Hofstetter Et Al JMR Constraining Ideas 2021

Article
Journal of Marketing Research

2021, Vol. 58(1) 95-114
Constraining Ideas: How Seeing Ideas of ª The Author(s) 2020
Others Harms Creativity in Open Innovation Article reuse guidelines:

sagepub.com/journals-permissions
DOI: 10.1177/0022243720964429
journals.sagepub.com/home/mrj
Reto Hofstetter, Darren W. Dahl, Suleiman Aryobsei,

and Andreas Herrmann
Abstract
Open innovation contests that display all submitted ideas to participants are a popular way for firms to generate ideas. In such
contest-based ideation, the authors show that seeing numerous competitive ideas of others harms, rather than stimulates,
creative performance (Study 1). Others’ competitive prior ideas interfere with idea generation, as new ideas need to be dif-
ferentiated from the preceding ones to be original. Exposure to an increasing number of prior ideas thus heightens individuals’
perceived constraints of expressing ideas and harms creative performance (Studies 2 and 3). Furthermore, creative performance
monotonically reduces with an increasing number of prior ideas (Study 4). A final study demonstrates that showing only a limited
number of ideas as well as grouping prior ideas offer actionable ways to reduce prior ideas’ harmful influence (Study 5). These
results illustrate viable ways to improve contest-based ideation outcomes merely by changing how competitive prior ideas are
presented.
Keywords
creativity, crowdsourcing, open innovation, innovation contests, user-generated content
Online supplement: https://doi.org/10.1177/0022243720964429
Over the past decade, many organizations have democratized Seeing the ideas of others is generally believed to stimulate
their innovation processes. Firms now involve consumers in the innovation, as prior knowledge can be reused, recombined,
product development process in many ways, ranging from and accumulated in novel ways to create new knowledge (Fur-
usability groups that test product prototypes to innovation con- man and Stern 2006; Murray and O’Mahony 2007). Indeed,
tests (Chesbrough 2003; Terwiesch and Ulrich 2009). Our many innovation examples illustrate how organizations can
research focuses on “open” innovation contests—contests that learn from and build on the discoveries of others. For instance,
allow anyone to participate, but also make all submissions trans- in the pharmaceutical industry, drugs such as insulin or peni-
parent and accessible to participants. Popular examples of plat- cillin have been improved as subsequent innovators have bet-
forms enabling such contests include OpenIdeo.com, tered previous technologies (Scotchmer 1991). In the software
Crowdspring.com, 99designs.com, or Atizo.com (now Hyve industry, open-source development such as the Linux operat-
crowd.com). Open contest platforms are the most common for- ing system shows that open sharing of collective knowledge
mat of creative crowdsourcing utilized in the business context
(Eyeka 2017). On each of these platforms, firms hosting contests
typically receive large numbers of solutions at minimal cost. As on the platform as of March 26, 2020. This revealed that a majority of 432,408
a result, these platforms have experienced dramatic growth in (77.60%) of these contests were open and transparent.
recent years and are increasingly used by firms to take advantage
of consumers’ latent creative potential. In many of these con- Reto Hofstetter (corresponding author) is Professor of Marketing and
Director of the Institute of Marketing and Analytics, University of Lucerne,
tests, the numerous submissions are transparent and can be Switzerland (email: reto.hofstetter@unilu.ch). Darren Dahl is Innovate BC
accessed by all participants during the contest.1 Professor of Marketing and Behavioral Science, Sauder School of Business,
University of British Columbia, Canada (email: darren.dahl@sauder.ubc.ca).
Suleiman Aryobsei is Senior Consultant, A.T. Kearney (International) AG,
1
On 99designs.com, firms can choose between blind (i.e., intransparent) or Switzerland (email: suleiman.aryobsei@atkearney.com). Andreas Herrmann
open (i.e., transparent) contests. To learn about the popularity of open contests, is Professor of Marketing and Director of the Institute of Customer Insight,
we counted the number of open/blind contests out of all 557,207 contests listed University of St. Gallen, Switzerland (email: andreas.herrmann@unisg.ch).
96 Journal of Marketing Research 58(1)
can indeed spur ongoing innovation. In ideation, popular by visually grouping similar ideas together reduces participants’
approaches such as brainstorming (Osborn 1953) or brainwrit- felt constraints in expressing their ideas. The remainder of this
ing (Paulus and Yang 2000) build on the concept of idea article is structured as follows. We first review the literature
communication and sharing. Thus, seeing the ideas of others addressing how others can influence individuals’ creativity in
can be viewed as a main advantage of open innovation, as it group brainstorming and ideation. Then, we present our concep-
can foster incremental and cumulative innovation (Dahlander tual framework and formal predictions specific to the context of
and Gann 2010). contest-based ideation. We test these predictions in a series of
Little is known, however, about the role of competition in five empirical studies. We conclude with a summary and a dis-
this context (Amabile 2018). Specific to contest-based idea- cussion of the implications and limitations of this research.
tion, it is unclear how seeing ideas of others (denoted as “prior
ideas”) influences the creativity of a participant in a contest in
which people compete against the manifest submissions of
Conceptual Framework and Hypothesis
others. Although the contest literature has investigated contest Development
format with an emphasis on the design of financial incentives Ideation has a long tradition in business practice. In 1957,
(e.g., Boudreau, Lacetera, and Lakhani 2011; Terwiesch and Osborn introduced brainstorming as an effective way to
Xu 2008), less is known about contest design in terms of the enhance the number and quality of ideas generated in groups.
information presented via user interfaces (UIs). Prior innova- In a typical brainstorming session, individuals are instructed to
tion contest research has found that moving from a blind inter- generate and verbally express as many creative ideas as possi-
face to an open and transparent one can change entries’ quality ble without criticizing them and to build on ideas others have
characteristics and contest outcomes (Wooten and Ulrich already suggested. The internet has taken group ideation
2015). However, the specific mechanisms of how others’ ideas online, and other approaches have emerged, such as electronic
influence creative performance, independent of potentially brainstorming (Gallupe and Cooper 1993), online innovation
confounding factors such as visible feedback from firms, par- communities (Bayus 2013; Stephen, Zubcsek, and Goldenberg
ticipant experience, or entry timing, have not been studied. Luo 2016), and online innovation contests (Terwiesch and Xu
and Toubia (2015) found that decomposing an uncompetitive 2008). In each of these approaches, other group members and
ideation task into a sequence of separate tasks that focus on and their ideas may harm or help creative performance in different
display only one specific common category of prior solutions at ways, as we discuss next (for an overview, see Table 1).
a time can improve the creative performance of knowledgeable
individuals. It is not clear if these results hold when individuals
The Harmful and Helpful Influence of Others in Group
see prior ideas from different categories at the same time,
compete against them, and have to think beyond them in order
Brainstorming and Ideation
to be original. Others can significantly harm the creative performance of indi-
We contribute to this literature by documenting the harmful viduals in group brainstorming and ideation. The mere presence
influence of seeing numerous prior ideas in the context of an of others can increase anxiety and excitement, causing a narrow-
innovation contest and by showing how individuals’ creative ing of attention and reduction in performance (social inhibition;
performance can be improved simply by changing the number Mullen, Johnson, and Salas 1991; Zajonc 1965). The need to
of displayed prior ideas, altering ideation instructions, or verbally express ideas to the group is a major limiting factor that
grouping competitive prior ideas. Importantly, we focus on reduces the productivity of group members, who are unable to talk
contests in which innovators compete against prior ideas for when others are talking and have to wait their turn to commu-
exclusive financial rewards. We find that when prior ideas are nicate their own ideas (production blocking; Diehl and Stroebe
competitively (vs. uncompetitively) presented, they will influ- 1987, 1991). Individuals may withhold their ideas, fearing nega-
ence one’s perceptions of how constrained the ideation task tive evaluations from others (evaluation apprehension; Camacho
will be. Indeed, being confronted with numerous prior ideas and Paulus 1995; Collaros and Anderson 1969) or match others’
as competitors, rather than pure sources of information or lower performance (social matching; Paulus and Dzindolet
inspiration, will heighten the perceived challenge of separating 1993). When performance is evaluated at the group (vs. individ-
one’s ideas from the pack, even when many more solutions are ual) level, social loafing may occur (Harkins 1987; Karau and
still possible. Therefore, we find that seeing more competing Williams 1993). Similarly, individuals may reduce their efforts
ideas harms participants’ creative performance. This effect and free ride on others when the perceived dispensability of their
increases monotonically with higher numbers of prior ideas work to the success of the group increases (Kerr and Bruun 1983).
and is defined for both novices and expert participants. These harmful influences have given rise to alternative
From a managerial perspective, finding ways to reduce the approaches, such as the nominal group technique, electronic
negative effects of prior ideas while simultaneously leveraging brainstorming, or online innovation communities. These strate-
their stimulating effects is crucial in ensuring the success of open gies eliminate brainstorming’s most vexing issues by letting par-
innovation contests. We find that exposing participants to a ticipants generate ideas individually, increasing accountability
limited number of prior ideas, framing instructions in a less through measuring individual performance, and by asking them
competitive manner, or changing the prior ideas’ presentation to type ideas instead of verbally expressing them (DeRosa, Smith,
Hofstetter et al. 97
Table 1. Overview of Harmful and Helpful Influences of Others on Individual Creative Performance in Brainstorming and Ideation.
Literature Direction Context Effect
Mullen, Johnson, and Salas (1991); Harmful Group brainstorming Social inhibition: The mere presence of others (as
Zajonc (1965) independent coactors) can be arousing, narrowing
attention and reducing creative performance
Diehl and Stroebe (1987, 1991) Harmful Group brainstorming Production blocking: Difficulties include not being able to
talk while others are talking, forgetting ideas while one
waits one’s turn to talk, thinking an idea is less relevant
after having to wait to talk, and trying to generate ideas
while others are talking
Paulus and Dzindolet (1993) Harmful Group brainstorming Social matching: Matching the ideas of low-performing
individuals
Camacho and Paulus (1995) Harmful Group brainstorming Evaluation apprehension: Fear of negative evaluation
Diehl and Stroebe (1987); Harkins (1987); Harmful Group brainstorming Social loafing: Pooling of ideas with others reduces
Karau and Williams (1993) accountability for one’s own performance, reducing effort
Kerr and Bruun (1983) Harmful Group brainstorming Free riding: Group members exert less effort as the
perceived dispensability of their efforts for group success
increases
Boudreau, Lacetera, and Lakhani (2011); Harmful Innovation contests Incentive effect: Increasing the number of participants
Taylor (1995) reduces winning chances
Bayus (2013); Duncker (1945); Jannson Harmful Design tasks; ideation Fixation: Individuals are shown to conform to a prior idea
and Smith (1991); Smith, Ward, and tasks; innovation and produce creative solutions that include features seen
Schumacher (1993); Kohn and Smith communities in the template prior idea(s)
(2011)
Stephen, Zubcsek, and Goldenberg (2016) Harmful Innovation communities Network structure effect: Higher connectivity among
ideators harms the innovativeness of ideas
Dennis and Valacich (1993); Dugosh et al. Helpful Nominal groups (with Cognitive stimulation: Cognitive stimulation occurs from
(2000); Nijstad and Stroebe (2006); exposure to ideas seeing prior ideas because they activate concepts and
Paulus and Yang (2000) of others) topics (i.e., images) that otherwise would not have been
activated, potentially leading to new ideas that contain or
combine such activated elements in new ways
Gallupe et al. (1992); Paulus et al. (2013) Helpful Electronic Number of prior ideas: The number of ideas that members
brainstorming can build on increases as a function of the number of ideas
exposed to
This article Harmful Innovation contests Idea competition effect: Exposure to a larger (vs. smaller)
number of competitive prior ideas increases the
perceived constraints of expressing one’s own ideas in
the mind of the participants, harming creativity
and Hantula 2007). Yet others’ ideas are still typically visible, However, in these studies, individuals did not compete against
which can result in fixation (Duncker 1945) when the template the prior ideas they saw.
ideas or prior ideas share common concepts (e.g., Smith, Ward,
and Schumacher 1993). In these situations, participants conform
to the ideas of others and produce creative solutions that include Contest-Based Idea Generation and Competing Prior
features seen in the prior idea(s) (Jannson and Smith 1991; Kohn Ideas’ Harmful and Helpful Influences
and Smith 2011). In innovation communities, friends tend to Contest-based idea generation differs from the previously dis-
suggest similar ideas, thereby reducing the inspiration realized cussed group ideation approaches in one particular and major
from seeing them (network structure effect; Stephen, Zubcsek, way: individuals’ performance is evaluated relative to the per-
and Goldenberg 2016). formance of other group members instead of a group’s overall
Despite these negative influences, seeing others’ ideas may performance (the sum of all performances), introducing com-
not always harm creative performance but may instead cogni- petition into the ideation task. Importantly, this evaluation typi-
tively stimulate creativity (Dennis and Valacich 1993; Nijstad cally happens only after all ideas have been submitted and the
and Stroebe 2006). Greater stimulation is also achieved with contest has ended (Harvey and Kou 2013). Contest participants
exposure to more prior ideas, which has resulted in the obser- are rewarded by rank, and only the top-ranked individuals
vation of a positive number of prior ideas effect—in other receive a financial reward, emphasizing the need to stand out
words, the more ideas individuals are exposed to, the better from the competition (Lazear and Rosen 1981). In sports, for
their creativity (Gallupe et al. 1992; Paulus et al. 2013). instance, this increase in the importance of good or improved
performance has been found to increase pressure among ath- choice in how to do a task) are less creative (Amabile 2018).
letes, who may “choke” under pressure, resulting in perfor- Importantly, externally imposed output constraints that harm
mance decrements (Baumeister 1984; Otten 2009). motivation are different from input constraints, which can
Competition can also increase stress levels, which may impede actually increase creative performance in uncompetitive idea-
performance on tasks that require undivided attention (Ellis tion (i.e., they trigger a more constructive cognitive process;
2006; LeBlanc 2009). In innovation contests, Boudreau, Lace- Moreau and Dahl 2005).
tera, and Lakhani (2011; Taylor 1995) show that increasing the In summary, we posit that competition frames the prior
number of competitors lowers winning chances, thereby dilut- ideas as a constraint to creative output, which in turn interferes
ing individual incentive to exert effort (incentive effect). Thus, with the free expression of ideas. Therefore, when prior ideas
the typically large numbers of ideas in online ideation result in are competitively presented, exposure to a growing number of
low probability of winning for an individual idea, which in turn prior ideas will increase the individual’s perceived constraint of
reduces individuals’ motivation to exert sufficient effort expression and will, in turn, lead to a reduction in creative
toward the innovation task. performance (i.e., an idea competition effect). Importantly,
In our research, we add to these identified effects by propos- we predict that these effects are independent of the objective
ing an additional negative outcome when contest participants chances of winning. More formally,
are exposed to competitive prior ideas (i.e., an idea competition
effect). Beyond the influence of a contest’s objective incentive H1a: In open innovation contests, where prior ideas are
properties, we propose that exposure to a larger (vs. smaller) competitively presented, exposure to a larger (vs. smaller)
number of competitive prior ideas increases the perceived con- number of prior ideas decreases the level of creative per-
straints of expressing one’s own ideas in the mind of the parti- formance realized (i.e., a negative effect of prior ideas).
cipants, thus harming their creative performance. This effect H1b: The negative effect of prior ideas on creative per-
occurs when telling participants that their performance will be formance is mediated by the perceived constraints of idea
compared with others (competition), which implies that they expression.
should not copy others’ ideas and differentiate their own ideas
for them to be perceived as innovative (constraint). Prior ideas may also help creative performances. Nijstad,
Indeed, when prior ideas are competitively presented, they Stroebe, and Lodewijkx (2002) showed how example ideas
will directly influence the assessed challenge of proposing stimulate the ideation process by presenting a series of such
creative ideas. We define “creative” as being the first in pro- ideas prior to participants’ own idea generation. Individuals
posing an idea and suggesting ideas that go beyond what others built on the examples by leveraging them in their own crea-
have already identified (i.e., proposing original ideas; Guilford tions. Dugosh et al. (2000) showed that greater stimulation can
1967). Some contest platforms even explicitly state that con- be expected if more example ideas are shown. Increased acti-
testants “must create their own unique implementations, and vation of previously stored knowledge through exposure to
must not merely attempt to replicate the decisions made by prior ideas can thus increase individuals’ cognitive ability to
[others]” and that “unique and original concepts must be generate numerous and original ideas (i.e., a number of prior
respected and only developed by the designer that introduced ideas effect; Dennis and Valacich 1993; Nijstad and Stroebe
them.” (see, e.g., 99designs 2018). Therefore, individuals are 2006; Paulus and Yang 2000).
likely to experience external constraints on their idea expres- The cognitive stimulation benefits gained from exposure to
sion due to the disallowance of copying, imitating, or utilizing prior ideas are more likely to be realized in uncompetitive
elements of original concepts that have already been submitted settings when the context emphasizes the informational value
(Hofstetter, Nair, and Misra 2020). of prior ideas and not their competitive nature. We argue that in
Being confronted with numerous prior ideas as competi- open innovation contests, where the competitive nature of the
tors, rather than ideas as sources of information or inspiration, prior ideas is salient, the heightened competition inherent in
will heighten the perceived challenge of separating one’s seeing more prior ideas draws attention away from their infor-
ideas from the pack, even when there is enough room for mational benefits. Increasing the saliency of competition in the
additional ideas. Individuals will feel less able to express their presentation of prior ideas can cause a shift away from their
own ideas and more constrained by the prior ideas (i.e., they informational aspects, thus reducing their positive influence
will feel that they cannot freely express ideas when incenti- (Deci and Ryan 1985). Therefore, we argue that seeing more
vized to care about the distinctiveness of those ideas. This in prior ideas can increase cognitive stimulation but only when
turn reduces the subjective probability of their being able to they are presented uncompetitively, without constraints. The
produce further ideas (i.e., the participant’s expectancy to do outcome of this increased stimulation is improved creative per-
so) and thus the tendency to perform well and persist in idea- formance (Figure 1 summarizes the hypotheses that form our
tion (Nijstad, Stroebe, and Lodewijkx 1999; Vroom 1964). conceptual model). Thus,
Indeed, we contend that fewer ideas will be generated and the
chances that high-quality ideas emerge will be reduced. This H2a: In open innovation contests, when prior ideas are
expectation is in line with prior research arguing that people presented uncompetitively, exposure to a larger (vs.
who feel constrained in their environment (e.g., by having less smaller) number of prior ideas increases the level of
H1b
Constraint of Idea
Expression (M1)
−
+
Number of rior H1a: − / H2a: + Creative
deas (X) Performance (Y)
+ −
+
− Cognitive Stimulation
(M2)
Competitive
Presentation (W) H2b
Moderators within competitive presentation:

