Rijksuniversiteit Groningen, The Netherlands
Faculty of Economics and Business
Date of Submission:
June 20
th, 2016
Collaboration Experience:
The Impact of Team Experience & Lone Wolves in R&D Projects
Master's Thesis
Author:
Carolin Laura Krause (S3001857)
MSc Business Administration
Strategic Innovation Management
Abstract
Table of Content
1 INTRODUCTION ... 4
2 THEORY AND BACKGROUND ... 8
3 HYPOTHESES ... 11
3.1 GENERAL TEAM EXPERIENCE ... 11
3.2 SPECIFIC TEAM EXPERIENCE ... 13
3.3 GENERAL & SPECIFIC TEAM EXPERIENCE ... 15
3.4 GENERAL TEAM EXPERIENCE AND SINGLE INVENTOR ... 15
3.5 LONE WOLF EXPERIENCE ... 16
4 METHODOLOGY ... 18
4.1 SAMPLE AND DATA ... 18
4.2 MEASUREMENTS... 19
4.2.1 Dependent variable: inventive performance... 19
4.2.2 Independent variables ... 20 4.2.3 Control variables ... 21 4.3 ANALYSIS ... 22 5 RESULTS ... 23 5.1 DESCRIPTIVE STATISTICS ... 23 5.2 HYPOTHESES TESTING ... 25 5.3 ROBUSTNESS ... 28
6 DISCUSSION AND CONCLUSIONS ... 31
1 Introduction
Continuous innovation has become essential for a firm’s long-term success in order to maintain competitive advantage. Therefore, firms increasingly focus on how to organize in-novation processes effectively in order to generate more successful creative ideas (Amabile, 1996; Woodman, Sawyer, & Griffin, 1993). This leads to the question whether teams or sin-gle inventors contribute more to a highly innovative outcome. Wuchty, Jones and Uzzi (2007) conducted an intensive research on team versus solo authors and observed that teams outper-form solo authors in generating knowledge. Their findings are in line with multiple studies from collaboration research emphasizing that inventors working together in teams perform better by creating more breakthrough inventions than lone inventors (e.g. Girotra, Terwiesch & Ulrich, 2007; Singh & Fleming, 2010; Taylor & Greve, 2006). The main underlying course is that teams are more successful due to their diverse knowledge pool and evaluation process-es for ideas (Reagans, Argote, & Brooks, 2005). On the contrary, team collaboration also in-corporates risks of miscommunication, social conflicts and could hinder selecting truly crea-tive ideas by blocking them (Taylor & Greve, 2006; Paulus & Nijstad, 2003). However, it is argued that profitable teamwork is highly influenced by an individual’s experience in collabo-rative and single R&D projects (e.g. Singh & Fleming, 2010; Taylor & Greve, 2006; Reagans et al., 2005). Especially experience deriving from collaboration varies, since inventors form distinct experiences based on the number of projects and social interactions with other team members involved.
Although research in team collaboration for innovation has intensely analysed the im-pact of social networks on performance with respect to trust, shared understanding, communi-cation, task conflict and its potential costs (Onal Vural, Dahlander, & George, 2013; West, 2003; Taylor & Greve, 2006), the role of distinct types of experience in the context of innova-tion received limited atteninnova-tion. Collaborainnova-tion experience shapes the inventor’s future social behaviour by establishing social bonds that effect communication and coordination within the team. Furthermore, it amplifies the shared understanding for a given task as the inventors learn about each other’s knowledge and develop the same technical jargon (Onal Vural et al., 2013).
Flem-ing, 2010). To our knowledge, so far no research has addressed the impact of previous team experience on a single inventor’s performance in future. This leads to the question, how the sudden absence of social bonds, task coordination and idea evaluation influences an individu-al’s ability to generate novel ideas. On the contrary, some individuals are predominately used to working alone. They have developed their own work routines and differ significantly from team inventors due to their unwillingness to collaborate in teams and to form social bonds (Dixon, Gassenheimer & Feldman Barr, 2003). Prior literature refers to this particular type of single inventor as ‘lone wolves’. In sum, each individual inventor is influenced by a set of distinct experiences from prior joint or lone project involvements. Moreover, experiences are shaped by previous social interactions with other team members or by the inventor’s prefer-ence to work alone. Thus, it is crucial to distinguish between distinct types of experiprefer-ence to evaluate their impact on the innovation process.
A third limitation of literature of experience in innovation concerns the inventors’ di-verse knowledge backgrounds. Various scholars confirmed that knowledge transfer increases productivity and leads to an organization’s long-term survival (e.g. Argote, Beckman, & Ep-ple, 1990; Argote & Ingram, 2000; Tsai, 2001). However, studies that focus on the transfer of diverse knowledge experience are scarce. One example is the research by Onal Vural et al. (2013), who explored the inventor’s scientific proximity that describes the technological background diversity based on previous experience between two inventors. However, the re-sults of the present literature do not account for experience with diverse team members or for the relationships between distinct forms of experience that affect the inventor’s ability to deal with knowledge diversity.
Finally, the literature on repeated collaboration experience with the same team mem-bers was analysed by various scholars, who investigated the positive effects of communica-tion and task coordinacommunica-tion advantages (e.g. Reagans et al., 2005; Liang, Moreland & Argote, 1995). Yet, repeated collaboration also involves risks such as established routines and de-creasing communication (e.g. Argote & Miron-Spektor, 2011; Inoue & Liu, 2015; Katz, 1982; Levitt & March, 1988; Onal Vural et al., 2013). Especially innovation projects require novel input that cannot be attained when sticking to old routines. Thus, it is important to investigate what kind of experience fosters the negative effects of routines in an innovation driven firm.
experi-creation of knowledge and innovations (Wuchty et al., 2007), empirical evidence on how to organize teams based on their distinct history of collaboration experience remains an open question. Answering this question and filling the presented research gaps could improve in-ternal workflows to transfer knowledge and expertise across the organization, resulting in more profitable innovations (Argote & Miron-Spektor, 2011; Argote & Kane, 2003; Singh & Fleming, 2010).Thus, this research aims to contribute to the literature on organizing for inno-vation by conducting an extensive empirical research on the team level. In addition, this study also addresses a novel field of research – namely lone inventors. Therefore, this study ad-dresses two sides of single projects. First, the impact of team collaboration experience on fu-ture projects when working alone. And in contrast, we examine the effect of ‘lone wolf expe-rience’ on collaborative projects. Lone wolves in joint R&D projects play a crucial role due to team member’s contrasting social behaviour that could lead to conflicts (Feldman Barr, Dix-on, & Gassenheimer, 2005).
In order to enhance the understanding of the impact of distinct team experience on a team’s inventive performance, this study distinguishes between the following three forms of experience: 1) General team experience highlights the quantity of an individual’s involve-ment in team projects and previous experience with other team members. 2) Specific
experi-ence is defined as repeated collaborations between two individual inventors working together
as a pair or in a team. 3) Lone wolf experience refers to the influence of single inventors in collaborations.
wolves have a negative influence on team performance. This prediction builds on literature on lone wolves social behaviour that is determined by the unwillingness to form social bonds and value contributions of others (Dixon et al., 2003).
