first-person opportunity beliefs as a re-sult. Third-person opportunity recognition is a nec-essary precondition for opportunity evaluation,and if the evaluation results in a first-person oppor-tunity belief, the entrepreneurial action of sellingproducts and services may be triggered. The pro-cess of opportunity evaluation is portrayed as anintensive cognitive process, during which informa-tion describing a person’s external environmentguides and updates his/her emergent cognitive rep-resentations (Shepherd et al., 2007). However, be-cause it is difficult to obtain unbiased observationsof individuals in prefounding situations (Shane &Khurana, 2003), little research exists on where ex-ternal information stimuli come from, how individ-uals evaluate them, and under which conditions asufficient reduction in response uncertainty isachieved to trigger entrepreneurial action (Grégoireet al., 2010). The few empirical studies buildingupon McMullen and Shepherd’s (2006) action the-ory of entrepreneurship employ experimental de-signs to examine how individuals judge opportu-nity scenarios (Fitzsimmons & Douglas, 2011;Grégoire et al., 2010; Haynie et al., 2009). Whileilluminating the cognitive processes of opportunityevaluation, such experimental designs tell littleabout where stimuli arise from in practice andwhich conditions facilitate the conversion of theseinto first-person opportunity beliefs and entrepre-neurial action.

In this article, we address this gap by exploringhow an individual’s exposure to external informa-

tion regulates the recognition and evaluation ofthird-person opportunities, on the one hand, andsubsequent entrepreneurial action, on the other.Specifically, we explore how the production anddistribution of information regarding “means” and“ends” in social systems regulates these processes.We suggest that in contexts characterized by fre-quent technological advances, potential entrepre-neurs are usefully thought of as embedded in notone, but two, interconnected information domains:one associated with technological advances (i.e.,“means”), and another associated with user needs(i.e., “ends”). The two domains exercise differentinfluences on opportunity evaluation and entrepre-neurial action. Specifically, we propose that tech-nological information regulates third-person op-portunity recognition and opportunity evaluationby individuals. We further posit that exposure toinformation about user needs drives entrepreneur-ial action, both in its own right and (especially)when combined with information about technolog-ical solutions.

We test our theoretical model using data from anonline user community. Online user communitiesare internet-based platforms for communicationand exchange among users interested in a givenproduct or technology (Baldwin & von Hippel,2011; Jeppesen & Frederiksen, 2006). They are typ-ically not confined to the boundaries of any givenorganization. Joining such communities is usuallyeasy, and there is usually little screening of com-munity members beyond self-selection based onthematic interest (Dahlander & O’Mahony, 2011;Lee & Cole, 2003). In online user communities,users of different products and services share anddebate experiences, news, and ideas regardingproduct and service features, uses, and improve-ments (von Krogh & von Hippel, 2006). Such com-munities facilitate collective experimentation andexploration around a given set of technologies andthus provide an ideal platform for the study of theeffect of technological and user information on op-portunity recognition and entrepreneurial action(Haefliger, Jaeger, & von Krogh, 2010; Lakhani &von Hippel, 2003). We study the Propellerhead on-line community, which has stimulated extensiveuser innovation and entrepreneurship in softwarefor music production, processing, and editing(Jeppesen & Frederiksen, 2006).

Our study offers four distinctive contributions.First, we provide a rare examination of the facilita-tors of opportunity evaluation and entrepreneurialaction by individuals in prefounding situations,

2013 1349Autio, Dahlander, and Frederiksen

thereby contributing to the action theory of entre-preneurship. Second, and specifically, we advanceunderstanding of how the production and distribu-tion of different kinds of information in social sys-tems facilitates individuals’ opportunity evaluationand entrepreneurial action. Third, our work con-tributes to the emerging research stream on userentrepreneurship (Shah & Tripsas, 2007) by movingbeyond descriptive and conceptual studies towardtheory testing. Finally, while earlier work showsthat user communities are important platforms forinnovation and experimentation (Dahlander &Frederiksen, 2012), we demonstrate how thesecommunities can also provide a fertile ground forentrepreneurial action.

In addition to providing empirically groundedinsight into the origins and operation of stimuli foropportunity evaluation and entrepreneurial action,our model also has broader implications for man-agement scholars. Entrepreneurial action is notconfined to new and small firms alone. For estab-lished firms to recognize and act upon diversifica-tion opportunities, they need to match the discov-ery of technological opportunities with theidentification and evaluation of unrealized userneeds (Alderson, 1965). Although the need to re-duce both supply and demand uncertainty is com-monly accepted by management scholars, it has notbeen widely tested empirically. Our model shouldtherefore have implications in contexts other thannew ventures—for example, corporate venturing.

THIRD-PERSON OPPORTUNITY RECOGNITIONAND EVALUATION AND FIRST-PERSON

OPPORTUNITY BELIEFS

Entrepreneurship is purposive social action thatrequires foresight, effort, and resources. Entrepre-neurial initiatives therefore reflect an individual’sconjecture that a first-person opportunity existsand is attractive—that is, if realized, the opportu-nity promises returns greater than the cost of itspursuit (Eckhardt & Ciuchta, 2008). To explicatehow such conjectures are formed, McMullen andShepherd (2006) drew on the psychology and phi-losophy of action to build an action theory of en-trepreneurship. Central to their conceptualizationwere the notions of judgment, response uncer-tainty, and doubt. Following Hebert and Link(1988) and Casson (1982: 22), they defined an en-trepreneur as someone who specializes in makingjudgmental decisions about the coordination ofscarce resources. Such decisions are made in re-

sponse to external stimuli. Whether or not an indi-vidual interprets an external stimulus as signaling afeasible and desirable course of entrepreneurial ac-tion will depend on judgments formed during op-portunity evaluation. Because of the uncertaintycharacteristic of future-oriented endeavors, entre-preneurial judgments are accompanied by doubt,which tends to block or delay action (Lipshitz &Strauss, 1997) by giving rise to hesitancy, indeci-sion, and procrastination (McMullen & Shep-herd, 2006).

Given that uncertainty inhibits entrepreneurialaction, it is important to consider how potentialentrepreneurs evaluate third-person opportunitiesand overcome uncertainty. McMullen and Shep-herd (2006) posited that the opportunity recogni-tion process consists of “attention” and “evalua-tion” stages. In the attention stage, individualsobserve third-person opportunities by virtue ofhaving been sensitized to a given set of problems(e.g., dissatisfaction with existing digital tuning al-gorithms, in our empirical context) and exposed toinformation describing potential solutions. Oncean individual recognizes a third-person opportu-nity, a first-person evaluation ensues, during whichthe individual identifies a defined course of action(i.e., the launch of an entrepreneurial venture toexploit the opportunity) and determines whetherthe identified course of action is feasible and desir-able. If the evaluation suggests a strong enoughcomposite of feasibility and desirability, action tocreate value by selling products and services willbe triggered.

This schematic illustration of the attention andevaluation stages is silent about how the evaluationis actually performed and what triggers entrepre-neurial action. In a follow-on work, Shepherd et al.(2007) elaborated on this aspect and presented amodel of the formation of opportunity beliefs byindividuals. They drew on coherence theory (e.g.,Rensink, 2002) to build a process model that por-trayed opportunity belief formation as either a cog-nitive bottom-up process driven by an individual’sformation of gestalt-like mental representations ofhis/her environment or a top-down process drivenby the individual’s preconceived knowledge andmotivation structures, which are updated againstinformation describing the environment. This sug-gests that the formation of first-person opportunitybeliefs is a cognitively intense process character-ized by interaction between an individual’s knowl-edge and motivation, and external information.

1350 OctoberAcademy of Management Journal

The centrality of external information as an ini-tiator and regulator of opportunity beliefs raises thequestion of whether and how information exposureinfluences the process. Because of its narrow focuson entrepreneurs’ cognitive processes, the actiontheory of entrepreneurship does not provide de-tailed elaboration of this aspect, beyond generalreferences to external stimuli and individuals’ ex-posure to information flows. We think this questionmerits attention, since individuals can only recog-nize and react to stimuli that they are exposed to,and they can only combine their knowledge andmotivation with environmental information thatthey are able to access. It is therefore important toconsider how the exposure to different kinds ofinformation influences opportunity evaluation andentrepreneurial action by individuals.

The effect of information on opportunity discov-ery is extensively discussed in research that drawson Austrian economics (Eckhardt & Shane, 2003;Kirzner, 1997; Shane & Venkataraman, 2000). In theAustrian portrayal, the distribution of informationin social systems regulates opportunity discovery.Individuals partly access, partly create this infor-mation through their interactions with others, asthey mutually learn about one another’s plans,preferences, and intentions (Kirzner, 1997). Thus,rather than regulating action, information regardingopportunities is produced through speculative ac-tions by market participants. Because of the wayinformation is partly embedded in social systems,partly produced through mutual interaction, it isnot fully knowable to any single individual ex ante,but rather, only gradually revealed through interac-tions with others (Hayek, 1945). New opportunities

are discovered through uncovering new means-ends relationships that, if fruitfully combined withan individual’s idiosyncratic skills, experience,and resources, trigger entrepreneurial action (Eck-hardt & Shane, 2003; Shane, 2000).

The opportunity discovery perspective suggeststwo important insights into how an individual’sexposure to external information influences oppor-tunity recognition and evaluation and subse-quently drives entrepreneurial action. First, giventhe way information relevant for the formation ofopportunity beliefs is embedded in social structure,an individual’s position within a social structureshould matter for first-person opportunity beliefsand entrepreneurial action. Second, the prevalentnotion of opportunities taking the form of newmeans-ends relationships implies that to embarkon entrepreneurial action, the individual needs tobe exposed to information of at least two kinds—information on both means and needs. Informationrelevant to each may reside in different parts of thesocial structure, and the individual’s position maytherefore matter differently for exposure to the twodifferent kinds of information.

In what follows, we build a model that explicateshow an individual’s exposure to information re-garding means and ends regulates the recognitionand evaluation of third-person entrepreneurial op-portunities and entrepreneurial action, respec-tively. Figure 1 presents our model, which distin-guishes between exposure to information residingin the technological domain and that in the socialdomain. We submit (following McMullen andShepherd [2006]) that an individual’s exposure totechnological advances will enhance the recogni-

FIGURE 1Conceptual Model

Technological Domain

Lead user attributes

Technological probing

H1 +

H2 +

Third-PersonOpportunity Recognition

+

First-PersonOpportunity Beliefs

Opportunity evaluation Entrepreneurial action

Social Domain

Community attention

Community spanning

+ H3

+ H4

a We expect a positive correlation between opportunity evaluation and entrepreneurial action, as illustrated with the positive sign.However, in the article we do not theorize about this relationship. Our estimations are able to empirically deal with this correlation.


tion of third-person opportunities and trigger op-portunity evaluation activities. We also suggest thatfor the individual to decide that opportunities rep-resent feasible and desirable first-person opportu-nities, information on user needs is necessary.However, we are not suggesting that entrepreneur-ial action should always be preceded by an empir-ically observable evaluation phase. As Shepherd etal. (2007) noted, individuals, especially when inthe “top-down” mode of opportunity belief forma-tion (a mode driven by established knowledgestructures rather than emergent cognitive represen-tations of environment, as is the “bottom-up”mode), can form first-person opportunity beliefsquite rapidly, even instantaneously.

