ArticlePDF Available

Superman or the Fantastic Four? Knowledge Combination and Experience in Innovative Teams

September 2008
Academy of Management Journal 49(4)

September 2008
49(4)

DOI:10.5465/AMJ.2006.22083029

Authors:

INSEAD

We examine how knowledge and experience affect both the mean and variance values of innovations from individuals and teams. We apply and extend theory on innovativeness and creativity to propose that holding multiple knowledge domains produces novel combinations that increase the variance of product performance; and that extensive experience produces outputs with high average performance. We analyzed innovations in the comic book industry, finding that innovations with extreme success and failure are affected by similar factors as high-performing innovations. Multi-member teams and teams with experience working together produced innovations with greater variation in value, but individuals were able to combine knowledge diversity more effectively than teams.

Results of Linear Regression Analysis of Level and Deviation of Performance a

…

Figures - uploaded by Henrich Greve

Content may be subject to copyright.

Content uploaded by Henrich Greve

Content may be subject to copyright.

Electronic copy available at: http://ssrn.com/abstract=948597

Tuck School of

Business at Dartmouth

Tuck School of Business Working Paper No. 2009-56

SUPERMAN OR THE FANTASTIC FOUR?

KNOWLEDGE COMBINATION AND

EXPERIENCE IN INNOVATIVE TEAMS

Alva Taylor

Dartmouth College – Tuck School of Business

Henrich Greve

Norwegian School of Management (BI)

This paper can be downloaded from the

Social Science Research Network Electronic Paper Collection:

http://ssrn.com/abstract=948597

Electronic copy available at: http://ssrn.com/abstract=948597Electronic copy available at: http://ssrn.com/abstract=948597

SUPERMAN OR THE FANTASTIC FOUR? KNOWLEDGE

COMBINATION AND EXPERIENCE IN INNOVATIVE TEAMS

ALVA TAYLOR

Dartmouth College

HENRICH R. GREVE

Norwegian School of Management BI

This study focuses on effects of knowledge and experience on both mean and variance

measures of individual and team innovations. We propose that multiple knowledge

domains produce novel combinations that increase the variance of product perfor-

mance and that extensive experience produces outputs with high average perfor-

mance. We analyzed innovations in the comic book industry, finding that innovations

with extreme success and failure were affected by factors similar to those affecting

high-performing innovations. Multimember teams and teams with experience working

together produced innovations with greater variation in value, but individuals were

able to combine knowledge diversity more effectively than teams.

If an organization were structuring a team for

developing an innovation, should the knowledge

and experience of this team be different from those

of a team for developing a high-quality product in

line with the company’s past products? Would the

answer differ if only a single person could be cho-

sen? The organizational innovation literature cur-

rently answers yes to the first question, as it makes

a clear distinction between activities that increase

variance in performance and activities that increase

mean performance. For example, new products that

are incremental extensions of a current product

will have a positive financial return (high mean

performance), but they will not offer significant

inordinate profits or losses. Product offerings that

are dramatically different from past products offer

the potential for extreme profits or extreme losses

(high variance in performance), but overall will

have a lower expected mean owing to the high

likelihood of failure. Another argument in the lit-

erature is that organizational exploration, which

introduces experiments of uncertain value into an

organization’s activities, is different from exploita-

tion, which maintains and refines current activi-

ties, and that the antecedents of exploration are less

well known (March, 1991). Yet both exploitation

and exploration can be seen as practices of combin-

ing knowledge, one using existing knowledge in

new ways and the other using existing knowledge

in well-understood ways (Schumpeter, 1934).

The tension between exploration and exploita-

tion is founded on the view that experimentation

with new alternatives slows improvements in ex-

isting ones. Conversely, improvements in existing

activities make experimenting with new ones less

attractive (Levitt & March, 1988). This argument is

based on the differing uses of knowledge. Its as-

sumption is that innovations arise from two

sources: (1) the knowledge available for an innova-

tive activity (e.g., Ahuja, 2000; Powell, Koput, &

Smith-Doerr, 1996) and (2) the ability of individu-

als and teams to apply the available knowledge

(Brown & Duguid, 1991; Tripsas, 1997; Von Hippel,

1988). Simply put, the more diverse the informa-

tion and knowledge that are applied, the more

novel is the output; and the deeper the use of ex-

isting knowledge, the less novel but more predict-

ably performing the output. Dysfunctions can occur

in both cases: too-diverse knowledge can result in

unwieldy and impractical outputs, while too-fo-

cused knowledge can result in “competence traps,”

in which new information is disregarded and teams

become locked in their old behaviors (Arthur, 1989;

March & Simon, 1958).

The literature on creativity provides a different

view of organizing for innovation by focusing on

how individuals and teams come to shape knowl-

edge in unique ways. Innovation consists of the

creative generation of a new idea and the imple-

mentation of the idea into a valuable product, and

thus creativity feeds innovation (Nijstod & De Dreu,

2002) and is particularly critical in complex and

interdependent work (Drazin, Glynn, & Kazanjian,

We are grateful for assistance from Paul Wolfson and

Jessica Sedgewick. We also thank Quintus Jett, Ruth

Wageman, Mary Tripsas, Judith White, two reviewers,

and Christina Shalley for helpful comments on drafts of

the paper.

娀 Academy of Management Journal

2006, Vol. 49, No. 4, 723–740.

723

Electronic copy available at: http://ssrn.com/abstract=948597Electronic copy available at: http://ssrn.com/abstract=948597

1999). Creativity can be viewed as the first stage of

the overall innovation process (Amabile, 1996;

West, 2002) or as intertwined with the implemen-

tation of innovative ideas (Paulus, 2002). In both

views, the creativity literature emphasizes novelty

(Morgan, 1953) but stresses that new ideas must be

useful and appropriate for the situation at hand

(Sternberg & Lubart, 1995). Thus, researchers in

this stream have seen creativity as constrained by

problems or tasks and have focused on perfor-

mance enhancement rather than on increased vari-

ance of innovation.

Like the innovations literature, the creativity lit-

erature frames innovative solutions as arising from

diverse knowledge, processes that allow for creativ-

ity, and tasks directed toward creative solutions

(Gilson & Shalley, 2004). However, it also suggests

that creativity requires application of deep knowl-

edge because individuals must understand a

knowledge domain to push its boundaries with any

nontrivial likelihood of success (Sternberg &

O’Hara, 2000), and that work practices can have a

significant impact on creative outcomes (Gilson,

Mathieu, Shalley, & Ruddy, 2005). Team creativity

likewise relies on tapping into the diverse knowl-

edge of a team’s members, which can be difficult. In

this literature, the distinction between structures

and processes that generate highly creative solu-

tions and those that generate high-performing but

less innovative solutions is less emphasized than it

is in the innovation literature.

Extending the innovation and creativity litera-

tures, this paper examines whether structures that

lead to variance-enhancing behaviors differ from

those that lead to higher mean performance. The

study also addresses an area that has received little

attention in both literatures: whether innovations

by teams have different causes than those by indi-

viduals. We focus on the impact of combining and

applying diverse knowledge. In organizations, the

availability of diverse knowledge is shaped and

constrained by the realities of collaborators, dead-

lines, resources, and workloads. Innovation is both

the creative development of novelty and its appli-

cation to generation of a new product (Amabile,

1996; West, 2002), with areas of expertise held by

individuals or teams within organizations accessed

and processed to generate innovations (Amabile,

1983; Ford, 1996; Simon, 1986). We acknowledge

that there are many sources of diverse experiences,

including the externally observable dimensions of

age, ethnicity, and gender (Milliken, Bartel, &

Kurtzberg, 2003); however, our emphasis on

knowledge use leads us to be principally concerned

with the less observable cognitive diversity that

arises from varied work, task, and organizational

experience.

We draw on cognitive perspectives on creativity,

innovation, and learning to develop a theoretical

model to predict both mean performance and vari-

ance in performance. We empirically tested our

model in a context specifically suited to our ques-

tions about knowledge-based activity: the creation

and publishing of comic books. Comic book cre-

ation is a commercial activity that has the useful

properties of being driven by intellectual property

and of having a collector valuation of output that

allows for empirical determination of the success of

the creative effort.

An added merit of the data is that it allows ex-

amination of the difference between single individ-

uals and individuals in teams acting in the same

innovative context. Despite the acknowledged im-

portance of individual cognition and of teams

(Kurtzberg & Amabile, 2001), little research inves-

tigates differences in the effects of individual ver-

sus team performance on commercial innovation.

Here we examine how individual and team knowl-

edge create radical innovations or incremental im-

provements. To ground our theoretical discussion,

in the next section we first describe the innovation

context we focused on; we then derive predictions

based on a cognitive perspective on innovation and

creativity. We divide the discussion into the avail-

ability of ideas to individual or to team, and the

ability of the individual or team to recognize and

implement the available ideas.

THE COMIC BOOK INDUSTRY

Innovation research requires that innovations be

understood relative to a context of interest, that the

setting be well understood, and that the data be

appropriate to the questions of interest. For our

study of individual and team innovation, the data

needed to allow us to identify specific creators,

individuals and teams of creators undertaking the

same commercial activities, and creators associated

with multiple products, as well as to objectively

measure the value of the output and the knowledge

domains and experiences of the creators. The

comic book publishing industry provides a unique

data opportunity that meets these empirical re-

quirements. It is also an industry where innovation

is valued and explicitly recognized, both by the

market and by the creators.

From the 1897 publication of the first American

comic book, The Yellow Kid, the comic industry

has evolved to the point that products based on the

intellectual property generated by the storytelling

pictures and words of one title such as Spider-man

724 AugustAcademy of Management Journal

Electronic copy available at: http://ssrn.com/abstract=948597

can generate over a billion dollars of revenues in

ticket and DVD sales. The intellectual property of

comics has been used by many other media, includ-

ing movies, television shows, children’s cartoons,

and video games. In addition, comic-sourced mate-

rials are one of few growing areas of the publishing

business and have become hotly contested prizes

for distributors such as Random House, HarperCol-

lins, Simon & Schuster, and the Time Warner Book

Group (Wyatt, 2006).

Innovations have always been critically impor-

tant to companies in the comic industry. For exam-

ple, the release of the world’s most famous comic,

Superman, in Action Comics #1, led to DC Comics

becoming a dominant publisher in the comic-pub-

lishing industry. Similarly, the early success of an-

other prominent comic-publishing company, Mar-

vel Comics Group, was based on the innovations of

the creators Stan Lee and Jack Kirby. Lee and Kirby

pioneered innovations such as depicting comic he-

roes that had real lives and problems (Sassienne,

1994). For example, their comic Fantastic Four had

a main character, The Thing, who was grumpy,

angry, and unhappy at having superpowers; their

character the Hulk battled his good and bad sides;

and Peter Parker in Spider-man was a geek under-

going all the troubles of a teenager while being a

superhero.