The results of the empirical analysis suggest that distinct types of collaboration experi-ence impact the performance of R&D projects. However, the findings significantly differed from the given predictions. We unexpected find that both general and team specific experi-ence have a curvilinear relationship with invention performance, thus only providing support for our second hypothesis. Third, the interaction effect between both team experiences was partially supported, but presented a negative effect on innovation performance. Fourth, the results showed no significant influence of general experience on the inventor’s performance in single projects. Finally, lone wolf experience positively affected innovation performance in collaboration. Against the prediction, this result provides truly novel contributions to research on lone wolves’ behaviour, indicating that experience in single projects can be beneficial for team collaboration.
2 Theory and Background
In order to understand the impact of team experience on innovation, this study relies on two complementary theoretical bases – namely social network perspective and collabora-tion experience. From the social network perspective, team collaboracollabora-tion for innovacollabora-tion de-mands knowledge, skills and abilities, which differ from an individual’s in terms of commu-nication, coordination and conflict solving skills (Stevens & Campion, 1994). It involves sev-eral underlying social factors for applying the knowledge effectively, such as trust, shared understandings and norms. Working collectively on an innovation project requires common team mental models. This implies that “[…] members share an understanding of the nature of the group's task, its task processes, how members are required to work together, and the or-ganizational context “(West, 2003, p.29). Those skills promote teamwork and indicate how much an individual will benefit from collaboration (West, 2003).
However, multiple individuals working together also implies collaboration costs aris-ing from social conflicts, shararis-ing information or miscommunication that emerge when knowledge is transferred between two parties (e.g. Paulus & Brown, 2003; Singh & Fleming, 2010). According to Onal Vural et al. (2013, p.123), “successful collaborations occur when the benefits from collaboration outweigh the costs of coordination that team members face.” Working in diverse teams compromises the risks of miscommunication since each individual possesses knowledge from distinct specialized domains. Highly-specialized technological knowledge, for example, might involve a particular technical jargon, which not all the mem-bers can refer to (Maznevski, 1994). Nevertheless, various studies emphasized the overall benefits of broader knowledge domains and collaboration in teams (e.g. Fleming, 2001; West, 2003; Taylor & Greve, 2006).
any social interaction to other experts. The results suggested that the ‘selection phase’ – shar-ing and evaluatshar-ing ideas with others – is crucial for generatshar-ing breakthrough inventions and thus, lone wolves will have more difficulties inventing truly novel outcomes than teams.
The second literature field of interest collaboration experience is defined as a single inventor’s prior interactions with novel tasks involving one or more team members (Argote & Miron-Spektor, 2011). According to Onal Vural et al. (2013), team experience emerges in two major advantages: coordination benefits through information and shared resources. The for-mer arise from enhanced communication and routines within the team. Experience can lead to the establishment of routines and processes that result in more efficient workflows and man-agement control (Argote & Kane, 2003; Porter, 1979). Routines are defined as a sequence of actions, executed by an individual or organization, so that this action becomes reproducible (Nelson & Winter, 1982).
Besides coordination benefits, shared resources such as knowledge are ingrained in the previous experience of all team members and result in a broader set of diverse expertise do-mains (Onal Vural et al., 2013; Taylor & Greve, 2006). Argote and Miron-Spektor (2011, p.1125) concluded that “[…] knowledge can be embedded in the active context of members, tools, and tasks and their networks” as well as in a firm’s culture (Weber & Camerer, 2003). Social interactions with other employees allow knowledge transfers that occur when individu-als or groups are influenced by the experience of other members (Argote & Ingram, 2000; Argote & Miron-Spektor, 2011). Knowledge transfer between units is greatly enhanced by the social ties between the organisation’s employees, as well as by communication (Levine, Hig-gins, & Choi, 2000; Stasser, Vaughan, & Stewart, 2000) and employee rotation (Almeida & Kogut, 1999). Additionally, inventor teams not only profit from sharing information across the organisation, but also from observing others while working (Reagans et al., 2005). Thus, inventors gain access to knowledge components from diverse expertise domains that can complement their own specific knowledge base. And in return, recombining those knowledge components can lead to a competitive advantage by creating novel, valuable inventions (e.g. Katila & Ahuja, 2002; Nerkar, 2003).
3 Hypotheses
This study follows the general proposition that three main types of experience impact collaboration experience: general team experience, specific team experience and lone wolf experience. The given dimensions interact with each other and shape a team’s or single inven-tor’s knowledge, skills as well as abilities in collaboration. The joint experiences of all team members lead to the innovation outcome – namely R&D project performance. Furthermore, we also investigate the impact of collaboration experience on a single inventor’s project per-formance.
3.1 General Team Experience
General team experience refers to the individual’s history of collaboration in teams
(Reagans et al., 2005) as well as the inventor’s experience with diverse team members. High experience in a certain task deepens the inventor’s specialized knowledge, which promotes learning by exploiting knowledge combination possibilities across projects (Dibiaggio, Na-siriyar, & Nesta, 2014). Performing the same task repeatedly leads to greater learning curves and levels of experimentation with diverse solutions, which in return decrease the time to per-form a certain task and reduce task complexity (Argote & Miron-Spektor, 2011; Argote & Kane, 2003; Reagans et al., 2005). The challenges to profit from experiences are to capture the attained knowledge and procedures, since it is not always obvious which previous experi-ence attributes relate to the current task (Argote & Kane, 2003).
tential outcomes. Thus, a moderate level of task conflict increases creativity and results in a more profound decision making process in a rather cooperative than competitive group envi-ronment (De Dreu & Weingart, 2003; Tjosvold, 1998). Furthermore, a moderate level of task challenge strengthens team members’ ambition to look for novel solutions (Amabile, Conti, Coon, Lazenby, & Herron, 1996; Onal Vural et al., 2013). This can lead to the creation of more unusual, truly novel ideas by challenging the status quo (Amabile et al., 1996). Thus, collaboration requires a work climate that encourages employees to experiment with new ide-as by providing autonomy and meaningful tide-asks (Onal Vural et al., 2013).
From the collaboration experience perspective, in this study general team experience includes also the inventor’s experience with diverse team members. Experience in team pro-jects fosters sharing novel knowledge and idea generation that is promoted by an increasing number of team members with a diverse professional background (Argote & Kane, 2003). Knowledge diversity is defined as cognitive diversity, which derives from task, field and or-ganizational experiences (Taylor & Greve, 2006). Thus, with each new team member
experi-ence, the inventor is able to extend their expertise across various diverse knowledge domains
(Argote & Miron-Spektor, 2011; Onal Vural et al., 2013). Previous literature points out that diverse experience of individual team members is important to minimize the risks of errors and promote learning (Argote & Miron-Spektor, 2011; Haunschild & Sullivan, 2002; Singh & Fleming, 2010). Yet, too high levels of knowledge diversity implies higher uncertainty about the optimal component combination and knowledge integration costs, which are linked to the required social mechanisms such as sharing information or conflicts (Katila & Ahuja, 2002; Taylor & Greve, 2006).
Wang, 2015; Taylor & Greve, 2006). In conclusion, inventors enhance their ability to apply their knowledge in collaborations due to an increasing general team experience.
Hypothesis 1: R&D project performance is positively influenced by general team experience.