HYPOTHESES

Exposure to the Technological Domain andThird-Person Opportunity Recognition

Opportunities evolving in the technology do-main represent observable third-person opportuni-ties for individuals. Technological progress con-sists of the creation of physical and knowledge-based artifacts that can be harnessed to serveinstrumental purposes (Niiniluoto, 1993). Becausetechnological advances exploit organized knowl-edge concerning physical reality, technological in-formation tends to be more structured than infor-mation concerning user tastes and preferences(Clark, 1985; Wade, 1996). It is organized aroundtechnological trajectories (Dosi, 1982), punctuatedby technological guideposts (Sahal, 1985), and con-strained by dominant designs (Abernathy & Utter-back, 1978; Anderson & Tushman, 1990). Thesecharacteristics make technological progress morepredictable than social progress, enabling individ-uals to predict, recognize, and contemplate third-person opportunities created by technologicaladvances.

Although technological progress is largely pre-dictable, not all individuals are equally positionedto access and evaluate information concerningtechnological advances. Two factors are particu-larly salient in this regard. First, individuals’ expe-rience-induced awareness of technological bottle-necks, along with their experience searching forsolutions, sensitizes them to information stimulithat signal third-person opportunities. These qual-ities resonate closely with the notion of lead users(Morrison, Roberts, & Midgley, 2004; von Hippel,1988). Second, technological probing enables indi-

viduals to participate in, and sometimes lead, thearticulation and creation of new technological ad-vances in a community (Morrison, Roberts, & vonHippel, 2000). In the following section, we elabo-rate how these characteristics facilitate an individ-ual’s recognition and evaluation of third-person op-portunities created by technological advances.

Lead User Attributes and Third-PersonOpportunity Recognition

Lead users are individuals whose goal fulfillmentis hampered by technological performance bottle-necks and who therefore expect significant benefitsfrom technological advances in a given solutiondomain (von Hippel, 1988). Since lead users’ abil-ity to meet desired ends is constrained by the tech-nological state of the art in this domain, they aremore likely than ordinary users to search for poten-tial solutions and be cognizant about and monitor-ing relevant technological advances. This scanningactivity makes lead users better aware than ordi-nary users of emerging technological trends andassociated third-person opportunities. Lead usersshould therefore be more likely than ordinary usersto engage in evaluation of third-person opportuni-ties for entrepreneurial action.

Sometimes lead users experience technologicalconstraints so acutely that they start developingtechnological solutions by themselves (Morrison etal., 2000). A literature has documented a multitudeof examples of lead users who have developedproduct modifications, add-ons, and completelynew products in a wide range of industries (Franke& Shah, 2003; von Hippel, 1988). To develop newtechnological solutions, lead users occasionally in-teract with others who are likely to possess the kindof state-of-the-art knowledge required to developsolutions. This activity should further enhancelead users’ recognition of third-person opportuni-ties, as well as the likelihood that they engage inopportunity evaluation.

Although lead users experience bottlenecks andperformance constraints more clearly than ordinaryusers, their estimations regarding the potential eco-nomic value of technological solutions remain sub-jective and may lack external validation. Whilelead users may be sensitized to technologicaltrends, they may lack access to endorsement andinformation from other users. However, the assess-ment of the economic potential associated withthird-person opportunities opened by technologytrends requires access to information regarding


other users’ needs (Companys & McMullen, 2007;Shane, 2000). Therefore, although early discoveryof third-person opportunities is likely to promptspeculation whether user-developed solutionsmight be useful to other users (Jeppesen & Laursen,2009), solution discovery alone may not be suffi-cient to trigger entrepreneurial action:

Hypothesis 1. Individuals with a high degree oflead user attributes are more likely to engage inopportunity evaluation.

Technological Probing and Third-PersonOpportunity Recognition and Evaluation

Communication among users in online commu-nities is organized around discussion “threads”:users open new discussion threads to signal andframe a new issue, problem, call for advice, or areaof exploration. We characterize this activity astechnological probing. We propose that technolog-ical probing provides informational advantagesthat facilitate the recognition of technological op-portunities, prompting evaluation of these. This isbecause technological probing makes users betteraware of technological trends and developments,and it also helps advance desired developmentswithin a user community.

Individuals active in technological probing canenjoy early access to information about emergingtechnological developments in online user commu-nities. By opening new discussions, technologicalprobers can receive early feedback about the viabil-ity of technological developments of interest forthem. This information advantage enhances theirability to recognize which developments are tech-nically possible and which developments a usercommunity supports. However, because technolog-ical probing explores emergent technological pathsrather than unrealized user needs, it does not nec-essarily yield information advantages regardinguser needs. Absent information about the potentialeconomic value of alternative technological devel-opment paths, the enhanced recognition of third-person opportunities in the technological domainshould encourage opportunity evaluation but notnecessarily entrepreneurial action.

Technological probing not only increases users’awareness of technological opportunities, but alsomay enhance the viability of the technological de-velopments probed. By highlighting given techno-logical issues and solutions, technological probersdraw the attention of their user community to

these, helping prompt technological developmentactivity around the issues probed. This way, tech-nological probers may act as opinion leaders whohelp steer the allocation of technological effort inthe community toward given technological issuesand solutions (see also Myers & Robertson, 1972).Although the opinion leadership effect is, in mostcases, an unintended consequence of technologicalprobing activity, the agenda-setting effect of suchprobing should nevertheless facilitate the emer-gence of technological solutions that are of interestfor the technological probers. The emergence ofsuch solutions should trigger opportunity evalua-tion. Summarizing, we predict:

Hypothesis 2. Individuals engaged in a highdegree of technological probing are more likelyto engage in opportunity evaluation.

Exposure to the Social Domain andEntrepreneurial Action

We argued that exposure to technological infor-mation imbues individuals with domain insightand stimulates opportunity evaluation. However,in the absence of strong preconceived knowledgeand motivation, third-person opportunities alonedo not necessarily prompt the conjecture that theopportunities observed represent actionable first-person opportunities for an individual. If responseuncertainty prevails, the individual is likely to hes-itate and forego or postpone action (McMullen &Shepherd, 2006). We propose that the individual’sexposure to information regarding user needs re-duces response uncertainty and lowers his/herthreshold for entrepreneurial action.

Our model contains two factors in the social do-main that affect individual propensity for entrepre-neurial action. First, we expect that communityattention received an individual receives fromother users facilitates insights regarding unrealizeduser needs. Second, we expect that communityspanning enables the individual to draw on analo-gous examples in related domains to validate ten-tative conjectures regarding user needs, and thus,judge the potential economic value of newdevelopments.

Community Attention and First-PersonEntrepreneurial Action

In the context of entrepreneurial action, an im-portant aspect of response uncertainty concerns the


scale of market demand for new products and ser-vices. We define market uncertainty as the sum ofdemand uncertainty, product uncertainty, and sup-plier uncertainty. Demand uncertainty concernsunderlying market demand (i.e., Is there a need fora solution like this?). Product uncertainty concernsthe attractiveness of a given solution relative toother solutions (i.e., Would other users choose thisparticular solution?). Supplier uncertainty involvesthe legitimacy of an individual as the supplier ofthe solution (i.e., Would other users buy this solu-tion from me?). Community attention should lowerall three aspects of market uncertainty. Attentionsignals that the individual’s actions resonate withothers in his/her user community and providesproof of widely experienced needs and emotionalcommitment to the eventual outcome, increasingindividual’s confidence that others will commit toa particular solution (Franke, Schreier, & Kai-ser, 2010).

The magnitude of consumer response is a well-established metric of market demand in the mar-keting literature (Rein, Kottler, & Stoller, 1987). Forexample, consumer goods companies routinely usethe speed and magnitude of returned coupons as areliable metric of consumer demand (Berry & Li-noff, 1999). As users share experiences, proposeand select solutions, and exchange ideas regardinga given product in user communities, their re-sponses reveal what they care about (Whittaker,Terveen, Hill, & Cherny, 1998). Therefore, re-sponses to online postings provide an importantsource of information on user needs. The higher thenumber of responses received, the greater the re-duction in demand uncertainty should be.

Individuals who engage in a conversation regard-ing new technological developments by respondingto postings are more likely to develop a feeling ofpsychological ownership toward any solution de-veloped as the outcome of the conversation (Frankeet al., 2010). This form of consumer empowermentcan be a potent tool for increasing users’ emotionalattachment to a given product (Fuchs, Prandelli, &Schreier, 2010). Franke et al. (2010) demonstratedthat proposing solutions to a problem shared in auser community increased respondents’ emotionalcommitment to the eventual solution and theirwillingness to pay—a dynamic referred to as the “Idesigned it myself” effect (Norton, 2009). Attentionfrom other users is therefore a promising sign ofother users’ likely commitment to any new devel-opments resulting from a conversation, thereby re-ducing market uncertainty. In addition, a high level

of community attention has been shown to enhancethe ability of an individual to solicit positive emo-tional responses from other users, thereby increas-ing confidence that other users will react positivelyto his/her offerings (Kurzman et al., 2007).

To summarize, a high level of community atten-tion to a focal individual’s expressed technologicalinterests informs the individual that a widespreadneed exists for the solution or issue under discus-sion; signals that other users will likely experiencepsychological ownership of the resulting solution;and suggests that the community is likely to accepta solution supplied by that individual. All thislowers market uncertainty and therefore, thethreshold for entrepreneurial action:

Hypothesis 3. Individuals who receive a highdegree of community attention are more likelyto engage in entrepreneurial action.

Community Spanning and EntrepreneurialAction

We propose that community spanning, definedas participation in several communities, helps re-duce market uncertainty and thus lower the thresh-old for entrepreneurial action. First, communityspanners gain informational advantages from par-ticipation in several communities (Tushman &Scanlan, 1981; Vaghely & Julien, 2010). Second,community spanning enables individuals to drawon parallels and analogies when judging the eco-nomic viability of technological options and antic-ipating demand patterns (Gaglio & Katz, 2001; Lee,Goodwin, Fildes, Nikolopoulos, & Lawrence,2007). Third, individuals exposed to diverse“thought worlds” are able to frame problems differ-ently (Dougherty, 1992), which helps reduce per-ception of market uncertainty in any givencommunity.