In the late 1950s up to the early 70s, innovation

was constrained as comic creators faced harsh cen-

sorship. Dr. Fredric Wertham, a child psychologist,

had convinced the U.S. Senate that comic books

were a major cause of juvenile delinquency. To

avoid federal regulation, the comic industry agreed

to self-censor, in 1954 instituting a Comic Code

Authority (CCA) to monitor and restrict content.

The censorship constrained innovation until Mar-

vel broke the rules in 1971, and soon other publish-

ers followed suit. The modern age of comics is

considered to have started after this period of

censorship.

The comic industry is composed of creators (e.g.,

artists and writers) producing comic titles that are

published by companies such as DC, Marvel, and

Vertigo, and distributed through retail stores and

direct sales to individuals. Like most innovative

processes, comic book creation can vary widely,

but in general it includes functions such as script-

ing, penciling, inking, and drawing. All of these

functions can be done by a single person or by

multiple people.

Both single creators and multiple creators strug-

gle with integrating the expertise necessary to pro-

duce a comic. It is the ability to combine the art,

writing, and structural form that generates an inno-

vative comic. The value of a comic is influenced by

the artwork, layout, arrangement, page composi-

tion, writing, and dialogue. For example, the fol-

lowing discussion of a creator’s work illustrates

how comic innovation is a combination of page

structure, art, and storytelling:

Like many innovative artists before him and since,

Bendis explores and expands his page layouts here.

Tied to the emotions of the main character, not

always but at certain key moments, the panels con-

tort to convey angst or fear. Also, Bendis thematic

keeps his “hero” characters round and simple, very

unthreatening, while the antagonists, the “Holly-

wood” types, are more angular and rarely open their

eyes, showing their narrow view approach of the

world. It’s a subtle but effective means of visually

reminding the reader who we like and who we

should be wary of in this. Oh, and one last thing, it’s

funny. It’s very funny. Bendis writes like how your

best friend tells a good story to you. He conveys

enough detail so you can follow the action, and

enough humorous anecdotes along the way to keep

you laughing . . . make this book a classic for mod-

ern humor comic books. (Messano, 2004)

Comics go from their creator to commercial re-

lease in several ways. A comic from an individual

creator can be created and sold to a publisher,

created and independently published, or created as

part of a contract with a publisher to produce a

comic. Comics are also created in teams, which

generally come about in one of two ways. The first

is that a publisher assigns creators to work together;

the second is that creators use their informal net-

works to work with desirable collaborators.

As in many new-product collaborations, the fre-

quency of teams continuing to work together is a

product of structure, success, and comfort. Creators

who work for the same publisher will often be

teamed together. Successful teams may also con-

tinue to work together, but it is the experience of

working together that is important because the per-

formance feedback of a comic is not immediate,

and the financial reward for a creator does not

differ substantially for an acknowledged innovative

comic.

Comics are grouped in market spaces called

genres. The genre creates a shared context, or set of

understandings shared by creator and audience

(Marks, 2004). A genre both prompts the expecta-

tions of the customer and provides a stylistic vo-

cabulary and grammar for the creator. For example,

Joe Zabel, an acclaimed comic creator who worked

on the comic American Splendor, provides the fol-

lowing description of the mystery comic genre in a

quote from a 1999 interview). It illustrates the set of

expectations, form, limitations, and knowledge

2006 725Taylor and Greve

represented when creating in a genre. It also hints

at work that crosses genres.

As a genre, the mystery exists within certain bound-

aries. First, a crime of some sort must have been

committed. The tension created by violence and

conflict are essential to the mystery’s appeal. Sec-

ond, the story must take place in a realistic setting.

That is to say, it’s not a fantasy or a science fiction

story. Mystery hybrids like The X-Files and Dirk

Gently, Holistic Detective aside, the essence of a true

mystery is believability. Third, there should be the

presence, in some form, of the unknown. This is the

distinction between a mystery story and the more

general category of crime fiction. (http://

amazingmontage.tripod.com/mcman.html)

The best-known genre in comics is the superhero

genre. Other popular genres are westerns, comedy,

fantasy, horror, adult, crime, science fiction, non-

fiction, and romance. Genres represent well-estab-

lished forms and product expectations, and transi-

tioning from one genre to another requires a creator

to learn and apply different domains of knowledge.

In much the same way as other products cross

market boundaries, comics can cross genre bound-

aries. Work that attempts to integrate multiple

genres has the promise of creatively crossing

boundaries and generating valuable products, as

well as the downsides of having to meet two sets of

genre domain requirements and potential rejection

by the market. Time’s list of the top ten comics of

2004 lauded several comics for their ability to cross

and bend genres (Arnold, 2004), but several lists of

the worst comics from that same year also included

comics that attempted to combine genres.

It is in this rich, creative, and knowledge-inten-

sive context that we tested our hypotheses on in-

novation and learning.

THE AVAILABILITY OF IDEAS

When investigating how knowledge is combined,

it is important to examine all innovations, not just

those that are successful, because the novelty of

innovations makes them uncertain, causing some

to fail. The distinction is important because activ-

ities that increase mean performance through ex-

ploitation are different from activities that increase

performance variance through exploration (March,

1991). For example, in the pharmaceutical indus-

try, where generating variations on existing prod-

ucts or new generic products may result in small

financial performance gains, many firms are will-

ing to gamble on potential breakthrough drugs that

have lower likelihoods of success, but when suc-

cessful result in major financial performance gains.

Whereas incremental improvements increase aver-

age performance by using existing knowledge, rad-

ical innovations are characterized by prior uncer-

tainty and posterior variance in performance.

Similarly, in variation-retention models of creativ-

ity, creative outputs are viewed as being highly

variable (positively and negatively) and facing later

selection into successful and unsuccessful work

(Campbell, 1960; Simonton, 1999b). Studies that

only examine high-valued outcomes cannot distin-

guish mean and variance effects on innovative per-

formance, leaving open the question of whether

individual and organizational factors differently af-

fect the mean or the variance.

The ability to generate novel, high-variance out-

comes is based on the availability of ideas. Idea

availability can be constrained by local search, in

which a limited set of options is considered accord-

ing to confidently held beliefs (Lyles & Schwenk,

1992). Broader search results in more idea variety

and can identify ways to combine knowledge that

challenge the beliefs that constrain innovative be-

havior (Greve & Taylor, 2000). The paradox is that

innovative experts also search locally to determine

what rules to break, while nonexperts search lo-

cally and conform to those rules (Weisberg, 1999).

For example, in new-product development, firms

attempt to structure the process to facilitate novel

combinations of knowledge. Hargadon and Sutton

(1997) described how the product development

firm IDEO used a formal archive of past design

information as a way to access old knowledge for

use in new combinations, but also as a record of

past design mistakes. This is one way the firm

combines efforts that have already demonstrated

appropriateness of design and commercial applica-

tion. The firm also invites designers not associated

with a focal project to provide feedback and advice

in the initial design stages. In this way IDEO at-

tempts to facilitate the combination of different

knowledge domains to generate creative product

designs, while locally searching through past

knowledge.

Diverse knowledge of multiple domains and

deep knowledge in a specific domain can both lead

to innovations. Environments in which diverse

knowledge domains are available are more likely to

produce new ideas and new combinations of ideas

that drive the creation of innovations. Diverse

knowledge provides more components useful for

making innovative combinations, which gives the

opportunity for significant advances, but also for

innovations that receive low evaluations because

the combinations have unanticipated flaws (Flem-

ing, 1999). However, the challenge of implement-

ing the diverse knowledge that arises from broad

search makes the generation of usable innovations

726 AugustAcademy of Management Journal

difficult. In situations in which creators combine

diverse knowledge domains, we expect innovative

behavior when the participants are able to effec-

tively combine the knowledge and have well-estab-

lished expertise (Brass, 1995; Perry-Smith & Shal-

ley, 2003). In situations in which creators draw on

a single domain in a practiced manner, we expect

incremental improvements instead for nonexperts,

and nonincremental improvements for experts.

Our theory of individual creativity yields similar

conclusions as those for teams. Several disposi-

tional factors affect individual creativity, including

personality traits and innate genius (e.g., Feist,

1999; Simonton, 1999a), but we focus on how in-

dividuals obtain diverse knowledge and become

willing to combine it. Amabile’s (1983, 1996) the-

ory of individual creativity has three components

that affect the availability of ideas: domain-relevant

skills, creativity-relevant skills, and task motiva-

tion. Exposure to a greater variety of ideas, includ-

ing ideas that are inconsistent with current beliefs,

increases individual creativity (Parnes & Noller,

1972). In individuals, diverse ideas become avail-

able through exposure from past experiences, espe-

cially high-commitment exposure such as actually

using a knowledge domain to create a new product.

Through their career histories, individuals can be-

come proficient in multiple knowledge domains

and motivated to combine them. Creativity-

relevant skills, such as the ability to suspend judg-

ment, using widely inclusive categories, perceiving

things differently than most people, and breaking

out of perceptual or cognitive patterns, add to the

breadth of ideas that individuals consider in their

choice sets. These skills add to the availability and

use of diverse knowledge for an innovative

solution.

It follows that teams with individuals who each

hold diverse knowledge domains will be likely to

combine them. In addition, teams allow individu-

als who do not hold diverse knowledge to become

exposed to it to through interaction with other

members (Nonaka & Takeuchi, 1995). Discussions

that start with differing viewpoints result in

broader search for information (Nemeth & Rogers,

1996) and more complex reasoning (Gruenfeld &

Kim, 1998), which can give new insights that help

develop innovations (Van Dyne & Saavedra, 1996).

Thus, teams whose members have and share di-

verse knowledge can obtain higher levels of indi-

vidual and team creativity (De Dreu & West, 2001).

Cognitive Diversity

Diversity of knowledge in teams has been termed

“deep-level diversity” (or “cognitive diversity”) to

distinguish it from diversity in surface characteris-

tics such as the demographic variables of age, gen-

der, and race (Harrison, Price, Gavin, & Florey,

2002; Jehn, Northcraft, & Neale, 1999). Although

the theory of knowledge combination holds that

diverse knowledge components generate perfor-

mance variance (Fleming, 1999), the information

processing perspective on team diversity holds that

greater cognitive diversity leads to higher perfor-

mance potential. Work group creativity is en-

hanced when a work environment provides rich

knowledge stimuli, sufficient resources, and a chal-

lenging workload (Amabile, Conti, Coon, Lazenby,

& Herron, 1996; West, 2002). Working in groups

exposes individuals to a broader set of perspec-

tives, and cross-fertilization of ideas results in more

creative outcomes (Perry-Smith & Shalley, 2003;

Tesluk, Farr, & Klein, 1997). When knowledge com-

ponents are diverse, uncertainty about the value of

each component increases the uncertainty of the

output, however, and so does uncertainty about the

optimal way to combine components. The result is

that multicomponent innovations have greater vari-

ance in quality evaluations, resulting in perceived

failures as well as in breakthroughs (Fleming, 1999;

Fleming & Sorenson, 2001).

A high number of creators is an important source

of diverse knowledge (West & Anderson, 1996).