3.2 Specific Team Experience
Although general team experience indicates with how many distinct team members inventors have worked together over time, the concept of specific team experience determines the repeated collaboration relationship between two inventors. These experiences differ great-ly from general experience since they amplify distinct social mechanisms and thus, result in a divergent set of collaboration experience. Regarding the social mechanisms of specific team experience, repeated collaboration improves the communication skills and promotes trust within the team. Forming social ties by increasing trust simulates the knowledge and infor-mation exchange between the team members (Weber & Cramerer, 2003; Reagans et al., 2005). Furthermore, team members are getting more confident to question ideas in terms of the presented task conflict (Mayer, Davis & Schoorman, 1995) and develop the same tech-nical jargon amplifying shared understanding (Maznevski, 1994).
However, beyond a certain point increased repeated team member experience has a negative impact when a team member leaves the group or is absent, causing difficulties to align the tasks (Argote & Kane, 2003). Reliance on previously established task division can harm the team’s project performance as well as an overemphasis on established routines that impede finding novel ideas due to less task conflict, search for familiar ideas and decreasing external communication with other organizational members (Katz, 1982; Skilton & Dooley, 2010). While routines can be beneficial when applying known problem-solving methods, they can also limit the search process for finding truly new combinations when applied in the wrong context (Argote & Miron-Spektor, 2011; Dibiaggio et al., 2014; Rodan & Galunic, 1998). This process is called superstitious learning, which “[…] occurs when the subjective experience of learning is compelling, but the connections between actions and outcomes are misspecified” (Levitt & March, 1988, p. 325). For example, a team’s commitment to prob-lem-solving routines that rely on previous successful collaboration experience might create a misleading impression of the right solution (Levitt & March, 1988). In contrast to general team experience, jointly established routines gain a greater importance in specific team expe-rience because of the stronger relational ties based on higher levels of trust between the two inventors. However, too close social bonds enhance the risk of reduced monitoring of the oth-er’s behaviour due to enhanced trust in their abilities (Langfred, 2004). Therefore, high levels of repeated collaboration are expected to have a negative impact on the project performance because team members get blind to incorrectly aligned routines. Furthermore, Katz (1982) and Inoue and Liu (2015) found out that repeated collaboration with the same team member decreases performance in the long run mainly due to decreasing project communication and new idea generation.
Hypothesis 2: Specific team experience has an inverted U-shape relationship with R&D pro-ject performance.
3.3 General & Specific Team Experience
Merging the insights gained from forming the first two hypotheses, general team expe-rience and specific team expeexpe-rience are expected to have positive interactive effects on pro-ject performance. Due to the higher risk of routines leading to superstitious learning and isola-tion, specific team experience with the same team members is predicted to be curvilinear. However, since increasing general team experience involves diverse projects and team mem-ber knowledge, the negative impact of repeated collaborations is likely to be damped. For example, inventors with high levels in both team experiences are probably more able to adjust their routines due to diverse collaboration experience. Additionally, they extract advantages from enhanced coordination due to the created transactive memory between two inventors. Therefore, the benefits of diverse knowledge sources as well as trust and coordination ad-vantages positively moderate the relationship between general-, specific team experience and performance.
Hypothesis 3: The interaction of general team - and specific team experience is positively related to innovation performance.
3.4 General Team Experience and Single Inventor
Therefore, this study predicts that increasing general team experience amplifies the risks that an individual inventor becomes less productive when working alone.
Hypothesis 4: A single inventor’s performance is negatively influenced by their general team experience.
3.5 Lone Wolf Experience
In contrast, there are also inventors who predominately and prefer to work alone. Lone
wolf behaviour is described as “[…] a psychological state in which one prefers to work alone
when making decisions and setting/accomplishing priorities and goals” (Dixon et al., 2003; p.205). This behaviour is driven by impatience for collaboration processes and unwillingness to form social bonds and to recognize the value of ideas from others (e.g. Blau & Boal, 1989; Griffeth, Gaertner & Sager, 1999; Ingram, Lee & Lucas, 1991). Although lone wolves are highly committed to their own tasks, they view others as less competent and are unable to develop trust relationships (Feldman Barr et al., 2003).
Feldman Barr et al. (2003) investigated lone wolf behaviour in sales teams and found a positive relationship between autonomy and lone wolves. They concluded that, in general, lone wolves show more autonomously conducted work processes (e.g. higher responsibility, goal commitment) and prefer to work alone. This tendency is driven by lack in trust and pa-tience towards others. These findings lead to the proposition that lone wolves are less efficient when working in teams (Blanchard, Bowles, Carew & Parisi-Carew, 2001; Ingram, Lee & Lucas, 1991).
Prior research of Singh and Fleming (2010) pointed out that single independent inven-tors perform less well compared to lone inveninven-tors within an organization, since they are iso-lated from any social interaction and solely build on their own prior experience. However, within an organization this is rarely the case as firms function as social networks and lone inventors could also receive feedback from other colleagues (Singh & Fleming, 2010). Never-theless, a pure lone wolf is expected to behave as a single independent inventor. Even though they have the opportunity to ask other organizational members for feedback, their unwilling-ness to recognize others’ opinions would neglect this option.
single-member projects hampers team collaboration on the social level due to the absence of trust and reduced communication as well as shared goal orientation.
4 Methodology
4.1 Sample and Data
This research was conducted by using an already existing dataset of U.S. patents of the National Bureau of Economic Research (NBER) U.S. Patent Citation Data File. The database contains about three million patents and 16 million citations from 1963 to 1999. The infor-mation includes among others: inventors’ names, geographical location, company name, ap-plication year, grant year, count of forward citation and information about the invention itself clustered by technological categories (Hall, Jaffe, & Trajtenberg, 2001).
Patents are commonly used when analysing inventive outcomes, because their stand-ardized application process makes them a robust measurement over time. This research relies on patent applications that were subsequently granted. Patents do not capture the total amount of inventions created by an organization, because companies commonly rely on multiple val-ue protection mechanisms, such as secrecy, to prevent knowledge spillovers (e.g. Cohen et al. 2000). However, they are still considered to be a reliable source. In addition, this research investigates the German automotive industry; a high-tech sector in which patents are com-monly used to protect novel technological components (Arundel & Kabla, 1998), especially since this market involves high competition. Consequently, the data by the NBER data file – which has been used by several authors – should provide a validated data source (e.g. Dibiag-gio, Nasiriyar & Nesta, 2014; Jones, 2009; Melero & Palomeras, 2015).
contribution to this research, since the number of previous collaborations and repeated collab-orations are presumably higher compared to other industry sectors.
We selected three major companies based in Germany that filed patents in the U.S. – namely BMW, Daimler (incl. DaimlerChrysler AG 1998-2007; Mercedes-Benz) and Volkswagen (incl. Porsche; Audi). The Volkswagen Group merged with Audi in the 1960s and fully acquired Porsche in 2012. Both subsidiaries are included in the Volkswagen Group dataset, although Porsche only became a part of the Volkswagen Group in 2012. The Daimler AG patents also include all patents filled for the DaimlerChrysler AG between 1998 and 2007, the time period in which both companies were merged. In addition, this data also covers Mercedes-Benz and various other subsidiaries majority-owned by Daimler AG. In order to link the subsidiaries to their parent company, information on their acquisitions were collected and we checked annual reports to match the parent company (stakes over 51%) with their subsidiaries. Afterwards, the dataset needed to be harmonized by correcting inventors’ names due to misspellings in the patent file. This was conducted by running a name-matching pro-gramme to identify high resemblance of the inventors’ names. This resulted in a dataset con-taining 4.831 filled patents for the three given companies between 1970 and 1999. However, after merging the datasets 1.455 patents had to be excluded due to missing data, leaving 3.376 patents for the further empirical analysis.