Community spanners have the potential to ob-serve related, yet different, social dynamics. Expo-sure to different communities enables individualsto more accurately recognize incipient and poten-tially influential social trends and foresee theirlikely outcomes (Vaghely & Julien, 2010). Commu-nity spanning also allows individuals to learn andreuse different practices, tricks, and referencesfrom different communities, strengthening the fo-cal individuals’ belief in the first-person viability ofthird-person opportunities.

Community spanning enables individuals todraw on analogies when evaluating opportunities.


Analogous reasoning provides a useful device forpredicting future developments in uncertain situa-tions (Gaglio & Katz, 2001), because it provides asource of surrogate experience that enables the in-dividuals to more accurately estimate the outcomesof socially complex developments (Autio, George,& Alexy, 2011). Demand patterns often unfold insimilar ways in related communities: the useful-ness of drawing on observed demand patterns inrelated contexts is well established in the area ofdemand forecasting (Lee et al., 2007). The ability todraw on analogous examples from other communi-ties therefore lowers demand uncertainty for a focalindividual.

Finally, community spanning exposes individu-als to different thought worlds and ways of framingproblems (Dougherty, 1992). An important aspectof third-person opportunities concerns the assess-ment of alternative approaches to opportunity pur-suit. By virtue of their ability to better reframe theproblem to which a given technological develop-ment is applied, community spanners are able tocircumvent seemingly difficult obstacles, therebysidestepping aspects of demand uncertainty. Whendemand uncertainty is lowered, entrepreneurial ac-tion is more likely to follow:

Hypothesis 4. Individuals who engage in a highdegree of community spanning are more likelyto engage in entrepreneurial action.

We hypothesized that exposure to a technologi-cal domain enhances the likelihood of opportunityevaluation, while exposure to a social domain en-hances the likelihood of entrepreneurial action.However, we are not formally hypothesizing a me-diating relationship here, as we recognize that theformation of first-person opportunity beliefs cansometimes be virtually instantaneous (Shepherd etal., 2007). Our approach is also consistent with thatof Haynie et al. (2009), who argued that since op-portunity evaluation gives rise to future-orientedcognitive representations of what would happen,

were the individual to pursue an opportunity, op-portunity evaluation is likely to proceed and beseparate from a decision to exploit. Thus, we rec-ognize that individuals can exhibit four combina-tions of opportunity evaluation and entrepreneur-ial action: (1) no evaluation of third-personopportunities and no entrepreneurial action; (2)evaluation of third-person opportunities withoutaction (i.e., “unrealized opportunities”); (3) entre-preneurial action with virtually instantaneous eval-uation (i.e., “opportunistic action”); and (4) evalu-ation combined with action (i.e., “premeditatedaction”). Figure 2 illustrates these combinations.

METHODS

Research Setting

To test our theoretical model, we use unique datafrom a public, unrestricted online user community,hosted by the Swedish music software firm Propel-lerhead, that focuses on digital music productionand editing software. This company has hosted awebsite since 2001 to allow users to communicatewith one another, solve problems, and tinker withits digital music production and editing platform.Members of the community create, produce, pro-cess, and record music with Propellerhead’s soft-ware and develop related complements. Over theyears, the community has spawned more than a 100product modifications and customized applica-tions that either added features and functionalitiesto the software platform or provided connectivityto other software products in useful and value-adding ways. Because of this, the Propellerheadcommunity has become an important forum for thediscovery and pursuit of entrepreneurial opportu-nities and seen more than 30 new ventures createdby community members.

Because the community has active contributingmembers from more than 35 countries, and becausethe online user community provides the main plat-

FIGURE 2Opportunity Evaluation and Entrepreneurial Action

Entrepreneurial Action

No Yes

No Opportunistic

action Opportunity Evaluation

Yes Unrealized

opportunities Premeditated

action


form for interaction between these individuals, thevast majority of all interactions between commu-nity members have taken place online, where theyhave been recorded and made available for post hocanalysis. This setting therefore provides a uniqueopportunity to track how an individual’s experi-ence and position within a community influencethird-person opportunity recognition and evalua-tion and entrepreneurial action.

Data

We adopted a multimethod strategy and “triangu-lated” data by using different data sources (Jick, 1979;Miles & Huberman, 1994). A multimethod inductive-deductive approach permitted us to better validateconstructs and to choose an appropriate theoreticallens for the study (Edmondson & McManus, 2007).Our data came from (1) interviews, (2) a survey, and(3) a web log (message archive) of all interactions inthe Propellerhead user community (henceforth, “thecommunity”), thus helping us avoid commonmethod bias (Doty & Glick, 1998). We used web logdata and content coding to track users’ informationexposure prior to the distribution of the survey,which collected data on opportunity evaluation andentrepreneurial action. We matched survey data withweb log data using a unique identifier for eachrespondent.

Interviews. We conducted 42 interviews with 19individuals in the community and Propellerhead em-ployees before the end of 2011. The interviews pro-vided contextual grounding for our theory develop-ment and supported the empirical validity of ourtheory and constructs. We sought to understand howthe community worked in terms of communicationbetween users, problem-solving processes, and theideation, creation, and dissemination of new prod-ucts and solutions within the community.

To obtain insight into the history and evolutionof the Propellerhead community, we conducted in-terviews with the CEO of Propellerhead as well aswith its product developers and the moderator ofthe online user community. We interviewed indi-viduals within the user community, oversamplingindividuals who had started entrepreneurial firmson the basis of their experiences in the communitybut also including individuals who had not pur-sued entrepreneurial opportunities.2 We con-

ducted interviews face-to-face and by telephoneand also corresponded by e-mail. We recorded andtranscribed each interview and, in most cases, tooknotes during and after interviews. We used theinterviews to guide our questionnaire design andquantitative data collection. In parallel with ana-lyzing our quantitative data, we conducted addi-tional interviews and built interpretations from thequalitative data. When building our interpreta-tions, we iterated back and forth between data,relevant literature, and embryonic theory (Dough-erty, 2002). To provide contextual grounding forour interpretations and theory and to illustrate thepossible direction of underlying causal mecha-nisms (Kim & Miner, 2007), our Results sectionintroduces a few qualitative findings after we pres-ent the quantitative analysis.

Survey. We collected data on entrepreneurial ac-tivities among community members with an onlinesurvey administered at the community’s website.The questionnaire was open for replies forone month and attracted answers from 280 of 966users who logged in during this period. Only userswho logged in to the community pages were pre-sented with the option to participate in the survey.After a review, we dropped five responses becauseof missing information on the dependent variables,leaving us with a survey sample of 275 valid an-swers and a 28.5 percent response rate. The surveyemployed previously validated multi-item scales tomeasure our theoretical constructs. Following Pod-sakoff, MacKenzie, Jeong-Yeon, and Podsakoff(2003) we protected anonymity, counterbalancedquestion order, and used several scale anchors andscale formats. We pretested the survey with a smallnumber of individuals in the community beforedistributing it.

Our checks for response bias (Armstrong & Over-ton, 1977) revealed that individuals who re-sponded to the survey were slightly more activethan the general population in the community. Asexplained later in the Analysis section and theAppendix, we controlled for self-selection bias bycomputing the inverse Mill’s ratio, using informa-tion about nonrespondents from the web log data.

Web log of interactions. To complement the sur-vey data, we obtained data from the community’smessage archives (web log) of all interactions in thecommunity from its 2002 inception to May 2008.All communications within the online communityare recorded in online archives and provide a valu-able, unbiased source of behavioral data on intra-community interactions. We gathered complete in-

2 We also conducted interviews with people in thecommunity who cannot be labeled as lead users.


formation from all participants, including thesender of each posting, timing, subject, messagecontent, whether the posting was a first thread in anew conversation, and who the recipient was. Theweb log contained all 151,092 messages posted inthe community by a total of 8,640 individuals. Thisunique data set enabled us to monitor an individ-ual’s engagement and role in the user communityrather than rely solely on self-reported data. Weused a unique identifier to link the web log datawith the survey data.

We also used the web log to code the content ofmessages (Weber, 1990) to verify the quality, use-fulness, and degree of technical specification ofquestions and answers provided by users. Codingallowed us to control for any effects these aspectsexacted on users’ opportunity evaluation and en-trepreneurial action.

Measures

Dependent variables. Using data from the surveyinstrument, we developed two dependent variablesthat reflect third-person opportunity recognition andevaluation and first-person entrepreneurial action, re-spectively. Because entrepreneurial action requirestime and effort, it is typically preceded by premed-itation and planning (Bird, 1988). As our proxy forthird-person opportunity recognition and evalua-tion, we used a set of questions designed to deter-mine whether an individual had undertaken tangi-ble action to evaluate opportunities, drawing ontheir activities and experience in the Propellerheadcommunity. We chose a proxy that covers bothopportunity recognition and evaluation becauseopportunity recognition is difficult to observe di-rectly. Observing evaluation was appropriate, be-cause evaluation must necessarily be accompaniedby recognition of the stimulus that is being evalu-ated. Although not all stimuli are subjected to eval-uation, we chose to capture tangible evaluation ac-tivities because these signal meaningful intent (seeMcMullen & Shepherd, 2006: 148–149). Our opera-tionalization of opportunity evaluation is consis-tent with the theory of the psychology of action,which distinguishes between “motivational” and“volitional” phases of the opportunity awarenessand evaluation continuum and positions planningactivities in a volitional preactivity phase (Gollwit-zer, 1996: 289).

Opportunity evaluation was coded as a dummy(“yes” � 1) if (1) a respondent stated that he/shehad “made written plans to start a business” or

“looked for resources to start a new business (e.g.,funding, business partners, business premises,etc.)” and (2) stated that “my community experi-ence was of importance” in that action.3 Theseitems were adopted from the Panel Study of Entre-preneurial Development (PSED), which is widelyused in the entrepreneurship research community(Reynolds, 2007). When recognizing a potential op-portunity for entrepreneurship, individuals oftenundertake precursory activities—planning activi-ties intended to determine the feasibility and desir-ability of the new venture without representing adefinite commitment to the venture (Carter, Gart-ner, & Reynolds, 1996; see also Gollwitzer, 1990).Our operationalization is close to the one used byDelmar and Shane (2003); Shane and Delmar(2004), and Dimov (2010), who also used a PSED-based measure of business planning as a proxy foropportunity evaluation. Although business plan-ning represents a volitional planning activity (Goll-witzer, 1996), it does not automatically lead to en-trepreneurial action: indeed, the PSED shows thatnearly two-thirds of nascent entrepreneurs neveractually start a new business (Reynolds, 2007).

