Who Am I? The Influence of Knowledge Networks on PhD Students’ Formation of a Researcher Role Identity

Abstract

Higher education institutes both foster the advancement of knowledge and address society's socioeconomic and environmental challenges. To fulfil these multiple missions requires significant changes to how the role of a researcher is perceived e.g. a researcher identity that is congruent with the objective of contributing to fundamental knowledge while also engaging with non-academic actors, broadly, and entrepreneurship, in particular. We argue that the early stages of an academic career—namely the PhD training trajectory—and the knowledge networks formed during this period have a major influence on the scientist’s future capacity to develop an appropriate researcher role identity. We draw on knowledge network and identity theories to investigate how the knowledge networks (i.e. business, scientific and career knowledge networks) of PhD students promote changes to, reinforce or conflict with the perception of a researcher role identity. Our longitudinal qualitative network study includes PhD students and their supervisors funded by the H2020 FINESSE project. At the network level, we show that scientific knowledge is distributed equally throughout young academics’ networks but that entrepreneurial (business) and career knowledge tend to be concentrated around certain individuals in these networks. On the PhD student level, we observe different pronunciations of the researcher role identity linked to students’ interactions with their knowledge networks. We distinguish identity conflicts due to misalignment between ego and alters which leads to withdrawal from the network. Our findings have practical implications and suggest that universities and PhD student supervisors should support PhD students to develop a researcher identity which is in line with the individual PhD student’s expectations.

Introduction

Since the early 1980s, universities have been important players in entrepreneurial ecosystems (Guerrero et al. 2016). Universities generate societal and economic impacts resulting from the transfer of academic research results and technologies (Cunningham et al. 2014; Mars et al. 2014; Mars and Moravec 2022; Thune 2010). Experienced researchers with a doctoral degree generally have high levels of independence and are key players in the university system based on their mastery of a versatile skillset which goes beyond research competence and includes the ability to manage tangible (e.g. financial) and intangible (e.g. information and knowledge) resources. In addition, in certain (science, technology, engineering, mathematics—STEM) disciplines such as photonics, which is the focus of the present study, researchers require expertise in different forms of academic engagement and academic entrepreneurship. What is meant by being ‘an independent researcher’ is associated with the formation of a researcher role identity by independent researchers (Jain et al. 2009; O’Kane et al. 2020; Pratt et al. 2006). In certain disciplines, this formation involves finding a balance between entrepreneurial engagement and academic research (Mars et al. 2014; Mars and Moravec 2022). Recent literature suggests that more attention needs to be paid to preparing researchers to perform these tasks alongside the development of a researcher role identity (O’Kane et al. 2020). Most authors claim that the formation of a researcher role identity starts at a senior level and is accompanied by various challenges (Jain et al. 2009; O’Kane et al. 2020; Pratt et al. 2006). However, research in higher education confirms that the formation of a researcher role identity begins during the PhD trajectory (Castelló et al. 2020; Dison 2004; Holley 2015; McAlpine et al. 2014; Mula et al. 2021; Smith and Hatmaker 2014). It has been shown that independent researchers in STEM fields are required increasingly to take on a researcher role identity which combines traditional research activities such as publication and research collaboration with entrepreneurial and commercialization activities (Casati and Genet 2014; Jain et al. 2009; Mars et al. 2014; Meek and Wood 2016).

A career in research requires setting up and managing professional networks (O’Kane et al. 2020). Each individual in a network possesses certain attributes, knowledge and functions. Also, networks provide access to different types of resources, both tangible and intangible (Villanueva-Felez et al. 2013) and influence network actors’ attributes, behaviors and expectations, which are important for the formation of a researcher role identity (Baker and Pifer 2011; Stuart and Sorenson 2005; Walker and Lynn 2013). At the micro-level, the actors’ individual agency might work to structure the network to align with the network actors’ expectations, which leads to the question of who shapes whom since the combined effects of individual agency and structural network configuration lead to various outcomes on an individual level, i.e., an inclination toward entrepreneurial engagement (Cunningham et al. 2014; Stuart and Sorenson 2005). For the researcher, engagement in a network provides a mechanism to access different types of knowledge which can influence the emergence and formation of different researcher role identities. Phelps et al. (2012: 3) define social networks that emphasize the content of the relationship as ‘knowledge networks’ i.e. “a set of nodes - individuals […] - that serve as heterogeneously distributed repositories of knowledge and agents that search for, transmit and create knowledge - interconnected by social relationships”.

Higher education researchers have investigated the influence of networks and relationships on the development of PhD students’ researcher identity but do not delve into the specific pronunciations of a traditional and/or entrepreneurial researcher role identity. In addition, the current management literature assumes that the formation of a researcher role identity begins only after the PhD trajectory (Jain et al. 2009; O’Kane et al. 2020). Work in this stream sees the formation of a researcher role identity as involving interaction between the individual researcher and his/her network. Social psychology research suggests that the PhD students’ self-perceptions, expectations and innate attributes and the expectations of their peers (which includes the supervision team) are central to this process (Stryker and Burke 2000; Walker and Lynn 2013). We contribute to the higher education and management literature by proposing a network perspective which in addition to considering internal processes such as self-perceptions, attitudes and values includes so far unexplored aspects of the researcher role identity formation process. Researcher role identity formation is an important stage in the individual’s career development which involves the preparation for expectations about independent researchers. Knowledge networks shed light on individual sources of knowledge and the extent to which they influence the PhD students’ researcher identity formation and their acceptance or denial of what it means to be a researcher. In the present paper, we investigate the role of knowledge networks during the process of researcher role identity formation by PhD students, by investigating the following research question: How do PhD students access and use different types of knowledge while forming a researcher role identity?

To answer our research question, we perform a longitudinal qualitative network analysis. We follow 15 PhD students in the field of photonics. We examine the first stages of the PhD students’ academic career and show how changes in the PhD students’ knowledge networks lead to changes in their researcher role identities. Our longitudinal approach includes five questionnaires asking about the PhD students’ knowledge networks, administered twice yearly. We also conducted 35 interviews with the PhD students and their respective (co-)supervisors to understand the dynamics that occur within their networks. To obtain a better understanding of the processes at play during the formation of a researcher role identity, we performed an in-depth analysis of four model cases.