4.2 Measurements
4.2.1 De pe nde nt variable : inve ntive performa nce
4.2.1 Dependent variable: inventive performance. To measure the performance
out-come of team collaborations, patent forward citations are used. The number of forward cita-tions illustrates how often the original patent’s knowledge was used for following invention purposes. Therefore, forward citations can be regarded as an indicator for the value of the invention and the incorporated knowledge (Hall, Jaffe & Trajtenberg, 2001; Kelley, Ali & Zahra, 2013).
contribution to the focal patent. Hence, all inventor-related variables were aggregated to the patent-level.
However, this study does not exclude truncation. Truncation describes the phenome-non that, for example, a patent filed in 1970 has a much higher chance for being cited until the end of the data collection in 1999 compared to a patent filed in 1998. The NBER data does not include forward citations before 1975 and after 1999, leading to truncation effects on both sides of the investigated time period. Consequently, we created application year dummies to control for the mentioned effects.
4.2.2 I nde pe nde nt variable s
4.2.2 Independent variables
4.2.2. 1 Gener al team experience
4.2.2.1 General team experience. The NBER patent file provided information on each
inventor’s name, linked to a team or a single inventor’s unique patent number. To measure general team experience, two key factors were considered: previous team experience and count of distinct team members the inventor has worked with before.
First, based the number of times an inventor’s name was mentioned on a filed patent, the accumulative count for each individual was taken. Thus, the number of filed patent per person indicates the number of projects the inventor was involved in and each of those pro-jects lead to new experience. Afterwards, the increasing experience with diverse team mem-bers was calculated. For this type of experience only the first joint project between two inven-tors was of interest. Furthermore, the median of the team members’ number of collaborations and experience with diverse team members was taken. The levels of individual experience within a team were quite unequally distributed, which supported the decision to take the me-dian rather than the average team experience. Both meme-dians combined resulted in the general team experience per patent.
4.2.2. 2 Specific te am ex perience
4.2.2.2 Specific team experience. In order to test hypotheses 2 and 3, the cumulative
Figure 1
1st Collaboration between Inventor C and D Previous specific dyad experience=0
A&B=3 A&C=1 A&D=2 B&C=1 B&D=2 C&D=0 Figure 2
2nd Collaboration between Inventor C and D Previous specific dyad experience=1
B&C=2 B&D=3 C&D=1
4.2.2. 3 Lone wolf experie nce
4.2.2.3 Lone wolf experience. The third independent variable reflects lone wolf
expe-rience. As mentioned earlier, this study also investigates the influence of a single inventor’s
experience on team performance. Therefore, the computation relies on the accumulative count of inventors’ previous experience of working alone.
Figure 3
Lone wolf inventor C’s experience (from working alone to collaboration)
cate the number of previous patents the focal patent builds on and therefore give information about the technology scope of the invention. The effect on the team’s ability to align several distinct knowledge domains is captured by the originality of backward citations; a patent with a high originality score indicates that the previous patents on which the focal patent builds belonged to a wide range of technological fields (Hall et al., 2001). As a result, backward cita-tions account for the amount of knowledge used from other patents, whereas originality pro-vides additional insights on the industry scope of those patent. Another potential influence factor is theage of backward citations since previous literature stated that older knowledge is
less useful in innovation projects (e.g. Ahuja & Lampert, 2001).
Moreover, the control variable team size illustrates the effects of varying team size. Following Lee et al. (2015) it can be expected that too many diverse team members could have a negative impact on the team’s performance.
Additionally, dummy variables were created to differentiate between the three studied companies (BMW; Daimler AG; Volkswagen). Furthermore, a dummy for lone wolfs was computed to capture the effect of single inventors. As already mentioned, truncation due to the data sample is considered by generating dummy variables for each application year. In this study application year of each subsequently granted patent is used.
4.3 Analysis
5 Results
5.1 Descriptive Statistics
The distribution of the performance variable in Fig.3 showed a highly biased on the right side of the graph. The high amount of zero forward citations (13.4%) is also due to the truncation of the dataset, because citations of more recent patents could not be gathered. About 85% of the analysed patents received fewer than ten forward citations, which resulted in a very low number of high performance breakthrough inventions.
Figure 4: Distribution of the performance variable
Regarding the analysed companies, Daimler AG produced the majority of patents (1835), followed by Volkswagen (1281) and finally BMW (260).
Moreover, analysing the team size (min=1; max=13) 40,2% patents were filed by sin-gle inventors, followed by dyad inventor teams (27,2%) and teams of three inventors (18,7%). In addition, the descriptive results show a decreasing trend on performance with increasing team size, which is in line with previous literature (e.g. Lee et al., 2014).
Examining the individual experience, the average number of projects was 7.5 per in-ventor. A closer investigation on the relationship between project experience (min=1; max=45) and performance indicated that the outliers are equally distributed between inventors low in experience and high in experience.
Table 1: Descriptive statistics full sample
Variables Mean S.D. 1 2 3 4 5 6 7 8
1 Performance 5,1108 6,1368
2 General Team Experience 2,3215 3,0665 -0,0200
3 Specific team experience 0,6493 1,7079 -0,0028 0,5566
4 Lone wolf experience 0,9207 1,9735 0,0296 0,3260 -0,0163
5 Years till granted 1,8634 0,9777 0,0740 -0,0246 0,0017 0,0016 6
Originality backward
cita-tions 0,2846 0,2678 0,0848 0,0224 0,0251 -0,0163 0,0953
7 Backward citations 5,6727 3,2590 0,0819 -0,0080 0,0069 0,0461 0,1157 0,3405 8
Average age backward
cita-tions 5,4950 3,6914 -0,0966 0,0787 0,0375 -0,0018 -0,1749 0,1176 0,0284
9 Team size 2,1742 1,3794 -0,0268 0,1373 0,3446 -0,1815 -0,0111 0,0485 0,0223 0,0809
Table 2: Descriptive statistics only including single inventor’s patents
Variables Mean S.D. 1 2 3 4 5 6
1
Single inventor's
perfor-mance (H4) 5,1268 5,9017
2 General Team Experience 2,3215 3,0665 0,0106
3 Years till granted 1,8634 0,9777 0,0687 -0,0246 4
Originality backward
cita-tions 0,2846 0,2678 0,0953 0,0224 0,0953
5 Backward citations 5,6727 3,2590 0,1196 -0,0080 0,1157 0,3405 6
Average age backward
citations 5,4950 3,6914 -0,0657 0,0787 -0,1749 0,1176 -0,0284
5.2 Hypotheses Testing
The results of the negative binominal regression analysis are presented in Table 3. Model 1 solely accounts for the control variables and serves as a baseline model. Apart from the mean age of knowledge from backward citations, all control variables had a positive in-fluence on performance.
Following, Model 2 tested for a positive linear relationship between general team ex-periences on a team’s inventive outcome. The results showed that the outcome of team col-laboration is not significantly influenced by general experience in cooperative innovation pro-jects or dependent on the number of distinct team members. Thus, hypothesis 1 cannot be supported.