We measure entrepreneurial action by capturingdefinite venture organizing activities (e.g., Delmar& Shane, 2003). We captured entrepreneurial ac-tion with a dummy variable that was assigned thevalue 1 if an individual reported that he or she had“sent invoices for products or services offered byyour company,” “done the paperwork and offi-cially registered the company,” or “sold productsand services through my company” and that “mycommunity experience was of importance” in theseactions.4 The statements focusing on sales andvalue creation capture a tangible validation that anopportunity actually exists to sell products andservices (Eckhardt & Shane, 2003).

3 To ensure that the survey respondents’ activities sur-rounding opportunity evaluation and entrepreneurial ac-tion were directly linked to their community participa-tion, we included an additional scale in our questionsprobing the extent to which community experience wasimportant (“not important,” “of little importance,”“somewhat important,” or “very important”) relative tothe question at hand.

4 For validation purposes, we also constructed an al-ternative objective measure of entrepreneurial action us-ing data on start-ups in the community. We were able toassess this for all cases except a handful, and in thosecases in which we had data, this measure was perfectlyaligned with our self-reported measure.


Independent variables. We based our scale oflead user attributes on the work of Morrison et al.(2004). Because we perceived a few methodologicalissues with their scale, we extended it to eightitems in the online survey to create a more compre-hensive measure.5 We used a Likert scale rangingfrom 1 (“strongly disagree”) to 5 (“strongly agree”).The internal reliability of this scale was good (� �.75). The lead user attribute scale was formed as thearithmetic mean of the different items.

Opening new online discussion threads can be apowerful means to shape the agenda of a user com-munity. We therefore measured an individual’stechnological probing activity in the community bythe total number of new threads started by theindividual divided by the overall activity (totalnumber of postings) of the individual. Similarly,we measured the community attention received bythe individual by using the average number of re-sponses received for each online posting of theindividual.

Finally, we argued spanning multiple communi-ties provides informational advantages that lowerperceived response uncertainty for entrepreneurialaction. To measure community spanning, we useda two-stage question in the online survey. Weasked, “Have you actively participated in other on-line communities NOT hosted by Propellerheadduring the last three years?” For those individualsthat ticked “yes,” a follow-up question queriedabout the names of other online communities. Weused Google and other search engines to identifyand code all communities mentioned. The contentcoding of these communities was carried out bytwo of the authors independently and cross-checked afterward. We reasoned that participationin multiple online communities that were techno-logically related to the focal community, as op-posed to online community engagement in general,

would provide more pertinent information advan-tages. We therefore only included those communi-ties that, similarly to Propellerhead, dealt with (1)software products for music production, recoding,and processes, or (2) addressed music editing forcreating sound tracks, music, or (3) hosted commu-nities for inspiration and downloading new sam-ples, or (4) hosted communities for downloadingand engaging with software products not directlyconnected to music production but that supported,for example, software programming or sound trackproduction. We excluded general social networkssuch as Facebook and Yahoo user groups.

Control variables. Longer tenure in the commu-nity would allow individuals more time to recog-nize opportunities and, presumably, give themmore experience in sourcing information from it.We used the web log data to control for communitytenure, measured as the number of years since anindividual’s first posting in the community.

To isolate the community attention effect, weneeded to rule out the effect of sheer number ofpostings in the community. The nature of a givenindividual’s contributions varied considerably.Some postings were more sophisticated, asking, forexample, about suggested solutions for difficulttechnical problems. Others were thank you or idlechatter messages. To isolate relevant postings, wemeasured the number of questions using contentcoding. Only postings requesting information andhelping to solve technical problems were countedas relevant. We also controlled for the number ofanswers by only counting when an individual pro-vided an answer to another individual’s question.Before initiating this content coding, we read over5,000 postings, which we divided into distinct cat-egories. We developed a coding scheme with fourdifferent categories of postings: (1) “question,” (2)“answer/response,” (3) “comment,” and (4)“other.” We developed the coding scheme by ac-tively debating alternative categorizations. Twocoders then each coded the messages indepen-dently. When evaluating the coders after the first200 posts, we had an agreement of 86 percent and aCohen’s kappa of .83. We also discussed thesources of disagreements and how they could beovercome. Once we had ensured this initial reli-ability among the two coders, we distributed sam-ples of 2,700 posts to each coder to make sure wehad some overlap (2,700 � 2 � 5,400). To ensurehigh intercoder reliability, we had an overlap ofposts that we evaluated throughout the process.These checks were similar to the first test of inter-

5 We used the following items: “I find out about newproducts and solutions earlier than others in the commu-nity”; “I would benefit by early adoption and use of newproducts”; “I have tested prototype versions of new prod-ucts”; “I use Propellerhead’s products more regularlythan other software related products”; “I have made sug-gestions for how to improve Propellerhead’s products”;“I often identify important ideas or trends related toPropellerhead’s community before other members in thecommunity”; “I am interested in doing prototype evalu-ation on products before public release”; and “I would bewilling to spend plenty of time to get access to novelinformation before others.”


coder reliability, suggesting that the coders’ ratingswere consistent over time. A kappa of .83 is con-sidered to show good intercoder reliability (Cohen,1960), especially for a nonestablished codingscheme. The coders were good in separating ques-tions from answers or responses. Indeed, for thesetwo categories, the two coders had an agreement of94 percent.

An individual’s belief in his or her ability tosucceed in specific situations (Bandura, 1977)could affect both dependent variables. Inspired bythe PSED II (Reynolds & Curtin, 2009), we sepa-rated items that measured an individual’s beliefthat he or she has the skills, abilities, experience,and contacts to start a business. A factor analysisfound that the items loaded on two factors. One welabeled as “skills,” and the other we labeled as“resource mobilization.” The first factor capturesindividuals’ belief they have sufficient skills to be-come entrepreneurs. We used items from theGlobal Entrepreneurship Monitor survey to form ascale of skills self-efficacy (� � .85): “I am confidentI would succeed if I started my own firm” and “Ihave the skills and capabilities required to succeedas an entrepreneur” (1 � “strongly disagree,” 5 �“strongly agree”). The second type of self-efficacycaptured individuals’ belief in their access to re-sources, which has been shown to affect propensityfor entrepreneurial action. We again used itemsfrom the Global Entrepreneurship Monitor surveyto form our resource mobilization self-efficacy scale(� � .91): “I have the right contact network to starta new firm,” “It would be easy for me to gather theresources necessary to get the new firm going (e.g.funding, employees, etc.),” “I have the right busi-ness contacts to make a new firm possible.” Sinceprior research has demonstrated an effect of age onentrepreneurial action (Evans & Leighton, 1989),we created three dummies to control for age: under30 years, between 30 and 40 years, and above40 years.

An individual’s level of education can shape op-portunity evaluation and entrepreneurial action.

Hence, we controlled for university degree by cod-ing a dummy 1 for a bachelor’s degree or higherlevel of education.6

The survey respondents were slightly more ac-tive in the online community than the nonrespon-dents in our survey, suggesting a degree of respon-dent self-selection. Restricting the analysis torespondents could potentially bias the results byignoring factors that encouraged individuals to self-select into the sample. To alleviate this problem,we first estimated the inverse Mill’s ratio as a se-lection parameter using the full data set and thenincluded this measure for individuals who re-sponded to the survey (Heckman, 1979). To calcu-late the selection parameter, we used available in-formation about the nonrespondents, namely, thenumber of answers and questions and their tenurein the community. We then controlled for self-selection bias by inserting the inverse Mill’s ratio asa control variable in regressions predicting userentrepreneurship (see the Appendix for details).

Estimation Strategy

As expected, the two dependent variables werecorrelated, as Figure 3 illustrates. Entrepreneurialaction was more likely among individuals who en-gaged in opportunity evaluation. But, as notedabove, we did not treat opportunity evaluation as anecessary condition for entrepreneurial action,which would have required the use of a selectionmodel approach.7 Because our theoretical modeldid not imply that the two forms of informationexposure operate in a mediation relationship, we

6 We excluded gender as a control in our analysis asonly 1.44 percent of community members were female.

7 A traditional Heckman approach would ignore thefact that some individuals undertake entrepreneurial ac-tion without opportunity evaluation. In this view, entre-preneurial action would always be preceded by opportu-nity evaluation.

FIGURE 3Actual Distribution of the Relation between Opportunity Evaluation and Entrepreneurial Action

Entrepreneurial Action

No Yes No 224 13 237 Opportunity Evaluation Yes 18 20 38

242 33 275


employed an analytical approach to consider thesechoices jointly.

To accommodate for the different pathways, weused a bivariate probit model to estimate opportu-nity evaluation and entrepreneurial action as a pairof seemingly unrelated regressions. Such a modelallows for two binary dependent variables that varyjointly. For example, it has been used to studymake-or-buy decisions, in which two choices areconsidered simultaneously (see, e.g., Oxley &Sampson, 2004). This specification allows for esti-mating the coefficients while adjusting for correla-tions between the error terms for opportunity eval-uation and entrepreneurial action. This techniqueallowed us to control for unobservable factors thataffect both dependent variables. The estimates aremore efficient because they account for correlationbetween the two equations (Greene, 1997). Indeed,we expected opportunity evaluation and entrepre-neurial action to be correlated, and we found evi-dence that the bivariate probit regression fitted thedata better than two separate probit equations.

RESULTS

The descriptive statistics used for the statisticalanalysis are presented in Table 1. Our respondentswere clustered in 31 countries, each representing adifferent set of institutional conditions for entre-preneurial action. Hence, there was a possibilitythat our explanatory variables would predict thedependent variables differentially across countries,causing nonnormality in the distribution of randomerrors and raising the prospect of type I errors. Wetherefore employed the Huber-White sandwich es-timate of variance to obtain robust variance esti-mates and adjust for within-cluster correlation.

Table 2 presents the results from bivariate probitregressions predicting opportunity evaluation andentrepreneurial action jointly. The ancillary param-eter rho (�), which captures the correlation of theresiduals from the two models, is significant, sug-gesting that the two equations were related and thata bivariate probit strategy was suitable for the data,rather than two separate probits (model 1: � � .77,Wald �2 � 93.3, p � .01; model 2: � � .79, Wald �2

� 88.1, p � .01). The results are presented in twocolumns, one for the analysis of opportunity eval-uation and one for the analysis of entrepreneurialaction. Model 1, the baseline, includes the controlsonly, and model 2 includes the independentvariables.