• High-/ ow-constraints instructions (Study 3)
• Expertise (Study 4)
• Grouping of ideas (Study 5)
Figure 1. Conceptual framework of how competition affects the dual role of prior ideas in open innovation contests.
creative performance realized (i.e., a positive effect of this is to happen (Fiske and Taylor 1991; Lewandowsky, Little,
prior ideas). and Kalish 2007). For example, expert accountants were found
H2b: The positive effect of prior ideas on creative perfor- to be less capable of using a new tax law replacing an existing
mance is mediated by cognitive stimulation felt by the law (Marchant et al. 1991), and compared with novices, expert
participant. bridge players found it more difficult to adapt to a modified
game with altered rules (Dane 2010; Frensch and Sternberg,
1989). Experts fixate on internal knowledge and known paths
Constrained Ideation Tasks and Domain Expertise to solutions (different from fixation on external examples;
There are considerable individual differences in how a con- Smith and Blankenship 1991; Youmans and Arciszewski
strained ideation task may undermine creativity. We explore 2014). Such fixation and inflexibility may be particularly
how domain experts and novices cope with the constraints harmful when developing maximally innovative (or radical)
arising from competitive prior ideas. Participants in ideation ideas that require flexible reorganization and combination of
contests are often novices (Terwiesch and Xu 2008), yet some concepts in a way that departs from established patterns.
participants may be considered experts due to prior knowledge Therefore, we expect that domain expertise will increase the
specific to the ideation domain and their prior experience in negative effect of seeing more prior ideas on creative perfor-
related ideation tasks (Luo and Toubia 2015). mance. External constraints will reduce experts’ subjective
Although there is a wide agreement that a certain level of probability of producing creative ideas due to (1) less flexibil-
expertise is necessary for creativity, greater expertise may not ity in integrating additional information and (2) a fixation on
known paths to solution (which are likely less valid with more
always be better (Coursey et al. 2019; Dane 2010; Mumford
competing prior ideas being seen). Although our theorizing and
et al. 2006; Stacey, Eckert, and Wiley 2002; Wiley 1998). By
conclusions specifically draw from domain expertise, similar
definition, experts possess a large body of domain knowledge
results may be expected for the related constructs of anticipated
and store substantial and meaningful chunks of domain-related
success, self-efficacy, and ideation expertise.
information. This also involves methodological knowledge—
including trained, habitual, and algorithmic ways to solve sim- H3: High (vs. low) domain expertise increases the nega-
ilar problems—and solutions that have worked in the past tive effect of prior ideas on creative performance.
(Amabile 2018; Dane 2010; Murray and Häubl 2007). These
knowledge structures can be quickly activated to produce solu-
tions following a path of least resistance (Moreau and Dahl Creative Performance in Open Innovation Contests
2005; Ward 1994). In open innovation contests, the frequently used winner-take-
However, when confronted with external constraints from all prize structure suggests that managers typically care most
numerous competitive prior ideas, experts are required to flex- about the quality of the top ideas they select for later processing
ibly integrate new information and deviate from their known and development (Hofstetter, Zhang, and Herrmann 2018).
solution paths. The higher the domain expertise, the less likely Managers often care about breakthrough or top ideas and the
maximum innovativeness that can be achieved (Fleming 2007). In testing our hypotheses, we controlled for a range of alter-
Therefore, we use the innovativeness of an individual’s best native explanations for harmful or helpful influences of prior
idea out of all generated ideas (top-box innovativeness) as our ideas mentioned by prior research (Table 1). Ideation was car-
key measure of performance. In previous research, creative ried out electronically, individually, and anonymously, which
performance often has been measured on two quality dimen- controlled for the possibility of social matching, mere presence
sions: an individual’s ability to produce ideas that are original of others, or network structure effects, and eliminating produc-
(or novel, unique; Gallupe et al. 1994) and useful (or appealing, tion blocking. Evaluation of the ideas happened at the end of
valuable; Dahl, Chattopadhyay, and Gorn 1999). An increase in the contest, which controlled for evaluation apprehension and
the level of originality achieved in the innovation process has possible interplay between simultaneous ideation and evalua-
been shown to have a direct positive effect on the success of a tion (Harvey and Kou 2013). We incentivized individual as
new product development effort (e.g., Dahl and Moreau 2002; opposed to group efforts, controlling for the possibility of
Goldenberg, Lehmann, and Mazursky 2001; Lilien at al. 2002). social loafing. Winning chances (i.e., the number of competing
However, originality alone may not guarantee market success; ideas) were fixed in the instructions across groups, and in Stud-
a successful outcome also requires that customers perceive ies 2–4, we explicitly state a fixed total number of prior ideas
products as useful (Dahl, Chattopadhyay, and Gorn 1999; Mol- (i.e., the total number of ideas already submitted to the contest).
dovan, Goldenberg, and Chattopadhyay 2011). Together, ori- This number is identical across both smaller and larger prior
ginality and usefulness define an idea’s innovativeness. In ideas conditions. Beyond objective winning chances, the Study
addition, in innovation contests the number of ideas also indir- 3 follow-up also controls for perceived winning chances.
ectly matters (i.e., creative fluency; Guilford 1967), as greater Finally, participants were randomly assigned to conditions (and
numbers increase chances for top ideas (Girotra, Terwiesch, contests in Studies 1 and 5) and could not move between
and Ulrich 2010; Terwiesch and Ulrich 2009). experimental conditions (or contests), thereby eliminating
In the open innovation contest context that we study, we self-selection concerns.
acknowledge that individuals’ creative performances can be
measured in multiple ways, but we emphasize top-box innova-
tiveness as the key metric. Nevertheless, in all studies we will Study 1: A Field Experiment on an Open
also report results for the alternative measures of top-box ori- Innovation Platform
ginality, top-box usefulness, total innovativeness, and number In Study 1 we investigate the overall impact of the number of
of ideas generated. To integrate multiple elements of creative prior ideas on creative performance within a competitive
performance in a single measure, we follow the approach of frame. We used data from a field experiment across six actual
Luo and Toubia (2015), who define creative performance as the open innovation contests hosted on the European open innova-
individual-level sum of the average innovativeness ratings.2 tion platform Atizo (now HyveCrowd). Atizo was founded in
Throughout the studies, we focus on top-box innovativeness Europe in 2007 and now has an online community of 98,000
in our analyses but also report all four additional measures. For registered participants (at November 2020). Participants are
dependent variables, means, and results of all reported studies, incentivized with financial rewards that are granted based on
see Table 2. creative performance. Contests on average attract 135 partici-
pants and 357 ideas over the course of the typical contest dura-
tion of 25 days. The typical Atizo user is 39 years old, has been
Overview of Studies and Controls a member for 2.43 years, and is male (74%). Examples of
Five experiments test our predictions, with Studies 1 and 5 innovation problems hosted by companies such as Nestlé (food
being field studies. Studies 2, 3, and 4 use a similar contest- and beverage) and BMW (car and motorcycle manufacturer)
based ideation paradigm, in which individuals perform an include new product and service ideas for yogurt, mobility
unusual uses task for a common household object (Guilford services, or creative marketing slogans for consumer goods.
1967) while seeing a sample of prior ideas at the same time,
mimicking the UI of typical ideation platforms. With their own Method
ideas, individuals compete for a financial reward that is granted
to those who generate the best ideas (for details, see Study 2). The experiment varied the number of prior ideas participants
All studies (except for Study 5) expose individuals to a random were exposed to when entering the contest. We randomly dis-
sampling of different prior ideas we provide in Web Appendix played a list of either 2 or 10 different prior ideas (i.e., smaller
A, Tables 1 and 2. We vary the number of prior (smaller vs. vs. larger number; we used larger numbers in later studies) out
larger) ideas in all studies, with differing numbers across of a set of at least 15 ideas per contest. These ideas were
studies. generated by a small subsample of five Atizo community mem-
bers before the contest started (we invited Atizo members to
suggest ideas by email; Web Appendix A shows the list of ideas
2
Specifically, let xij be the average innovativeness score an idea i of individual used to sample from; Web Appendix A, Figure 1, shows how
j received and Nj be the total number of ideas submitted by j; then creative the ideas were displayed). Clickstream data of Atizo commu-
PN j
performance is calculated as Yj ¼ x .
i¼1 ij nity members revealed that 90% of them view nine or fewer
Table 2. Means Across Experimental Conditions over All Studies.
Experimental Top-Box Top-Box Top-Box No. of Total Guilford’s