Second, specific team experience was found to have a significant (p1=0.014; p2=0.007) curvilinear relationship to the dependent variable (Model 3, 4). First, this study tested whether specific team experience had a positively linear relationship on performance (model 3). Since these results were insignificant, Model 4 illustrates the hypothesized curvi-linear effect (β1=0.0575; β2= -0.00749). Therefore, repeated collaboration with the same members appeared to be positive until a certain point is reached and the negative effects out-weighed the benefits of task coordination. This result is also confirmed by the utest syntax that reflects the exact relationship between the dependent and independent variables on a giv-en interval (see Appgiv-endix C).
In the next step, the interaction effect of specific team experience and general team experience on performance was analysed (Model 5). Neither general nor specific team experi-ence showed a significant effect on performance. However, the moderator variable (gen-eral*specific) showed a significant negative interaction effect on the relationship (p=0.017; β= -0.00584). Thus, increasing general and specific team experience seems to have the oppo-site effect on team performance.
Model 6 tested for the interaction impact of general experience on a single inventor’s performance. Yet, there were no significant relationships to be found. Consequently, hypothe-sis 4 got rejected, suggesting that increased general team experience does not affect the inven-tor’s performance when working alone.
Table 3a: Robust negative binomial regressions
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7
All patents All patents All patents All patents All patents Single inventor’s patents
All Patents General Team experience (H1, H3, H4) 0.00807 0.00890b 0.00632b 0.0126 0.0111 0.0000106b
[0.00614] [0.00723] [0.00729] [0.00731] [0.00957] [0.00672]
Specific team experience (H2) -0.00300 0.0575* 0.0431
[0.0143] [0.0291] [0.0261]
Squared specific team experience (H2) -0.00749*
[0.00296]
Interaction effect general; specific team experience (H3)
-0.00584* [0.00246]
Lone wolf experience (H5) 0.0307***
[0.00928] Years till granted -0.0513* -0.0525* -0.0525* -0.0514* -0.0544* -0.0466 -0.0566*
[0.0259] [0.0258] [0.0258] [0.0259] [0.0257] [0.0360] [0.0261] Originality backward citations 0.278*** 0.276*** 0.276*** 0.267*** 0.271*** 0.274* 0.282*** [0.0758] [0.0759] [0.0759] [0.0759] [0.0759] [0.117] [0.0760] Backward citations 0.0220*** 0.0221*** 0.0222*** 0.0227*** 0.0228*** 0.0367*** 0.0215***
[0.00587] [0.00587] [0.00587] [0.00586] [0.00588] [0.00906] [0.00587] Average age backward citations -0.0181** -0.0181** -0.0181** -0.0178** -0.0177** -0.0161 -0.0168**
[0.00556] [0.00556] [0.00556] [0.00556] [0.00555] [0.00834] [0.00556] Team size 0.0747*** 0.0754*** 0.0763*** 0.0744*** 0.0733*** 0.0785*** [0.0196] [0.0197] [0.0202] [0.0201] [0.0201] [0.0197] Observations 3373 3373 3373 3373 3373 1354 3373 Pseudo R-squared 0.020 0.020 0.020 0.021 0.021 0.018 0.021 Log Likelihood -8998.6 -8997.7 -8997.7 -8994.6 -8995.0 -3621.1 -8992.9 a
5.3 Robustness
In order to check the robustness of the results additional analyses have been conduct-ed, which can be found in Table 4. First, since hypothesis 1 did not show the assumed positive linear relationship on performance, the sensitivity analysis tested whether their correlation could be curvilinear. The results in Model 8 suggested an inverted U-shape relationship (p1=0.025; p2=0.039). Hence, performance is positively affected by general team experience positively only until a certain degree (β1=0.0313; β2=-0.00227) (full results see Appendix A). Following, we conducted another utest syntax to confirm the results (see Appendix D).
Second, due to truncation the dependent variable indicated a drastic increase of zero forward citations within the last 5 years of the dataset, accounting for approximately 57,1% of patents with zero forward citations. Although the research already controlled for truncation by using year dummy variables, the underlying expectation was that experience might have a greater positive impact by reducing the zero performance outcomes. Therefore, the timeframe was limited to the years up until 1994 and the findings for all hypotheses were found to be robust. In addition, a robustness test including solely breakthrough inventions (5% highest performance outcome) was conducted. The findings suggest no significant results for any of the stated hypotheses. Hence, consistent with earlier studies (e.g. Singh & Fleming, 2010), this finding suggests that factors driving the creation of breakthrough inventions differ largely from those related to incremental inventions (see Appendix B).
Furthermore, Model 10 indicates that reducing the sample size by one firm changes the results. Since Daimler AG produced the majority of patents (and zero performance pa-tents), it was reasonable to exclude this firm from the analysis to check the robustness of our findings. Against the expectations, the findings could not contribute to the robustness of the previous results of the analysis. Interestingly, solely specific team experience showed a signif-icant negative impact (p=0.025; β= -0.0671) on performance (full results see Appendix B).
hypothesis 2. The findings are presented in Model 11-13 and implicate that specific team ex-perience seems to increase the search for new ideas from distinct industries, at least until a certain value is reached.
Table 4: Robustness test a
a Significance levels are indicated by *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in brackets. Firm, year and lone wolf dummies included in all models. b
General team experience added as control variable.
Model 8
Full Sample
Model 9
Full Sample (only breakthrough H2)
Model 10
Full Sample (only breakthrough H2)
Model 11
Full Sample (Poisson regression baseline model)
Model 12
Full Sample (Poisson regres-sion H2)
Model 13
Full Sample (Poisson regression H2)
General Team Experience 0.0313* 0.0220 b 0.0194b 0.00112b 0.000227b
[0.0140] [0.0129]b [0.0131]b [0.00175]b [0.00177]b
Squared general team experience -0.00227*
[0.00110]
Specific team experience (H2) -0.0671* 0.0359 -0.000312 0.0170***
[0.0299] [0.0633] [0.00321] [0.00654]
Squared specific team experience (H2)
-0.0160* -0.00204***
[0.00753] [0.000671]
Years till granted -0.0523* -0.0215 -0.0194 0.0223*** 0.0222*** 0.0223***
[0.0257] [0.0412] [0.0414] [0.00447] [0.00512] [0.00511]
Originality backward citations 0.275*** 0.226* 0.218
[0.0758] [0.112] [0.112]
Backward citations 0.0225*** 0.0284** 0.0286** 0.0275*** 0.0278*** 0.0278***
[0.00586] [0.00910] [0.00907] [0.00132] [0.00134] [0.00134]
Average age backward citations -0.0179** -0.00416 -0.00397 0.0101*** 0.0105*** 0.0105***
6 Discussion and Conclusions
6.1 Findings
Existing literature has analysed social structures in teams (e.g. trust, coordination, communication) and the impact of experience on knowledge diversity and establishment of routines (e.g. Argote & Miron-Spektor, 2011; Reagans et al., 2005; Taylor & Greve, 2009). Nevertheless, prior research studied experience mainly on the surface without differentiating between distinct types of an individual’s past collaboration experience. Furthermore, present literature lacks severely in empirical investigations on the behaviour of lone wolves in collab-oration. This study aimed to fill this research gap by analysing the impact of general and spe-cific team experience as well as lone wolf experience in collaboration. The initial proposition was that general team experience has a positive impact on the performance of R&D projects. However, the findings were not significant and suggest that increasing involvement in innova-tion projects as well as collaborainnova-tion with diverse team member do not lead to a higher overall team performance. Interestingly, the results of the robustness test suggest that the relationship is rather curvilinear (inverted U-shape). This finding indicates that the risks of combining too diverse knowledge backgrounds might increase collaboration costs to a point where additional experience or interaction with diverse team members inhibits the positive effects of general team experience (e.g. Katila & Ahuja, 2002; Taylor & Greve, 2006).