Theoretical Background

How the social network is configured has an effect on its members’ attitudes, behaviors and outcomes and their interactions with one another (O’Kane et al. 2020; Stuart and Ding 2006; Walker and Lynn 2013). Individual attitudes, behaviors and interactions are crucial for forming and consolidating a role identity. The notion of role identity is derived from sociology and psychology theories. In general, “identity is the internal component of role/identity, role is the external component” (Burke and Tully 1977: 883). More specifically, an individual’s ‘identity’ is based on their self-perception and the perceptions of others in their environment. Stets et al. (2020: 200) define identity as “a set of self-meanings derived from being a distinct person, role player, or member of a group or category”. Thus, identity develops by giving meanings to certain social positions and is activated and validated through interactions with the actors of their social networks. These processes of activation and validation occur through finding or creating situations that allow the identity to be expressed and negotiated with other individuals (Stryker and Burke 2000). When individuals interact with others with similar expectations about an identity, this provides validation for that identity which then becomes more prominent. This results in the creation of identity hierarchies in which certain identities emerge as more dominant and are nourished through interactions within the social network (Stets et al. 2020). However, if such interactions reveal differences in identity expectations, the result can be: (1) a loss of prominence of an emerging identity or (2) a breakdown in network relations and the prevailing social structure (Stryker and Burke 2000). Identity is complemented by personal characteristics, attributes and meanings (Fearon 1999). Each individual has several identities related to different social positions (Stets et al. 2020; Wang et al. 2022). These include assigned identities e.g., child, parent, etc. and chosen professional identities e.g., teacher, researcher, etc. (Burke and Tully 1977). The translation of negotiated expectations and the individual’s self-perception of an identity into a respective behavior requires the individual to take on a role. What the individual believes is the behavior expected in a particular setting will be represented by his/her behavior. Thus, a role embodies the behavioral expectations and obligations attached to a particular identity, and includes the individual’s interpretation of what constitutes that particular identity (Herrmann and Jahnke 2012; Stets et al. 2020; Walker and Lynn 2013). In this paper, our use of the term ‘role identity’ explicitly links identity (internal), i.e. the individual’s self-perception, and role (external), i.e. its translation into a respective behavior. This allows an examination of both the components of a role and their interrelations (Burke and Tully 1977; Jain et al. 2009).

Scholars in higher education have done substantial work in investigating various perspectives of researcher identity formation by PhD students (Castelló et al. 2020; Mula et al. 2021). In the management literature, identity theory is used as a theoretical basis to study career development and career choices (Hall 1968; Ibarra 1999; Jain et al. 2009; Pratt et al. 2006). Some authors subscribe to the view that individuals invoke multiple role identities over a lifetime and that not all role identities persist (McAlpine et al. 2014; Stryker and Burke 2000). The individual’s interactions with the environment validate and foster role identity, thereby linking it to the individual’s social network (Baker and Pifer 2011; Pratt et al. 2006; Stets et al. 2020). Those embarking on a PhD trajectory have certain expectations about what it means to be a researcher. During the course of their doctoral research trajectory, PhD students exchange ideas and expectations about their particular researcher role identity with supervisors, colleagues and peers. Initially, expectations about meaning comprise specific research knowledge e.g., theory, design and practice (Murakami-Ramalho et al. 2013). Later, these expectations may refer to academic reputation in the case of traditional researchers and/or may lead to the development of a commercializable output in the case of entrepreneurial researchers (Lam 2010; Sauermann and Roach 2012). Researchers displaying an equal interest in both activities present a hybrid researcher role identity (Jain et al. 2009). Higher education literature provides interesting findings on the crucial role of supervisors in actively supporting PhD students to develop their researcher role identity through informal and formal discussions about role expectations (Barnes and Austin 2009). Additionally, conferences as professional networking and development events provide opportunities to facilitate the validation of researcher role identities through social interactions with peers (Baker and Pifer 2011; Murakami-Ramalho et al. 2013). The initial role identity is reinforced, challenged and/or adapted as the result of social interactions. Social network theory allows us to scrutinize these interactions on several levels. Social network analysis can be applied to study the characteristics of the actors, the individual’s position in the network, and the relationships and the information flows among the individuals in the network, i.e. the types of knowledge exchanged among the actors and the frequency of their interactions (Coviello 2005; Phelps et al. 2012).

Thus, social networks shape the researcher role identity which supports the argument that the structural configuration of the network shapes the person (Burt 2004; Ibarra 1999; Tasselli et al. 2015). However, some studies propose the reverse effect of ‘individual agency’ (O’Meara et al. 2014; Tasselli et al. 2015). In our context, individual agency which is derived from philosophy, social psychology and sociology can be defined as the power or freedom to choose certain relationships and ignore others (Mars and Rhoades 2012; Mehra et al. 2001; Tasselli et al. 2015). This definition is less applicable in a workplace setting since colleagues are imposed to a certain extent. In this case, the scenario is one of co-evolution in which individual agency and structural configuration influence each other (Reinholt et al. 2012; Tasselli et al. 2015). The types of relationships created -i.e. the networks built- are influenced by personality traits. Especially, stable personality traits such as self-monitoring and the so-called ‘Big 5’ (openness, conscientiousness, extraversion, agreeableness and neuroticism) personality traits are decisive for the individual’s network creation behavior (Mehra et al. 2001; Sánchez et al. 2011; Shane 2003; Tartari 2013). High self-monitors present behavior that will lead to acceptance within the social network, while low self-monitors tend to be truer to themselves and their individual beliefs and values. High self-monitors more often adopt the role of a boundary-spanner connecting knowledge clusters (Mehra et al. 2001; Snyder 1974; Tartari 2013; Tasselli et al. 2015). In the context of researcher role identity formation, high self-monitors are more receptive to the role expectations of their environment and will be more likely to adapt to these expectations; low self-monitors are more likely to act regardless of these outside expectations (Tasselli et al. 2015). Self-monitoring is closely related to knowledge networks and researcher role identity formation where it can be a tool enabling micro-level analysis. For instance, low self-monitors tend to regard contacts with similar values as more important and struggle with their identity if expectations are not aligned. High self-monitors may be more inclined to develop a researcher role identity that is aligned with the expectations of the knowledge network. Self-monitoring also sheds light on how PhD students structure their networks to access knowledge resources, bringing us back to the central research question.

Data and Method

Empirical Setting

The research was conducted within the FINESSE project, a Marie Sklodowska-Curie innovative training network (ITN), funded by the European Commission (EC). The goals of an ITN which are not mutually exclusive are defined by the EC as being to “train a new generation of scientists”, i.e. to contribute to the state-of-the-art in their research domain and “to convert knowledge and ideas into products and services for economic and social benefit”¹. The FINESSE project works on developing new distributed sensing technologies, and searching for (commercial) applications of those technologies. FINESSE is characterized by an explicit application-oriented goal which is reflected in the individual PhD student’s goals and research plans. The project consortium includes 9 universities (in Belgium, France, Germany, Israel, Spain (2), Sweden, Switzerland, United Kingdom (2)) and 13 companies which ensures training, collaboration and researcher mobility during the development of the technology. The participation of companies allows PhD students to perform field tests and facilitates rapid feedback on the commercial potential of FINESSE research results. The FINESSE project presents a homogenous landscape of photonics expertise and provides participating PhD students with similar access to relevant resources. The FINESSE network includes star scientists and facilitates exchanges and collaborations with these individuals. Research mobility is mandatory for all FINESSE PhD students and is aimed at providing them with the possibility to acquire experience in both academic and industry settings. It also facilitates collaboration, knowledge exchange and interaction between PhD students from different universities (De Moortel et al. 2021). During the period of training, the PhD students participate in four one-week intensive trainings on photonics which links academic research to industrial applications and extends the PhD students’ knowledge in that domain. A fifth training event on the topic of entrepreneurship and technology transfer familiarizes the PhD students with opportunities for research commercialization.