Individuals have different cognitive strategies and

career experiences, leading to variation in knowl-

edge and problem-solving approaches that can help

teams identify and use multiple knowledge compo-

nents. A high number of creators also increases the

likelihood that creative processes such as consid-

ering exceptions, challenging well-worn scripts, or

playing with ideas will occur (Amabile, 1996). The

role of career experience in generating unique in-

dividual stocks of knowledge is especially impor-

tant for work done in creative or problem-solving

teams in settings such as consulting (Hansen,

1999), product development (Hargadon & Sutton,

1997), and creative industries (Miller & Shamsie,

2001). In such work, individuals develop their own

knowledge and network ties with other knowledge-

able persons, both of which help them to retrieve

and apply knowledge components useful for a

given task. As a result, the size of a team drives

diversity in knowledge and ability to innovate

(Jackson, 1996). Thus, we predict a positive rela-

tion between team size and the variance of product

evaluations:

Hypothesis 1. As the number of creators in-

creases, they are more likely to generate prod-

ucts that have extreme (best or worst)

outcomes.

2006 727Taylor and Greve

It has been argued that communication in large

groups has a process cost that reduces group out-

puts (Kurtzberg & Amabile, 2001; Steiner, 1972).

An integration of the influences from a broader set

of inputs and more costly communication in large

teams would seem to predict a curvilinear or even

an inverted-U-shaped relation between team size

and innovativeness. This proposition is most rele-

vant for teams that are larger than those that ob-

served in this study, only 1 percent of which had

more than six members, so we were not able to

evaluate this argument with the available data.

Holding constant the effect of team size, teams in

which the members have had exposure to more

diverse knowledge will have access to more knowl-

edge components and will as a result be more cre-

ative, but they will also have greater potential for

team conflict (Williams & O’Reilly, 1998). Because

the diversity literature has focused on problem-

solving rather than creative tasks, the research mea-

sures the mean of performance rather than its vari-

ance, and often seeks to separate the effects of deep-

level and surface diversity on performance and to

investigate how conflict mediates this relation (e.g.,

Harrison et al., 2002; Jehn et al., 1999). The primary

effect of task conflict is to reduce team performance

(De Dreu & Weingart, 2003), whereas high perfor-

mance in cognitively diverse teams is possible

when a socially cohesive and participatory envi-

ronment allows members to freely apply and con-

tribute their knowledge (Chatman, Polzer, Barsade,

& Neale, 1998; De Dreu & West, 2001). Thus, di-

verse teams will have both highly positive and

highly negative outcomes from their innovation

attempts.

For individuals, combining diverse experiences

does not have the coordination or access problems

that arise in teams, so an individual can have more

integrated, diverse knowledge without the interper-

sonal conflicts present in teams. As a result, an

individual creator is less likely to make compro-

mises in the creative process. Although negative

extremes might occur, they will be due to the in-

trinsic risks of experimentation and not to difficul-

ties in communicating or reaching agreement. As a

result, individual knowledge combination should

give even greater performance variation than that of

teams. These arguments generate the following

hypotheses:

Hypothesis 2. As the knowledge diversity of

creators increases, they are more likely to gen-

erate products that have extreme (best or

worst) outcomes.

Hypothesis 3. Increased knowledge diversity

has a stronger effect on extreme (best or worse)

outcomes for individual creators than for

teams.

Innovating from Available Ideas

Having available ideas is only one part of gener-

ating useful innovations: The ideas must be turned

into innovations under the organizing structure of

the work. One dimension of the structure of work—

being organized for creativity—is taken for granted

in the innovations literature, as it is the goal of the

process. The creativity literature has stressed that

having both the task motivation for creativity and

creative processes in place are important in gener-

ating creative outcomes (Amabile, 1983; Drazin et

al., 1999).

Organizations in creative industries value inno-

vativeness as a way to generate occasional (but

highly profitable) blockbuster products, even

though it is recognized that many creative products

are not highly valued in the market. Because cre-

ative products are a goal and are recognized to

generate variance in product evaluations, high vari-

ance in outputs is a measure of creative teams ef-

fectively using their diverse knowledge. As noted

earlier, creators are allowed a degree of control over

team selection and continuation in creative indus-

tries and may dissolve a team that experiences con-

flict or a creative drought. As a result, creative

works that are brought into production tend to

come from teams whose members are comfortable

working with each other, which is a selection

mechanism that reduces the effects of team conflict

on the observed innovations.

An important characteristic of team composition

is the experience that the members have in working

together. Teams go through a process of socializa-

tion that makes communication easier as members

adapt to each other (Katz, 1982). The result is that

low or diverse member tenures in teams create

communication difficulties (Pfeffer, 1983), which

increase the negative effects of member diversity on

communication (Pelled, Eisenhardt, & Xin, 1999).

Conversely, teams with sufficient experience to

have established efficient communication can more

easily utilize member diversity (Harrison, Price, &

Bell, 1998; Harrison et al., 2002). When teams work

on projects of short duration, team experience is

often better expressed as the number of projects

that a given team has worked together on than as

the duration of prior work together, because it is

repetition of the creative phase in the start of a

project that improves cooperation. Also, teams are

more likely to re-form for new projects when their

members are satisfied with a previous work pro-

cess, so the gain in collaboration from experience is

728 AugustAcademy of Management Journal

amplified by selectivity in assembling teams that

work well together.

Thus, teams with many prior

collaborative projects have better communication

and are thereby more likely to fulfill the goal of

obtaining more creative outputs. These teams are

also more likely to develop standardized practices

for operation, which result in higher mean perfor-

mance outcomes (Gilson et al., 2005). For the same

reasons, teams with many prior collaborative

projects should be able to raise the average quality

of their outputs. These arguments lead to the fol-

lowing predictions:

Hypothesis 4. As the team experience of cre-

ators increases, they are more likely to generate

products that have extreme (best or worst)

performance.

Hypothesis 5. As the team experience of cre-

ators increases, they generate products with

higher mean performance.

Learning by refining existing products and pro-

cedures increases mean performance rather than

increasing the dispersion of performance—indeed,

procedures that encourage incremental improve-

ments tend to reduce the dispersion of performance

(Benner & Tushman 2002; March, 1991). Incremen-

tal improvements occur through learning by doing,

which is a process of making minor experiments,

observing the results, and adopting those experi-

ments that are seen to improve performance (Ar-

gote, Beckman, & Epple, 1990). Experience leads to

opportunities to experiment and improve, but op-

portunities come at a decreasing rate as the poten-

tial for improving performance through incremen-

tal changes runs out. In consequence, efficiency

improves proportionally to the total volume of pro-

duction, as is captured in the well-documented

learning curve (Argote, 1999). Learning by doing is

subject to decay and obsolescence (Argote et al.,

1990), so it is most effective when recent produc-

tion is high. It follows that a work unit with a full

workload is likely to perform better than one that is

underemployed.

Although the learning curve is well documented

for cost reductions in industrial production, it is

not known whether learning also improves creative

performance. Against the learning curve, one may

argue that the processes of experimentation and

routinization are unlikely to positively affect such

unique tasks as developing a new product that

changes the frontiers of knowledge; further, one

may argue that tastes change too fast for learning to

have much value over time. An observation in favor

of learning by doing in creative tasks is the finding

that heavy (but not excessive) workload improves

creativity in organizations (Amabile et al., 1996;

West, 2002), as learning by doing would predict.

The theoretical argument against learning curves in

creative work also seems weak, as it discounts the

usefulness of cumulative knowledge for creative

tasks and the ability of creators to anticipate or

follow shifting tastes. It is more parsimonious to

maintain the usual learning curve in this context as

well, leading to the following prediction:

Hypothesis 6. Creators with heavy workloads

generate products with higher mean

performance.

Learning curves have different shapes, depend-

ing on the nature of the learned task. Tasks can be

learned at different speeds as a learner gains expe-

rience, and they can be forgotten at different speeds

if the learner has a low workload (Argote et al.,

1990). Thus, the relative importance of total cumu-

lated experience and recent workload in determin-

ing performance will differ with type of task. If

creative tasks are highly dependent on a set of basic

skills that can be applied in many situations, then

total experience will matter most. If creative tasks

are instead dependent on volatile knowledge such

as shifting audience tastes, then recent workload

will matter more. If we assume that there is a set of

learnable basic skills useful in creative tasks, we

can also hypothesize the following effect of total

experience:

Hypothesis 7. As the tenures of creators in-

crease, they generate products with higher

mean performance.

Investing in work facilities can also increase ef-

ficiency. In industrial production, this observation

leads to the obvious positive relation between

amount and quality of production machinery and

production efficiency. For creative work, a techni-

cal and administrative support system that relieves

creators of routine tasks has the same effect, as does

organizational slack devoted to improving the

knowledge of the innovators (Cohen & Levinthal,

1990). These considerations predict:

Hypothesis 8. As organizational resources in-

crease, creators generate products with higher

mean performance.

This argument can be verified by the data. We found

that first-time collaborations were more likely to be dis-

solved when the product received a low evaluation, but

after the first collaboration there was no longer a relation

between product success and group re-formation.

2006 729Taylor and Greve

METHODS

Data and Sample

We analyzed data on comic books published

from 1972 through 1996. We chose 1972 as the

beginning year for our data, as the censorship of

comics by CCA, and hence constrained innovation,

ended in 1971. We ended the study in 1996 for two

reasons. First, we wanted substantial time to have

elapsed from the end of our data set that early

valuations resulting from company marketing

strength or speculation would no longer affect the

perceived market value of a comic. Second, after

1996, the comic bubble generated by an influx of

speculators (a consequence of widespread recogni-

tion of comic art as valuable) burst, and the indus-

try saw massive structural changes. Many publish-

ers went out of business, and even Marvel filed for

bankruptcy (Duin & Richardson, 1998). To avoid

this noise in the data, we stopped the study before

these structural changes occurred.

The ability of the comic-trading market to per-

ceive and respond to differences in product quality

made these data particularly attractive. The collec-

tor values of comics in the data represent retrospec-

tive market judgments of quality and innovative-

ness, with the impact of temporary perturbations

caused by nonqualitative product gimmicks greatly

reduced by the time span between publication and

collector value assessment. The collector value of a

comic book is a natural version of Amabile’s (1982,

1996) consensual assessment technique for creativ-

ity, as collectors provide independent expert judg-

ments of products, which are compared against

each other rather than against an absolute ideal.

The evaluators of comics are increasingly adult; a

survey by the largest comic distributor, Diamond

Comics, revealed that the average age of (avid)

comic readers is 34.

The unit of analysis was the individual comic

book. The price and comic data we used came from

the electronic source Comicbase, which we tapped

in 2004. Supplemental data for identifying authors

and publishers were obtained from the 2003 Over-

street Comic Book Price Guide and the 2002 Stan-

dard Catalogue of Comic Books. The value infor-

mation in the Comicbase data is drawn primarily

from analysis of actual sales transactions collected

from online auctions, dealer sales, shop sales, and

convention sales. For a reliability check, we com-

pared the prices our primary source listed for a

random sample of 250 comics to prices in another

data source, the Overstreet Comic Book Price

Guide. The correlation between the two sources

was .99. The data on each comic book used in our

analysis include the date issued, its creators, and

the organization that published it. We began the

data collection by identifying all comic books pub-

lished in the United States between 1972 and 1996.