The first two hypotheses test the determinants ofopportunity evaluation, which are shown in the leftcolumn of model 2 in Table 2. Consistently withHypothesis 1, we observe that lead user attributeshad a positive and significant effect on opportunityevaluation (p � .01). Lead users have deep knowl-edge of product functionalities and features andalso perceive themselves to be ahead of markettrends. They therefore thought themselves able toclearly see the value-creating potential obtained byaddressing various technical opportunities. Theconnection between an individual’s lead user attri-butes and opportunity evaluation was also spelledout in our interviews.

In support of Hypothesis 2, we observe a positiverelationship between technological probing activi-ties and opportunity evaluation (p � .05). Someusers, by virtue of discovering new problems andrelated solutions and communicating these withinthe community, are in a position to steer collectivelearning through aspirational search (Smith & Cao,2007). The resulting influence on the collectivemind-set helped create momentum for these users’discoveries. One interviewee commented on howtechnological probing can enhance opportunityrecognition:

Some sort of competition on setting agendas doestake place in the community. What type of techno-logical development is cool, etc. Sometimes uncon-scious but slowly proliferating into a creation of apotential commercial opportunity.

Tests of Hypotheses 3 and 4 are displayed in theright column of model 2. Hypothesis 3 proposesthat community attention has a positive associationwith entrepreneurial action. This hypothesis issupported (p � .01). Our interviewees explainedthat users attract attention from their peers in thecommunity by coming up with novel and innova-tive ideas and engaging in technical problem solv-ing. Individuals who achieved these activities ob-tained the social validation signals required toembark on entrepreneurial action:

I always felt that if I have this status [in the commu-nity] people will allow me to set up my business. . . .They love someone who shows entrepreneurialspirit. So, when I initially set up my company Iwrote something on the community board and a lotof people wished me luck and started listening tomy music and using my products. So, it helped.

We also find support for Hypothesis 4, whichstates that community spanning should be posi-


TA

BL

E1

Des

crip

tive

Sta

tist

icsa

Var

iabl

eM

ean

s.d

.M

inim

um

Max

imu

m1

22

34

57

89

1011

1213

14

1.O

pp

ortu

nit

yev

alu

atio

n0.

130.

340

1

2.E

ntr

epre

neu

rial

acti

on0.

120.

320

1.4

9

3.L

ead

use

rat

trib

ute

s3.

410.

761.

385

.22

.13

4.T

ech

nol

ogic

alp

robi

ng

0.15

0.22

01

.1.1

1�

.03

5.C

omm

un

ity

atte

nti

on4.

234.

740

46.5

.07

.2.2

2.1

3

6.C

omm

un

ity

span

nin

g2.

32.

170

10.0

9.2

2.2

4.0

4.3

1

7.T

enu

rein

com

mu

nit

y1.

551.

840

5�

.01

.06

.19

�.2

1.2

2.1

5

8.1

�n

um

ber

ofco

ded

answ

ersb

0.26

0.61

03.

29.0

2.1

6.1

1�

.08

.24

.17

.35

9.S

elf-

effi

cacy

:S

kill

s3.

881.

051

5.2

1.0

8.2

2.0

4.0

0�

.01

.04

�.0

6

10.

Sel

f-ef

fica

cy:

Res

ourc

em

obil

izat

ion

3.01

1.24

15

.27

.18

.22

.04

.02

.08

�.0

1�

.06

.61

11.

Un

iver

sity

deg

ree

0.62

0.49

01

.04

.00

�.0

8.0

6�

.02

.02

�.0

9.0

0�

.01

.04

12.

Pro

du

ctex

per

ien

ce4.

260.

852

5.0

4.1

1.3

1.0

5.1

9.2

2.2

�.0

4.0

8.0

9.0

1

13.

Age

:U

nd

er30

year

s0.

380.

490

1.0

0�

.03

.08

�.0

4�

.07

�.0

8�

.06

�.0

5�

.06

�.1

7�

.25

�.0

7

14.

Age

:30

–40

year

s0.

440.

50

1.0

6.0

6.0

4.0

9.0

3.0

7.0

8.0

5.1

4.1

9.1

9.1

�.6

9

15.

Age

:A

bove

40ye

ars

0.18

0.38

01

�.0

7�

.05

�.1

3�

.06

.06

.01

�.0

2.0

0�

.11

�.0

4.0

9�

.04

�.3

6�

.41

an

�27

5.b

Log

arit

hm

.

tively associated with entrepreneurial action (p �.01). In an interview, one user said:

Being active in other related communities on soft-ware or music production is useful also for my en-trepreneurial activities. You can check out if youridea has interest beyond the users in the Propeller-head community, you may pick up an idea for howto improve or develop new products or features foryour software, music or your business in general.

Overall, the statistical analysis provided supportfor our conceptual model. Lead user attributes andtechnological probing facilitate opportunity evalu-ation, and community attention and community

spanning affect entrepreneurial action. These find-ings suggest that exposure to technological domaininformation drives third-person opportunity recog-nition, while exposure to a social domain confersinformational advantages and reduces perceiveddemand uncertainty, thereby lowering the thresh-old for entrepreneurial action.

An additional advantage of our empirical strat-egy is that it allows us to develop marginal effectsof the likelihood of an individual locating in any ofthe four cells presented in Figure 2. The ancillaryparameter rho shows that the two outcome vari-ables are correlated, which is accounted for in our

TABLE 2Bivariate Probit Regressions Predicting Opportunity Evaluation and Entrepreneurial Actiona

Variables

Model 1 Model 2

OpportunityEvaluation

EntrepreneurialAction

OpportunityEvaluation

EntrepreneurialAction

Control1 � number of answers to questionsb 0.18† 0.48** 0.11 0.34*

(0.10) (0.13) (0.14) (0.14)Self-efficacy: Skills 0.11 �0.13 0.05 �0.12

(0.08) (0.11) (0.07) (0.14)Self-efficacy: Resource mobilization 0.37** 0.37** 0.35** 0.34**

(0.06) (0.09) (0.07) (0.08)University degree 0.16 �0.04 0.22 �0.05

(0.16) (0.28) (0.22) (0.33)Product experience 0.05 0.29* �0.15 0.15

(0.14) (0.14) (0.14) (0.14)Age: 30–40 years �0.18 �0.06 �0.13 �0.12

(0.20) (0.31) (0.20) (0.30)Age: Above 40 years �0.52** �0.28† �0.49† �0.36†

(0.19) (0.17) (0.28) (0.19)Inverse Mill’s ratio 0.11 0.08 0.16 �0.15

(0.36) (0.32) (0.45) (0.34)IndependentH1: Lead user attributes 0.50** 0.06

(0.08) (0.05)H2: Technological probing 0.51* 0.61

(0.25) (0.38)H3: Community attention 0.01 0.03**

(0.03) (0.01)H4: Community spanning 0.04 0.11**

(0.03) (0.02)Constant �3.09** �3.26** �4.04** �3.03*

(0.79) (1.26) (0.93) (1.28)Log pseudo-likelihood �164 �154df 16 24n 275 275

a Unstandardized coefficients are shown, with Huber-White robust standard errors clustered by nation in parentheses.b Logarithm.

† p � .10* p � .05

** p � .01Two-tailed tests.


estimation strategy. Table 3 presents the marginaleffects of locating in any of the four cells in Figure2 based on the results from model 2 in Table 2,holding all other variables at their means. The mar-ginal effects extend our insights about why differ-ent users end up in different cells. Users that eval-uate third-person opportunities but do not engagein entrepreneurial action are often lead users, butnone of the other marginal effects are significant(see Table 3). One lead user explained how heevaluated opportunities but decided not to start abusiness:

I have a lot of entrepreneurial ideas and actuallyengaged in real activities to see if something couldbe done about them. I evaluated the options toachieve financial return from doing something likeselling this product. But I found it wasn’t worth it; itwas just too time consuming.

In contrast, as shown in Table 3, lead user attri-butes had a negative effect on opportunistic action(i.e., entrepreneurial action without exhibiting ob-servable evaluation activities). This is consistentwith our reasoning. Further, Table 3 shows thatcommunity attention and community spanningboth exhibited a positive marginal effect on oppor-tunistic action. A quote from a user who explainedhow he became an entrepreneur illustrates themechanism for this path: “The only planning activ-ity was that I wanted to get out and network. Therest of it fell into place and through some flexibilityI was able to make the most of some opportunitiesthat came up.”

We also found that lead user attributes, techno-logical probing, and community spanning all had apositive marginal effect on planned behavior (i.e.,entrepreneurial action associated with opportunityevaluation). For community spanning, this path re-

flects our reasoning that users exposed to multiplethought worlds are able to draw on analogous ex-periences to both evaluate opportunities and actupon them.

Robustness Checks

We conducted several robustness checks. First,we reestimated the regressions using two separateprobit regressions, confirming that a bivariate pro-bit specification is a better fit with our data. Sec-ond, we tried an alternative measure of communityattention, the Bonacich centrality measure, whichcaptures whether a given individual is connectedto other central individuals. We were able to testthis measure because we had a full matrix of allinteractions between all members in the commu-nity. Third, we checked whether including the in-verse Mills ratio affected our results. None of thesemodifications altered our results.

DISCUSSION

The past decade has witnessed a significantamount of research and theorizing on opportunitydiscovery, evaluation, and pursuit. Much of thisresearch has focused on the sources of opportuni-ties (e.g., Eckhardt & Shane, 2003; Shane, 2000),debated the nature of opportunities (e.g., Alvarez &Barney, 2007; Companys & McMullen, 2007; Shane& Venkataraman, 2000), and studied the drivers ofopportunity discovery (e.g., Foss & Foss, 2008).More recently, the pendulum has started to swingaway from an interest in opportunities themselvestoward the study of conditions that prompt indi-viduals to perceive opportunities, evaluate them,and act upon them (Choi & Shepherd, 2004; Gré-

TABLE 3Marginal Effects for Cell Membership Based on Regressionsa

VariablesNo

EntrepreneurUnrealized

OpportunitiesOpportunistic

ActionPlannedBehavior

H1: Lead user attributes �0.06** 0.05** �0.02* 0.03**H2: Technological probing �0.12* 0.03 0.04 0.05*H3: Community attention 0.00 0.00 0.003* 0.00H4: Community spanning �0.02** 0.00 0.01** 0.01**

a n � 275. The marginal effects of the independent variables with all other variables held at their means are shown. For instance, thecolumn for planned behavior is Pr(opportunity evaluation � 1, entrepreneurial action � 1). These values are based on the results frommodel 2 in Table 2.