The principal investigators of FINESSE all provide scientific photonics expertise i.e. in the science and technology of light. All have published extensively on fundamental and applied research in photonics and all have commercialization experience based either on spin-off creation or the commercialization of patents and contribute valuable business/commercial knowledge to the project network.

Data Collection

Our primary data source is a longitudinal online survey (n=15) administered five times over a six-month cycle from May 2018 to April 2020 (Figure 1).

Fig. 1

Data collection timeline

The first of these surveys asked the PhD students about their prior exposure to entrepreneurship, including personal engagement in entrepreneurship, knowledge obtained from family and friends engaged in entrepreneurial activities, and training. This survey allowed us to assess the PhD students' self-monitoring levels based on the questions proposed by Snyder (1974). The third part of the survey focused on the PhD students' knowledge network. We collect ego-network data based on the responses to a question asking each PhD student to name direct important contacts and specify the knowledge obtained from/ exchanged with that contact. The respondents were asked to choose among seven statements in the three knowledge categories of scientific, business and career knowledge (Table 1). The PhD students were asked to rate the perceived value of each type of knowledge exchanged on a 5-point Likert scale.

Table 1 Seven statements on the types of knowledge

The second survey allowed the PhD students the opportunity to update their previous responses and to add and rate new contacts. The data generated by our survey allowed us to identify the PhD students’ knowledge sources and the type and value of the knowledge exchanged within the ITN project.

The second source of data was interviews (n=35) with: (1) the participating PhD students (n=16) and (2) their supervisors (n=19). In the first part of the PhD student interviews, we asked about the involvement of peers in their research and how they coped with any problems during their PhD trajectory. We also asked about preferences for fundamental research and/or technology applications (Roach and Sauermann 2010) which revealed the PhD students’ researcher role identities. We also included a question about mobility and the impact on their research activities and outcomes. The second part of the interviews focused on the further development of their research and possible outcomes of the PhD training. The responses to these questions provided information on the PhD students’ perceptions of their role as PhD researchers. The researcher role includes what the PhD students perceived to be their responsibilities, values and duties and the strategies adopted to fulfil these and the expectations of the environment (Table 2).

Table 2 Number and duration of the interviews

We conducted our second type of interview with the PhD supervisors involved in the FINESSE project (2). During these interviews, we focused on determining what the supervisors tried to provide and produce during the PhD students’ trajectory. We asked about the supervisors' research position, how they thought their students were progressing and their understanding of the researchers’ role. These interviews enrich our analysis in two ways: (1) they provided validation of the different knowledge resources and more in-depth information on the types of knowledge available in the network, and (2) described how they facilitated access to knowledge and network resources and facilitated exchanges with the different actors and the PhD students. Finally, we asked about their vision of the FINESSE network, including questions about the structure of the FINESSE project consortium, which led to a discussion of research mobility and the opportunities provided by the FINESSE network. It allowed us to crosscheck the information provided by the PhD students on the knowledge networks.

We use secondary sources to familiarize ourselves with the empirical setting. These secondary data sources include in-depth information about the project from the granted project proposal. This helps us to better understand the overall and specific goals of the project with regard to scientific and commercial achievements. This also allows us to uncover connections predating the project. We asked the PhD students about the mobility periods and destinations. The latter indicated which PhD students had tested their technologies in an industry setting and gained business experience and knowledge, and which had remained in an academic environment and followed a more fundamental scientific path. We additionally consulted the PhD students’ CVs for details on their educational background and work experience. Finally, we consulted the PhD students’ personal career development plans (PCDPs). These plans were codified at the beginning of the PhD training and updated every year and provided information on the students’ long- and short-term career goals.

Data Analysis Process

Our longitudinal qualitative network study involved integrating the individual PhD student ego-networks into one network map per survey cycle and highlighting overlapping connections among ego-alters (Figure 2). We consider our study to be dynamic since the participants were able to add contacts at each survey round and/or to label previously identified contacts as ‘latent’. Using this method, we gathered information on the duration of each relationship since contacts could be identified, then disappear and reappear along the PhD research process. These relationships were aimed at establishing tie content i.e. type and value of the knowledge exchanged (Coviello 2005). We characterized the FINESSE network as multi-layered with each layer representing one type of knowledge (Figures 4, 5, 6). That is, a particular alter could be linked to PhD students in more than one layer. We employed social network analysis to describe and characterize the knowledge networks quantitatively and supported our interpretation using qualitative data derived from the 35 in-depth interviews with the PhD students and their supervisors.¹ After mapping the ego networks, we scrutinized the interview data to try to explain the network formation. We adopted a deductive approach to the qualitative analysis to identify certain theoretical concepts such as ‘knowledge networks’ and ‘researcher role identity’. We went back and forth in our interview data to identify text strings consisting of keywords or short sentences. Some interesting patterns emerged that call for further analysis. We found several references to feedback from and interaction with various individuals in the close and extended work environment. Using the PhD student interviews, we identified statements about interactions with supervisors, colleagues and peers, knowledge exchanges, perceptions of the researchers’ role and the opportunities provided by the FINESSE network. From the supervisor interviews, we identified expectations about the PhD students and their work, the knowledge provided by supervisors, statements about entrepreneurial/scientific activities and the opportunities afforded by the FINESSE network. To avoid biases and to obtain trustworthiness in the analysis and outcome, two individuals familiarized themselves with the data collected and conducted independent, simultaneous analyses and comparisons of the findings.

Fig. 2

Network evolution (Survey 1 to 5)

After analyzing the whole network and its evolution over time, we conducted an in-depth analysis of four selected PhD students and their networks.² We chose distinct cases of researcher role identity development through e.g., reinforcement of the existing role identity and negotiation with or replacement with alternative role identities. Our analysis shows that HI1_PhD and K1_PhD are low self-monitors, while AB1_PhD and AB2_PhD are high self-monitors (Table 3).