Study Group N Innovativeness Originality Usefulness Ideas Innovativeness Originality
1 2 prior ideas 263 3.533 2.627 4.549 1.662 5.423 —

10 prior ideas 277 3.376* 2.379** 4.442y 1.444** 4.683* —
Main finding: Study 1 shows that seeing more prior ideas reduces creative performances of contestants across measures in actual
ideation contests.
2 5 prior ideas, competitive 98 3.054 3.382 3.603 4.592 12.719 24.89

50 prior ideas, competitive 106 2.822*** 3.048*** 3.003*** 2.906*** 7.997*** 18.71**
5 prior ideas, uncompetitive 103 3.056 3.361 3.503 4.893 13.436 27.17
50 prior ideas, uncompetitive 93 3.101 3.606* 3.487 5.720* 15.564* 32.60*
Moderator (manipulated): Competitive (vs. uncompetitive) presentation

Main finding: Study 2 provides direct evidence for the proposed framework by showing that seeing prior ideas as competitors
(competitive) rather than sources of inspiration (uncompetitive) introduces the prior ideas’ harmful influence.
3 5 prior ideas, high CE 254 2.672 2.855 3.360 4.339 9.277 —

50 prior ideas, high CE 232 2.510** 2.711* 3.113** 3.233*** 6.943*** —
5 prior ideas, low CE 253 2.675 2.893 3.321 4.158 9.133 —

50 prior ideas, low CE 245 2.791* 3.037* 3.224 4.155 9.099 —
Moderator (manipulated): High versus low CE within competitive presentation

Main findings: Study 3 provides moderated mediation evidence for perceived constrained expression as mediator of the prior
ideas’ harmful influence. It supports the harmful influence of prior ideas after controlling for a depletion of the idea pool and the
comparability of prior ideas between smaller and larger numbers.
Study 3 5 prior ideas 260 2.789 2.949 2.745 4.069 9.533 —

follow-up 50 prior ideas 240 2.664* 2.816* 2.629* 3.3*** 7.703*** —
Mediators: Perceived constrained expression, perceived competence, perceived self-efficacy, perceived winning chances,
perceived competitiveness
Main finding: Exploratory mediation provides supporting evidence for constrained expression as mediator. It supports the harmful
influence of prior ideas after controlling for a depletion of the idea pool, the comparability of prior ideas between smaller and
larger numbers, and after controlling for functional fixedness (in two ways).
4 3 prior ideas 347 3.631 3.609 3.793 2.870 13.036 —

10 prior ideas 337 3.465** 3.493* 3.699* 2.325*** 10.338*** —
100 prior ideas 316 3.421*** 3.415*** 3.609*** 2.193*** 9.046*** —
Moderators (measured): Domain expertise, ideation expertise, creative self-efficacy, innovativeness, lead user, emergent
consumerism, personal involvement, achievement orientation
Main finding: First, Study 4 provides evidence that effect persists also for larger (compared with previous studies) numbers of ideas.
Second, it shows that expertise moderates the prior ideas harmful influence such that the prior ideas more strongly negatively
influence creative performance for those high in domain expertise (also ideation expertise, lead user–ness, creative self-efficacy,
and involvement).
5 Open 278 4.363 3.260 4.766 1.309 5.625 —

Restricted 273 4.485* 3.775*** 4.971* 1.729*** 7.557*** —
Grouped 277 4.616*** 3.679*** 4.935y 1.744*** 7.711*** —
Main finding: Study 5 validates the implications of prior studies on the design of a mock-up ideation platform’s UI. Restricting and
grouping competitive prior ideas increases creative performances of individuals.
Total 4,252
*p < .05.
**p < .01.
***p < .001.
y
p < .10.
Notes: CE ¼ constraint of idea expression. Statistical tests are always relative to the smaller number of prior ideas within condition or study. For Study 1, we
ran linear regressions with project controls and clustered standard errors (at the project level). For all other studies, we used ANOVAs. We only calculated
Guilford’s originality for Study 2. In Study 4, the difference between 10 and 100 is significant for novices (F(1, 994) ¼ 5.00, p < .05), but not for experts (F(1, 994) ¼
.01, p > .10).
ideas prior to posting their own ideas, suggesting that ten prior significant negative effect of the prior ideas manipulation on
ideas is appropriate for this platform.3 top-box innovativeness (M2 ¼ 3.53, SD ¼ 1.18 vs. M10 ¼ 3.38,
To control for potential confounds arising from selective SD ¼ .97; b ¼ .24, p < .05), top-box originality (M2 ¼ 2.63,
entry, participants were randomly assigned to one of six similar SD ¼ 1.88 vs. M10 ¼ 2.38, SD ¼ 1.55; b ¼ .41, p < .01), and
ideation contests (e.g., ideas for novel pillows, ideas for viral marginally on top-box usefulness (M2 ¼ 4.55, SD ¼ 1.18 vs.
videos featuring an eyeglasses brand) and they could only see M10 ¼ 4.44, SD ¼ 1.14; b ¼ .12, p < .10) and the remaining
and participate in the contest to which they had been assigned two alternative measures (for additional Study 1 results, see
(the website was personalized accordingly). In the contest, we Table 2 and Web Appendix A).
instructed participants about the specific innovation problem
and provided a short background of the firm that elucidated the Discussion
problem. We also told participants that original ideas would
win 50 CHF (about US$50), and that their ideas would compete In support of H1a, the field experiment provides initial evidence
with all other submitted ideas for these rewards. In all experi- that exposure to an increasing number of competitive prior
mental conditions, we told them that 30% of all participants ideas in an open innovation contest has a negative effect on
would win such a reward. Such high percentages of winners are creative performance. Importantly, the negative effect identi-
not uncommon for contests on Atizo, with some contests fied is not driven by objective winning chances, as they have
rewarding more than 50% of the participants. We invited Atizo been fixed across conditions. We find no evidence of a positive
community members to participate in the contests by email. A stimulation effect of prior ideas in Study 1. We believe that the
total of 2,015 individuals registered for the contests and 540 absence of this effect is driven by the highly competitive nature
decided to participate (26.80%), of which 263 (48.70%) were of Atizo’s innovation contests and the competitive presentation
assigned into the two prior ideas condition and 277 (51.30%) in of prior ideas (e.g., the competitive nature of the task is made
the ten prior ideas condition. These participants submitted 850 evident, contestants’ profiles and contest rewards are promi-
ideas in total. The average entrant submitted 1.57 ideas (SD ¼ nently displayed). We test this further in a more controlled
1.61, min ¼ 1, max ¼ 21). We sent all submitted ideas to the Study 2. Study 2 also addresses the small number of prior ideas
firms, which then returned their evaluations to us (we do not in Study 1 (two vs. ten), which could be seen as a limitation
have any information about the number of raters used or their given that contests often feature many more ideas.
demographics). Each firm only saw and evaluated the origin-
ality of ideas related to their own contest, using a seven-point
Likert scale. The firms were blind to the experimental condi-
Study 2: Constraints Implied by Competitive
tion and the participant who generated the idea. One hundred Presentation Trigger Ideas’ Harmful
sixty-one ideas that ranked highest in originality received a Influence
reward. We augment these data with information about useful- In Study 2, we explicitly test the parallel processes outlined in
ness; we had two marketing experts (doctoral students in mar- H1 and H2 and the (in)dependence of these effects with respect
keting; one male, one female) evaluate the ideas on seven-point to the competitive nature of the prior ideas’ presentation in the
Likert scales and used the average as our measure (a ¼ .89). open innovation contest. Consequently, we varied the number
We then calculated an idea’s innovativeness as (originality þ of prior ideas to which participants were exposed and investi-
usefulness)/2 (Argo and Tu 2013; results are comparable for gate their influence on creative performance when the prior
the product of originality and usefulness). We utilized the five ideas were (vs. were not) competitively presented.
distinct dependent variables to test our hypotheses. Impor-
tantly, we included project fixed effects in all analyses to con-
trol for contest and evaluation differences across competitions.
Method
The experiment had a 2 2 between-subjects design that varied
the number of prior ideas (smaller: 5 out of 100 vs. larger: 50 out
Results of 100) and the nature of the prior ideas’ presentation (compet-
We pooled all data across contests to increase power, but we itive vs. uncompetitive). We displayed more prior ideas in this
included project controls (and interactions of projects with the study compared with Study 1 to test whether the negative influ-
number of prior ideas, which are not significant in any of the ence of prior ideas also manifests within larger numbers of those
analyses) to account for differing performance levels across ideas, and for comparability with previous experiments that
projects. In support of H 1a , regression analysis shows a exposed individuals with up to 60 ideas as part of the Guilford
task (Nijstad, Stroebe, and Lodewijkx 2002). Most participants
(84%) correctly recalled the number of prior ideas they were
3
To further validate this manipulation, we conducted a pretest among 50 U.S. confronted with in a recognition check question (n.s. between
individuals from Amazon Mechanical Turk (MTurk). Individuals who were competitive conditions). We invited U.S. individuals from
exposed to ten ideas (vs. two) indicated agreement toward the response “There
are a lot of prior ideas” (7) relative to the response “There are few prior ideas” MTurk and offered them a fixed and a variable (bonus) amount
(1) on a seven-point semantic differential (M10 ¼ 5.00, M2 ¼ 2.96, diff ¼ 2.04; for their participation in the study. We told them that by parti-
t(48) ¼ 4.04, p < .01). cipating in the study they had entered into a competition for an 1
USD bonus payment that would be paid out depending on their task. We measured cognitive stimulation using four items (e.g.,
creative performance (in addition to the fixed payment). We “After having viewed the others’ ideas, I am aware of many
fixed the winning chances by telling them that typically 5% (5 different concepts and topics related to bricks”; a ¼ .93; see
out of 100 ideas) win the bonus in all experimental conditions. Web Appendix B). Factor analysis revealed that all items load
We told participants in all conditions that there are currently 100 onto the same factor and that only one factor was retained
ideas submitted and that we would show them a random sam- (eigenvalue ¼ 2.35, proportion ¼ 1.07, w2(3) ¼ 950.68, p <
pling of either 5 or 50 different ideas out of these 100 prior ideas. .001). We measured perceived constraint of idea expression
In the uncompetitive condition, we told study participants with three items based on Ryan and Deci’s (2000) Basic Need
that the prior ideas were a sampling of example ideas generated Satisfaction Scale (Gagné 2003; e.g., “In this innovation task, I
by previous participants and that the winner would be deter- don’t feel free to express my ideas”; a ¼ .91; see Web Appen-
mined by individually assessing the innovativeness of their dix B). Participants then completed the actual innovation task.
new ideas, independent of the prior ideas displayed. Partici- We used a version of Guilford’s unusual uses task (Guilford
pants who generated innovative ideas would win the bonus. In 1967; Guilford, Merrifield, and Wilson 1958), a divergent
the competitive condition, we told participants that their ideas thinking task, to measure the creative performance of partici-
were competing with the prior ideas and that we would com- pants. Divergent thinking tasks are frequently used to estimate
pare their creative performance against other participants. the potential for creativity and have been found to be reliable
Those who generated different and innovative ideas (compared and also predictive of real-world creative performance. Guil-
with other competitors) would win the bonus.4 ford’s task requires participants to produce varied responses to
We also indicated that each prior idea was a randomly chosen a question that has multiple alternatives within a limited time
idea of another competing participant, and that their own ideas frame. In this study (and Studies 3 and 4), we asked participants
would compete against the ideas we showed them (competitive to generate as many innovative ideas for unusual uses of a brick
as possible.
condition) or that the ideas they would see were randomly cho-
A total of 400 U.S. respondents from MTurk participated in
sen from previous participants (uncompetitive condition).5 In
the study (Mage ¼ 36.82 years, SD ¼ 11.95; 46.50% female).
each condition, we told participants that prior ideas represented
Altogether, they submitted 1,794 ideas for uses for a brick. The
a sampling of different prior ideas of participants who may or
mean respondent submitted 4.49 ideas (SD ¼ 2.84, min ¼ 1,
may not have submitted more than one idea. Participants were
max ¼ 16). We calculated five different outcome measures as
then asked to submit as many innovative ideas as they could.
follows: we had MTurk raters evaluate the innovativeness, ori-
Before participants were asked to submit their ideas, we
ginality, and usefulness of each idea on five-point Likert scales
exposed them to the sampling of prior ideas. Participants either
(one separate item for each dimension; e.g., 1 ¼ “Not at all
saw a smaller number of different prior ideas (5 different ran-
innovative,” and 5 ¼ “Extremely innovative”). We used ten
domly picked ideas) or a larger number of different prior ideas raters for each dimension (asinnovativeness > .60, asoriginality >
(50 different randomly picked ideas). The ideas and their order .70, asusefulness > .75). These ratings allowed us to calculate the
were randomized between participants. The presented prior five creative performance measures. For the sake of compar-
ideas were sampled out of 100 different ideas collected in a ability, we also calculated Guilford’s total originality measure,
separate pretest with participants from the same population (for but only for this study (as described in Web Appendix C). Note
a list of these ideas and the presentation format, see Web that results are comparable if we calculate innovativeness as
Appendix A, Table 2). the sum or product of originality and usefulness.
After participants had seen these prior ideas, we asked them Mean top-box innovativeness equaled 3.00 (SD ¼ .362, min
to report in random order their level of cognitive stimulation6 ¼ 1.24, max ¼ 3.87). Mean cognitive stimulation equaled 4.24
and perceived constraint of idea expression in the innovation (SD ¼ 1.53, min ¼ 1, max ¼ 7), and mean perceived constraint
of expression equaled 2.38 (SD ¼ 1.50, min ¼ 1, max ¼ 7).
4 Note that cognitive stimulation and perceived constraint of
Note that this manipulation of competition implies a constraint on idea
expression.
expression are uncorrelated (r ¼ .08, p > .10) (we provide
5
We ran a pretest among 88 participants from MTurk. We randomly assigned correlations across studies in Web Appendix D).
individuals to one of the two task descriptions (45 to the uncompetitive and 43 Finally, we checked for possible external fixation, which
to the competitive group) and then measured how they perceived competition can harm creative performance beyond the perceived constraint
using three items—“I think the task is very competitive,” “I think the of idea expression. Here, fixation may occur in the form of
competition in this task is very intense,” and “Competition is fierce in this
task”—on seven-point Likert scales (a ¼ .84). The competitive description was functional fixedness (Duncker 1945), which refers to the ten-
perceived as significantly more competitive than the uncompetitive description dency to perceive an object only in terms of its most common
(Mhigh ¼ 5.39, Mlow ¼ 4.67; d ¼ .72, t(86) ¼ 2.68, p < .01). We also measured use. Individuals may not be able to think beyond the prior ideas
how well respondents understood what they would have to do in the task based they see and accordingly may functionally fixate on them.
on the description (“I understand what I would have to do in this task”) and When experiencing functional fixedness, individuals will
found high levels of comprehension (Munderstand ¼ 6.01, SD ¼ 1.28, min ¼ 1,
max ¼ 7). likely restate prior ideas. Although fixation occurs typically
6
Asking about stimulation before the task may affect results. We omitted this when only very few common concepts are displayed and
measure in Studies 1, 3, and 5. should vanish with increasing numbers of prior ideas (Jannson
moderated parallel mediation model with 10,000 bootstrapped