* p � .05** p � .01

Two-tailed tests.


goire et al., 2010; Haynie et al., 2009). Yet, fewempirical studies examine the drivers of opportu-nity recognition and entrepreneurial action in pre-founding situations. We contribute by building andtesting a model of how an individual’s exposure todifferent kinds of information affects opportunityevaluation and entrepreneurial action. Our find-ings suggest that the individual’s exposure to tech-nological information increases the likelihood thathe or she will recognize third-person opportunitiesas manifested in evaluation, whereas exposure toinformation about user needs exhibits significantinfluence on first-person opportunity beliefs asmanifested in entrepreneurial action. Several im-portant implications emerge from these findings.

Our most important finding is that opportunityevaluation and entrepreneurial action are regulatedby different external stimuli. This supports the po-sition that third-person opportunity recognitionand entrepreneurial action are distinct processes(Grégoire et al., 2010; Haynie et al., 2009; McMul-len & Shepherd, 2006). In received research onopportunity discovery, the two processes havebeen typically conflated into a single discovery orevaluation event (Kirzner, 1997). On the otherhand, the “structuration” and “effectuation” per-spectives, while recognizing opportunity evalua-tion and entrepreneurial action as distinct pro-cesses, nevertheless have emphasized theirintertwined and iterative character, implying thatthe two are subject to the same external stimuli(Sarason, Dean, & Dillard, 2006; Sarasvathy, 2001).We find that the two processes are not only dis-tinct, but also, at least partly responsive to differentinfluences.

Our model also supports McMullen and Shep-herd (2006), by highlighting that reduction in re-sponse uncertainty is central for triggering entre-preneurial action. Mere exposure to third-personopportunities will not stimulate action if responseuncertainty prevails. In our model, this reductionin response uncertainty is driven by (1) an individ-ual’s exposure to alternative thought worlds(through participation in different online commu-nities) and (2) the attention given to the individu-al’s statements in the focal online community. Bothof these are social mechanisms that emphasize therole of information obtained through social interac-tions with others. Given that social trends tend tobe less predictable than technological trends, thisrole of social mechanisms emphasizes the intersub-jective character of response uncertainty reduction,

in which beliefs regarding the commercial viabilityof third-person opportunities emerge through so-cial interactions (Giddens, 1984; Shepherd etal., 2007).

Our model also highlights the importance of ex-posure to technological development as an impor-tant driver of third-person opportunity recognitionand evaluation. Although both Austrian and actiontheories routinely point to technological develop-ment as a source of opportunities, few studies haveempirically examined how an individual’s connec-tivity with the technological domain facilitatesthird-person opportunity recognition. We ad-dressed this gap by considering the effect of tech-nological probing and lead user attributes. Wefound that exposure to technological developmentsfacilitates evaluation actions but does not necessar-ily support entrepreneurial action directly. Thisfinding suggests that opportunity evaluation is afuture-oriented activity that informs but is separatefrom entrepreneurial action (Haynie et al., 2009).Being at the vanguard of a given technology enablesindividuals to perceive trends earlier than othersand also to evaluate emerging opportunities as theysee them. In keeping with the action theory of en-trepreneurship, our findings also support the im-portance of problem sensitization to the recogni-tion of third-person opportunities arising fromtechnological advances.

Although our theoretical model provides a wel-come empirical test of the McMullen and Shepherd(2006) model, it also extends their framework inimportant ways. The focus of their model and itssubsequent tests (Grégoire et al., 2010; Haynie etal., 2009) has been predominantly on the motiva-tional, goal-intention-forming stage of the theory ofthe psychology of action (Gollwitzer, 1996: 289).8

Our model extends their theory to a volitional plan-ning stage, which nevertheless usually precedesthe entrepreneurial action itself. As business plan-ning activity signals awareness of third-person op-portunities and a reasonable supposition that thesemight constitute feasible and desirable first-personopportunities, our empirical test highlights a stageat which opportunity awareness induces volitionalevaluation. Although the evaluation activity itselfmay feed escalation of commitment, our test high-lights the importance of demand uncertainty reduc-

8 We thank Jeffrey McMullen for sharing his insightand for explaining the relationship between his actiontheory and the theory of the psychology of action.


tion for the transition from volitional evaluation tovolitional action. Our model, therefore, helps pin-point the threshold between evaluation and entre-preneurial action while also highlighting stimulithat facilitate crossing this threshold.

By demonstrating the different effects of externalinformation exposure on opportunity evaluationand entrepreneurial action, respectively, we extendthe analysis of opportunity recognition fromindividual-centric considerations toward thebroader social system within which individuals areembedded. Therefore, our study provides a bridgebetween the action and network perspectives toentrepreneurial opportunity evaluation and entre-preneurial action. Our model is consistent with thestructural paradigm of network research, in thesense that an entrepreneur’s network is seen asregulating his/her exposure to information regard-ing opportunities (Aldrich & Kim, 2007; Ma, Huang& Shenkar, 2011; Podolny, 2001). However,whereas the emphasis in the network tradition ison self-reinforcing, power- and resource-based con-straints on entrepreneurial action (Hallen, 2008;Zaheer & Soda, 2009), our model highlights infor-mational facilitators of action. In short, whereasreceived network studies have focused on whetheran entrepreneur can act, our model highlights con-ditions under which the entrepreneur will act. Acombination of power-based and cognitive expla-nations could provide more nuanced insight intothe varied influences of social networks on entre-preneurial action.

We also contribute by moving the conversationabout the role of users in innovation (von Hippel,1988) toward understanding users as entrepre-neurs. The internet makes it possible to sustaingeographically disparate communities where userscreate and exchange ideas. Several studies haveexplored why users are motivated to share ideaswith other users (von Krogh & von Hippel, 2003).Although it is frequently assumed that sharing pre-vents commercialization, sharing may also encour-age commercialization by allowing a user commu-nity to weigh in on the developments. Users withaccess to shared ideas can build products and ser-vices around identified needs in user communities.We show how interactions in user communitiesmay actually lower the threshold for entrepreneur-ial action: Individuals who both attract communityattention and span multiple external communitieswill, in effect, reduce their demand uncertainty.Collecting a similar amount and quality of knowl-

edge would presumably be more time consumingand more expensive in an “offline” context. Ourstudy highlights the implications for understand-ing the social patterns that unfold in online com-munities and how they provide opportunities foruser entrepreneurship.

Thus far, research on user entrepreneurship hasfocused on describing the phenomenon and explor-ing sector-level regulators of user entrepreneurshipwhile overlooking individual-level influences(Baldwin, Hienerth, & von Hippel, 2006; Lettl, Hie-nerth, & Gemuenden, 2008). Our study has ad-dressed this oversight with a large-sample empiri-cal analysis that draws on unique microlevel datadescribing individuals’ communication patternsand networks. We add to research that has demon-strated cross-industry variance in user entrepre-neurship (Shah & Tripsas, 2007) by showing thatwithin a given industry, the exposure of individualusers to different knowledge domains regulates thepropensity of users to move from innovation toentrepreneurship. We also demonstrated that expo-sure to social information affects entrepreneurialaction: While community attention and communityspanning regulated entrepreneurial action, theyhad no significant effect on opportunityrecognition.

Von Hippel (1988) highlighted the informationadvantages that lead users enjoy as a result of theirsensitization to technological bottlenecks. Becauseof their heightened awareness of technological bot-tlenecks, lead users are better positioned than or-dinary users to monitor and even anticipate tech-nological developments. Our study shows thatsuch sensitization can, when combined with com-munity spanning and community attention, alsopropel entrepreneurial action via reduction in re-sponse uncertainty. This is an interesting findingthat may offer utility in, for instance, predictingwhere technology entrepreneurs are more likely toemerge in social networks.

Finally, our findings may have implications formanagerial practice. First, our findings highlightthe importance of online communities as platformsfor technological and economic coordinationamong geographically dispersed users. Specifi-cally, our findings suggest that some communityparticipants, by virtue of their position relative tothe technological and economic domains, may bemore likely than others to identify viable techno-logical paths and even embark on entrepreneurialaction to exploit these. By monitoring user commu-


nities and other similar online communities, there-fore, platform owners might gain valuable, perhapseven predictive, insight into where technologicalinnovation is more likely to occur within a com-munity and what forms it might take. Our study hashighlighted some structural determinants of suchactivities, but there might be others that our inquiryhas not covered. This appears a promising avenuefor further research, one with potentially signifi-cant implications for practice. Second, our findingsmay have implications for corporate venturing.Many established firms seek to encourage intrapre-neurial (i.e., corporate entrepreneurial) activityamong their employees. Our findings imply thatconnectivity with social domain and technologicaldomains, perhaps in the form of active participa-tion in boundary-spanning communities of prac-tice, could prove a useful inducement for the de-velopment of commercially viable initiatives.Third, our findings on the role of technologicalprobing highlight the potential utility of activecommunity participation not only for gaining infor-mation, but also for shaping the directions of col-lective development efforts. Thus, our study pro-vides support for the effectiveness of activeparticipation in online practitioner communities asboth a form of “disembodied experimentation”(Keil, Autio, & George, 2008) and a device withwhich firms can subtly steer technological comple-ments and externalities.

Limitations and Future Research

Although our conceptual model emphasizesthe joint effects of exposure to technological andsocial information on opportunity evaluation andentrepreneurial action, we have been careful notto imply a mediation effect, in which discernibleopportunity evaluation must always precede en-trepreneurial action. By so doing we have madean allowance for Austrian or Kirznerian discov-ery (e.g., Kirzner, 1973, 1997), in which the real-ization that a third-person opportunity exists isfollowed almost immediately by opportunitypursuit. Indeed, while our empirical model sup-ports the distinction between opportunity evalu-ation and entrepreneurial action, a number of ourrespondents exhibited entrepreneurial actionwithout engaging in evaluation activities—a cir-cumstance we operationalized in Figure 1. Thismay be a sign of opportunistic behavior, in whichindividuals react opportunistically to the market

needs they detect. However, we cannot excludethe possibility that such individuals may beaware of technological opportunities without un-dertaking any volitional action to evaluate them,as specified in our operationalization. After all,our empirical context was an online user commu-nity, where recognition of technological ad-vances could be taken as given. We expected tofind a positive correlation between opportunityevaluation and entrepreneurial action, and ourmodeling strategy confirmed this expectation.

We tested our hypotheses with a data set of in-teractions in one online community in the softwareindustry, a decision that could have implicationsfor the generalizability of our results. Central to ourargument is the idea that information about userneeds reduces demand uncertainty and shapes en-trepreneurial action. A scope condition of our the-ory is that this is more likely to apply when aknowledge frontier is gradually evolving and someindividuals are better positioned than others to re-ceive signals about user needs.