Table 3 Self-monitoring scale

Based on this classification, we expect to see high self-monitors in boundary-spanning positions, bridging distant individuals and engaging in knowledge sharing and exchange with a broad range of network actors. We expect high self-monitors to be shaped more by the network in line with the argument that structural configuration shapes individual behavior, and low self-monitors to be more likely to be influenced by individual actors in line with the idea that individual agency shapes the network. HI1_PhD and K1_PhD indicated an interest in pursuing an academic career and hence are classified as traditional researchers. Student AB1_PhD indicated involvement in research in both academia and industry as future career possibilities; this individual is classified as a traditional hybrid. We classified AB2_PhD as an entrepreneurial researcher based on the expression of interest in starting a business. The classification was aimed at defining a starting point for each of the selected students and tracking the development of their respective researcher role identities. We triangulated the data with information obtained from the interviews.

Findings

Network Evolution

Figure 2 depicts the evolution of the knowledge network derived from the five surveys.

Over time, the network does not expand much beyond those individuals linked directly to the FINESSE project. Since the main (knowledge) resources needed to perform research tasks are present and accessible within the FINESSE network, the PhD students tended first to search for knowledge and resources within the consortium. The new edges that emerge in networks 2 to 5 are mainly the result of mandatory research visits and point to an increase in collaborative work among PhD students. Figure 2 shows that during the study period, several individuals did not connect to the ‘main’ network component; there were some isolated links from the ‘islands’ to the main network but they were always temporary. The supervisors displayed different reactions to the lack of collaboration between research groups. For instance, supervisor M, whose group is isolated from the main network, regards research islands as the logical consequence of a homogenous consortium in terms of research focus:

Supervisor M: “This [collaboration as a FINESSE team, red.] did not happen. There were lots of individual teams. Many do FBGs [Fiber Bragg Gratings, a photonics technology platform, red.]. Those who know FBGs don’t really learn from each other because the principle is set”.

This quote indicates that within the project knowledge exchange was considered less relevant due to the high level of specialization of the laboratories involved and the smaller need for complementary knowledge to address the research topics. However, Supervisor L, who also stated an awareness of the homogenous research focus, had expected more collaboration among research labs using similar techniques.

Supervisor L: “There are now two or three PhD students doing shape sensing and with various approaches. And I was expecting to see differences, advantages, drawbacks. And then no. Nothing came out and I’m surprised. […]They should compare their work. I don't understand why they didn't speak together”.

Our secondary data show that some of the separate individuals work in closely related research fields, focused on different applications but they did not establish any relationships during the project.

Finally, in the context of network evolution entrepreneurship training between the second and third survey cycles did not lead to an extension of the network to contacts in business-related domains.

Knowledge Distribution

Coviello (2005) mentions the importance of the relationship ‘content’, i.e. the type and value of the knowledge exchanged between individuals within the network. Our results show that the knowledge distribution in the network varies over time and that although scientific knowledge is the most prominent type of knowledge exchanged we observe a decrease in its perceived value over time (see Fig. 3). As the PhD students become more expert in their research field, information from colleagues and supervisors is assessed more critically and compared more closely to already acquired knowledge. Hence, dependence on supervisors’ knowledge decreases as identified by supervisor E in the following quote (Bastalich 2017; Delamont et al. 1999):

Supervisor E: “In the third year, then they know at least as much as I do. And then the discussion is on an equal level, "Okay, shall we do this or shall we--" it is like we, it is no longer, "Can you do this?" It is like, "What shall we do?" It is on a "we" level. They suggest something and I will say, "Well, this will probably result in this”. It is different. It is much more mature. This is a typical evolution”.

Fig. 3

Knowledge distribution (scientific, business, career)

We observe that business knowledge is the least exchanged type of knowledge and by the fifth survey had very little value. Business knowledge was barely exchanged and rated as only slightly valuable. The interviews showed also that the importance of commercialization of research results was valued very differently. Supervisors and some of the PhD students already had some ideas about the possible applications of their technology at the start of the research period.

K2_PhD: “You can do research, and you do research that directly impacts society. Now, it just does not stay in the lab”.

AB2_PhD: “But he’s also telling us that we want our technology to be used by society”.

MN1_PhD: “I wanted my PhD… after my PhD, come up to something useful for society”.

Over the project’s lifetime, the main focus of both PhD students and supervisors was scientific achievements rather than commercialization of the research results. This is due undoubtedly to the fact that the aim of the ITN and the PhD students’ are to complete a doctoral dissertation which contributes to the state-of-the-art in the focal research domain.

Finally, we noted that career knowledge was rated equally important at the beginning and the end of the PhD trajectory. At the beginning of their academic career, the PhD student obtains information from various individuals in the working environment about career opportunities and some seek confirmation of their choice to embark on doctoral research. Toward the end of the PhD trajectory, the topic of ‘careers’ becomes much more important with the greater awareness of the PhD students of their abilities and the need to plan towards establishing a stable career path.

Scientific Knowledge Network

Flows of scientific knowledge (Fig. 4) are distributed evenly within the network and are rated mostly as valuable. The high value attributed to this type of knowledge indicates an alignment in expectations about researcher identity and shows that all the students involved had managed to activate their researcher role identity during these knowledge exchanges. This is as expected given that the students’ main task is to perform research and graduate as a PhD. Less valuable knowledge flows mainly through ties that are external to the project or the research group and include former colleagues, family members and friends. These external ties tend not to have the same scientific insights into the student’s research topic and provide rather general advice.

Fig. 4

Scientific knowledge network evolution

Scientific knowledge exchanges within research groups tend to be more intensive than exchanges between individuals in different research groups; exchanges based on trustworthy relationships lead to ‘strong ties’ which facilitate higher levels of knowledge exchange (Granovetter 1973; Levin and Cross 2004; Villanueva-Felez et al. 2013). As part of the project’s work program, PhD students must complete extended research stays at partner institutions, and we observed that this increased the value of scientific knowledge exchanges outside the host institution. We observed also that close colleagues, i.e. people working in the same lab, and supervisors score high on weighted in-degree measures.³ This measure helps to identify contacts that are (1) mentioned by several PhD students and (2) score high on the importance of different knowledge categories. PhD students search for advice and confirmation mainly from hierarchically higher-positioned individuals, in particular, recognized, established and leading researchers (European Commission 2017).

During the exchange of scientific knowledge, the individuals touch upon various topics, e.g., experiments, results, publications, collaborations, conferences, etc. about both the PhD students, colleagues and supervisors had expectations. These exchanges allow the PhD student to negotiate his/her expectations about the traditional researcher identity with the supervisor. In most cases, the PhD students receive validation of their expectations regarding these aspects of the researcher role identity. However, some students identified fellow PhD students as important sources of scientific knowledge—mostly affiliated with renowned institutions with highly regarded and influential supervisors. Figure 4 shows that some research groups are better connected and engage in more knowledge exchanges. Our secondary data indicate that these groups collaborated intensively prior to the project. PhD students active in the ‘core’ of the network benefit more from pre-existing collaborations, and hence, enjoy better access to non-redundant knowledge sources. A PhD student from a peripheral research group told us that:

MN1_PhD: “In my case, there was not any collaboration established between [A, home institution, red.] and the other partners. So that means that you have to build a relation right now, which can lead to something beneficial for the PhD student within the limited amount of time or not”.