estimates (Hayes 2015; Preacher and Hayes 2008; Preacher,
Rucker, and Hayes 2007; for regressions, see Web Appendix
E, Table 6). We first report the indirect effect over constraint of
idea expression followed by the total effect over cognitive
stimulation (Figure 3). Mediation analyses in the following
studies are performed in the same fashion. In support of H1b,
we find that increasing the number of prior ideas significantly
increases constraint of idea expression in the competitive con-
dition (b ¼ .78, SE ¼ .21, p < .001). Constraint of idea expres-
sion is negatively related to top-box innovativeness (b ¼ .09,
SE ¼ .02, p < .001). The conditional indirect effect is nega-
tively significant in the competitive condition (oM1 ¼ .07
[.13, .03]b) and not significant in the uncompetitive condi-
tion (oM1 ¼ .02 [.01, .05]); moreover, the index of moderated
mediation is significant and negative (oM1 ¼ .09b [.16,
.03]). Cognitive stimulation shows a different pattern. In sup-
port of H2b, we find that increasing the number of prior ideas
Figure 2. More prior ideas discourage creative performance when positively influences cognitive stimulation in the uncompeti-
presented competitively (Study 2). tive condition (b ¼ .48, SE ¼ .21, p < .05). However, cognitive
Notes: Error bars represent standard errors. stimulation is not significantly related to top-box innovative-
ness (b ¼ .01, SE ¼ .02, p > .05), resulting in a nonsignificant
and Smith 1991), we found that 34 (1.90%) submitted ideas indirect effect in both conditions (oM2(Competitive) ¼ <.01
were identical to the prior ideas. We find more identical ideas [.01, .02]; oM2(Uncompetitive) ¼ <.01 [.01, .02]).
in the 50 (vs. 5) prior ideas group (mean number of ideas that Web Appendix E, Table 7, reports the indirect effects over
constraint of idea expression and cognitive stimulation for all
are copies: M50 ¼ .17, SD ¼ .49; M5 ¼ .01, SD ¼ .07; F(1, 396)
five creative outcomes (plus Guilford’s originality measure).
¼ 15.19, p < .001), but this number does not differ signifi-
Importantly, the indirect negative effect via constraint of idea
cantly by competitive presentation (Mcompetitive ¼ .07, SD ¼
expression in the competitive condition is significant across all
.32; Muncompetitive ¼ .10, SD ¼ .39; F(1, 396) ¼ .80, p > .10).
measures. Cognitive stimulation appears to primarily affect the
Results. First, utilizing analysis of variance (ANOVA), we
number of ideas, total innovativeness, and Guilford’s original-
find a main effect of presentation such that participants are
ity measure, but not the maximum quality that an individual
more creative in the uncompetitive (vs. competitive) condition
can achieve, thus offering only partial support for H2b.
(F(1, 396) ¼ 16.44, p < .001; Figure 2). The main effect sup-
ports our notion of the detrimental influence of seeing compet-
itive ideas per se. The interaction between number of prior Discussion
ideas and their presentation is significant (F(1, 396) ¼ 15.96, Our results show that prior ideas can harm or stimulate people’s
p < .001). In line with H1a, we find that when prior ideas are creative performance depending on whether the prior ideas
presented competitively, increasing the number of prior ideas shown in a creative task are presented competitively. In support
significantly reduces the creative performance of participants of H1a and H1b, an increase in the shared number of prior ideas
(M5 ¼ 3.05, SD ¼ .25; M50 ¼ 2.82, SD ¼ .45; F(1, 396) ¼ 22.88, increases perceived constraint of idea expression and decreases
p < .001; Cohen’s d ¼ .63). Concerning H2a, in the uncompe- the resulting creative performance when they are competitively
titive condition, creative performance is not significantly differ- presented. Results for H2a and H2b are mixed. Although we do
ent between seeing 50 vs. 5 prior ideas (M5 ¼ 3.06, SD ¼ .33; not find a positive influence of seeing more prior ideas on top-
M50 ¼ 3.10, SD ¼ .31; F(1, 396) ¼ .82, p > .10; d ¼ .14). box innovativeness (and usefulness), we find that higher num-
Table 2 shows all means across all measures. We see a bers of prior ideas can stimulate the generation of more ideas,
consistent negative effect from exposure to more prior ideas higher top-box originality, and an increase both total innova-
across all five measures in the competitive condition. Although tiveness and Guilford’s originality measure when they are pre-
we failed to find a significant effect in the uncompetitive con- sented uncompetitively. We return to these mixed findings in
dition for top-box innovativeness, we find a positive effect for the “General Discussion” section.
top-box originality, number of ideas, and total innovativeness, We performed a series of follow-up robustness check stud-
partially supporting H2a. For means and ANOVA for the med- ies to this study, which we report in Web Appendix F (i.e., three
iating variables, see Web Appendix E. follow-up studies). Most importantly, these checks suggest that
To test whether constraint of expression and cognitive sti- the prior ideas’ harmful influence remains after changing the
mulation mediate the role of prior ideas on participants’ crea- incentive scheme from top 5% to winner takes all (see Web
tive performance as outlined in H 1b and H 2b , we ran a Appendix F, “Contest Incentive Study”), that creative
H1b: M1(Competitive) = −.07a

= .0
Constraint of Idea
Expression (M1)
.18 n.s. (.78***) −.09***
−.97***
Number of prior .11 (−.33***) Top Box
ideas (X) Innovativeness (Y)
−.58***
.48* (.37m)
.01 n.s.
Cognitive Stimulation
−.11 n.s. (M2)
Competitive
Presentation (W) H2:
<
Figure 3. Results of mediation analysis (Study 2).

*p < .05.
**p < .01.
***p < .001.
y
p < .10
a
Significant at 5% level based on bootstrapped confidence intervals.
b
Effects are significant for alternative measures of creative performance (number of ideas generated, total innovativeness, and Guilford’s
originality).
Notes: Numbers in parentheses are effects under competitive presentation. Results after including demographic controls are comparable; standard errors are
bootstrapped with 10,000 replications.
performance reduces monotonically with increasing numbers competitive frame) can alleviate the negative influence of com-
of prior ideas (see Web Appendix F, “Number of Levels of petitive prior ideas. This outcome would provide process evi-
Prior Ideas Study”), and that competition saliency moderates dence supporting H1b and also demonstrate an approach to
the effect (Web Appendix F, “Competition Salience Study”). reducing the negative effects of exposing individuals to com-
Finally, we find some evidence for possible external fixa- petitive prior ideas. In this study we also control for two alter-
tion, as more ideas were identical to the prior ones in the larger native explanations for the effect: that is, (1) the pool of
(vs. smaller) number of prior ideas group in the competitive available ideas naturally depletes with more ideas (idea pool
condition. Such fixation may explain the reduction in creative depletion) and (2) the 50 (vs. 5) ideas group may contain better
performance. However, in the uncompetitive condition where top ideas just by chance (idea comparability).
individuals copied prior ideas to the same extent, the effect on
creative performance is not significantly negative (for top-box
innovativeness and usefulness) and even reverses (for top-box Method
originality, number of ideas, total innovativeness, and Guil-
The study design paralleled that of Study 2 (within the compe-
ford’s originality measure). We interpret this as evidence
tition condition). We again use Guilford’s unusual uses task for
against external fixation in explaining the prior ideas harmful
a brick to measure creative performance. We manipulated the
influence. In Study 3 (and follow-up), we make an effort to
number of competitively presented prior ideas (5 vs. 50) and
explicitly control for functional fixedness and alternative pro-
used either high- or low-constraining instructions (high vs. low
cess explanations.
constraints) in a 2 2 between-subjects design. Most partici-
pants (89%) correctly recalled the number of prior ideas they
Study 3: Lowering the Perceived Constraint were shown in a recognition check question (n.s. between com-
petitive conditions). In the high-constraint idea expression con-
of Idea Expression Mitigates the Negative
dition, respondents received the following instructions when
Effect seeing the prior ideas: “Importantly, you cannot express your
In Study 3, we experimentally manipulate the level of felt own ideas the way you want and as freely as you like because
constraint of idea expression and use moderation analysis to you are not allowed to copy ideas of others that have already
test its fundamental role in our process explanation (Spencer, been submitted.” Individuals in the low-constraint idea expres-
Zanna, and Fong 2005). Specifically, we test whether reducing sion condition received these instructions: “Importantly, you
felt constraints through less constraining instructions (within a are not allowed to copy ideas of others that have already been
submitted. However, note that you can express your own ideas A total of 984 U.S. respondents from MTurk participated in
the way you want and as freely as you like.” After seeing the the study (Mage ¼ 37.36 years, SD ¼ 11.43; 53.46% female).
ideas, participants responded to the perceived constraint of idea Altogether, respondents submitted 4,523 ideas. The mean
expression scale (a ¼ .93), followed by the unusual uses task. respondent submitted 3.99 ideas (SD ¼ 2.41, min ¼ 1, max
As a manipulation check, perceived constraint of idea ex- ¼ 17) and the mean top-box innovativeness score equaled 2.66
pression is significantly higher with the high- (vs. low-) con- (SD ¼ .64, min ¼ 1, max ¼ 4.67; rater asinnovativeness > .71,
straints instructions (Mhigh ¼ 3.05, Mlow ¼ 2.20; t(982) ¼ 7.82, asoriginality > .75, asusefulness > .79).
p < .001).
Results and Discussion