We cannot preclude the possibility that individ-uals who intend to become entrepreneurs behave ina certain way to manipulate their positions in asocial structure, creating a problem of reverse cau-sality. Although we cannot fully rule out the alter-native causation, we consider this explanation tobe unlikely for two reasons. First, community at-tention reflects actions by others more than those ofthe focal individual. We measure others’ responsesto an individual’s postings, and it is difficult toaffect the views of other community members. Sec-ond, it is not immediately apparent why entrepre-neurial actions would drive individuals to partici-pate in several related user communities.Nevertheless, to elaborate on the chain of eventsmerits further research attention.

Finally, although we controlled for a number ofpossible alternative explanations and used con-tent analysis to check the face validity of ourmeasures, we have not closely examined the con-tent of the communication among users overtime. Future studies could productively incorpo-rate more extensive content analysis into studiesof user entrepreneurship from online communi-ties. This could shed light on, for example, howdifferent types of individuals frame their mes-sages, their use of specific words, how these fac-tors are associated with opportunity evaluationand entrepreneurial action, and, in particular,how user entrepreneurs engage in communica-


tion in the community to discover opportunitiesand mobilize resources. Finally, research on userentrepreneurship could be improved by addingdynamics into the various empirical models. Forexample, by adding aspects of individuals’ net-work positions over time, the general evolutionof a community network, and the exact timing ofentrepreneurial actions (i.e., the time that a firmwas established or first sale was executed),would offer rich insights into how and why userentrepreneurship happens.

Conclusion

Our study breaks new ground for the understand-ing of entrepreneurship using a large-scale empiri-cal study, in particular, for entrepreneurial actiontheories and studies of user entrepreneurs. Expo-sure to technological and social information shapesopportunity evaluation and entrepreneurial action,albeit in different ways. These findings createbroad implications for entrepreneurship and man-agement scholars by highlighting the critical im-portance of reducing demand uncertainty and em-pirically testing how this occurs throughcommunity attention and spanning multiple com-munities. More people seem to engage in differenttypes of online communities each year. Our studyenriches the understanding of how these commu-nities nurture the development of user entrepre-neurs by showing that the exchange of and expo-sure to technological information, althoughimportant, is not enough.

REFERENCES

Abernathy, W., & Utterback, J. 1978. Patterns of indus-trial innovation. Technology Review, 80(7): 40–47.

Alderson, W. 1965. Dynamic marketing behavior.Homewood, IL: Irwin.

Aldrich, H. E., & Kim, P. H. 2007. Small worlds, infinitepossibilities? How social networks affect entrepre-neurial team formation and search. Strategic Entre-preneurship Journal, 1: 147–165.

Alvarez, S. A., & Barney, J. B. 2007. Discovery and cre-ation: Alternative theories of entrepreneurial action.Strategic Entrepreneurship Journal, 1: 11–26.

Anderson, P., & Tushman, M. L. 1990. Technologicaldiscontinuities and dominant designs: A cyclicalmodel of technological change. Administrative Sci-ence Quarterly, 35: 604–633.

Armstrong, J. S., & Overton, T. S. 1977. Estimating non-response bias in mail surveys. Journal of MarketingResearch, 14: 396–402.

Autio, E., George, G., & Alexy, O. 2011. Internationalentrepreneurship and capability development—Qualitative evidence and future research directions.Entrepreneurship Theory & Practice, 35: 11–37.

Bae, H., & Gargiulo, M. 2004. Partner substitutability,alliance network structure, and firm profitability inthe telecommunications industry. Academy of Man-agement Journal, 47: 843–859.

Baldwin, C., Hienerth, C., & von Hippel, E. 2006. Howuser innovations become commercial products: Atheoretical investigation and case study. ResearchPolicy, 35: 1291–1313.

Baldwin, C., & von Hippel, E. 2011. Modeling a paradigmshift: From producer innovation to user and opencollaborative innovation. Organization Science, 22:1399–1417.

Bandura, A. 1977. Self-efficacy: Toward a unifying the-ory of behavioral change. Psychological Review, 84:191–215.

Berry, M., & Linoff, G. 1999. Mastering data mining: Theart and science of customer relationship manage-ment. New York: Wiley.

Bird, B. 1988. Implementing entrepreneurial ideas: Thecase for intention. Academy of Management Re-view, 13: 442–453.

Carter, N. M., Gartner, W. B., & Reynolds, D. P. 1996.Exploring start-up event sequences. Journal of Busi-ness Venturing, 11: 151–166.

Casson, M. 1982. The entrepreneur: an economic the-ory. Totowa, NJ: Barnes & Noble.

Choi, Y. R., & Shepherd, D. A. 2004. Entrepreneurs’ de-cisions to exploit opportunities. Journal of Manage-ment, 30: 377–395.

Clark, K. B. 1985. The interaction of design hierarchiesand market concepts in technological evolution. Re-search Policy, 14: 235–251.

Cohen, J. 1960. A coefficient of agreement for nominalscales. Educational and Psychological Measure-ment, 20: 37–46.

Companys, Y. E., & McMullen, J. S. 2007. Strategic en-trepreneurs at work: The nature, discovery, and ex-ploitation of entrepreneurial opportunities. SmallBusiness Economics, 28: 301–322.

Dahlander, L., & Frederiksen, L. 2012. The core and cos-mopolitans: A relational view of innovation in usercommunities. Organization Science, 23: 988–1007.

Dahlander, L., & O’Mahony, S. 2011. Progressing to thecenter: Coordinating project work. OrganizationScience, 22: 961–979.


Delmar, F., & Shane, S. 2003. Does business planningfacilitate the development of new ventures? Strate-gic Management Journal, 24: 1165–1185.

Dimov, D. 2010. Nascent entrepreneurs and ventureemergence: Opportunity confidence, human capital,and early planning. Journal of Management Stud-ies, 47: 1123–1153.

Dosi, G. 1982. Technological paradigms and technologi-cal trajectories: A suggested interpretation of the de-terminants and directions of technical change. Re-search Policy, 11: 147–162.

Doty, H. D., & Glick, W. H. 1998. Common methods bias:Does common methods variance really bias results?Organizational Research Methods, 1: 374–406.

Dougherty, D. 1992. Interpretive barriers to successfulproduct innovation in small firms. OrganizationScience, 3: 179–202.

Dougherty, T. J. 2002. An update of clinical laser medi-cine & surgery. Journal of Clinical Laser Medicineand Surgery, 20: 3–7.

Eckhardt, J. T., & Ciuchta, M. P. 2008. Selected variation:The population-level implications of multistage se-lection in entrepreneurship. Strategic Entrepre-neurship Journal, 2: 209–224.

Eckhardt, J. T., & Shane, S. A. 2003. Opportunities andentrepreneurship. Journal of Management, 29: 333–349.

Edmondson, A., & McManus, S. 2007. Methodological fitin management field research. Academy of Manage-ment Review, 32: 1155–1179.

Evans, D. S., & Leighton, L. S. 1989. Some empiricalaspects of entrepreneurship. American EconomicReview, 79: 519–535.

Fitzsimmons, J. R., & Douglas, E. J. 2011. Interactionbetween feasibility and desirability in the formationof entrepreneurial intentions. Journal of BusinessVenturing, 26: 431–440.

Foss, K., & Foss, N. J. 2008. Understanding opportunitydiscovery and sustainable advantage: The role oftransaction costs and property rights. Strategic En-trepreneurship Journal, 2: 191–207.

Franke, N., Schreier, M., & Kaiser, U. 2010. The “I de-signed it myself” effect in mass customization. Man-agement Science, 56: 125–140.

Franke, N., & Shah, S. 2003. How communities supportinnovative activities: An exploration of assistanceand sharing among end-users. Research Policy, 32:157–178.

Fuchs, C., Prandelli, E., & Schreier, M. 2010. The psy-chological effects of empowerment strategies on con-sumers’ product demand. Journal of Marketing, 74:65–79.

Gaglio, C. M., & Katz, J. A. 2001. The psychological basisof opportunity identification: Entrepreneurial alert-ness. Small Business Economics, 16(2): 95–111.

Giddens, A. 1984. The constitution of society: Outline ofthe theory of structuration. Berkeley: University ofCalifornia Press.

Gollwitzer, P. M. 1990. Action phases and mind-sets. InE. T. Higgins & R. M. Sorrentino (Eds.), Handbook ofmotivation and cognition: Foundations of socialbehavior, vol. 2: 53–92. New York: Guilford.

Gollwitzer, P. M. 1996. The volitional benefits of plan-ning. In P. M. Gollwitzer & J. A. Bargh (Eds.), Thepsychology of action: 287–312. New York: Guilford.

Granovetter, M. 1985. Economic actions and social struc-ture: The problem of embeddedness. AmericanJournal of Sociology, 91: 481–510.

Greene, W. H. 1997. Frontier production functions. InM. H. Pesaran & P. Schmidt (Eds.), Handbook ofapplied econometrics: Microeconomics (vol. 2): 81–166. Oxford, UK, and Cambridge, MA: Blackwell.

Grégoire, D. A., Shepherd, D. A., & Schurer Lambert, L.2010. Measuring opportunity-recognition beliefs. Or-ganizational Research Methods, 13: 114–145.

Haefliger, S., Jaeger, P., & von Krogh, G. 2010. Under theradar: Industry entry by user entrepreneurs. Re-search Policy, 39: 1198–1213.

Hallen, B. 2008. The causes and consequences of theinitial network positions of new organizations: Fromwhom do entrepreneurs receive investments? Ad-ministrative Science Quarterly, 53: 685–718.

Hayek, F. A. 1945. The use of knowledge in society.American Economic Review, 35: 519–530.

Haynie, J. M., Shepherd, D. A., & McMullen, J. S. 2009.An opportunity for me? The role of resources inopportunity evaluation decisions. Journal of Man-agement Studies, 46: 337–361.

Hebert, R. F., & Link, A. N. 1988. The entrepreneur:Mainstream views and radical critiques (2nd ed.).New York: Praeger.

Heckman, J. J. 1979. Sample selection bias as a specifi-cation error. Econometrica, 47: 153–161.

Jeppesen, L. B., & Frederiksen, L. 2006. Why do userscontribute to firm-hosted user communities? Thecase of computer-controlled music instruments. Or-ganization Science, 17: 45–63.

Jeppesen, L. B., & Laursen, K. 2009. The role of lead usersin knowledge sharing. Research Policy, 38: 1582–1589.

Jick, T. D. 1979. Mixing qualitative and quantitativemethods: Triangulation in action. AdministrativeScience Quarterly, 24: 602–611.

Keil, T., Autio, E., & George, G. 2008. Corporate venture


capital, disembodied experimentation, and capabil-ity development. Journal of Management Studies,45: 1475–1505.