Business Knowledge Network

Figure 5 depicts the flows of business knowledge within the network. Business knowledge refers to knowledge about the commercial potential of new technologies and knowledge about (academic) entrepreneurship.

Fig. 5

Business knowledge network evolution

Business knowledge networks and scientific knowledge networks show distinctly different structures. Exchanges of business knowledge tend to be much less intensive than exchanges of scientific knowledge.

Another distinct characteristic is that business knowledge tends to be confined mainly within research groups compared to between groups. However, interactions involving business knowledge activate a hybrid role identity, adding an entrepreneurial dimension to the researcher role identity. Since the amount and value of business knowledge is distributed heterogeneously throughout the network, some PhD students enjoy more opportunities to foster and validate this other role identity and thus, to develop a hybrid role identity. Two of the research groups in our network show high concentrations of business knowledge flows within the group, according to our survey respondents. A senior PhD (C2_PhD) student working in the laboratory of supervisor C confirmed that these business-related interactions occur regularly: “So we discuss the ideas [with] the supervisors. For example, yesterday I was talking with [C3_Post-Doc, red.] and we talked about a new application of a sensor. And we did a brainstorming to try to find any application”. The supervisors active in these labs confirmed this and highlighted the focus on applied and impactful research:

Supervisor D: “We do applied research, we do not do basic research because I mean, well, it is our field”. […] “at the end, what we are doing is-- what we will do right now, is try to apply this technology […] or even to extend it from a different point of view to an application in order to get something”.

Supervisor C: “I make sensors. I can make signal processing, but then, I have to apply [this measurement technique, red.]. So I need chemistry or […] some engineers”.

In the context of business knowledge, we observed an interesting dynamic between supervisor K and his/her PhD student K2_PhD. K2_PhD was described as a source of business knowledge by respondents, and s/he considered it important for science to have societal impact. However, K2_PhD has no outgoing edges in networks 1-4, meaning that s/he does not consider any other individual as a source of business knowledge. In other words, K2_PhD possesses valuable business knowledge but seems not to have access to a business knowledge network. However, this PhD student appears to be aligned with the researcher role identity expected by her/his supervisor K: “I do not push for that [entrepreneurship, red.] because I think it is not our main business. It is making research” and “I know that people are reading our paper, they take it as reference. So, we have somehow responsibility to publish something solid. Quick and solid”.

K2_PhD is the only case identified of a difference between a PhD student and a supervisor with regard to the perception of the researcher role. Based on the PCDP, at the start of the PhD trajectory, this PhD student was characterized as a hybrid researcher. Despite the fact that the student made no attempts to access business knowledge, the misalignment between the perceptions of supervisor K and student K2_PhD led to the student experiencing inner conflict and, in turn, to the student not activating this hybrid role identity within the network. For example:

K2_PhD: “Still sometimes I think, okay, was doing a PhD the right choice or not? Because I want to do research, but I’m not sure […] if I’m in the right sector. Because so many people in the academic field, they care about just publishing and being published in major journals, and you just care about the fact that you have a paper which is cited by millions of people, but for me, I do not feel this is what I wanted from a PhD. So I want to publish a paper, but not just because I want to publish a paper. I want to publish it because I want it to be the base of something bigger. Something that would benefit other people or maybe me also. Not just to do something for the sake of, I do not know, publishing or something”.

Student K2_PhD is an example of preferences being undermined by the work environment, leading to a severe researcher role identity conflict. The student indicated that this questioning of the researcher role identity was due to the misalignment between the student’s and the supervisor’s expectations. The student decided to adapt to the supervisor’s expectations, hence, abandoning efforts to be entrepreneurial resulting in an adaption to the role of traditional researcher (Stryker and Burke 2000). This is an example of professional imprinting (Marquis and Tilcsik 2013):

K2_PhD: “[another PhD, red.] basically did not follow what [my supervisor, red.] wanted. So s/he was asking him/her if s/he can somehow work on something else. And of course, [my supervisor, red.] did not want this. So, in my case, I’m just a donkey following what [my supervisor, red.] wants”.

The above case shows how much the supervisor can influence the development of researcher role identities. We have shown that vertical interactions with supervisors are perceived as more valuable and provide the platform to activate a researcher role identity. This means that business-related topics must be introduced by the supervisor who should try to create an environment that affords the student the freedom to activate and foster role identities that are in line with his/her expectations and perceptions.

Career Knowledge Network

Career knowledge refers to both knowledge about the current career as a PhD and future career possibilities. This results in the activation and fostering of a hybrid researcher role identity through interactions in the career knowledge network. Figure 6 shows that the career knowledge and the business knowledge networks are similar in terms of knowledge flows.

Fig. 6

Knowledge network evolution of four selected students

Our network graphs show that career knowledge and business knowledge exchanges are mainly intra-institutional. The first survey showed a higher concentration of highly valuable knowledge flows distributed across the network; the other four surveys revealed a scarce distribution of links across the networks, although some specific institutes emerged as career knowledge hubs. Those institutes that acted as hubs operated in the same way in the case of business knowledge networks.

It would seem that business and career knowledge are interrelated. PhD students looking actively for career information tend to consult those individuals in their network whom they consider being sources of valuable business knowledge. Since it seems that most of our research subjects focus on downstream research and have access to both academic and industry actors, this promotes conversation about future career possibilities. We observe that PhD students (peers) act as important sources of career knowledge but are used less as sources of business knowledge. These horizontal interactions are important for exchanges of career knowledge; the cognitive and social distance between peers working in the same lab or on the same research project is small. These types of interactions involve expectations about what constitutes the researcher role in different career positions, within and outside academia. PhD students activate their researcher role identities and negotiate career possibilities based on their understanding of these role identities. When making career choices or when in doubt, they engage actively in internal discussions about what it means to be a researcher.

In-Depth Analysis of Selected PhD Students

A closer inspection of the evolution of knowledge networks in our four selected cases reveals a few differences and many similarities. Figure 6 depicts the number of contacts for each survey (S1-S5), split across three knowledge categories (S-Scientific, B-Business and C-Career), represented in the left bars.

Contact versatility is indicated by the overlaps among the knowledge categories, represented on the right-hand side bars. We see that K1_PhD has the largest network, with almost 70 contacts in survey 2, and HI1_PhD has the smallest knowledge network, with a maximum of 9 contacts in survey 2. All four PhD students have contacts that provide them with highly valuable scientific knowledge.