For each of these, we then captured the creator and

company data. We then examined the data, remov-

ing duplicates and other inconsistencies. Finally,

we eliminated comics released with marketing gim-

micks such as gifts to allow comparability. After

the clean-up, a total of 4,485 comic books with 234

publishers remained. Not surprisingly, Marvel and

DC, the two largest comic publishers, accounted for

almost 54 percent of all the comics in our data.

Thus, the findings may be weighted toward the

experiences of creators in these large firms.

We analyzed the collector value of comics as

reported in the Comicbase data for the year 2003,

leaving at least five years between the publication

of a comic and its evaluation. We standardized

these values to have a mean of zero and a standard

deviation of unity in each year. The evaluation of

comic values differs a great deal, so the standard-

ized values show extreme outliers to the right-hand

side (very innovative comics with very high collec-

tor value). On the left-hand side there are also out-

liers, but less extreme ones, as comics cannot have

a negative value. Figure 1 shows the distribution of

values at four five-year intervals starting in 1977.

The distributions place most of the observations in

the center (“normal” comics), but with left and

right tails of unusually low and high values,

respectively.

Measures

Dependent variable. As noted earlier, our de-

pendent variable was the within-year standardized

collector market value of a comic in the year 2003.

We also obtained the collector values from 2000

through 2002 and used these to verify that the

values were stable. We found that the pairwise

correlations of adjacent years were on average .92

for all the comics. This correlation is quite high,

given that values may change as a result of reeval-

uation of the artistic value of a comic as well as in

response to new information about its scarcity and

demand. We examined the correlations of values

for the last five years of comics in the data (1992–

97) separately to check for instability in the evalu-

ation of recent comics but found that these corre-

lations were also high: the adjacent-year correlation

was .97 on average.

To further test that the upper values were comics

perceived to be innovative, we sent a survey to 20

industry experts identified in the Overstreet and

Wizard comic guides asking for a list of which 25

published comics in the modern comic era they

730 AugustAcademy of Management Journal

considered most innovative. Thirteen of the experts

responded; the average percentage of agreement

was 76 percent for the top 25 and 85 percent for the

top 10. This is an impressive level of agreement,

given that the raters had the entire universe of

published comics to draw from. We also calculated

an interrater agreement between the possible com-

binations using Cohen’s kappa (Cohen, 1960; Lan-

dis & Koch, 1977). The results showed significant

agreement between the lists, with mean Ks of .46

for the top 25 and .67 for the top 10 most innovative

comics (Z ⬍ .000 for both). We generated a list of 50

comics by combining the lists. The innovations

included comics such as the 1987 Maus: A Survi-

vor’s Tale, the first comic nominated for the Na-

tional Book Critics Award and a winner of the

Pulitzer Prize; the 1962 Amazing Fantasy #15, the

first depiction of a hero with personal problems

(Spider-man); and the 1982 Love and Rockets,

which showcased ethnic sensibility and stylistic

sophistication. We compared the survey results to

the price data. Of the 50 innovative comics identi-

fied by the experts, 64 percent were in the top 10

percent of the comic values in our data, and the

innovative comics were on average at the top 13.3

percent of standardized values.

Independent variables. To test Hypothesis 1, we

included number of creators, measured as a count

of the writers and artists involved in a comic. A

creator can act as both writer and artist, but typi-

cally a team of creators develops a comic. Adding

team members on a comic often adds different per-

spectives to the creative process. The use of differ-

ent views in the process is illustrated by this quote

from an interview with Geoff Johns, a writer for the

Superman comic among others. “I love co-writing.

It’s a blast to sit in a room or talk over dinner or just

throw around ideas with another writer. It’s the

best thing in the world if you’re working with

someone you really gel with creatively and person-

ally. . . sometimes we fight over it and do it all over

again eight times” (Hays, 2005). Some of the qual-

itative data suggested that a creator acting as both

writer and artist might denote a special creative

process. For example, award-winning comic cre-

ator James Sturm stated: “Working alone allows

you to have a single focus—the art, writing, every-

thing is purely directed by your vision, without

having to make compromises” (personal interview,

November 2003). To address this possibility, we

also included an indicator value for single creator.

To test Hypothesis 2, we coded diversity of back-

grounds as the number of genres in which creators

had worked over the year prior to the publication of

a focal comic. This variable, genre experience, was

a count of all genres that the creators on a team had

worked with in their careers, omitting double-

counts of genres that more than one creator had

worked with. This variable captured a team’s range

of genre knowledge domains. Comicbase catego-

FIGURE 1

Value of Comics for Four Years

2006 731Taylor and Greve

rized comics into 22 genres but associated some

comics with multiple genres (e.g., a graphic super-

hero novel). To test Hypothesis 3, we interacted

this variable with the indicator variable for single

creator.

To test Hypotheses 4 and 5, we calculated team

experience as the number of times a creative team

had worked together previously. That is, we kept

track of the team number of collaborations rather

than the number of collaborations between the

members of a team (who might have collaborated as

parts of different teams). We also tried a measure

counting all collaborations between members of a

team, which gave the same findings. In these data,

the two measures correlated nearly perfectly (.96).

To test Hypothesis 6, we included a workload vari-

able, the average number of comics that the creators

of each comic had produced in the past year. To

test Hypothesis 7, we measured the number of

years elapsed since each team member’s first pub-

lished comic and took the highest as our variable

for tenure. Though the data we analyzed start in the

modern period (1972), we based the tenure and

genre variables on data going back to 1930 to cor-

rectly measure experience before 1972. To test Hy-

pothesis 8, we measured organizational resources

as the sum of the standardized values of the num-

ber of comics published, total circulation, and

number of affiliated creators per year for each pub-

lisher (

␣

⫽.92; average r ⫽ .80). We constructed this

measure using a data set in which each firm-year

contributed one observation in order to avoid over-

weighting the large firms.

Initial circulation of a comic was used as a con-

trol for scarcity effects on comic value. Because

multigenre comic titles might be evaluated differ-

ently than single-genre titles, we entered the vari-

able genre count, which equaled the number of

genres in which a focal comic was classified. We

created an indicator variable with a value of 1 if the

average workload of creators had significantly in-

creased (more than fivefold) over the previous year.

We included the highest-value comic any of the

creators had made in the past in order to control for

creator ability differences and attempts to repeat

past successes (Amburgey & Miner, 1992).

To control for observations lost through missing

data on creators or publishers, we computed an

inverse Mills ratio from a logit model predicting

complete data and included it in the model (Lee,

1983). Finally, we included indicator variables for

five-year spans (1978–82, 1983–87, 1988 –92, and

1993–96, with 1973–77 as the omitted category). To

keep the tables brief, the coefficient estimates for

these indicator variables are not shown. We lagged

all independent variables by a year.

Model Specification

Because our focus on innovations as extreme val-

ues was novel, we tried three modeling approaches

to test the hypotheses. The first approach uses lin-

ear regression analysis on the standardized level of

performance and deviation from mean perfor-

mance. We standardized performance to have a

mean of 0 and a standard deviation of 1 in each

year. Thus, it expressed performance relative to

other comics in the same year. We expressed the

performance deviation as the absolute deviation

from the mean of the standardized variable (which

is 0). The second approach was linear regression on

percentiles, in which we rank-ordered the observa-

tions according to ascending performance within

each year and then calculated the rank percentile.

This approach has assumptions about performance

similar to those of linear regression, except that it

focuses on rank rather than on absolute level of

performance. The third was linear regression anal-

ysis deleting the top and bottom 1 percent of obser-

vations as a check for influential outliers. These

approaches gave consistent results except as noted

in the text, and we present the results of the first

approach, linear regression on standardized perfor-

mance, with no deletions.

Table 1 provides the descriptive statistics and the

correlation matrix. As expected, the variables with

high correlations were single creator and multiple

creators, workload increase and workload average,

and publisher size and single creator. None of the

other variables had high correlation coefficients.

Preliminary analysis showed that the results were

stable when subsets of the variables were omitted

from the model.

RESULTS

We first report results of analyzing the data with-

out distinguishing individual creators and multi-

person creative teams and then show models that

distinguished the effects of individual creators. Ta-

ble 2 shows the result of the analyses of perfor-

mance level and deviation. Because Hypotheses 1

through 5 concern innovative (variance-increasing)

behaviors, we first describe model 2, which exam-

ines the performance deviation from the mean. Hy-

pothesis 1 states that having many creators in-

creases the chance of producing extreme outcomes.

This hypothesis is supported, as seen in the posi-

tive and highly significant coefficient for number of

creators in the regression on the standardized de-

viation of performance. However, it is not sup-

ported in the regression on rank deviation, so the

finding is sensitive to the treatment of extreme out-

732 AugustAcademy of Management Journal

liers in the distribution. Next, Hypothesis 2 states

that diversity of creator background increases the

performance variance. The variable for genre expe-

rience had a significant, positive effect, in support of

the hypothesis. Hypothesis 3, stating that this effect is

stronger for individual creators, is tested in model 4.

Hypothesis 4 states that increased team experi-

ence increases variance in performance. The posi-

tive and significant coefficient estimate of team

experience strongly supports this hypothesis.

These findings show that teams with diverse

knowledge and much experience working together

produce comics with extremely high or extremely

low performance, in support of our theory of the

effect of knowledge combination on innovative-

ness. The analysis of deviation from the mean thus

provides strong support for three of our hypotheses

on factors that increase the likelihood of teams

making innovative products with extreme (high or

low) evaluations.

TABLE 1

Descriptive Statistics and Correlations

Variable Mean s.d. 12345678910111213

1. Circulation 0.02 0.01

2. Number of genres 1.19 0.41 ⫺.17

3. Workload increase 0.11 0.32 ⫺.09 .10

4. Lowest-value

comic

⫺0.46 1.39 .05 .02 .11

5. Single creator 0.27 0.45 ⫺.17 .12 ⫺.06 ⫺.17

6. Number of creators 2.17 1.31 .05 .11 .12 .07 ⫺.55

7. Genre experience 2.56 1.54 ⫺.02 .21 .02 ⫺.02 ⫺.11 .19

8. Team experience 3.35 6.53 .10 ⫺.14 ⫺.14 ⫺.01 ⫺.31 ⫺.02 ⫺.06

9. Highest-value

comic

4.66 18.27 .04 ⫺.07 ⫺.08 ⫺.08 ⫺.02 .01 ⫺.00 .03

10. Workload average 14.73 10.41 .19 ⫺.10 ⫺.46 ⫺.14 .12 ⫺.15 .04 .11 .09

11. Tenure 16.63 11.14 ⫺.03 .15 ⫺.12 ⫺.12 .14 .16 .05 ⫺.07 ⫺.04 .17

12. Tenure squared/

100

4.01 5.26 ⫺.08 .20 ⫺.09 ⫺.12 .21 .11 .05 ⫺.12 ⫺.05 .18 .96

13. Publisher size 0.74 0.98 .25 ⫺.10 .00 .02 ⫺.33 .07 .07 .17 ⫺.05 .12 ⫺.27 ⫺.30

14. Inverse Mills ratio 0.52 0.23 .22 ⫺.22 ⫺.13 ⫺.07 ⫺.01 ⫺.17 .03 .10 .15 .30 .28 .02 .02

Correlation coefficients with a magnitude greater than .03 are significant at the .05 level.