Additional Controls A positive interaction term implies that the prior ideas influ-
Importantly, we adjusted the sampling of prior ideas to avoid ence top-box innovativeness differently in the high- versus
potential confounds (in Study 3, Study 3 follow-up, and Study low-constraints conditions (F(1, 983) ¼ 11.68, p < .001;
4).7 First, when larger numbers of prior ideas are displayed, the Table 2). In the high-constraints condition, we replicate the
prior ideas can constrain expression simply because of poten- finding that increasing the number of prior ideas significantly
tial overlap with the individuals’ own ideas. Therefore, the pool reduces top-box innovativeness (M5 ¼ 2.76, SD ¼ .62; M50 ¼
of available ideas naturally depletes with more ideas (idea pool 2.51, SD ¼ .67; F(1, 980) ¼ 7.77, p < .01; d ¼ .25). In the
depletion). Although the redundancies found in parallel idea- low-constraints condition, however, the effect reverses and
tion are limited, some ideas are more frequently suggested, and becomes significantly positive (M5 ¼ 2.68, SD ¼ .65; M50 ¼
redundancies do exist (Kornish and Ulrich 2011). Aiming to 2.79, SD ¼ .62; F(1, 447) ¼ 4.17, p < .05; d ¼ .18). Perceived
separate a possible idea pool depletion effect from the prior constraint of idea expression increases more strongly with the
ideas’ harmful influence, we removed 19 common ideas from number of prior ideas in the high- (vs. low-) constraints condi-
the set of 100 prior ideas, resulting in 81 prior ideas (Web tion (interaction: F(1, 980) ¼ 8.23, p < .01). In the high-
Appendix A, Table 2). We define common ideas as ideas that constraints condition, increasing the number of prior ideas
were mentioned by at least 5% of other individuals (following significantly increases perceived constraints (M 5 ¼ 2.56,
Guilford [1967]; percentages were determined by an analysis SD ¼ 1.62; M50 ¼ 3.59, SD ¼ 2.00; F(1, 980) ¼ 46.87,
of 7,405 ideas [4,404 original] submitted by 1,840 individuals p < .001; d ¼ .58). In the low-constraints condition, however,
in a large-scale pretest). For example, we removed prior ideas the effect is attenuated but remains significant (M5 ¼ 1.99,
such as “paperweight,” “pencil holder,” or “weapon” for SD ¼ 1.44; M50 ¼ 2.41, SD ¼ 1.62; F(1, 980) ¼ 8.13, p <
unusual uses of a brick.8 .01; d ¼ .28). Cognitive stimulation does significantly
We also adjusted how the prior ideas were sampled. On increase with the number of prior ideas (M5 ¼ 4.12, SD ¼
average, because of the randomization, the sampled ideas will 1.51; M50 ¼ 4.40, SD ¼ 1.61; F(1, 980) ¼ 7.78, p < .01;
not differ in their means between the smaller and larger number d ¼ .18), but the interaction with high/low constraints is not
of prior ideas groups. However, they will differ in their significant (F(1, 980) < .01, p ¼ .94).
extremes because the more the ideas that are sampled, the
higher the chance that the sample contains extreme ideas.
Therefore, the groups will differ in their maximum creative Discussion
performance scores, reducing the comparability of the idea
This study again shows a reversal, corroborating the notion that
groups (idea comparability). This may confound our analysis,
prior ideas have a harmful negative influence and discounting
as innovators may be particularly sensitive to the best prior
the idea that external fixation is a potential process explanation
(most innovative) ideas in a competition, resulting in a poten-
for the identified effects. In a Study 3 follow-up (see Web
tially inflated negative effect of seeing more ideas. To control
Appendix H), we explicitly measure similarity to the prior
for this possibility, we used a search algorithm (see Web
ideas (as an indication of functional fixedness) and remove
Appendix G) that ensures comparability between the number
ideas that overlap with prior ideas in the analysis. In this
of prior idea groups on a set of creative performance measures,
follow-up, we also performed an exploratory mediation analy-
including top-box scores in this study (and the Study 3 follow-
sis testing for the role of self-efficacy, perceived competence,
up and Study 4).
perceived chances of winning, and perceived competitiveness
as additional factors potentially influencing the identified pat-
7
We thank an anonymous reviewer for bringing up these possibilities. tern of effects. The results of this follow-up study corroborate
8
A back-of-the-envelope calculation shows that any remaining idea pool our mechanism. Taken together, our results provide clear evi-
depletion effect is very small. From our pretest, in the 50 ideas group at dence that perceived constrained expression is the key mechan-
least 4,354 ideas would still be available for discovery, while in the 5 ideas ism underlying the harmful influence of prior ideas on creative
group at least 4,399 would be available. The idea pool depletion effect would
thus be maximally 1.03% (45/4,354). As this is much less than the 36.72% performance. Importantly, note that Study 2 manipulates com-
reduction in number of ideas or 7.60% in maximum innovativeness in Study 2, petition (implying constraints) and Study 3 manipulates con-
idea pool depletion does not appear to be a first-order concern. straints directly (holding competition constant). Combined,
these studies suggest that prior ideas have a negative effect only Results
when both competition and constraints are present.
Results from an ANOVA show that the number of ideas sig-
nificantly influences creative performance (F(2, 997) ¼ 9.27, p
< .001). Seeing more ideas significantly reduces top-box inno-
Study 4: Creative Performance vativeness when participants saw 10 (vs. 3) prior ideas (M3 ¼
Monotonically Falls When Seeing More Ideas 3.63, SD ¼ .58; M10 ¼ 3.47, SD ¼ .68; F(1, 997) ¼ 10.57, p <
for Both Novices and Experts .01; d ¼ .26) as well as when participants saw 100 (vs. 3)
prior ideas (M100 ¼ 3.42, SD ¼ .74; F(1, 997) ¼ 16.36, p <
The fourth study uses three levels of prior ideas (3 vs. 10 vs. .001; d ¼ .32; results comparable for other measures, see
100) and tests whether domain expertise moderates the harmful Table 2).
influence of seeing more competitive prior ideas (H3). As we We turn to regression analysis for interactions with the
have discussed, the harmful influence of seeing more ideas is domain expertise measure, and estimate two separate interaction
likely to be more pronounced among those high in domain terms with the 10 vs. 3 ideas and 100 vs. 3 ideas comparisons (in
expertise, which we directly measure through an adapted scale. the same model). We find significant interactions across crea-
In addition to domain expertise, we also measured other tive performance measures when comparing 10 vs. 3 ideas for
exploratory moderators for individual differences, and we sum- domain expertise (e.g., bexpertise 10 (vs. 3) ¼ .11, p < .05). The
marize these measures and results in Web Appendix I. The negative interactions show that the prior ideas’ harmful influ-
results show that the negative effect is visible for both novices ence is more pronounced among individuals high in this mea-
and experts, and creative performance monotonically falls with sure, in line with H3. This result does not change after adding our
more competitive prior ideas. idea similarity control (functional fixedness) to the regression.
Interestingly, we do not find any significant interaction when
comparing 100 vs. 3 ideas. This shows that both experts and
novices are negatively influenced in a similar way (if the number
Method
of ideas is large enough), but experts are harmed more quickly
The study design again paralleled that of Study 3, with the with increasing numbers of prior ideas (for all coefficient esti-
main difference being the inclusion of the trait measures in mates, see Web Appendix I, Table 15). Mediation analysis again
random order at the beginning of the survey. We measure confirms the importance of perceived constraint of idea expres-
domain expertise drawing on Mishra, Umesh, and Stem sion as mediator (see Web Appendix I).
(1993; a ¼ .95; see Web Appendix I, Table 14). We again
used the unusual uses task for a brick to measure creative
performance, and manipulated the number of competitively Discussion
presented prior ideas (here, we used 3 vs. 10 vs. 100 out of
Replicating our previous findings, we again show that seeing
140) in a three-cell between-subjects design. As in Study 3
more competitive prior ideas harms creative performance in a
and its follow-up, we control for idea pool depletion and monotonic fashion. The negative effect, however, is shown to
comparability. For idea comparability, we used our search be moderated by the level of domain expertise of the partici-
algorithm to find comparable combinations of 3, 10, and pant, providing support for H3. Those higher in domain exper-
100 ideas. For this purpose, we first increased the size of our tise are more severely affected by seeing more competitive
prior ideas pool from 100 to 140 by adding 40 original ideas prior ideas. Mediation analysis suggests that experts feel sim-
from a pretest (original ideas, stated by <5%). We again ilar constraints compared with nonexperts, but they are more
removed the 19 common ideas (Web Appendix A, Table 2). strongly affected by the perceived constraint of idea expres-
Most participants (55%) correctly recalled the number of prior sion. Remarkably, at the same time we find that experts fixate
ideas they were confronted with in a check question, but the less on the externally provided prior ideas, and external fixation
percentages differed significantly for the 100 ideas group does not explain these results. This points to internal rather than
(59%; w2 ¼ 6.03, p < .05; results are robust to adding recall external sources of fixation, in line with the reasoning for H3.
as a control). We note that the negative effect magnifies for related types
After seeing the ideas, participants responded to the per- of individual difference (ideation expertise, creative self-
ceived constraint of idea expression (a ¼ .93) and cognitive efficacy, involvement, lead user–ness) but not for others (inno-
stimulation (a ¼ .92) scales followed by the unusual uses task. vativeness, emergent consumer, achievement orientation; for
A total of 1,000 U.S. respondents from MTurk participated in full analysis of these exploratory measures, see Web Appendix
the study (Mage ¼ 37.04 years, SD ¼ 11.51; 51.30% female). I). The former measurement approaches are related to domain
Altogether, respondents submitted 3,667 ideas. The mean expertise in that they are also high in how knowledge and
respondent submitted 3.67 ideas (SD ¼ 2.55, min ¼ 1, max expertise relate to ideation, providing convergent validity on
¼ 16) and the mean top-box innovativeness score equaled 3.51 H3. The latter approaches are more related to individual’s moti-
(SD ¼ .67, min ¼ 1, max ¼ 5). Ten raters each again evaluated vation instead of knowledge, suggesting discriminant validity.
ideas’ innovativeness, originality, and usefulness (as > .73). The specific reasons for these differences across individual
measures may be worth exploring in future research. Note that an individual’s categorization and the format of presentation
in this study, measuring the moderators early in the process (Fiske and Taylor 1991; Morales et al. 2005). This results in
may be a limitation, as the measurement may influence later fewer cognitive interferences as semantically similar concepts
creative performance. Expertise could also be measured in are processed and a train of thought can be maintained (Nijstad,
other ways (e.g., by doing an unrelated creative task). Our Stroebe, and Lodewijkx 2002). We propose that visually group-
current measure may also tap anticipated success in the task ing prior ideas with conceptually similar content into a categor-
rather than expertise in a substantive domain. Although our ized format (instead of separately) will have a direct impact on
theorizing and conclusions are specific to domain expertise, the creative performance of participants. Visual grouping of
similar results may be expected for the related constructs of competitive prior ideas can help creative performance because
anticipated success, self-efficacy, and ideation expertise. it reduces the signal of competition and facilitates the processing
Future research could attempt to explore this distinction. In of prior ideas, mitigating constraint of idea expression.
Study 5, we test a direct intervention for reducing the negative Unlike the field setting used in Study 1, we do not fix the
impact of competitive prior ideas by changing the way they are number of prior ideas, and we provide a predefined number of
presented in innovation contests. rewards instead of a percentage (i.e., the top five ideas win
instead of top 5%), mirroring typical prizes of such contests
(Hofstetter, Zhang, and Herrmann 2018). We created an actual
Study 5: Optimizing the UI on an Ideation
open innovation platform on a publicly available internet
Platform domain that mimicked the standard platform functionalities
It is difficult to fully change the competitive nature of open (i.e., viewing the contest brief, viewing prior ideas, and sub-
innovation contests (i.e., to mitigate the negative outcomes that mitting one’s own ideas). Individuals who accessed the plat-
the competitive presentation of prior ideas entails). Therefore, form could participate in a real open innovation contest with
Study 5 tests if the presentation of prior ideas can alleviate their actual monetary incentives. The platform’s UI was persona-
harmful influence within a competitive setting. We adjusted the lized based on user HTTP cookies.
UI of a mock-up open innovation platform, offering a realistic
ideation-contest setting in which individuals propose ideas to
actual ideation problems. We tested three interfaces—all ideas
Method
listed (listing), restricted number of ideas listed (restricted), and Each user was randomly assigned one of the three UIs. In the
grouped ideas (categorized)—across three different contests basic version of the UI (listing), the prior ideas were presented
initiated by the same institution. separately in an ordered list tile layout with four columns and
The open interface mimics a typical way prior ideas are nine rows. A maximum total number of 36 ideas could thus be
presented. Similar UIs are used by many open innovation con- displayed on one page. If there were more ideas, they were
test platforms including Atizo, 99designs, or Crowdspring. displayed on other pages that were accessible over a paging
Restricting the number of prior ideas follows from the finding functionality. The ideas were sorted by submission date in des-
that competitive prior ideas can constrain creative perfor- cending order (i.e., the most recent idea was shown first). When-
mances. A simple improvement for ideation platforms may ever a new idea was submitted, it was automatically displayed as
be to simply restrict the visibility of ideas to a smaller number the first idea and all previous ideas moved down by one position.
than the actual submitted ideas. In the restricted version of the UI (restricted), only a random
We also tested whether grouping ideas in an actionable way selection of four ideas out of all submitted ideas was displayed.
to adjust the UI would improve creative performances of parti- These four ideas were randomly sampled for each individual
cipants. We draw this approach from the human–computer inter- and then fixed such that they stayed the same for the user,
action literature, which highlights that the structural independent of any newly submitted ideas. Individuals could
presentation of information can have a powerful influence on not access more than these four ideas. In the grouped version of
performance (Shneiderman 2000). We explore whether the pre- the UI (grouped), an independent moderator combined ideas
sentation format utilized in an open innovation contest can miti- into taxonomic groups. The moderator supervised the contest
gate the harmful influence of competitive prior ideas, and live and grouped ideas in real time, leaving some ideas grouped
instead reduce felt constraints that hinder creative performance. and others not. The ideas that were not grouped were displayed
Specifically, we investigate the impact of presenting prior ideas on the left side of the screen in a 2 9 tile layout, and grouped
in groups instead of separately, as an alternate way to organize ideas in multiple colors indicating their group membership
information. Ideas in open innovation contests frequently share were displayed on the right side in a second 2 9 tile layout.
commonalities and elements of overlap. As such, they can often On each side, ideas and groups were sorted by submission date
be grouped into taxonomic categories (Kornish and Ulrich in descending order (Web Appendix J, Figure 7).
2011). Individuals process new stimuli based on their prior We randomly assigned participants to one of three similar
knowledge of the stimuli’s category (Sujan 1985) and will often ideation contests about university activities and services (e.g.,
naturally organize products into taxonomic or goal-based cate- ideas for activities during breaks, ideas for how to promote
gories (Barsalou 1985; Rosch et al. 1976). These stimuli can be university programs, ideas for food offerings) and they could
processed more (less) easily when there is a (no) match between only see and participate in their assigned contest (controlling
Break Ideas Promotion Ideas Food Ideas
150 150 200