Vertical exchanges of scientific knowledge involve hierarchically higher-positioned researchers while horizontal exchanges involve other PhD students. In contrast, exchanges of business knowledge are vertical and business knowledge networks include fewer nodes. Interactions related to business topics such as the commercial potential of the PhD students’ research involve fewer contacts than do scientific knowledge-related interactions. At all levels, the knowledge networks of the four students yielded many opportunities to activate and foster their researcher role identities. The ratio of business contacts and their consistency during the data-gathering period are particularly high in the networks of AB1_PhD and AB2_PhD.

Reinforcement of the Traditional Researcher Role Identity

During the interview, K4_Post-Doc said that “most of the discussion is about publishing. Most of the conflicts we have in the lab are about who is publishing first and second author”. This statement shows the focus on reputation and merit in the context of contributions within the scientific community and frames the environment of K1_PhD. The initial perception of K1_PhD about what it means to be a researcher is reflected by the fact that s/he is always focused on the characteristics of traditional researchers (Lam 2010). This focus is exemplified by the student’s knowledge network. In all five surveys, we observe business knowledge contacts but their number decreases along the PhD trajectory. Additionally, the overall value of the exchanged knowledge appears to be low. Only one contact scored 3 (out of a possible 5) for business knowledge and this contact left the network after survey 2. The PhD student seems not to have found a replacement which would indicate that business knowledge was not relevant to his/her traditional researcher role identity formation. In terms of scientific focus, the network provided this student with various opportunities to activate his/her researcher role identity: In the interview, K1_PhD said that: “For me, I think when they have really good results and they want to publish it as soon as possible”. These interactions provided the PhD student with validation of a researcher role identity focused on scientific output. The significantly smaller business compared to the scientific knowledge network allowed for fewer exchanges by this student which was confirmed also by her/his reference to a focal interest in purely scientific activities such as publishing, and no real preference for entrepreneurship even when research results were likely commercializable:

K1_PhD: “I think the perfect situation, perfect outcome is, I really find some innovative idea to improve the technique. Maybe even a new technique.

I: And what would you do with this idea?

K1_PhD: I will continue research”.

During the interview, this student clearly stated a lack of interest in business-related discussions with supervisors:

I: “But did you ever talk, for example, with [your supervisors] or someone else about [patents or entrepreneurship, red.]? […]

K1_PhD: No, no. No entrepreneurship. […]

I: And are you interested in doing this? […]

K1_PhD: Not so much. Yes, not so much. But it’s just personal interest”.

We can presume that this student was not upset by the lack of business-related interactions and had assumed the identity of traditional researcher. This student is an example of perfect alignment with expectations about what it means to be a researcher. K1_PhD is a low self-monitor and stays true to him/herself regardless of the expectations of others in the environment. The network in which K1_PhD is involved perfectly represents her/his beliefs and values and works to reinforce the role identity of traditional researcher. This case is one of co-evolution of researcher role identity and knowledge network. Since K2_PhD did not join the lab until after we had conducted the interview we have no data on whether vertical interaction with K2_PhD in any way challenged his/her expectations about the researcher role identity.

K1_PhD intended to continue a career in academia and following graduation was appointed a postdoctoral researcher in the same laboratory, thus that student’s objective to become a traditional researcher was realized. It is a demonstration of his/her satisfaction with the scientific focus of the network and that the researcher role identity includes some level of ‘free will’, allowing the researcher role identity to be managed by the individual.

From a Traditional to Entrepreneurial Hybrid Researcher Role Identity

The difference between AB1_PhD and AB2_PhD in terms of the number of scientific contacts is pertinent since both are members of the same research group and have the same supervisors. AB1_PhD was a traditional researcher at the start of the doctoral trajectory, with previous education oriented solely toward traditional research activities. Similar to the other PhD students, AB1_PhD was introduced to the network at the beginning of the PhD trajectory; the fact that this network was imposed is characteristic of many work environments. AB1_PhD knowledge network includes contacts that cover all three knowledge categories. In contrast to K1_PhD, for AB1_PhD the knowledge gained from business contacts was considered very valuable throughout the student’s PhD trajectory. AB1_PhD and AB2_PhD belong to one of the business knowledge clusters described earlier. Our secondary data show that students AB1_PhD and AB2_PhD worked closely with an academic entrepreneur (AB3_Post-Doc) whose spin-off is located within the lab. Both students confirmed regular discussions about possible applications and declared a strong interest in downstream research which was something that was missing from their undergraduate programs. Also, during their interviews, supervisors A and B emphasized the importance of ‘science for society’:

Supervisor A: “It makes us slightly more realistic in the sense that somehow we are quite selective in terms of what we develop. If you see the scientific literature, there is loads and loads of very complicated experiments to do, very useless stuff. […] So, if something does a beautiful measurement, but it is not practical you’re not going to do anything with that anyway. So why do it”.

AB3_Post-Doc: “Since day one, this laboratory has been focused not only on pure fundamental research and then published but also on taking your results, taking your technology, and actually building something that is useful for mankind”.

The evolution of AB1_PhDs’ knowledge network shows that this student became a valuable source of scientific knowledge for the other PhD students in the network. This boundary-spanning role assumed by AB1_PhD in the overall knowledge network is due in part by that students’ high self-monitoring level (Mehra et al. 2001; Snyder 1974; Tartari 2013; Tasselli et al. 2015). In survey 4, the number of business contacts of AB1_PhD reached its maximum, representing over a third of that student’s total contacts. We suggest that access to both important scientific and business knowledge allowed him/her to explore the various components of the researcher role identity, including business as well as scientific topics. AB1_PhD exemplifies the situation where the structural configuration of the network shaped the researcher role identity which evolved from a traditional researcher to an entrepreneurial hybrid. This apparent change in the perception of the researcher role identity has been described in the literature as ‘professional imprinting’ (Marquis and Tilcsik 2013). Survey 5 shows that AB1_PhD’s perceptions changed during the PhD trajectory. S/he experienced an addition of an entrepreneurial layer to his/her role identity, during the PhD: AB1_PhD: “At first I was more motivated to do more ‘fundamental’ or theoretical research. Over the PhD, I started to value research more readily applicable to the industry from working in collaboration with industry partners”. S/he now presents as an entrepreneurial hybrid with a secondary commercial persona (Jain et al. 2009). That student’s perception of what it means to be a researcher was focused on scientific characteristics, namely, knowledge acquisition: “A researcher is someone who tries to develop some body of knowledge or field, usually with some rigor”. This explains the difference in the size of his/her scientific network compared to that of PhD student AB2_PhD. AB1_PhD’s case confirms that the development of researcher role identities is a continuous process and thus can change and be managed while being influenced by one’s knowledge networks. With regard to future career aspirations, AB1_PhD stated in the PCDP the desire to “continue research and development of photonic technologies through either industrial R&D or furthering a long-term academic career”. S/he realized that the first goal was to continue a career as a postdoctoral researcher.