The results of specific team experience are in line with the proposed hypothesis that repeated collaboration with the same team member first increases a team’s performance due to coordination and trust benefits (e.g. Reagans et al., 2005). However, at high levels of spe-cific team experience trust in each other’s abilities and jointly established routines lead to the negligence of monitoring in terms of superstitious learning and in return, decrease the team’s performance level. This leads to the suggestion that inventors should limit the number of pro-jects involving the same team members, although social ties and previous successful experi-ence are tempting to collaborate repeatedly.
perfor-to a lower performance outcome. Since general as well as specific team experience have an inverted U-shape relationship with performance, it would seem reasonable that higher levels of both experience lead to a negative effect. This contributes to the given suggestion that indi-vidual project member have to be carefully selected based on their joint collaboration history. Too high levels of general and specific team experience could severely harm the collaboration process, since team members reduce monitoring each other’s behaviour (Langfred, 2004). Furthermore, this study investigated in a relatively novel research field the role of sin-gle inventors within an organizational environment. We expected that a sinsin-gle inventor’s per-formance is negatively affected by general team experience, since increasing team work weakens the ability to work alone. However, the findings including the robustness tests (ex-cept for the last one) could not verify a significant effect.
Prior research of Feldman Barr et al. (2005) suggested that inventors who predomi-nately work alone are less efficient in collaboration. The results present the opposite effect, indicating that increasing experience in working alone has a significant positive impact on a team’s inventive performance. It would seem that increasing lone wolf experience does not simultaneously imply that the inventor does not value ideas from others or avoids social inter-actions. However, it is more likely that lone wolves contribute positively to the innovation’s outcome due to their strong work commitment. Lone wolves with intrinsic social skills may be able to convince other team member to become more committed and to strive for higher innovative outcomes. Additionally, lone wolves are less likely to become free rider in team collaboration since their attitude towards work greatly differs (Feldman Barr et al., 2005). Moreover, lone wolves are suspicious towards ideas of others and thus, the overall task con-flict is amplified that contributes positively to the idea evaluation process.
6.2 Theoretical Implications
The given results indicate that distinct types of lone and joint experiences impact the performance in R&D projects and provide several contributions to the existing literature on social networks and collaboration experience in innovation by giving new implications that enhance the understanding of team member composition based on their diverse experience background.
cer-tain degree. In contrast to prior research (e.g. Reagans et al., 2005; Singh & Fleming, 2010), the results suggest that experience in innovation projects and work experience with diverse inventors, do not consequently lead to a higher project performance. Based on the presented literature (Katila & Ahuja, 2002; Taylor & Greve, 2006) the negative impact of too high lev-els of general team experience would have to arise from an individual’s incapability to absorb new diverse knowledge or from misaligning previous successful practices with the current project situation (Ellis, 1965). Given the results, team member should not be exposed to inor-dinate distinct knowledge domains, providing them the chance to develop specialized knowledge in a certain domain as well. Moreover, the inventor’s ability to detect falsely con-nected experience and routines need to be intensified. For example Heimeriks (2010) ana-lysed superstitious learning in alliance portfolios and concluded that firms have to motivate employees to experiment with various solutions to overcome their dependency on routines.
Furthermore, the findings on the impact of specific team experience complement the presented literature. From the social network perspective, too high levels of trust and strong social relationships dampen monitoring each other’s behaviour. To overcome the disad-vantages of decreasing monitoring and superstitious learning, teams should learn to experi-ment and share successful but also unsuccessful experience. By doing so they reduce the risks to rely on standardized routines and stay flexible in their decision-making process (Heimeriks, 2010).
With regards to the negative moderation effect of specific team experience and general team experience on performance, the results are in line with the first two hypotheses that indi-cated that both experience turn negative beyond a certain point. With increasing general and specific team experience inventor may become unable to absorb the diverse knowledge in-volved. This could lead to ‘information overload‘ that also hampers the team’s ability to mon-itor each other’s behaviour sufficiently (Katila & Ahuja, 2002). Thus, this underlines again the need for mechanisms to identify when established routines are overemphasized and lead to superstitious learning.
In contrast to the anticipated social behaviour of lone wolves being unwilling to rec-ognize the value of team member’s ideas and strong mistrust in others (Dixon et al., 2003), this study provides findings that suggest a positive impact of lone wolves’ involvement in teams. Feldman Barr, Dixon and Gassenheimer (2003) investigated lone wolf behaviour in sales teams and found a positive relationship between autonomy and lone wolves. They con-cluded that lone wolves show in general more autonomously conducted work processes (e.g. higher responsibility; goal commitment) and prefer to work alone. However, when working in teams they are still able to value other opinions and ideas (Dixon et al., 2003). Another possi-ble explanation is the typical high work commitment of lone wolves that could motivate oth-ers to increase project commitment (Meyer, Becker & Vandenberghe, 2004). Nevertheless, high commitment increases also the risk to attract free riders in the team since others rely on the lone wolf’s effort in accomplishing the given tasks. Thus, these finding add great value to the present literature by showing that autonomously operating lone wolves are actually able to collaborate successfully in teams despite their social tendency to work alone. Additionally, growing involvement in single inventor projects even increases the overall project perfor-mance. Therefore, firms should encourage inventors also to engage in projects alone over time to enhance their autonomous working behaviour and feelings of responsibility.
6.3 Managerial Implications
carefully align team members by balancing their distinct general and specific team experience backgrounds.
Finally, firms should also support lone wolf projects since the experience in working alone highly contributes to later team collaborations. Lone wolves show high work commit-ment and by providing them a chance to pursue their own projects ideas, they could feel a higher organizational appreciation of their work (Amabile, 1993). Thus, this could foster their intrinsic motivation to achieve higher performance also when working in teams.
6.4 Limitations
This study implies several limitations that are mainly based on the used dataset. First, the patent sample solely includes patents of three firms from one industry sector in Germany. Therefore, the results of this study could be dependent on the chosen automotive industry or even on cultural influence factors since the majority of patents were produced in Germany. Nevertheless, this industry sector relies largely on patents as appropriation mechanisms (Di Bitonto, 2014) and thus should be a reliable source for innovation processes. Yet, future re-search should pursue to replicate the study in other industries and countries.
Furthermore, the chosen sample may not pay sufficient attention to the underlying so-cial mechanisms of team experience. Since the patents solely indicate the final performance outcome of the innovation project but do not account for the individual’s effort involved and therefore, it is not feasible to identify individual’s effort contributions in collaboration. For example, a project that is highly successful, as represented by a high number of forward cita-tions, need not necessarily have had a very equal distribution of tasks, leading to mistrust among the team members. So far, increasing trust and social ties can be solely anticipated based on repeated joint engagement in future.