TABLE 2

Results of Linear Regression Analysis of Level and Deviation of Performance

Variables Model 1: Level Model 2: Deviation Model 3: Level Model 4: Deviation

Circulation 10.16** (1.77) 4.62** (1.71) 10.13** (1.77) 4.59** (1.71)

Genre count ⫺0.04 (0.05) ⫺0.01 (0.05) ⫺0.05 (0.05) ⫺0.01 (0.05)

Workload increase ⫺0.22** (0.06) ⫺0.14* (0.06) ⫺0.21** (0.06) ⫺0.13* (0.06)

Highest value comic 0.02** (0.00) 0.02** (0.00) 0.02** (0.00) 0.02** (0.00)

Single creator ⫺0.12* (0.06) ⫺0.04 (0.05) ⫺0.22* (0.09) ⫺0.16

†

(0.08)

Number of creators 0.00 (0.02) 0.04* (0.02) 0.01 (0.02) 0.05** (0.02)

Genre experience 0.040** (0.012) 0.027* (0.012) 0.031* (0.014) 0.016 (0.013)

Genre experience ⫻ single creator 0.044 (0.028) 0.053* (0.027)

Team experience 0.011** (0.003) 0.011** (0.003) 0.011** (0.003) 0.010** (0.003)

Workload average ⫺0.004

†

(0.002)

⫺0.003 (0.002) ⫺0.004

†

(0.002)

⫺0.003 (0.002)

Tenure ⫺0.01 (0.01) ⫺0.01 (0.01) ⫺0.01 (0.01) ⫺0.01 (0.01)

Tenure squared 0.01 (0.01) 0.00 (0.01) 0.01 (0.01) ⫺0.00 (0.01)

Publisher size ⫺0.18** (0.04) ⫺0.07

†

(0.04)

⫺0.18** (0.04) ⫺0.07

†

(0.04)

Inverse Mills ratio 0.50** (0.10) 0.23* (0.10) 0.51** (0.10) 0.23* (0.10)

Constant ⫺0.25

†

(0.14)

0.08 (0.14) ⫺0.24

†

(0.14)

0.09 (0.14)

0.15 0.11 0.15 0.12

n ⫽ 4,485. Standard errors are in parentheses.

†

p ⬍ .10

* p ⬍ .05

** p ⬍ .01

Two-sided tests.

2006 733Taylor and Greve

Next, model 1, on the level of performance, was

used to test our predictions on learning made in

Hypotheses 5 through 8. First, the statement in

Hypothesis 5 that the team experience of creators

increases average performance is supported, as the

coefficient estimate of team experience is positive

and highly significant. Hypothesis 6, stating that a

heavy workload increases performance level, is not

supported. Instead, the coefficient estimate is neg-

ative and marginally significant, contrary to predic-

tion. Tenure has an insignificant effect on the level

of performance, indicating a lack of support for

Hypothesis 7. Finally, publisher size has a negative

and significant effect on performance, contrary to

the prediction made in Hypothesis 8, that organi-

zational resources increase performance. The anal-

ysis strongly supports a positive effect of team ex-

perience but provides highly mixed results for the

other hypotheses on mean-increasing effects in

comic book creation.

These models also provided some interesting

findings that were not predicted. Genre experience

had a significant and positive effect on the level of

performance, suggesting that broad expertise on a

creative team increased mean performance in addi-

tion to having the predicted effect of increasing the

variance of performance. A similar mean-increas-

ing effect was also found for the highest-value past

comic. In these data, the role of expertise in jointly

spurring creativity and raising average performance

is so strong that it overwhelms the theorized trade-

off between exploration and exploitation (March,

1991). A trade-off may still exist within each indi-

vidual team, but the teams differ so much in knowl-

edge and experience that the factors that predict

level of performance resemble those that predict

variation. Also, a workload increase reduces both

the level of performance and (for the standardized

variable only), its deviation. Thus, the creators with

high workloads had lower performance, and in-

creased workloads resulted in lower-quality, less

innovative comics.

Next, models 3 and 4 present the analysis that

takes into account the difference between individ-

ual creators and multiperson creative teams. As

before, the analysis supports the three hypotheses

on factors that increase the variance of perfor-

mance, but model 4 provides additional detail.

Now, the main effect of genre experience on ex-

treme outcomes (Hypothesis 2) is not significant,

but the interaction of genre experience and individ-

ual creators is positive and significant. This finding

is in support of Hypothesis 3, stating that knowl-

edge diversity in an individual has a stronger effect

than knowledge diversity in a team. Combined,

these findings are a clear demonstration of process

loss: genre experience held by a single individual

has an effect on innovation, but genre experience

dispersed over team members does not.

To show the substantive effect of these interac-

tions, we graph them in Figures 2a–2b. In both

panels, genre experience is on the horizontal axis

and is graphed from 0 to the mean plus two stan-

dard deviations. The performance of a team with no

genre experience (a team of novices) is set to 0, and

the regression coefficients of the single-creator in-

dicator, genre experience, and their interactions are

used to compute the curves. Figure 2a shows that

single creators start with a lower level of perfor-

mance, as the negative indicator variable indicates,

and then have a marginally higher improvement as

they gain experience, leading the curves to cross at

five genres. However, the interaction effect is not

significant in the performance-level equation, so

we cannot be sure that the climb really is higher.

Figure 2b shows the curves for the deviation of

performance, where the interaction effect is statis-

tically significant as well as larger in magnitude.

Although single creators start out with lower inno-

vativeness, their curve crosses that of teams at three

genres. Hence, genre experience is especially valu-

able for increasing individual innovativeness. Con-

versely, combining diverse knowledge from multi-

ple persons results in process costs that offset the

benefits.

We performed two additional analyses to explore

the conditions under which experience affects in-

novative output. First, we suspected that experi-

ence without diverse knowledge might not increase

the variance of outputs, and might even decrease it.

Using an interaction of experience and an indicator

variable for whether a team had experience from

only one genre, we found that the first of these

conjectures was true. This interaction variable had

a negative and significant coefficient in the analy-

ses of performance deviation and performance

level, and the coefficient was exactly large enough

to cancel out the positive main effect of experience.

Experience increases variation when combined

with diverse knowledge, but not otherwise. Sec-

ond, we thought that large organizations might

have lower performance because their formaliza-

tion prevented learning from experience in teams

and thus entered an interaction of size and experi-

ence to test this conjecture. The findings were un-

supportive, as the coefficient estimates on perfor-

mance were insignificant.

DISCUSSION AND CONCLUSION

This paper follows the research tradition of using

cultural industries to investigate competitive dy-

734 AugustAcademy of Management Journal

namics (Hirsch, 1972), as in work on music (All-

mendinger & Hackman, 1996; Bougan, Weick, &

Binkhorst, 1977); book and magazine publishing

(Levitt & Nass, 1989; Thornton & Ocasio, 1999);

film (Mezias & Mezias, 2000; Miller & Shamsie,

1996); and radio broadcasting (Greve, 1995; Greve

& Taylor, 2000). We use the comic-publishing in-

dustry to examine the relationship between cre-

ative and learning processes and subsequent com-

mercially recognized value. We distinguish factors

that increase the performance variance and factors

that increase the mean performance level.

Uniquely, we examine exploration activities that

result in outcome extremes, as seen in increased

variance of performance, rather than just enhanced

performance.

We found higher variance of performance in

teams with multiple members, experience from

multiple genres, and a history of working together.

The positive effect of genre experience on variance

was largely attributable to single-member teams,

however, suggesting that individuals are capable of

more creative integration of diverse experiences

than teams are. These findings were supportive of

our hypotheses. We also found that large organiza-

tions and high workloads reduced the variance,

although production of high-value output in the

past predicted high variance. We found a higher

level of performance in individuals and teams with

experience from multiple genres, and team experi-

ence working together also increased performance.

We also found that large organizations and high

workloads reduced the level of performance, and

single creators had lower performance than teams,

on the average. The results show an interesting

pattern. Innovation and creativity as variance-

increasing activities have clearly distinguishable

causes in line with predictions, but there was much

less support for the hypotheses on mean-increasing

activities. In addition, some of the factors hypoth-

esized to increase performance variance also in-

creased the mean.

The results substantiate our argument that com-

bining knowledge requires a deep understanding of

knowledge, rather than information scanning or ex-

posure. The variables that reflect a deep under-

standing—experience in diverse knowledge do-

mains, team experience, and previous innovation

experience—drove the increase in variance behav-

ior. Number of creators, the variable that captures

FIGURE 2a

Interaction Effects on Performance Level

FIGURE 2b

Interaction Effects on Performance Deviance

2006 735Taylor and Greve

exposure but not understanding, provided mixed

results. The findings suggest that it is not enough to

have access to new knowledge; commitment and

significant experience in a knowledge domain are

also needed to generate innovations (Tripsas,

1997). The results on multiple creators suggest that

without a deep understanding by the participants,

variance may be increased at the cost of lower mean

performance. On the other hand, knowledge-build-

ing experience, often considered the bane of inno-

vation, was an important positive factor.

Our investigation provides an interesting obser-

vation on the difference between exploration and

exploitation. Four of the variables included in the

model increased variance: diverse knowledge

(cross-genre experience), team experience, previ-

ous innovation success, and number of creators

(though the latter had somewhat mixed effects).

The first three of these also increase the mean per-

formance. This finding suggests that although ex-

ploration and exploitation may be two different

processes, experience affects both positively. Ap-

plication of diverse knowledge is a variance-increas-

ing process, yet applying experience is necessary for

both exploration and exploitation. It may be that

knowledge-intensive activities are united by the need

for understanding but differ in the direction of the

effort. Knowledge combination is inherently difficult,

and it occurs most easily when a team has past expe-

rience working together. This finding is consistent

with accounts in the team formation literature of

stages of team activity in which much of the early

activity is focused on learning how to work together

rather than on the team task. It also suggests that the

dichotomy between exploration and exploitation at

the organizational level is driven not by the differing

knowledge assets used, but by the differing goals and

expectations for the tasks. An organization may trade

off exploitation and exploration when assigning a

research and development team the goal of making a

radical innovation versus improving an existing tech-

nology or product, but a team composition that gives

high performance on an exploration task also gives

high performance on an exploitation task. It is not the

selection of people that determines the degree of ex-

ploration, but what they are asked to do.

Our findings on mean-increasing variables are

generally weak, with many insignificant effects.