Cumulative Number of Ideas
150
100 100
100
50 50
50
0 0 0
0 20 40 60 80 0 20 40 60 80 100 0 20 40 60 80 100
Nth Contest Entrant
Open Restricted Categorized
Figure 4. Restricted and categorized (vs. open) presentation of prior ideas produces more ideas in three different innovation contests (Study 5).
Notes: The figure shows the mean creative performances of individuals in the different experimental conditions across the three contests. It shows the cumulative
number of ideas over the course of the contest from the first to the last person entering the contest. At position x ¼ 20, for instance, the 20th person entered the
contest.
for self-selection). We also told participants that the five most As in Study 1, we calculated innovativeness as (originality þ
original ideas would win €20, and that their ideas competed usefulness)/2. We also focus on top-box innovativeness in this
against all other submitted ideas for these rewards. In all three study as our key variable of interest. Importantly, we included
UIs, participants were informed about the total number of prior project controls in all analyses to control for contest and eva-
ideas submitted (up to 557 ideas in the largest contest, and 386 luation differences across competitions.
and 376 in the smaller contests, respectively), fixing objective
winning chances. They could submit as many ideas as they Results and Discussion
wanted to. Idea titles could be up to 60 characters long and
content up to 180 characters long. We pool the data across the three contests for the analysis (N1
We invited a total of 2,744 individuals from an online panel ¼ 244, N2 ¼ 321, N3 ¼ 263). First, we find that the three
(Clickworker.de) and 828 (30.17%) chose to participate, inde- experimental groups differ significantly (F(2, 825) ¼ 10.21, p
pendent of the experimental treatment (w2 ¼ 2.20, p ¼ .33). < .001). Top-box innovativeness is significantly higher in the
These participants submitted 1,319 ideas in total. The average restricted (vs. open) condition (Mrestricted ¼ 4.49, SD ¼ .51;
entrant submitted 1.59 ideas (SD ¼ 1.35, min ¼ 1, max ¼ 13). Mopen ¼ 4.36, SD ¼ .78; F(1, 825) ¼ 4.75, p < .05; d ¼ .19)
and in the categorized (vs. open) condition (Mcategorized ¼ 4.62,
Each contest lasted three days. After the contest ended, all
SD ¼ .66; F(1, 825) ¼ 20.42, p < .001; d ¼ .35). Table 2
ideas’ originality and usefulness were evaluated by two mar-
shows ANOVA results for all of our five key measures of
keting experts; as raters, they were blind to the UI version and
creative outcomes. Figure 4 shows the cumulative number of
the author of the idea on a seven-point Likert scale (1 ¼ “Not
ideas generated in the three groups over the course of each
original/useful at all,” and 7 ¼ “Extremely original/useful”).
contest. We can observe that the cumulative number of ideas
The two raters showed significant agreement in all contests.9
increases more steeply, resulting in higher total numbers for the
restricted and grouped (vs. listing) UIs and providing visual
9 support for the aforementioned results. These results indicate
Originality: kbreak ideas ¼ .12, z ¼ 5.87, p < .01; kpromotion ideas ¼ .13, z ¼
5.46, p < .01; kfood ideas ¼ .12, z ¼ 6.46, p < .01; usefulness: kbreak ideas ¼ .14, z the potential for optimizing UIs in open innovation contests, as
¼ 5.87, p < .01; kpromotion ideas ¼ .05, z ¼ 2.32, p < .05; kfood ideas ¼ .07, z ¼ both the restricted and grouped interfaces significantly
2.83, p < .01. increased creative performance of participants. We report
further evidence in favor of grouping in Web Appendices K, L, Implications

and M (i.e., three follow-up studies), highlighting the relevance
Our results have implications for research related to open inno-
of this possibility for future research. Another fruitful direction
vation, creativity, ideation, contest theory, and design. Fore-
could be to look at combinations of grouped and restricted UIs.
most, the importance of competitive framing in open
innovation is established in our conceptualization. Indeed,
competitive framing is found to be a powerful force in directing
General Discussion how prior ideas are processed cognitively. Second, we define
In this article, we show that in open innovation contests (i.e., the constraint of idea expression on the part of the participant as
competitions), exposing participants to too many ideas reduces an important mechanism in driving our effects, contributing to
the creativity of the ideas generated. Importantly, competitive the discussion of how constraints influence creativity (Moreau
framing of the prior ideas is shown to trigger this negative and Dahl 2005). Other mechanisms may also be in play, as we
effect. Competing against prior ideas introduces the importance found a significant indirect effect via perceived winning
of distinguishing one’s ideas from prior ones, thereby increas- chances. Future research is warranted in order to investigate
ing felt constraints of idea expression in an ideation task fea- their role in how competitive prior ideas influence creative
turing numerous prior ideas. This finding extends the performances. The finding that the prior ideas’ harmful influ-
observation that individuals decrease their efforts when win- ence is robust across different types of innovators and aggra-
ning chances are lower (Boudreau, Lacetera, and Lakhani vated by domain expertise adds to the discussion of how
2011), as we find that even after fixing objective winning expertise can impede creativity (Dane 2010).
chances (Studies 2–4) and controlling for perceived winning Theoretically, our results also broaden the discussion related
chances (in Study 3 follow-up), seeing more prior ideas dis- to the optimal design of innovation contests (e.g., Terwiesch
courages creative performance in innovation contests. Interest- and Xu 2008; Wooten and Ulrich 2015). The extant literature
ingly, this effect is reduced when similar ideas are grouped has emphasized extrinsically motivating features of contest
together (effectively reducing the number of ideas). design, such as the reward size and structure, or the number
Our findings also show that without a competitive frame, of entrants as key drivers of contest outcomes (e.g., Cason,
external ideas are less constraining and can actually increase Masters, and Sheremeta 2010; Terwiesch and Xu 2008). How-
creative performance. This result is in line with prior research ever, our results highlight that contest participants also react to
showing positive effects when seeing more uncompetitive other subtler signals that inform the processing of ideas. Visual
ideas (Gallupe et al. 1992; Paulus et al. 2013). Here, we did grouping represents one such signal, and our investigation of its
not find a positive effect on top-box innovation or usefulness influence on contest participant’s ideation outcomes provides
(similar to Paulus et al. 2013), but benefits did identify for an impetus for more research in this vein.
originality and the number of solutions identified. We conjec- Our findings provide a substantive contribution to those
ture that the uncompetitive framing of prior ideas—and the organizations that are considering how best to organize idea-
stimulation it provides—better facilitates novel thinking but tion. Given the negative effects of seeing competitively posi-
not necessarily an effective identification of solutions. tioned prior ideas, it seems reasonable to conclude that standard
Across studies, the identified effects are consistent yet small brainstorming sessions and innovation contests may undermine
in magnitude on top-box measures (but larger in the number of an individual’s creativity if people feel that the sessions are
generated ideas). Given that managers of innovation contests competitive and see many other ideas before they themselves
typically care most about the quality of top ideas, even small generate ideas. The implication for such intraorganization idea
changes in quality can have a lasting harmful or helpful impact generation is to engage in brainwriting exercises which let
on the organization (Fleming 2007; Girotra, Terwiesch, and people think on their own before engaging in teams, exhausting
Ulrich 2010). We believe this finding is actionable for future to the extent possible the individual’s own ideas well before
research to both build on and refine—for example, how to considering the ideas of others. When considering others’ com-
effectively leverage prior ideas using an uncompetitive framing petitive ideas electronically, we showed that negative effects
approach is a question with great potential for creative ideation. can be addressed simply through interface design. Prior ideas
Importantly, all our studies manipulate the number of prior should be displayed with strategic intent (instead of simply
ideas exposed to participants, which raises the following ques- presenting all information last-in first-out) to improve results.
tion: Would it be best to not show any prior ideas and, instead, Managing the interaction with participants in this way incurs
host only blind contests? The small number of prior ideas we little cost and thus is an actionable strategy in improving the
display in Study 4 (three ideas) and the Study 2 follow-up (one integrity and effectiveness of innovation efforts. Future
idea; Web Appendix F) suggest that showing any larger num- research should explore other possibilities here—that is, are
ber of competitive prior ideas will harm creativity. However, there other ways that the communication and visual presenta-
future research could further explore the value of not showing tion of innovation contests can shape the motivation and beha-
any competing prior ideas at all. vior of participants?
Limitations and Future Research We also do not explore the various specific motives of par-
ticipants entering contests or how goal proximity may influ-
This research opens a number of paths for future exploration.
ence results. Indeed, future research could take a goal theory
First, future research could work to provide a deeper under-
perspective on contest-based ideation. We also did not expli-
standing of how external constraints in open innovation are
citly investigate self-selection of individuals into contests
impacting creative performance in the conceptual model we
depending on the number of prior ideas and their nature (there
have defined. For instance, is there a distinction in felt con-
was no significant self-selection in our studies). Self-selection
straints that represents the belief that one can perform the
may be an important mechanism to consider given that indi-
behavior versus felt constraints in the belief that one’s abilities
viduals can typically choose from a wide range of different
will produce a certain outcome (i.e., outcome expectation)? contests on open innovation platforms. People may also be
These two representations have been found to be empirically discouraged from participating at all in an innovation contest
related (Eccles and Wigfield 1995), but they may produce dif- when seeing large numbers of prior ideas, potentially harming
ferent creative outcomes. Second, the role of competition in creative outcomes of contests even more. How prior ideas
creativity tasks can be further explored following the call of inform entry decisions is certainly an essential topic for future
Amabile (2018). Competition may introduce good or bad research.
stress, affecting the creative performance of competitors in Finally, the nature of the innovation problem (and how it
different ways. Seeing more competing ideas can induce stress, may moderate the identified effects) poses another productive
which may acerbate the already harmful influence of said ideas, direction for future research. In our analyses, we focused on
as stress negatively affects memory, attention, and cognition low complexity ideation problems with many possible solu-
(Ellis 2006; LeBlanc 2009). Third, a limitation to our analyses tions. Negative effects of prior ideas may magnify for innova-
is the limited number of prior ideas we showed in this study. It tion problems for which fewer possible solutions exist (i.e., for
could be that cognitive stimulation turns into cognitive over- problems with a reduced solution space). For more complex
load beyond a certain point, resulting in a negative influence of problems, positive stimulation effects of prior ideas may
prior ideas even when presented uncompetitively (see, e.g., become more relevant. These questions point to the nascent
optimal levels of stimulation, Steenkamp and Baumgartner state of understanding that both academia and management has
1992; Yerkes and Doson 1908). Such an effect would further with respect to the democratization of innovation that is being
reinforce the negative influence of prior ideas, possibly even seen today. Our research efforts seek to add to the growing
when the ideas are uncompetitive. We did not test this possi- body of knowledge here and validate the importance and value
bility, which could be a viable direction for future research. of this trend in innovation.
Fourth, an opportunity exists to better define what types of
ideas and idea groupings might best facilitate individuals’ cog-
nitive efforts and positively stimulate participants in contests. Acknowledgments
For example, would bizarre ideas that are impossible to actu-
The authors thank Reto Aebersold, Markus Christen, Melanie Clegg,
ally be realized be useful in provoking individuals to produce Emanuel de Bellis, Andreas Gerber, Marek Gorny, Christian Hirsig,
something creative? Or, would it be better to show individuals Joey Hoegg, Thomas Ott, and seminar participants at the Wharton
more conservative ideas and let them build from a more tradi- School of University of Pennsylvania, University of Miami, Univer-
tional knowledge base? Note that artificial intelligence could sity of Louisville, University of St. Gallen, University of Lucerne,
be leveraged here to automate the grouping of ideas or the ideal Hebrew University, and Università della Svizzera italiana for helpful
stimulation and feedback to ideators. Fifth, other individual discussions. They thank the reviewers and participants from EMAC
factors driving prior ideas’ harmful influence should be further and ACR for their valuable comments.
explored. We did not find significant mediation for perceived
competence, self-efficacy, competitiveness, and winning
chance, although these measures were significantly influenced Associate Editor
by the number of prior ideas. Other individual factors such as Donald Lehmann
demographics could also be further explored (e.g., Baer and
Kaufman 2008). A better understanding of these factors and
their relation to individuals’ performance in contest-based Declaration of Conflicting Interests
ideation offers a potentially fruitful area of future research. The author(s) declared no potential conflicts of interest with respect to
Although we found similar results for self-reported experts and the research, authorship, and/or publication of this article.
novices, these negative competition effects may not extend to
highly professional creatives who regularly produce creativity
for pay and are thus less prone to negative emotional stimula- Funding
tion effects. There are many examples of brilliant creative con- The author(s) disclosed receipt of the following financial support for
tent within highly competitive domains such as advertising, the research, authorship, and/or publication of this article: This
product design, architectural design, news story illustrations, research was partly funded by Innosuisse – Swiss Innovation Agency
books, and even the writing of scientific papers. (grant 12747.1 PFES-ES).
References Deci, Edward L. and Richard M. Ryan (1985), Intrinsic Motivation