Reinforcement of the Entrepreneurial Researcher Role Identity

At the start of his/her PhD trajectory AB2_PhD identified as an entrepreneurial researcher and was focused on applicability and benefits for society, despite having followed the same undergraduate program as AB1_PhD, which was one that focused on a traditional research pathway. AB2_PhD aimed eventually to start a business and during the interview, several times referred to the importance of producing research outputs beneficial to society:

AB2_PhD: “That is why if you have something that is more- that you can fabricate, that you can pass to society, we try to do it. […] [supervisor A, red.] is always telling that. We want that our technology is implemented. […] We all think the same way. But [supervisor A, red. ] is also telling us that we want our technology to be used by society”.

This extract shows that this AB2_PhD’s expectations were in line with those of his/her social network and also with the work environment which allowed the PhD student to develop in his/her chosen direction. AB2_PhD’s knowledge network which includes (some overlapping) contacts from all three categories which the student considered highly valuable allowed him/her to explore both the scientific and entrepreneurial components of research. As a high self-monitor, AB2_PhD adapted effectively to the expectations of the environment which provided the opportunity to choose among the business knowledge contacts in the network s/he was introduced to at the beginning. This student did not make any efforts to grow this network, which suggests that the contacts it included were satisfactory and allowed reinforcement of the entrepreneurial researcher role identity. We characterize this situation as co-evolution; the inclusion of both scientific and business knowledge contacts in the initial knowledge network shaped the PhD student. Eventually, s/he favored those contacts that reinforced the entrepreneurial researcher role identity. His/her network includes a majority of contacts used to obtain career knowledge, which perhaps explains his/her entrepreneurial bent. However, this entrepreneurial desire is more uncertain than remaining in academia and requires more specific knowledge. Scientific knowledge remains important with the importance of business knowledge peaking in survey 2. AB2_PhD does not seem to activate his/her entrepreneurial role identities horizontally. Business knowledge contacts are in more senior positions in a pre-eminently hierarchical structure. In survey 5, AB2_PhD reconfirmed his/her entrepreneurial researcher role identity through the following statement: “[being a researcher means being, red.] responsible to envision helpful applications of [the research, red.] to the benefit of society”. We can conclude that the network provided the student with sufficient possibilities to reinforce his/her entrepreneurial researcher role identity.

Researcher Role Identity Discrepancies Leading to Network Abandonment

The evolution of HI1_PhD’s network is noteworthy. In survey 3, this student did not identify any valuable knowledge contacts and overall had the smallest network among the four selected students. The type of knowledge exchanged evolved mainly around scientific and career topics with no reference to business knowledge after survey 2. The self-monitoring assessment shows that s/he is a low self-monitor which explains that PhD student’s knowledge network. Research shows that low self-monitors tend to stay true to themselves and do not bend under the expectations of the environment (Tartari 2013; Tasselli et al. 2015). HI1_PhD did not see his/her beliefs and values represented in the initial knowledge network and created a new network that was aligned with his/her expectations. This is an example of how student agency allowed the structuring of a personal knowledge network. In the interview, this student described the difficult process involved in reinforcing his/her researcher role identity and identifying useful social network contacts:

HI1_PhD: “I have some colleagues. They sometimes help me with something but not that much. In the end, I’m doing this work on my own”.

As a consequence, s/he developed a self-perception of the researcher role which was based on independent work due to a misalignment with the expectations from the environment. This direction toward solo working shows that s/he abandoned the social network as depicted in Figure 3. This decision was based on the student’s everyday working experience and contrasts with the idea that researchers “share knowledge with different audiences and collaborate with people with different backgrounds (cultural, professional)" (HI1_PhD). This statement is in line with those students who highlighted the importance of collaboration, and unlike HI1_PhD worked in a collaborative working environment involving regular consultation with their networks. HI1_PhD confirms the findings in Stryker and Burke (2000) that social structures vanish in the face of differences in expectations and also demonstrates the micro-level impact of human agency on the network (Mehra et al. 2001; Tartari 2013). Eventually, Student HI1_PhD adapted and identified a knowledge network setting that allowed activation of the researcher role identity with some degree of satisfaction:

HI1_PhD: ”Since s/he got more time [now, red. ], we communicate almost every day. It is just more like a scientific discussion rather than supervision. I guess we just have different opinions about what PhD supervision is about. I did adapt, and this kind of collaboration is satisfying for me”.

S/he described this adaptation to a new social structure and accepted that it might require some change to expectations. S/he found that even in the early stages of an academic career, researcher role identities assume primacy over social structures. HI1_PhD reinforced his/her traditional researcher role identity but had to commit to solo working which became central to the researcher role identity. This was evident from the results of survey 5 when we asked about the PhD trajectory. S/he again underlined that the most valuable thing s/he had learned was that s/he should “not rely on others. The working conditions and the team are often far away from perfect. If you cannot do the job alone - do not start it”. Her/his position in the business knowledge network made her/him an important contact for other PhD students seeking career knowledge although she was the only student who does not mention these career-related interactions. We can assume that his/her valuable scientific and business knowledge and position as a peer created an environment conducive to other PhD students activating and challenging different role identities. However, HI1_PhD’s career aspirations and plans did not change and survey 5 shows that despite the difficulties s/he had faced, the traditional researcher role identity had been reinforced and she/he had identified a new social network in which expectations were more or less aligned. In survey 5 this student stated that what it meant to be a researcher was “freedom. You are able to do whatever you feel like today. You organize your work yourself”. This is likely based on the student’s extensive experience of working independently.

Discussion

Motivated by the extant literature on identity theory and network theory, our study reveals the differences in the access to and use of knowledge networks by PhD students, and the influence of these knowledge networks on researcher role identity formation. Management science research focuses mostly on the formation of a (hybrid) researcher role identity by senior researchers; it overlooks the importance of early career stage dynamics emerging during the doctoral trajectory highlighted by higher education scholars (Baker and Lattuca 2010; Castelló et al. 2020; McAlpine et al. 2014; Pifer and Baker 2013). However, the higher education literature focuses on only the traditional researcher identity and does not examine the entrepreneurial aspects of a researcher role identity. Our work extends both the higher education and management literature and proposes an ego-network perspective on the researcher role identity of PhD students. We enriched the framework developed by Stets et al. (2020) on the necessity of networks in activating and fostering the researcher role identities. Our qualitative network analysis provides a detailed perspective on PhD students’ researcher role identity formation taking account of distinct preferences for scientific and business-related topics influenced by interactions with network members. We have shown that pre-existing expectations about the researcher role identity can be reinforced or changed by the network and that networks can be abandoned if the differences in researcher role identity expectations become too great. Our findings further exhibit that scientific knowledge remains the most important type of knowledge accessed by PhD students from their respective networks. We found that individuals assume different network positions, and some PhD students with stronger links to other research groups and collaboration experience can become significant suppliers of knowledge. We identified the emergence of business knowledge clusters within the scientific networks accessible to young PhD students but show that not all PhD students in the network have equal access to diverse knowledge resources.