Kim, J.-Y., & Miner, A. S. 2007. Vicarious learning fromthe failures and near-failures of others: Evidencefrom the U.S. commercial banking industry. Acad-emy of Management Journal, 50: 687–714.

Kirzner, I. M. 1973. Competition and entrepreneurship.Chicago: University of Chicago Press.

Kirzner, I. 1997. Entrepreneurial discovery and the com-petitive market process: An Austrian approach. Jour-nal of Economic Literature, 35: 60–85.

Kurzman, C., Anderson, C., Key, C., Lee, Y. O., Moloney,M., Silver, A., & Van Ryn, M. W. 2007. Celebritystatus. Sociology Theory, 25: 347–367.

Lakhani, K. R., & von Hippel, E. 2003. How open sourcesoftware works: “Free” user-to-user assistance. Re-search Policy, 32: 923–943.

Lee, G. K., & Cole, R. E. 2003. From a firm-based to acommunity-based model of knowledge creation: Thecase of the Linux kernel development. OrganizationScience, 14: 633–649.

Lee, W. Y., Goodwin, P., Fildes, R., Nikolopoulos, K., &Lawrence, M. 2007. Providing support for the use ofanalogies in demand forecasting tasks. InternationalJournal of Forecasting, 23: 377–390.

Lettl, C., Hienerth, C., & Gemuenden, H. G. 2008. Explor-ing how lead users develop radical innovation: Op-portunity recognition and exploitation in the field ofmedical equipment technology. IEEE Transactionson Engineering Management, 55: 219–233.

Lipshitz, R., & Strauss, O. 1997. Coping with uncertainty:A naturalistic decision-making analysis. Organiza-tional Behavior and Human Decision Processes,69: 149–163.

Ma, R., Huang, Y.-C., & Shenkar, O. 2011. Social net-works and opportunity recognition: A cultural com-parison between Taiwan and the United States. Stra-tegic Management Journal, 32: 1183–1205.

McMullen, J. S., & Shepherd, D. A. 2006. Entrepreneurialaction and the role of uncertainty in the theory of theentrepreneur. Academy of Management Review,31: 132–152.

Miles, M. B., & Huberman, M. A. 1994. Qualitative dataanalysis: An expanded source book (2nd ed.). Lon-don: Sage.

Morrison, P. D., Roberts, J. H., & Midgley, D. F. 2004. Thenature of lead users and measurement of leadingedge status. Research Policy, 33: 351–362.

Morrison, P. D., Roberts, J. H., & von Hippel, E. 2000.Determinants of user innovation and innovation

sharing in a local market. Management Science, 46:1513–1527.

Myers, J. H., & Robertson, T. S. 1972. Dimensions ofopinion leadership. Journal of Marketing Research,9: 41–46.

Niiniluoto, I. 1993. The aim and structure of appliedresearch. Erkenntnis, 38(1): 1–21.

Norton, M. J. 2009. The IKEA effect: When labor leads tolove. Harvard Business Review, 87(2): 30.

Oxley, J. E., & Sampson, R. C. 2004. The scope andgovernance of international R&D alliances. StrategicManagement Journal, 25: 723–749.

Podolny, J. M. 2001. Networks as the pipes and prisms ofthe market. American Journal of Sociology, 107:33–60.

Podsakoff, P. M., MacKenzie, S. B., Jeong-Yeon, L., &Podsakoff, N. P. 2003. Common method biases inbehavioral research: A critical review of the litera-ture and recommended remedies. Journal of Ap-plied Psychology, 88: 879–903.

Polidoro, F., Ahuja, G., & Mitchell, W. 2010. When thesocial structure overshadows competitive incen-tives: The effects of network embeddedness on jointventure dissolution. Academy of ManagementJournal, 54: 203–223.

Rein, I., Kottler, P., & Stoller, M. 1987. High visibility.New York: Dodd, Mead.

Rensink, R. A. 2002. Change detection. In S.T. Fiske,D. L. Schacter, & C. Zahn-Wexler (Eds.), Annualreview of psychology: 245–277. Palo Alto, CA: An-nual Reviews.

Reynolds, P. D. 2007. New firm creation in the UnitedStates: A PSED I overview. Foundations and Trendsin Entrepreneurship, 3: 1–149.

Reynolds, P. D., & Curtin, R. T. (Eds.). 2009. New firmcreation in the United States: Initial explora-tions with the PSED II data set. New York:Springer.

Sahal, D. 1985. Technological guideposts and innovationavenues. Research Policy, 14: 61–82.

Sarason, Y., Dean, T., & Dillard, J. F. 2006. Entrepreneur-ship as the nexus of individual and opportunity: Astructuration view. Journal of Business Venturing,21: 286–305.

Sarasvathy, S. D. 2001. Causation and effectuation: To-ward a theoretical shift from economic inevitabilityto entrepreneurial contingency. Academy of Man-agement Review, 26: 243–263.

Shah, S. K., & Tripsas, M. 2007. The accidental entrepre-neurs: The emergent and collective process of userentrepreneurship. Strategic Entrepreneurship Jour-nal, 1: 123–140.


Shane, S. 2000. Prior knowledge and the discovery ofentrepreneurial opportunities. Organization Sci-ence, 11: 448–469.

Shane, S. 2001. Technological opportunities and newfirm creation. Management Science, 47: 205–220.

Shane, S., & Delmar, F. 2004. Planning for the market:Business planning before marketing and the contin-uation of organizing efforts. Journal of BusinessVenturing, 19: 767–785.

Shane, S., & Khurana, R. 2003. Bringing individuals backin: The effects of career experience on new firmfounding. Industrial and Corporate Change, 12:519–543.

Shane, S., & Venkataraman, S. 2000. The promise ofentrepreneurship as a field of research. Academy ofManagement Review, 25: 217–226.

Shepherd, D. A., McMullen, J. S., & Jennings, P. D. 2007.The formation of opportunity beliefs: Overcomingignorance and reducing doubt. Strategic Entrepre-neurship Journal, 1: 75–95.

Smith, K. G., & Cao, Q. 2007. An entrepreneurial perspec-tive on the firm-environment relationship. StrategicEntrepreneurship Journal, 1: 329–344.

Tushman, M. L., & Scanlan, T. J. 1981. Boundary span-ning individuals: Their role in information transferand their antecedents. Academy of ManagementJournal, 24: 289–305.

Vaghely, I., & Julien, P.-A. 2010. Are opportunities rec-ognized or constructed? An information perspectiveon entrepreneurial opportunity identification. Jour-nal of Business Venturing, 25: 73–86.

von Hippel, E. 1988. The sources of innovation. NewYork: Oxford University Press.

von Krogh, G., & von Hippel, E. 2003. Special issue onopen source software development. Research Pol-icy, 32: 1149–1157.

von Krogh, G., & von Hippel, E. 2006. The promise ofresearch on open source software. Management Sci-ence, 52: 975–983.

Wade, J. 1996. A community-level analysis of sourcesand rates of technological variation in the micropro-cessor market. Academy of Management Journal,39: 1218–1244.

Weber, R. P. 1990. Basic content analysis (2nd ed.).London: Sage.

Whittaker, S., Terveen, L., Hill, W., & Cherny, L. 1998.The dynamics of mass interaction. Proceedings ofthe Conference of Computer Supported Coopera-tive Work. New York: ACM.

Zaheer, A., & Soda, G. 2009. Network evolution: Theorigins of structural holes. Administrative ScienceQuarterly, 54: 1–31.

APPENDIX

Because we coded all 5,000 posts in the communitybefore distributing the survey, we could compare re-spondents and nonrespondents. Respondents weremore likely to answer questions and less likely to posequestions in the Propellerhead community. Individu-als who posed questions were likely to respond to thesurvey, but this variable was not correlated with op-portunity evaluation and entrepreneurial action. Wethus used the number of questions posed as an exclu-sion restriction variable. Obviously, for the nonrespon-dents we had less information than for the respon-dents, so we used the information available to predictthe likelihood of responding to the survey (see, e.g.,Bae and Gargiulo [2004] and Polidoro, Ahuja, andMitchell [2010] for a similar approach). The idea thatindividuals who pose questions are not associatedwith the entrepreneurial outcomes was also confirmedin interviews with community members. Table A1shows the mean numbers of the variables and results oft-tests of the differences.

Next we used a probit regression to predict the prob-ability an individual would respond to the survey.Table A2 shows the correlations between the variables.Table A3, which shows the results from the regression,illustrates that the number of questions has a negativeeffect on the probability of responding to the survey(community newcomers were slightly less likely to

TABLE A1

Comparing Respondents and Nonrespondentsa

Variable RespondentsNon-

respondents t

1 � number of answersto questions

0.77 0.41 **

1 � number of questions 0.48 0.81 **Tenure in community 1.62 1.51

a Respondents, n � 280; nonrespondents, n � 686.** p � .01

Two tailed tests.

TABLE A2

Correlationsa

Variable 1 2

1. 1 � number of answers to questions2. 1 � number of questions .323. Tenure in community .22 �.07

a n � 966.


take part in the survey). From this result we developedthe inverse Mill’s ratio included in Table 3. Includingthe inverse Mill’s ratio in our regressions did not alterthe direction of coefficients or the significance of ourresults.

Erkko Autio ([email protected]) is chair of en-trepreneurship and technology transfer at Imperial Col-lege Business School, London. He is also a visiting pro-fessor at Aalto University. His research focuses oncomparative entrepreneurship, national systems of entre-preneurship, and international entrepreneurship. He re-ceived his Ph.D. from Helsinki University of Technology.

Linus Dahlander ([email protected]) is an asso-ciate professor at ESMT European School of Managementand Technology. He is currently studying how networksunfold when people can choose their collaboration part-ners and consequences for innovation outcomes. He re-ceived a Ph.D. from Chalmers University of Technologyand was a postdoctoral fellow at Stanford University.

Lars Frederiksen ([email protected]) is professor atthe Innovation Management Group, School of Businessand Social Sciences, Aarhus University. He received aPh.D. in business administration from Copenhagen Busi-ness School. His current research interests include evo-lution of knowledge producing communities, entrepre-neurship, and user innovation.

TABLE A3

Probit Regression Predicting Response to Surveya

Variable Model A

Tenure in community �.01(.02)

1 � number of answers to questions .44(.09)**

1 � number of questions �.81(.10)**

Constant �.35(.07)**

Log-likelihood �545.49Selected observations 280

a Robust standard errors are in parentheses. n � 966.** p � .01


INFORMATION EXPOSURE, OPPORTUNITY EVALUATION, AND ... · herd, 2006; Shane, 2001; Shane &...

Documents

Transcript of INFORMATION EXPOSURE, OPPORTUNITY EVALUATION, AND ... · herd, 2006; Shane, 2001; Shane &...