99designs (2018), “Concept Originality Policy” (accessed December and Self-Determination in Human Behavior. New York: Plenum.
19, 2018), https://support.99designs.com/hc/en-us/articles/ Dennis, Alan R. and Joseph S. Valacich (1993), “Computer Brain-
204109569. storms: More Heads Are Better Than One,” Journal of Applied
Amabile, Teresa M. (2018), Creativity in Context: Update to the Psychology, 78 (4), 531–37.
Social Psychology of Creativity. New York: Routledge. DeRosa, Darleen M., Carter L. Smith, and Donald A. Hantula (2007),
Argo, Jennifer and Ke (Christy) Tu (2013), “The Impact of Compar- “The Medium Matters: Mining the Long-Promised Merit of Group
Interaction in Creative Idea Generation Tasks in a Meta-Analysis
isons with Others on Creativity Outcomes,” in Advances in Con-
of the Electronic Group Brainstorming Literature,” Computers in
sumer Research, Vol. 41, Simona Botti and Aparna Labroo, eds.
Human Behavior, 23 (3), 1549–81.
Duluth, MN: Association for Consumer Research.
Diehl, Michael and Wolfgang Stroebe (1987), “Productivity Loss in
Baer, J. and J.C. Kaufman (2008), “Gender Differences in Creativity,”
Brainstorming Groups: Toward the Solution of a Riddle,” Journal
Journal of Creative Behavior, 42 (2), 75–105.
of Personality and Social Psychology, 53 (3), 497–509.
Barsalou, Lawrence W. (1985), “Ideals, Central Tendency, and Fre-
Diehl, Michael and Wolfgang Stroebe (1991), “Productivity Loss in
quency of Instantiation as Determinants of Graded Structure in
Idea-Generating Groups: Tracking Down the Blocking Effect,”
Categories,” Journal of Experimental Psychology: Learning,
Journal of Personality and Social Psychology, 61 (3), 392–403.
Memory, and Cognition, 11 (4), 629–54.
Dugosh, Karen Leggett, Paul B. Paulus, Evelyn J. Roland, and Huei-
Baumeister, Roy F. (1984), “Choking Under Pressure: Self-
Chuan Yang (2000), “Cognitive Stimulation in Brainstorming,”
Consciousness and Paradoxical Effects of Incentives on Skillful
Performance,” Journal of Personality and Social Psychology, 46
Duncker, K. (1945), “On Problem-Solving (L. S. Lees, Trans.),” Psy-
(3), 610–20.
chological Monographs, 58 (5), i113. https://doi.org/10.1037/
Bayus, B.L. (2013), “Crowdsourcing new Product Ideas over time: An
h0093599.
Analysis of the Dell IdeaStorm Community,” Management Sci-
Eccles, Jacquelynne S. and Allan Wigfield (1995), “In the Mind of the
ence, 59 (1), 226–44.
Achiever: The Structure of Adolescents’ Academic Achievement
Boudreau, Kevin J., Nicola Lacetera, and Karim R. Lakhani (2011),
Related-Beliefs and Self-Perceptions,” Personality and Social Psy-
“Incentives and Problem Uncertainty in Innovation Contests: An
chology Bulletin, 21 (3), 215–25.
Empirical Analysis,” Management Science, 57 (5), 843–63. Ellis, Aleksander P.J. (2006), “System Breakdown: The Role of Men-
Camacho, L. Mabel and Paul B. Paulus (1995), “The Role of Social tal Models and Transactive Memory in the Relationship Between
Anxiousness in Group Brainstorming,” Journal of Personality and Acute Stress and Team Performance,” Academy of Management
Social Psychology, 68 (6), 1071–80. Journal, 49 (3), 576–89.
Cason, Timothy N., William A. Masters, and Roman M. Sheremeta, Eyeka (2017), “The State of Crowdsourcing in 2017,” research report
(2010), “Entry into Winner-Take-All and Proportional-Prize Con- (accessed October 22, 2020), https://en.eyeka.com/resources/
tests: An Experimental Study,” Journal of Public Economics, 94 reports.
(9), 604–11. Fiske, Susan T. and Shelley E. Taylor (1991), Social Cognition: From
Chesbrough, Henry (2003), “The Logic of Open Innovation: Manag- Brains to Culture, 2nd ed. New York: McGraw-Hill, 16–15.
ing Intellectual Property,” California Management Review, 45 (3), Fleming, Lee (2007), “Breakthroughs and the ‘Long Tail’ of
33–58. Innovation,” MIT Sloan Management Review, 49 (1), 69–74.
Collaros, Panayiota A. and Lynn R. Anderson, (1969), “Effect of Frensch, Peter A. and Robert J. Sternberg (1989), “Expertise and
Perceived Expertness upon Creativity of Members of Brainstorm- Intelligent Thinking: When Is It Worse to Know Better?” Advances
ing Groups,” Journal of Applied Psychology, 53 (2, pt. 1), 159–63. in the Psychology of Human Intelligence, 5, 157–88.
Coursey, Lauren E., Ryan T. Gertner, Belinda C. Williams, Jared B. Furman, Jeffrey L. and Scott Stern (2006), Climbing Atop the
Kenworthy, Paul B. Paulus, and Simona Doboli (2019), “Linking Shoulders of Giants: The Impact of Institutions on Cumulative
the Divergent and Convergent Processes of Collaborative Creativ- Research (No. w12523). Cambridge, MA: National Bureau of Eco-
ity: The Impact of Expertise Levels and Elaboration Processes,” nomic Research.
Frontiers in Psychology 10 699. Gagné, Marylène (2003), “The Role of Autonomy Support and Auton-
Dahl, Darren W., Amitava Chattopadhyay, and Gerald J. Gorn (1999), omy Orientation in Prosocial Behavior Engagement,” Motivation
“The Use of Visual Mental Imagery in New Product Design,” and Emotion, 27 (3), 199–223.
Journal of Marketing Research, 36 (1) 18–28. Gallupe, R. Brent and William H. Cooper (1993), “Brainstorming
Dahl, Darren W. and Page Moreau (2002), “The Influence and Value Electronically,” MIT Sloan Management Review, 35 (1), 27.
of Analogical Thinking During New Product Ideation,” Journal of Gallupe, R. Brent, William H. Cooper, Mary-Liz Grisé, and Lana M.
Marketing Research, 39 (1), 47–60. Bastianutti (1994), “Blocking Electronic Brainstorms,” Journal of
Dahlander, Linus and David M. Gann (2010), “How Open Is Innova- Applied Psychology, 79 (1), 77–86.
tion?” Research Policy, 39 (6), 699–709. Gallupe, R. Brent, Alan R. Dennis, William H. Cooper, Joseph S.
Dane, Erik (2010), “Reconsidering the Trade-Off Between Expertise Valacich, Lana M. Bastianutti, and Jay F. Nunamaker Jr.
and Flexibility: A Cognitive Entrenchment Perspective,” Academy (1992), Electronic Brainstorming and Group Size. Academy of
of Management Review, 35 (4), 579–603. Management Journal, 35 (2), 350–69.
Girotra, Karan, Christian Terwiesch, and Karl T. Ulrich (2010), “Idea Luo, Lan and Olivier Toubia (2015), “Improving Online Idea Gener-
Generation and the Quality of the Best Idea,” Management Sci- ation Platforms and Customizing the Task Structure on the Basis of
ence, 56 (4), 591–605. Consumers’ Domain-Specific Knowledge,” Journal of Marketing,
Goldenberg, Jacob, Donald R. Lehmann, and David Mazursky (2001), 79 (5), 100–114.
“The Idea Itself and the Circumstances of Its Emergence as Pre- Marchant, Garry, John Robinson, Urton Anderson, and Michael
dictors of New Product Success,” Management Science, 47 (1), Schadewald (1991), “Analogical Transfer and Expertise in Legal
69–84. Reasoning,” Organizational Behavior and Human Decision Pro-
Guilford, Joy Paul (1967), The Nature of Human Intelligence. New cesses, 48 (2), 272–90.
York: McGraw-Hill. Mishra, Sanjay, U.N. Umesh, and Donald E. Stem Jr. (1993),
Guilford, Joy Paul, Philip R. Merrifield, and Robert C. Wilson (1958), “Antecedents of the Attraction Effect: An Information-Processing
Unusual Uses Test. Orange, CA: Sheridan Psychological Services. Approach,” Journal of Marketing Research, 30 (3), 331–49.
Harkins, Stephen G. (1987), “Social Loafing and Social Facilitation,” Moldovan, Sarit, Jacob Goldenberg, and Amitava Chattopadhyay
Journal of Experimental Social Psychology, 23 (1), 1–18. (2011), “The Different Roles of Product Originality and Useful-
Harvey, Sarah and Chia-Yu Kou (2013), “Collective Engagement in ness in Generating Word-of-Mouth,” International Journal of
Creative Tasks: The Role of Evaluation in the Creative Process in Research in Marketing, 28 (2), 109–19.
Groups,” Administrative Science Quarterly, 58 (3), 346–86. Morales, Andrea, Barbara E. Kahn, Leigh McAlister, and Susan M.
Hayes, Andrew F. (2015), “An Index and Test of Linear Moderated Broniarczyk (2005), “Perceptions of Assortment Variety: The
Mediation,” Multivariate Behavioral Research, 50 (1), 1–22. Effects of Congruency Between Consumers’ Internal and Retai-
Hofstetter, Reto, Harikesh Nair, and Sanjog Misra (2020), “Can Open lers’ External Organization,” Journal of Retailing, 81 (2), 159–69.
Innovation Survive? Imitation and Return on Originality in Crowd- Moreau, C. Page and Darren W. Dahl (2005), “Designing the Solution:
sourcing Creative Work,” Stanford University Graduate School of The impact of Constraints on Consumers’ Creativity,” Journal of
Business Research Paper (January), https://ssrn.com/abstract¼3 Consumer Research, 32 (1), 13–22.
Mullen, Brian, Craig Johnson, and Eduardo Salas (1991),
133158.
“Productivity Loss in Brainstorming Groups: A Meta-Analytic
Hofstetter, Reto, John Z. Zhang, and Andreas Herrmann (2018),
Integration,” Basic and Applied Social Psychology, 12 (1), 3–23.
“Successive Open Innovation Contests and Incentives: Winner-
Mumford, Michael D., Cassie Blair, Leslie Dailey, Lyle E. Leritz, and
Take-All or Multiple Prizes?” Journal of Product Innovation Man-
Holly K. Osburn (2006), “Errors in Creative Thought? Cognitive
agement, 35 (4), 492–517.
Biases in a Complex Processing Activity,” Journal of Creative
Jannson, David G. and Steven M. Smith (1991), “Design Fixation,”
Behavior, 40 (2), 75–109.
Design Studies, 12, 3–11.
Murray, Fiona, and Siobhán O’Mahony (2007), “Exploring the Foun-
Karau, Steven J. and Kipling D. Williams (1993), “Social Loafing: A
dations of Cumulative Innovation: Implications for Organization
Meta-Analytic Review and Theoretical Integration,” Journal of
Science,” Organization Science, 18 (6), 1006–21.
Personality and Social Psychology, 65 (4), 681–706.
Murray, Kyle B. and Gerald Häubl (2007), “Explaining Cognitive
Kerr, Norbert L. and Steven E. Bruun (1983), “Dispensability of
Lock-In: The Role of Skill-Based Habits of Use in Consumer
Member Effort and Group Motivation Losses: Free-Rider Effects,”
Choice,” Journal of Consumer Research, 34 (1), 77–88.
Nijstad, Bernard A. and Wolfgang Stroebe (2006), “How the Group
Kohn, Nicholas W. and S.M. Smith (2011), “Collaborative Fixation:
Affects the Mind: A Cognitive Model of Idea Generation in Groups,”
Effects of Others’ Ideas on Brainstorming,” Applied Cognitive Personality & Social Psychology Review, 10 (3), 186–213.
Psychology, 25 (3), 359–71. Nijstad, Bernard A., Wolfgang Stroebe, and Hein F.M. Lodewijkx
Kornish, Laura J. and Karl T. Ulrich (2011), “Opportunity Spaces in (1999), “Persistence of Brainstorming Groups: How Do People
Innovation: Empirical Analysis of Large Samples of Ideas,” Man- Know When to Stop?” Journal of Experimental Social Psychology,
agement Science, 57 (1), 107–28. 35 (2), 165–85.
Lazear, Edward P. and Sherwin Rosen (1981), “Rank-Order Tourna- Nijstad, Bernard A., Wolfgang Stroebe, and Hein FM Lodewijkx
ments as Optimum Labor Contracts,” Journal of Political Econ- (2002), “Cognitive Stimulation and Interference in Groups: Expo-
omy, 89 (5), 841–64. sure Effects in an Idea Generation Task,” Journal of Experimental
LeBlanc, Vicki R. (2009), “The Effects of Acute Stress on Perfor- Social Psychology, 38 (6), 535–44.
mance: Implications for Health Professions Education,” Academic Osborn, Alex F. (1953). Applied Imagination. Oxford, UK: Scribner’s.
Medicine, 84 (10), 25–33. Otten, Mark (2009), “Choking vs. Clutch Performance: A Study of
Lewandowsky, Stephan, Daniel Little, and Michael L. Kalish (2007), Sport Performance Under Pressure,” Journal of Sport and Exercise
“Knowledge and Expertise,” in Handbook of Applied Cognition, Psychology, 31 (5), 583–601.
2nd ed., F. T. Durso, ed. Chichester, UK: John Wiley & Sons, Paulus, Paul B. and Mary T. Dzindolet (1993), “Social Influence
83–109. Processes in Group Brainstorming,” Journal of Personality and
Lilien, Gary L., Pamela D. Morrison, Kathleen Searls, Mary Sonnack, Social Psychology, 64 (4), 575.
and Eric von Hippel (2002), “Performance Assessment of the Lead Paulus, Paul B., Nicholas W. Kohn, Lauren E. Arditti, and Runa M.
User Idea-Generation Process for New Product Development,” Korde (2013), “Understanding the Group Size Effect in Electronic
Management Science, 48 (8), 1042–59. Brainstorming,” Small Group Research, 44 (3), 332–52.
Paulus, Paul B. and Huei-Chuan Yang (2000), “Idea Generation in Steenkamp, Jan-Benedict E.M, and Hans Baumgartner (1992),
Groups: A Basis for Creativity in Organizations,” “The Role of Optimum Stimulation Level in Exploratory Consumer
Organizational Behavior and Human Decision Processes, 82 Behavior,” Journal of Consumer Research, 19 (3), 434–48.
(1), 76–87. Stephen, Andrew T., Peter Pal Zubcsek, and Jacob Goldenberg (2016),
Preacher, Kristopher J. and Andrew F. Hayes (2008), “Asymptotic and “Lower Connectivity Is Better: The Effects of Network Structure
Resampling Strategies for Assessing and Comparing Indirect on Redundancy of Ideas and Customer Innovativeness in Interde-
Effects in Multiple Mediator Models,” Behavior Research Meth- pendent Ideation Tasks,” Journal of Marketing Research, 53 (2),
ods, 40 (3), 879–91. 263–79.
Preacher, Kristopher J., Derek D. Rucker, and Andrew F. Hayes Sujan, Mita (1985), “Consumer Knowledge: Effects on Evaluation
(2007), “Addressing Moderated Mediation Hypotheses: Theory, Strategies Mediating Consumer Judgments,” Journal of Consumer
Methods, and Prescriptions,” Multivariate Behavioral Research, Research, 12 (1), 31–46.
42 (1), 185–227. Taylor, Curtis R. (1995), “Digging for Golden Carrots: An Analysis of
Rosch, Eleanor, Carolyn B. Mervis, Wayne D. Gray, David M. Research Tournaments,” American Economic Review, 85 (4),
Johnson, and Penny Boyes-Braem (1976), “Basic 872–90.
Objects in Natural Categories,” Cognitive Psychology, 8 (3), Terwiesch, Christian and Karl T. Ulrich (2009), Innovation Tourna-
382–439. ments: Creating and Selecting Exceptional Opportunities. Boston:
Ryan, Richard M. and Edward L. Deci (2000), “Intrinsic and Extrinsic Harvard Business School Press.
Motivations: Classic Definitions and New Directions,” Contempo- Terwiesch, Christian and Yi Xu (2008), “Innovation Contests, Open
Innovation, and Multiagent Problem Solving,” Management Sci-
rary Educational Psychology, 25 (1), 54–67.
ence, 54 (9), 1529–43.
Scotchmer, Suzanne (1991), “Standing on the Shoulders of Giants:
Vroom, Victor H. (1964), Work and Motivation. New York: John
Cumulative Research and the Patent Law,” Journal of Economic
Wiley & Sons.
Perspectives, 5 (1), 29–41.
Ward, Thomas B. (1994), “Structured Imagination: The Role of Cate-
Shneiderman, Ben (2000), “Creating Creativity: User Interfaces for
gory Structure in Exemplar Generation,” Cognitive Psychology, 27
Supporting Innovation,” ACM Transactions on Computer-Human
(1), 1–40.
Interaction (TOCHI), 7 (1), 114–38.
Wiley, Jennifer (1998), “Expertise as Mental set: The Effects of
Smith, Steven M. and Steven E. Blankenship (1991), “Incubation and
Domain Knowledge in Creative Problem Solving,” Memory &
the Persistence of Fixation in Problem Solving,” American Journal
Cognition, 26 (4), 716–30.
of Psychology, 104 (1), 61–87.
Wooten, Joel O. and Karl T. Ulrich (2015), “The Impact of Visibility
Smith, Steven M., Thomas B. Ward, and Jay S. Schumacher (1993),
in Innovation Tournaments: Evidence from Field Experiments,”
“Constraining Effects of Examples in a Creative Generation task,”
SSRN (February 19), https://papers.ssrn.com/sol3/papers.cfm?
Memory & Cognition, 21, 837–45.
abstract_id¼2214952.
Spencer, Steven J., Mark P. Zanna, and Geoffrey T. Fong (2005),
Yerkes, Robert M. and John D. Dodson (1908), “The Relation
“Establishing a Causal Chain: Why Experiments Are Often More of Strength of Stimulus to Rapidity of Habit-Formation,”
Effective Than Mediational Analyses in Examining Psychological Journal of Comparative Neurology and Psychology, 18 (5),
Processes,”Journal of Personality and Social Psychology, 89 (6), 459–82.
845–51. Youmans, Robert J. and Thomaz Arciszewski (2014), “Design Fixa-
Stacey, Martin K., Claudia M. Eckert, and Jennifer Wiley (2002), tion: Classifications and Modern Methods of Prevention,” AI
“Expertise and Creativity in Knitwear Design,” International Jour- EDAM, 28 (2), 129–37.
nal of New Product Development and Innovation Management, Zajonc, Robert B. (1965), “Social Facilitation,” Science, 149 (3681),
4 (1), 49–64. 269–74.