The typical experience effects appear weak or ab-

sent in the creative task of generating comic books;

these effects may exist but be absorbed by other

variables, such as past innovation experience. The

impact of organization also was not as expected, as

the results show that larger organizations have

poorer overall performance. Generalizing this re-

sult is problematic given the dominance of Marvel

and DC as publishers in the field we examined,

because it may be indicative of the negative impact

of these two organizations on the performance of

creators. This interpretation was suggested, prior to

the empirical analysis, by interviews with creators

who had worked both independently and with

larger organizations, and who felt that the con-

straints imposed by these organizations reduced

the quality of their work. The findings proved them

right.

The results also highlight the methodological

contribution of the study. By analyzing perfor-

mance levels, past innovation research has only

looked at the upper tail of the impact of undertak-

ing innovative behavior. Exploration is variance-

inducing, and that variance can yield both positive

and negative innovative outcomes. The empirical

results support this argument, as the conditions

that caused positively evaluated innovations also

caused negatively evaluated innovations. Studies

that examine only positive results of innovative

activities meld the results of variance- and mean-

increasing activities, and thus they fail to provide

insights on either one. In addition, explicitly mod-

eling the negative impact of exploration more ac-

curately allows empirical measurement of the risks

associated with innovation.

This research has some limitations that we hope

to address in future work. First, although the task of

producing comics involves technical skills such as

writing, layout, inking, and coloring, it does not use

technology in the same way as products that re-

quire engineering. Nor does this task occur under

the limitations for safe and ethical experimentation

that face, for example, medical teams developing

new procedures. In recognition of the possible role

of constraints on innovations, we chose not to an-

alyze comics from the censorship period, and we

believe findings from an uncensored period are

more generalizable. Censorship is a constraint that

operates directly on creative output, however, and

is thus different from the constraints on the creative

process imposed by technology and safety con-

cerns. Innovations in comics may be a relatively

unconstrained creative task, so additional research

is needed to establish whether technological,

safety, or other constraints would modify these

findings.

Second, an important theoretical issue is that of

partial team replacement. There is a difference be-

tween a team becoming effective as a result of past

interactions of the team as a whole and a team

becoming effective as a result of past dyadic inter-

actions of members in previous teams. Individual

members can be replaced without fear of worse

outcomes if the latter process is the more impor-

736 AugustAcademy of Management Journal

tant, but not if the former is. Unfortunately, these

data showed a dominant pattern of replacing all

members or none, so measures of dyadic and full-

team experience correlated too highly for these ef-

fects to be distinguished. A context in which partial

replacement is more common may be necessary to

address the difference between these processes of

team learning.

Third, the data may be susceptible to effects from

the historical period under investigation. In any

data based on individual career outcomes, it is

worthwhile checking for influences from specific

cohorts and periods. We found the distribution of

career cohorts and time periods of the creative

works to be quite even, and thus there were no

numerically influential cohorts or time periods in

the data. Like other cultural products, however,

comic books are available for other creators to

study and mimic, and the broad agreement on

which comics are most successful created a shared

history that influenced these creators. We believe

that the long duration of this study—1972 to

1996—provides reasonable insurance against

shared-history effects because the history available

to the creators changed substantially during this

period.

Future research should focus on innovations as

novelty and examine more closely how organiza-

tions create extreme outcomes. This project would

require research designs to identify all innovative

products, not just the successful ones. Such designs

will give more incisive analysis because they will

allow empirical tests of the theoretical distinctions

between mean-increasing and variance-increasing

influences on innovation, which may be important

for work on team diversity. The tension between

diverse teams containing more information and

having greater difficulty using it has persistently

troubled the diversity literature (Williams &

O’Reilly, 1998). It could be that a focus on average

outcomes has exacerbated these difficulties be-

cause the net effect of diversity is unclear when one

is measuring average performance, but its effect

should be increased variance in performance be-

cause of the contrast between diverse teams that

function well and those that do not.

More important, however, may be the suggestion

that the real driver of innovation is combining di-

verse knowledge. Our focus on the genre experi-

ence of creators was empirically very powerful, as

it supported our predictions on performance devi-

ance and gave an unexpected finding on perfor-

mance level. This pattern of findings indicates that

future research should focus on concrete measures

of the career experiences of team members rather

than on surface-level diversity. Just as comic cre-

ators get expertise from working in different genres,

members of product development teams gain ex-

pertise from working on products and technologies.

Future research should investigate how career ex-

periences yield expertise in specific knowledge do-

mains that affects the innovative performance of

teams.

These findings have important implications for

managers organizing new-product development

teams. Most important is the effect of diverse

knowledge on the production of innovations. Com-

bining knowledge domains is not just a strong lever

for generating variance; in our data it also raised the

level of performance. Clearly, our findings imply

that individuals’ career experiences should be con-

sidered when teams are assembled, and they also

imply that careers should be managed so that an

organization has a broad stock of knowledge to

choose from. One career management issue is

whether to encourage specialization or broad expe-

rience. We found that although combining knowl-

edge domains appeared to affect the level of perfor-

mance in teams as well as in individuals, we could

only prove the effect for individuals. Although we

cannot tell for sure with these data, it is reasonable

to infer that the process cost implied by this finding

means that a given set of knowledge domains will

be less efficiently combined the more persons are

needed to assemble the domains in a team. In prod-

uct development, specialization can be costly. In

addition, when managers staff new-product teams

with cross-functional and cross-knowledge indi-

viduals, it is essential that the included members

have deep understandings of their respective

knowledge domains. Finally, our results show that

teams that have previously worked together are

superior to newly assembled teams. These findings

suggest that when seeking innovation in knowl-

edge-based industries, it is best to find one “super”

individual. If no individual with the necessary

combination of diverse knowledge is available, one

should form a “fantastic” team, with each team

member having deep knowledge and experience

working with the other team members.

Finally, the findings suggest that managers do

not have to make a trade-off between exploration

and exploitation when assembling teams. The char-

acteristics that increased exploration (extreme out-

comes) also increased exploitation (higher level of

outcomes). It is not team composition, then, but

rather the task and context given to a team that

creates a trade-off between exploration and exploi-

tation in product development. It may be the pro-

cess of setting goals and prefiltering options that

results in performance differences, not a difference

2006 737Taylor and Greve

in the core number of new ideas or proposed

options.

We used fine-grained data to reveal processes

that have been implicitly assumed but not explic-

itly tested in the literature on innovation through

combining knowledge. Our findings are consistent

with, and extend, many of the results found in the

prior research on team innovation grounded in psy-

chology and organizational behavior. Uniquely, we

were able to measure the value of different creative

team configurations with a real commercial out-

come—the collector value of a comic. In addition to

documenting the role of knowledge combination in

creating innovations, this research also raises inter-

esting questions of a possible dampening effect of

large organizations on innovative behavior and per-

formance level. We expect further work on explo-

ration through variance-increasing activities to

yield additional new insights.

REFERENCES

Ahuja, G. 2000. Collaboration networks, structural holes,

and innovation: A longitudinal study. Administra-

tive Science Quarterly, 45: 425–455.

Allmendinger, J., & Hackman, J. R. 1996. Organizations in

changing environments: The case of East German

symphony orchestras. Administrative Science

Quarterly, 41: 337–369.

Amabile, T. M. 1982. Social psychology of creativity: A

consensual assessment technique. Journal of Per-

sonality and Social Psychology, 43: 997–1013.

Amabile, T. M. 1983. The social psychology of creativity:

A componential conceptualization. Journal of Per-

sonality and Social Psychology, 45: 357–376.

Amabile, T. M. 1996. Creativity in context. Boulder, CO:

Westview.

Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron,

M. 1996. Assessing the work environment for cre-

ativity. Academy of Management Journal, 39:

1154–1184.

Amburgey, T. L., & Miner, A. S. 1992. Strategic momen-

tum: The effects of repetitive, positional and contex-

tual momentum on merger activity. Strategic Man-

agement Journal, 13: 335–348.

Argote, L. 1999. Organizational learning: Creating, re-

taining, and transferring knowledge. Boston: Klu-

wer Academic.

Argote, L., Beckman, S. L., & Epple, D. 1990. The persis-

tence and transfer of learning in industrial settings.

Management Science, 36(2): 140 –154.

Arthur, W. B. 1989. Competing technologies, increasing

returns, and lock-in by historical events. Economic

Journal, 99: 116 –131.

Benner, M. J., & Tushman, M. L. 2002. Process manage-

ment and technological innovation: A longitudinal

study of the photography and paint industries. Ad-

ministrative Science Quarterly, 47: 676–706.

Bougan, M., Weick, K., & Binkhorst, D. 1977. Cognition

in organizations: An analysis of the Utrecht Jazz

Orchestra. Administrative Science Quarterly, 22:

606–639.

Brass, D. J. 1995. Creativity: It’s all in your social net-

work. In C. M. Ford & D. A. Gioia (Eds.), Creative

action in organizations: 94–99. Thousand Oaks,

CA: Sage.

Brown, J. S., & Duguid, P. 1991. Organizational learning

and communities of practice: Toward a unified view

of working, learning, and innovation. Organization

Science, 2: 40 –57.

Campbell, D. T. 1960. Blind variation and selective re-

tention in creative thought as in other knowledge

processes. Psychological Review, 67: 380– 400.

Chatman, J. A., Polzer, J. T., Barsade, S. G., & Neale, M. A.

1998. Being different yet feeling similar: The influ-

ence of demographic composition and organization-

al culture on work processes and outcomes. Admin-

istrative Science Quarterly, 43: 749–780.

Cohen, J. 1960. A coefficient of agreement for nominal

scales. Educational and Psychological Measure-

ment, 20: 37– 46.

Cohen, W. M., & Levinthal, D. A. 1990. Absorptive ca-

pacity: A new perspective on learning and innova-

tion. Administrative Science Quarterly, 35: 128 –

152.

De Dreu, C. K. W., & Weingart, L. R. 2003. Task versus

relationship conflict and team effectiveness: A meta-

analysis. Journal of Applied Psychology, 88: 741–

749.

De Dreu, C. K. W., & West, M. A. 2001. Minority dissent

and team innovation: The importance of participa-

tion in decision making. Journal of Applied Psy-

chology, 86: 1191–1201.

Drazin, R., Glynn, M. A., & Kazanjian, R. K. 1999. Mul-

tilevel theorizing about creativity in organizations: A

sensemaking perspective. Academy of Manage-

ment Review, 24: 286 –307.

Duin, S., & Richardson, M. 1998. Comics between the

panels. Milwaukie, OR: Dark Horse Comics.

Feist, G. J. 1999. Influence of personality on artistic and

scientific creativity. In R. J. Sternberg (Ed.), Hand-

book of creativity: 272–296. New York: Cambridge

University Press.

Fleming, L. 1999. Recombinant uncertainty in technolog-

ical search. Management Science, 47: 117–132.

Fleming, L., & Sorenson, O. 2001. Technology as a com-

plex adaptive system: Evidence from patent data.

Research Policy, 30: 1019 –1039.

Ford, C. M. 1996. A theory of individual creative action

738 AugustAcademy of Management Journal

in multiple social domains. Academy of Manage-

ment Review, 21: 1112–1142.