Lastly, the collected data only includes patents filed from 1970 till 1999 and thus, the data is rather out dated. The rise of new communication and coordination technologies (e.g. email, intranet) during the last 17 years could have improved collaboration in organizations. However, face-to-face communication still plays an important role in innovation projects, hence our findings should still be generalizable to current work environment situations.
col-guishing between experiences is relevant for R&D project success and thus, provides a foun-dation for future research
6.5 Future Research
The findings suggest that future research should continue investigating how diverse types of collaboration experience impact the performance of joint and single innovation pro-jects. Some of the results were quite surprising and need further in-depth analyses, especially with regard to general team experience and lone wolf experience. Future studies should exam-ine why general team experience turn negative beyond a certain point and which social mech-anisms as well as collaboration experience contribute to this outcome. One possible explana-tion would be that inventors are restricted in the capacity to absorb new knowledge from an increasing number of diverse team members or developed individual routines that hamper their ability to search for new knowledge (Argote & Miron-Spektor, 2011; Levitt & March, 1988). Alternatively, they could also develop a lone wolf characteristic behaviour such as mistrust of others’ opinions due to unsuccessful collaboration experience. However, although this study contributed to the research on lone wolf experience in collaboration, scholars should continue examining their behaviour, especially since lone wolves seem to have a posi-tive impact on the team’ performance.
Although, patent data is an adequate indicator for innovation activities in firms, it is less applicable when investigating social mechanisms. Thus, future research should include in-depth field studies that are more appropriate to identify the underlying causes on how lone wolves add advantages to collaboration. Moreover, scholars should differentiate between dis-tinct types of lone wolves, because their tendency to work alone does not necessarily imply that they do not appreciate others’ work.
From the social network perspective, another suggestion for future research involves the impact of experience on role allocation in teams. Especially, increasing specific team ex-perience may lead to a fixed distribution of roles that hampers the generation of novel knowledge and fosters reliance on established routines.
coordination as well as communication mechanisms. Additionally, other scholars found out that purely virtual teams need more time to develop transactive memory systems in order to build trust relationship and to coordinate their tasks efficiently (Kanawattanachai & Yoo, 2007). In conclusion, it would be highly interesting for future research to investigate the im-pact of virtual team tools on the inventor’s collaboration and lone wolf experience by incorpo-rating findings on virtual teams.
6.6 Conclusion
References
Ahuja, G., & Lampert, C. M. (2001). Entrepreneurship in the large corporation: A longitudi-nal study of how established firms create breakthrough inventions. Strategic
Manage-ment Journal, 22(6-7), 521–543.
Albrecht, T. L., & Hall, B. J. (1991). Facilitating talk about new ideas: The role of personal relationships in organizational innovation. Communications Monographs, 58(3), 273-288.
Almeida, P., & Kogut, B. (1999). Localization of knowledge and the mobility of engineers in regional networks. Management science, 45(7), 905-917.
Amabile, T. M. (1993). Motivational synergy: Toward new conceptualizations of intrinsic and extrinsic motivation in the workplace. Human resource management review, 3(3), 185-201.
Amabile, T. M. (1996). Creativity in Context: Update to" The Social Psychology of
Creativi-ty”. Boulder, CO, US: Westview Press
Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron, M. (1996). Assessing the Work Environment for Creativity. Academy of Management Journal, 39(5), 1154–1184. Argote, L. (1993). Group and organizational learning curves: Individual, system and
envi-ronmental components. British journal of social psychology, 32(1), 31-51.
Argote, L., & Ingram, P. (2000). Knowledge Transfer: A Basis for Competitive Advantage in Firms. Organizational Behavior and Human Decision Processes, 82(1), 150–169. Argote, L., & Kane, A. A. (2003). Learning from direct and indirect experience in
organiza-tions. In Paulus, P. B., & Nijstad, B. A. (Eds.), Group creativity: Innovation through collaboration. Group creativity: Innovation through collaboration (277-303). Oxford, UK: Oxford University Press.
Argote, L., & Miron-Spektor, E. (2011). Organizational Learning : From Experience to Knowledge Organizational Learning : From Experience to Knowledge. Organization
Argote, L., Beckman, S. L., & Epple, D. (1990). The persistence and transfer of learning in industrial settings. Management science, 36(2), 140-154.
Arundel, A., & Kabla, I. (1998). What percentage of innovations are patented? Empirical es-timates for European firms. Research policy, 27(2), 127-141.
Blanchard, K., Bowles, S., Carew, D., & Parisi-Carew, E. (2001). High five! The magic of
working together. New York, NY, US: Morrow.
Blau, G., & Boal, K. (1989). Using job involvement and organizational commitment interac-tively to predict turnover. Journal of management, 15(1), 115-127.
Boerner, S., Linkohr, M., & Kiefer, S. (2011). Top management team diversity: positive in the short run, but negative in the long run? Team Performance Management, 17(7/8), 328–353.
Bottler, S. (2011). Wirtschaft – Das IHK-Magazin für München und Oberbayern. Retrieved 2 June, 2016, from
https://www.muenchen.ihk.de/de/WirUeberUns/Publikationen/Magazin-wirtschaft- /Aktuelle-Ausgabe-und-Archiv2/Magazin-01-2011/Betriebliche-Praxis/Befristete-Arbeitsverhaeltnisse-Bewaehrtes-Mittel-in-bewegten-Zeiten
Carlile, P. R. (2002). A pragmatic view of knowledge and boundaries: Boundary objects in new product development. Organization science, 13(4), 442-455.
Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learn-ing and innovation. Administrative science quarterly, 35(1), 128–52.
De Dreu, C. K., & Weingart, L. R. (2003). Task versus relationship conflict, team perfor-mance, and team member satisfaction: a meta-analysis. Journal of applied Psychology,
88(4), 741.
Di Bitonto, S. (2014). The Automotive Industry in Germany. Germany Trade & Invest, 14. Retrieved 2 June, 2016 from
Dibiaggio, L., Nasiriyar, M., & Nesta, L. (2014). Substitutability and complementarity of technological knowledge and the inventive performance of semiconductor companies.
Research Policy, 43(9), 1582–1593.
Dixon, A. L., Gassenheimer, J. B., & Feldman Barr, T. (2003). Identifying the Lone Wolf : A Team Perspective. Journal of Personal Selling & Sales Management, 23(3), 205–219. Ellis, H. (1965). The Transfer of Learning. New York, NY, US: Macmillan.
Feldman Barr, T., Dixon, A. L., & Gassenheimer, J. B. (2005). Exploring the “Lone Wolf” Phenomenon in Student Teams. Journal of Marketing Education, 27(1), 81–90.
Fleming, L. (2001). Recombinant Uncertainty in Technological Search. Management Science,
47(1), 117–132.
Fleming, L., & Sorenson, O. (2001). Technology as a complex adaptive system: Evidence from patent data. Research Policy, 30(7), 1019–1039.
Girotra, K., Terwiesch, C., & Ulrich, K. T. (2007). Idea generation and the quality of the best idea. Management science, 56(4), 591-605.
Griffeth, R. W., Gaertner, S., & Sager, J. K. (1999). A model of employees' adaptive re-sponse. Human Resource Management Review, 4, 577-590.
Griliches, Z. (1990). Patent Statistics as Economic Indicators: A Survey.Journal of Economic
Literature, 28(4), 1661-1707.
Hall, B. H., Jaffe, A. B., & Trajtenberg, M. (2001). The NBER patent citation data file: Les-sons, insights and methodological tools. NBER Working Paper No. 8498, 1–73.
Haunschild, P. R., & Sullivan, B. N. (2002). Learning from complexity: Effects of prior acci-dents and inciacci-dents on airlines' learning. Administrative Science Quarterly, 47(4), 609-643.