Hofstetter Et Al JMR Constraining Ideas 2021

Uploaded by

Copyright:

Available Formats

Hofstetter Et Al JMR Constraining Ideas 2021

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hofstetter Et Al JMR Constraining Ideas 2021

Uploaded by

Copyright:

Available Formats

Article

Journal of Marketing Research

Others Harms Creativity in Open Innovation Article reuse guidelines:

Reto Hofstetter, Darren W. Dahl, Suleiman Aryobsei,

Literature Direction Context Effect

Moderators within competitive presentation:

Table 2. Means Across Experimental Conditions over All Studies.

Experimental Top-Box Top-Box Top-Box No. of Total Guilford’s

1 2 prior ideas 263 3.533 2.627 4.549 1.662 5.423 —

2 5 prior ideas, competitive 98 3.054 3.382 3.603 4.592 12.719 24.89

Moderator (manipulated): Competitive (vs. uncompetitive) presentation

3 5 prior ideas, high CE 254 2.672 2.855 3.360 4.339 9.277 —

5 prior ideas, low CE 253 2.675 2.893 3.321 4.158 9.133 —

Moderator (manipulated): High versus low CE within competitive presentation

Study 3 5 prior ideas 260 2.789 2.949 2.745 4.069 9.533 —

4 3 prior ideas 347 3.631 3.609 3.793 2.870 13.036 —

5 Open 278 4.363 3.260 4.766 1.309 5.625 —

moderated parallel mediation model with 10,000 bootstrapped

H1b: M1(Competitive) = −.07a

Figure 3. Results of mediation analysis (Study 2).

Results and Discussion

Break Ideas Promotion Ideas Food Ideas

150 150 200

Nth Contest Entrant

Open Restricted Categorized

further evidence in favor of grouping in Web Appendices K, L, Implications

References Deci, Edward L. and Richard M. Ryan (1985), Intrinsic Motivation

You might also like