Gilson, L. L., & Shalley, C. E., 2004. A little creativity

goes a long way: An examination of teams’ engage-

ment in creative processes. Journal of Management,

30: 453– 470.

Gilson, L. L., Matheiu, J. E., Shalley, C. E., & Ruddy, T. M.

2005. Creativity and standardization: Complemen-

tary or conflicting drivers of team effectiveness?

Academy of Management Journal, 48: 521–531.

Greve, H. R. 1995. Jumping ship: The diffusion of strategy

abandonment. Administrative Science Quarterly,

40: 444 – 473.

Greve, H. R., & Taylor, A. 2000. Innovations as catalysts

for organizational change: Shifts in organizational

cognition and search. Administrative Science Quar-

terly, 45: 54 –80.

Gruenfeld, D. H., & Kim, P. H. 2004. Power and single-

mindedness in U.S. Supreme Court justices. Manu-

script, Graduate School of Business, Stanford Uni-

versity, Stanford, CA.

Hansen, M. T. 1999. The search-transfer problem: The

role of weak ties in sharing knowledge across organ-

ization subunits. Administrative Science Quarterly,

44: 82–111.

Hargadon, A., & Sutton, R. I. 1997. Technology brokering

and innovation in a product development firm. Ad-

ministrative Science Quarterly, 42: 716–749.

Harrison, D. A., Price, K. H., & Bell, M. P. 1998. Beyond

relational demography: Time and the effects of sur-

face- and deep-level diversity on work group cohe-

sion. Academy of Management Journal, 41: 96 –

107.

Harrison, D. A., Price, K. H., Gavin, J. H., & Florey, A. T.

2002. Time, teams, and task performance: Changing

effects of surface- and deep-level diversity on group

functioning. Academy of Management Journal, 45:

1029–1045.

Hays, J. 2005. Superman’s Busiek & Johns on one year

later: “Can’t tell you!” (Silverbullet Comic Books

Interviews). http://www.silverbulletcomicbooks.com/

features/113696176268529.html.

Herriott, S. R., Levinthal, D., & March, J. G. 1985. Learn-

ing from experience in organizations. American

Economic Review, 75: 298 –302.

Hirsch, P. M. 1972. Processing fads and fashions: An

organization-set analysis of cultural industry sys-

tems. American Journal of Sociology, 77: 639– 659.

Jackson, S. E. 1996. The consequences of diversity in

multi-disciplinary work teams. In M. A. West (Ed.),

Handbook of work group psychology: 53–76.

Chichester, U.K.: Wiley.

Jehn, K. A., Northcraft, G. B., & Neale, M. A. 1999. Why

differences make a difference: A field study of diver-

sity, conflict, and performance in work groups. Ad-

ministrative Science Quarterly, 44: 741–763.

Katz, K., 1982. The effects of group longevity on project

communication and performance. Administrative

Science Quarterly, 27: 81–104.

Kurtzberg, T. R., & Amabile, T. M. 2001. From Guilford to

creative synergy: Opening the black box of team

level creativity. Creativity Research Journal, 13:

285–294.

Landis, J. R., & Koch G. G. 1977. The measurement of

observer agreement for categorical data. Biometrics,

33: 159 –174.

Lee, L. F. 1983. Generalized econometric models with

selectivity. Econometrica, 51: 507–512.

Levitt, B., & March, J. G. 1988. Organizational learning. In

W. R. Scott (Ed.), Annual review of sociology, vol.

14: 319 –340. Palo Alto, CA: Annual Reviews.

Levitt, B., & Nass, C. 1989. The lid on the garbage can:

Institutional constraints on decision making in the

technical core of college-text publishers. Adminis-

trative Science Quarterly, 34: 190–207.

Lyles, M. A., & Schwenk, C. R. 1992. Top management,

strategy and organizational knowledge structures.

Journal of Management Studies, 29: 155–174.

March, J. G. 1991. Exploration and exploitation in organ-

izational learning. Organization Science, 2: 71– 87.

March, J. G., & Simon, H. 1958. Organizations. New

York: Wiley.

Marks, T. 2004. Chapter one: What is a genre (Genre

Criticisms, Spare Parts Comics). http://www.

sparepartscomics.com/genre/genre1.html.

McCloud, S. 1993. Understanding comics. New York:

HarperCollins Publishers (by permission of Kitchen

Sink Press).

Mezias, J. M., & Mezias, S. J. 2000. Resource partitioning,

the founding of specialist firms, and innovation: The

American feature film industry, 1912–1929. Organi-

zation Science, 11: 306 –322.

Miller, D., & Shamsie, J. 1996. The resource-based view

of the firm in two environments: The Hollywood

film studios from 1936 to 1965. Academy of Man-

agement Journal, 39: 519 –543.

Milliken, F. J., Bartel, C. A., & Kurtzberg, T. 2003. Diver-

sity and creativity in work groups: A dynamic per-

spective on the affective and cognitive processes that

link diversity and performance. In P. Paulus & B.

Nijstad (Eds.), Group creativity: 32– 62. New York:

Oxford University Press.

Morgan, D. N. 1953. Creativity today: A constructive

analytic review of certain philosophical and psycho-

logical Work. Journal of Aesthetics and Art Criti-

cism, 12: 1–24.

Nemeth, C. J., & Rogers, J. 1996. Dissent and the search

for information. British Journal of Social Psychol-

ogy, 35: 67–76.

Nijstad, B. A., & De Dreu, C. K. W. 2002. Creativity and

2006 739Taylor and Greve

group innovation. Applied Psychology: An Interna-

tional Review, 51: 400 – 406.

Nisbett, R. E., & Ross, L. 1980. Human inference: Strat-

egies and shortcomings of social judgment. Engle-

wood Cliffs, NJ: Prentice-Hall.

Nonaka, I., & Takeuchi, H. 1995. The knowledge-creat-

ing company: How Japanese companies create the

dynamics of innovation. New York: Oxford Univer-

sity Press.

Parnes, S. J., & Noller, R. B. 1972. Applied creativity: The

creative studies project–Part II: Result of the two-

year program. Journal of Creative Behavior, 6: 164–

186.

Paulus, P. B. 2002. Different ponds for different fish: A

contrasting perspective on team innovation. Applied

Psychology: An International Review, 51:

394–399.

Pelled, L. H., Eisenhardt, K. M., & Xin, K. R. 1999. Ex-

ploring the black box: An analysis of work group

diversity, conflict, and performance. Administrative

Science Quarterly, 44: 1–28.

Perry-Smith, J. E., & Shalley, C. E. 2003. The social side of

creativity: A static and dynamic social network per-

spective. Academy of Management Review, 28:

89–106.

Pfeffer, J. 1983. Organizational demography. In L. L.

Cummings & B. M. Staw (Eds.), Research in organ-

izational behavior, vol. 5: 298 –357. Greenwich, CT:

JAI Press.

Powell, W. W., Koput, K. W., & Smith-Doerr, L. 1996.

Interorganizational collaborations and the locus of

innovation: Networks of learning in biotechnology.

Administrative Science Quarterly, 41: 116–145.

Sassiene, P. 1994. The comic book. London: Ebury Press

Limited.

Schumpeter, J. A. 1976. Capitalism, socialism, and de-

mocracy (3rd ed.). New York: Harper & Row.

Simon, H. A. 1986. Rationality in psychology and eco-

nomics. Journal of Business, 59(4): S209 –S224.

Simonton, D. K. 1999a. Origins of genius. New York:

Oxford University Press.

Simonton, D. K. 1999b. Creativity as blind variation and

selective retention: Is the creative process Darwin-

ian? Psychological Inquiry, 10: 309–328.

Steiner, I. D. 1972. Group processes and productivity.

New York: Academic Press.

Sternberg, R. J., & Lubart, T. I. 1995. Defying the crowd:

Cultivating creativity in a culture of conformity.

New York: Free Press

Sternberg, R. J., & O’Hara, L. A. 2000. Intelligence and

creativity. In R. J. Sternberg (Ed.), Handbook of in-

telligence: 609– 628. New York: Cambridge Univer-

sity Press.

Tesluk, P. E., Farr, J. L., & Klein, S. R. 1997. Influences of

organizational culture and climate on individual cre-

ativity. Journal of Creative Behavior, 31: 27– 41.

Thornton, P., & Ocasio, W. 1999. Institutional logics and

the historical contingency of power in organizations:

Executive succession in the higher education pub-

lishing industry, 1958–1990. American Journal of

Sociology, 105: 801– 843.

Tripsas, M. 1997. Surviving radical technological change

through dynamic capability: Evidence from the type-

setter industry. Industrial and Corporate Change,

6: 341–377.

Van Dyne, L., & Saavedra, R. 1996. A naturalistic minor-

ity influence experiment: Effects on divergent think-

ing, conflict, and originality in work-groups. British

Journal of Social Psychology, 35: 151–172.

Von Hippel, E. 1988. Sources of innovation. New York:

Oxford University Press.

Weisberg, R. W. 1999. Creativity and knowledge: A chal-

lenge of theories. In R. J. Sternberg & L. A. O’Hara

(Eds.), Handbook of creativity: 226 –250. Cam-

bridge, U.K.: Cambridge University Press.

West, M. A. 2002. Sparkling fountains or stagnant ponds:

An integrative model of creativity and innovation

implementation in work groups. Applied Psychol-

ogy: An International Review, 51: 355– 424.

West, M. A., & Anderson, N. R. 1996. Innovation in top

management teams. Journal of Applied Psychology,

81: 680 – 693.

Williams, K. Y., & O’Reilly, C. A. 1998. Demography and

diversity in organizations: A review of 40 years of

research. In B. M. Staw & R. I. Sutton (Eds.), Re-

search in organizational behavior, vol. 20: 77–140.

Greenwich, CT: JAI Press.

Wyatt, E. 2006. Publishers find growth in comics. New

York Times, February 13: C1.

Alva H. Taylor (alva.taylor@dartmouth.edu) is an asso-

ciate professor in the strategy group at the Tuck School of

Business at Dartmouth. He received his Ph.D. in business

from Stanford University. His research examines the in-

ternal processes and cognitive aspects of innovation, or-

ganizational learning from competition, and managerial

risk taking.

Henrich R. Greve (henrich.greve@bi.no) is a professor of

strategy at the Norwegian School of Management BI. He

received his Ph.D. in business from Stanford University.

His research examines organizational learning from net-

works and from performance feedback, and determinants

of innovations in new and existing organizations.