Hausman, J., Hall, B. H., & Griliches, Z. (1984). Econometric Models for Count Data with an Application to the Patents-R&D Relationship. Econometrica, 52(4), 909-38.
Hilbe, J. M. (2011). Negative binomial regression. Cambridge, UK: Cambridge University Press.
Ingram, T. N., Lee, K. S., & Lucas, G. H. (1991). Commitment and involvement: Assessing a salesforce typology. Journal of the Academy of Marketing Science, 19(3), 187-197. Inoue, H., & Liu, Y. Y. (2015). Revealing the intricate effect of collaboration on innovation.
PLoS ONE, 10(3), 1–16.
Jones, B.F. (2009). The Burden of Knowledge and the "Death of the Renaissance Man": Is Innovation Getting Harder? The Review of Economic Studies, 76(1), 283-317
Kanawattanachai, P., & Yoo, Y. (2007). The impact of knowledge coordination on virtual team performance over time. MIS quarterly, 31(4), 783-808.
Katila, R., & Ahuja, G. (2002). Something Old, Something New: A Longitudinal Study of Search Behavior and New Product Introduction. Academy of Management, 45(6), 1183– 1194.
Katz, R. (1982). The Effects of Longevity on Project Communication and Performance.
Administrative Science Quarterly, 27, 81–108.
Kelley, D. J., Ali, A., & Zahra, S. A. (2013). Where Do Breakthroughs Come From? Charac-teristics of High‐Potential Inventions. Journal of Product Innovation
Manage-ment, 30(6), 1212-1226.
Kratzer, J., Leenders, R. T. A., & van Engelen, J. M. (2005). Keeping virtual R&D teams cre-ative. Research-Technology Management, 48(2), 13-16.
Langfred, C. W. (2004). Too much of a good thing? Negative effects of high trust and indi-vidual autonomy in self-managing teams. Academy of management journal, 47(3), 385-399.
groups. Organizational behavior and human decision processes, 82(1), 88-101. Levitt, B., & March, J. G. (1988). Organizational learning. Annual review of sociology, 14,
319-340.
Liang, D. W., Moreland, R., & Argote, L. (1995). Group versus individual training and group performance: The mediating role of transactive memory. Personality and Social
Psy-chology Bulletin, 21(4), 384-393.
Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organization-al trust. Academy of management review, 20(3), 709-734.
Maznevski, M. L. (1994). Understanding our differences: Performance in decision-making groups with diverse members. Human relations, 47(5), 531-552.
Melero, E., & Palomeras, N. (2014). The Renaissance Man is not dead! the role of generalists in teams of inventors. Research Policy, 44(1), 154–167.
Meyer, J. P., Becker, T. E., & Vandenberghe, C. (2004). Employee commitment and motiva-tion: a conceptual analysis and integrative model. Journal of applied psychology, 89(6), 991.
Monge, P. R., Cozzens, M. D., & Contractor, N. S. (1992). Communication and motivational predictors of the dynamics of organizational innovation. Organization Science, 3(2), 250-274.
Moreland, R. L., Argote, L., & Krishnan, R. (1996). Socially shared cognition at work: Trans-active memory and group performance. In Nye, J. L. & Brower, A. M. (Eds), What's
so-cial about soso-cial cognition? Research on soso-cially shared cognition in small groups
(57-84). Thousand Oaks, CA, US: Sage Publications, Inc.
Nelson, R. R., & Winter, S. G. (1982). The Schumpeterian tradeoff revisited. The American
Economic Review, 72(1), 114-132.
Nerkar, A., & Roberts, P. W. (2004). Technological and product‐market experience and the success of new product introductions in the pharmaceutical industry. Strategic
Onal Vural, M., Dahlander, L., & George, G. (2013). Collaborative benefits and coordination costs: Learning and capability development in science. Strategic Entrepreneurship
Journal, 7(2), 122–137.
Paulus, P. B., & Nijstad, B. A. (Eds.), Group creativity: Innovation through collaboration. Oxford, UK: Oxford University Press.
Paulus, P. B., & Brown, V. R. (2003). Enhancing ideational creativity in groups. In Paulus, P. B., & Nijstad, B. A. (Eds.), Group creativity: Innovation through collaboration (110-136). Oxford, UK: Oxford University Press.
Porter, M. E. (1979, October 22). Experience curve. Wall Street Journal, 30.
Reagans, R., Argote, L., & Brooks, D. (2005). Individual Experience and Experience Working Together: Predicting Learning Rates from Knowing Who Knows What and Knowing How to Work Together. Management Science, 51(6), 869–881.
Rodan, S., & Galunic, D. C. (1998). Resource Recombinations in the Firm : Knowledge Structures and the Potential for Schumpeterian Innovation. Strategic Management
Journal, 19(12), 1193–1201.
Schumpeter, J. A. (1934). The Theory of Economic Development. Cambridge, MA, US: Har-vard University Press.
Simons, T. L., & Peterson, R. S. (2000). Task conflict and relationship conflict in top man-agement teams: the pivotal role of intragroup trust. Journal of applied psychology,
85(1), 102.
Singh, J., & Fleming, L. (2010). Lone Inventors as Sources of Breakthroughs: Myth or Reality? Management Science, 56(1), 41–56.
Skilton, P. F., & Dooley, K. J. (2010). The Effects of Repeat Collaboration on Creative Abrasion. Academy of Management Review, 35(1), 118–134.
Stasser, G., Vaughan, S. I., & Stewart, D. D. (2000). Pooling unshared information: The bene-fits of knowing how access to information is distributed among group members.
Stata Data Analysis Examples- Negative Binomial Regression. UCLA: Statistical Consulting Team. from http://www.ats.ucla.edu/stat/stata/dae/nbreg.htm (accessed May 8, 2016). Stevens, M. J., & Campion, M. A. (1994). The knowledge, skill, and ability requirements for
teamwork: Implications for human resource management. Journal of management,
20(2), 503-530.
Taylor, A. & Greve, H.R. (2006). Superman or the fantastic four? Knowledge combination and experience in innovative teams. Academy of Management Journal, 49(4), 723-740. Tjosvold, D. (1998). Cooperative and competitive goal approach to conflict:
Accomplish-ments and challenges. Applied Psychology, 47(3), 285-313.
Tsai, W. (2001). Knowledge transfer in intraorganizational networks: Effects of network posi-tion and absorptive capacity on business unit innovaposi-tion and performance. Academy of
management journal, 44(5), 996-1004.
Virany, B., Tushman, M. L., & Romanelli, E. (1992). Executive succession and organization outcomes in turbulent environments: An organization learning approach. Organization
Science, 3(1), 72-91.
Weber, R. A., & Camerer, C. F. (2003). Cultural conflict and merger failure: An experimental approach. Management science, 49(4), 400-415.
Wegner, D. M. (1987). Transactive memory: A contemporary analysis of the group mind. In Mullen, B., & Goethals, G. R. (Eds.), Theories of group behavior (185-208). New York: Springer.
West, M. A. (2003). Innovation Implementation in Work Teams. In Paulus, P. B., & Nijstad, B. A. (Eds.), Group creativity: Innovation through collaboration (245-276). Oxford, UK: Oxford University Press.
Woodman, R. W., Sawyer, J. E., & Griffin, R. W. (1993). Toward a Theory of Organizational Creativity. Academy of Management Review, 18(2), 293–321.