740 AugustAcademy of Management Journal

Knowledge diversity and technological innovation: The moderating role of top management teams

Article

Full-text available

Mar 2024
TECHNOVATION

Thriving in future work: knowledge management and innovation perspectives

Article

Full-text available

Apr 2024
Knowl Manag Res Pract

The future of work and business is undergoing rapid transformation, driven by dynamic shifts in technology, the economy, climate, and demographics. Scholars emphasise the necessity for proactive adaptation to these changes to ensure the thriving of both individuals and organisations in future business landscapes. Recent research has increasingly underscored the crucial role of knowledge in fostering both individual growth and organisational innovation. Particularly, innovative knowledge is deemed essential for future success, as it can enable individuals and organisations to navigate complex environments effectively. However, gaps persist in understanding how knowledge, especially innovative knowledge, can be harnessed effectively to benefit stakeholders. This raises the need for deeper insights into its implications for individual and organisational thriving in the future of work. In this piece of writing, we showcase a series of studies that aim to theoretically and empirically address the intersection of knowledge, innovation, technology, thriving, and careers. By doing so, our intention is to ignite continued scholarly interest in enriching our understanding of how to facilitate thriving and effective individuals and organisations through a knowledge-driven approach, and consequently to spark new ways of thinking to tackle future challenges.

Unlocking Creativity: The Impact of Inventors’ Knowledge Complementarity and Substitutability in Moderating Structural Holes

Article

Full-text available

Feb 2024

This study delves into the intricate dynamics of network structures and their influence on creativity within innovation networks, emphasizing the pivotal role of knowledge diversity. It introduces the novel concept of “network content view,” a perspective focusing on the knowledge characteristics of network actors, namely knowledge complementarity and substitutability. Our research posits that these knowledge dimensions critically modulate the relationship between structural holes and creative output. We assert that both knowledge complementarity and substitutability enrich the advantages provided by structural holes in networks. Complementarity enhances the creative process by facilitating the integration of disparate knowledge domains, while substitutability ensures efficient assimilation and application of similar knowledge elements. The study’s empirical analysis, utilizing data from 33 pharmaceutical companies, employs a negative binomial model to explore these relationships. Our findings reveal that knowledge complementarity and substitutability significantly bolster the creative benefits derived from structural holes. However, the balance and interaction between these knowledge types optimally leverage the potential of structural gaps for fostering creativity. The research contributes to the existing literature by highlighting the importance of network content alongside structure in stimulating innovation. Practically, this study guides managers and inventors in capitalizing on network positions and knowledge attributes to cultivate a fertile environment for creativity. It underscores the need for strategic management of knowledge resources within networks, aligning with the imperatives of the knowledge economy. Thus, this research enhances theoretical understanding and offers practical insights for fostering innovation in information-intensive settings like the pharmaceutical industry.

Exaptation within structure: the challenge of external resources and risk-taking for public R&D institutions

Preprint

Full-text available

May 2024

Giulio Toscani

The Barcelona Supercomputing Center (BSC) emerges as a prominent research & development (R&D) entity in high-performance computing, facilitating interdisciplinary collaborations that integrate computational methodologies with empirical research. Through interviews with 20 BSC experts, this study elucidates the organizational dynamics of publicly funded institutions. It reveals an internal exaptation, an adaptation as pivotal mechanism, and the requisite for external exaptation to address challenges related to recruitment alignment with productivity metrics, budget constraints, and reluctance to assume leadership roles. The BSC employs exaptation tactics, facing obstacles in enlarging its workforce and bolstering support services; nonetheless, BSC's fluid environment mandates a willingness to assume risks in order to enact pioneering resolutions for enduring expansion and effective personnel administration.

Building effective inside-outside sales rep dyads: A collaboration perspective

Article

Jul 2023

Business-to-business (B2B) firms regularly engage in collaborative selling where an outside sales (OS) representative (rep) interfaces with customers and an inside sales (IS) rep supports the OS reps through remote selling. While anecdotal evidence abounds, there is little empirical research examining factors driving successful IS-OS co-selling, as evidenced by objective sales performance. The authors use an organizational behavior lens and theories of shared team experiences and member knowledge diversity to posit that the collaboration experience (length and intensity) and product knowledge diversity affect customer sales outcomes. Further, they unpack how the efficacy of dyadic attributes should be contingent on the length of the customer–firm relationship and the customer’s product need complexity. Using field data and identification strategies suitable to their setting, the authors confirm that an IS–OS dyad’s collaboration experience and product knowledge diversity have a positive effect on customer-level sales outcome. However, they find a nuanced interplay of IS-OS dyadic attributes with customer characteristics. For example, customers with more complex product needs buy more when dyadic collaboration is long or intensive but not when the dyad possesses diverse product knowledge. The conceptual framework and empirical results together enable sales managers to match IS-OS dyads to customers they can serve profitably.

The synergistic effect of corporate environmental responsibility and technological base on firm performance: From the perspective of knowledge recombination

Article

Apr 2024

The research constructs a theoretical framework including the relationship between the interaction of corporate environmental responsibility and technological base and firm performance and the mediating role of knowledge recombination. The research empirically studies the proposed framework by using the data of 258 firms listed on the Shanghai and Shenzhen Stock Exchanges from 2008 to 2019. The results show that the interaction between corporate environmental responsibility and technological base has a positive effect on firm performance and both the two dimensions of knowledge recombination that are knowledge recombination reuse and knowledge recombination creation could mediate this relationship. These results not only figure out the interaction between corporate environmental responsibility and technological base and its influencing mechanisms on firm performance, which is knowledge recombination and further broaden the application of the knowledge‐based view, but also provide important managerial implications for firms to acquire superior performance through corporate environmental responsibility and technological base.

How geographic diversity and collaborative breadth prevent knowledge leakage during open innovation processes

Article

Jun 2023
J Knowl Manag

Purpose Drawing upon insights from knowledge-based theory and the learning perspective, this study aims to explore safeguarding strategies in open innovation. Geographic diversity and collaborative breadth can effectively protect proprietary innovations that limit knowledge leakage concerns. Design/methodology/approach Using a cross-industry sample from the Taiwanese Technological Innovation Survey III, which covered 1,519 firms, the authors investigate the conditions under which partnership portfolios affect radical innovation. Findings The findings suggest that the partnership portfolio has an inverted U-shaped influence on radical innovation and that this relationship is moderated by geographic diversity and collaborative breadth. This work identifies a balance in the tension between diverse partnership portfolios and knowledge leakage with regard to open innovation activities. Practical implications This study provides senior managers with an indication of the relationships between partnership portfolios and innovative knowledge protection, identifying the geographic diversity and collaborative breadth that serve as safeguards to prevent leakages of a firm’s innovative knowledge. Originality/value This study makes an original contribution to the empirical exploration of innovation knowledge protection and provides new insights into the field of open innovation. The authors, thus, balance the tension between partnership portfolios and knowledge leakage.

Affiliation within structure: the challenge of resources for public R&D institutions like the Barcelona supercomputing center

Preprint

Full-text available

Mar 2024

Giulio Toscani

The Barcelona Supercomputing Center (BSC) emerges as a prominent research & development (R&D) entity in high-performance computing, facilitating interdisciplinary collaborations that integrate computational methodologies with empirical research. Through interviews with 20 BSC experts, this study elucidates the organizational dynamics of publicly funded institutions. It reveals an internal exaptation affiliation and the requisite for external exaptation to address challenges related to recruitment alignment with productivity metrics, budget constraints, and reluctance to assume leadership roles. The BSC employs exaptation strategies, encountering challenges in workforce expansion and support services. The organization's dynamic context necessitates innovative solutions for sustainable growth and efficient talent management.

The Influence of Entrepreneurial Leadership and Team Creativity. A Systematic Literature Review. Faculty of Management at the Seeburg Castle University

Thesis

Full-text available

Mar 2024

Amila Balic

Entrepreneurial leadership is a well-known leadership style that sets itself apart from others due to its exceptional qualities. The significant interest in entrepreneurial leadership is evident from the multitude of articles written on the subject. The focus of this literature review is to examine what is already known about the impact of entrepreneurial leadership on team creativity, how it can be fostered by entrepreneurial leadership, what challenges may arise and how they can be overcome, and finally, what areas need further research.

The impact of managerial autonomy and founding-team marketing capabilities on the relationship between ambidexterity and innovation performance

Article

Feb 2024

Creativity: It's all in Your Social Network

Chapter

Jan 1995

Daniel Brass

A Coefficient of Agreement for Nominal Scales

Article

Apr 1960
EDUC PSYCHOL MEAS

Jacob Cohen

Rationality in Psychology and Economics

Article

Jan 1987

H.A. Simon

TASK VERSUS RELATIONSHIP CONFLICT: A META-ANALYSIS.

Article

Aug 2002

This study provides a meta-analysis of research on the associations between relationship conflict, task conflict, team performance, and team member satisfaction. Consistent with past theorizing, results reveal strong and negative correlations between relationship conflict and team performance, and team member satisfaction. In contrast to what has been suggested in both academic research and introductory textbooks, however, results also reveal strong and negative (instead of the predicted positive) correlations between task conflict and team performance, and team member satisfaction. Some of the variation in study findings could be attributed to task type, such that conflict had stronger negative relations with team performance in highly complex (decision-making, project, mixed) than in less complex (production) tasks. In addition, some of the variation could be attributed cross-cultural differences, and study characteristics. Implications are discussed.

Understanding Comics

Article

Jan 1994

Scott McCloud

Exploration and Exploitation in Organizational Learning

Article

Jan 1991

J.G. March

Institutional logics and the historical contingency of power in organizations: Executive succession in the higher education publishing industry, 1958-1990

Article

Nov 1999

This article examines the historical contingency of executive power and succession in the higher education publishing industry. We combine interview data with historical analysis to identify how institutional logics changed from an editorial to a market focus. Event history models are used to test for differences in the effects of these two institutional logics on the positional, relational, and economic determinants of executive succession. The quantitative findings indicate that a shift in logics led to different determinants of executive succession. Under an editorial logic, executive attention is directed to author-editor relationships and internal growth, and executive succession is determined by organization size and structure. Under a market logic, executive attention is directed to issues of resource competition and acquisition growth, and executive succession is determined by the product market and the market for corporate control.

A Theory of Individual Creative Action in Multiple Social Domains

Article

Oct 1996

Cameron M. Ford

Creative and habitual actions represent competing behavioral options that may be simultaneously influenced by multiple domains of social action. This article integrates psychological and sociological descriptions of creativity and conformity to present a theory of individual creative action within organizational settings composed of intertwined group, organizational, institutional, and market domains. This theory contributes to the innovation literature by illustrating how intentional action and evolutionary processes that legitimize action interact to facilitate creativity and innovation.

Beyond relational demography: Time and the effects of surface- and deep-level diversity on work group cohesion

Article

Feb 1998

We examined the impact of surface-level (demographic) and deep-level (attitudinal) diversity on group social integration. As hypothesized, the length of time group members worked together weakened the effects of surface-level diversity and strengthened the effects of deep-level diversity as group members had the opportunity to engage in meaningful interactions.

Applied Creativity: The Creative Studies Project -- Part II

Article

Jan 1972

Superman or the Fantastic Four? Knowledge Combination and Experience in Innovative Teams

Abstract and Figures

Recommended publications

Creative and Innovative Group Effort in Reducing High Failure Rate in Mathematics

Visualisation to Support the Configuring of Mechanical Fasteners

Ways of Knowing: A Nurse and a Physician Discuss Clinical Decisions, Actions, and Lessons

The Corporate Culture Myth