Our in-depth analysis of four selected cases revealed interesting differences in the role played by knowledge networks on researcher role identity formation by PhD students. What differentiates the cases is the ease of settlement on a certain researcher role identity; some PhD students struggled (e.g., HI1_PhD) whereas others were supported by the network to activate and foster their researcher role identity (e.g., AB1_PhD, AB2_PhD and K1_PhD). The case of HI1_PhD not only supports the finding that individuals abandon social structures if discrepancies about role identity emerge (Stryker and Burke 2000) but also proves that low self-monitors are more likely to change the structural components of their networks if misalignments over expectations emerge (Mehra et al. 2001; Tartari 2013; Tasselli et al. 2015). The case of AB1_PhD exemplifies the shift from traditional to entrepreneurial hybrid identity. The PhD student’s knowledge network was very diverse which allowed experimentation with all the components of a researcher role identity. The lived beliefs and values of network members had a strong influence on this PhD student’s perceptions and allowed him/her to exchange both scientific and business knowledge. This observation supports the literature on professional imprinting by showing that prominent features of the environment are imprinted on the focal entity AB1_PhD (Azoulay et al. 2017; Marquis and Tilcsik 2013). K1_PhD was presented with a knowledge network which included a range of different knowledge contacts and after the main business knowledge contact disappeared from the network, s/he made no efforts to find a replacement but instead shifted to a science focus which helped him/her to reinforce the traditional researcher role identity. At the micro-level, this PhD presents us with an example where co-evolution of the structural configuration of the knowledge network and enactment of individual agency (Mars and Rhoades 2012; O’Meara et al. 2014; Tasselli et al. 2015). We also extract from our results that some PhD students were able to separate behavioral expectations from identity. For instance, HI1_PhD enacted behavior in line with the expectations of the environment (e.g. entrepreneurial—patenting) but described his/her identity as traditional. We found also that the prominence of scientific or business-related preferences varies among PhD students which highlights the diversity among different laboratories. Some focus more on upstream research, others focus on downstream research or pursue both simultaneously. What matters is the quality of the outcomes and the space they provide for PhD students’ various researcher role identities (Jain et al. 2009; Olmos-Peñuela et al. 2015), and the. This opportunity allows students to better negotiate their expectations about these researchers’ role identities. This provides a good foundation for a future career in both academia or industry. In our study, the students' plans and aspirations about future careers did not change dramatically during the course of their PhD trajectory.

Contributions and Future Research

Our research holds several theoretical implications. First, we show that a focus on PhD students is needed in the management literature on researcher role identity, as we confirm that the formation of researchers’ role identity takes place during the PhD trajectory. In this stage, the PhD students begin vertical and horizontal negotiation about expectations related to what it means to be a researcher, and begin to create a hierarchy of role identities (Brenner et al. 2014; Pifer and Baker 2016; Stets et al. 2020). This identity hierarchy plays a crucial role in determining a preference for fundamental or applied research as such, i.e., the formation of a hybrid researcher role identity (Jain et al. 2009; Pifer and Baker 2016). The career aspirations of PhD students are affected by their interactions with members of their knowledge networks.

Second, we provide evidence of the relation between identity theory and network theory which supports and extends the findings of Stets et al. (2020) and Walker and Lynn (2013). Specifically, we showed that networks provide a platform for interactions and continuous activation and negotiation of identities. Differences in PhD students’ networks lead to distinct researcher role identities depending on the characteristics of the respective individuals within the knowledge network and the student’s level of self-monitoring. Some parts of the PhD students’ knowledge networks were established at the start of the project; some contacts were added by the PhD students at later stages. Some studies point to a threefold paradox regarding the interrelation of structural network configurations and individual agency (Stryker and Burke 2000; Tasselli et al. 2015): (1) Role identities are affected by changes in the network, i.e. role identity changes with changes to the individuals in the network (Burt 2004; Ibarra 1999), which is in line with work on professional imprinting (Azoulay et al. 2017; Marquis and Tilcsik 2013); (2) PhD students reflect continuously on their role identity and choices and access knowledge from certain individuals which allows changes to their networks in order to foster and validate their role identity (the idea of ‘free will’); (3) Mutual influence of individuals and networks (co-evolution) triggers changes to researcher role identities and the structural components of the networks.

Although the collected data refer to work-related networks which are considered less likely to be fully selected by the egos (Tasselli et al. 2015), we distinguished between these three scenarios based on the level of self-monitoring of the PhD students. This enriches our findings and adds the layer of individual agency. Hence, our findings provide evidence not only of the dependency of knowledge networks and the formation of different researcher role identities as outcomes but also explain these differences using the self-monitoring scales which show whether the individual prefers alignment to or development independent of the expectations of the environment (Tartari 2013; Tasselli et al. 2015).

Additionally, our research has practical implications for the management and support of PhD students. We highlighted the role of the supervisors and their influence on PhD students’ perceptions of the researcher role identity on several levels. Our analysis distinguished between PhD students who received little supervisory support and those who were supported strongly on scientific- and business topics. In the network studied, scientific knowledge exchange was prioritized and the interaction between institutions and supervisors to foster collaboration benefited the PhD students. Nevertheless, we found that business knowledge was confined to certain clusters and mostly advantaged those PhD students linked to these clusters. A PhD trajectory should prepare the student for a career in academia and in industry (Mars et al. 2014). We hope that our findings will encourage universities and individual supervisors to support PhD students to develop researcher role identities that are in line with this expectation rather than being in line only with the supervisors’ preferences. It is important to remember that supervisors should create an environment that allows the PhD student to activate the various components of the researcher role identity, otherwise students will question their expectations and their perceptions of that identity.

Our study has some limitations which suggest valuable avenues for future research. First, it would be interesting to extend this study by administering large-scale surveys of PhD students in a range of faculties. This would allow us to depict different pronunciations of researcher role identities including in disciplines with scarcer commercialization activities. This would help to identify tools and strategies that would increase entrepreneurial engagement and especially in the humanities or social science fields. Self-reported surveys could be complemented by surveys of the network contacts referred to by the students. This would extend the network beyond the ego-network and provide deeper insights into the indirect sources of knowledge or triadic structures. Our study has provided some interesting findings related to PhD students involved in an ITN and would benefit from comparison with other types of PhD students, especially those not introduced to a network at the start of their trajectory.

Third, the fact that the researchers are part of the same project and know the research subjects might result in the PhD students feeling obliged to respond which would render our results not comparable with data derived from voluntary responses. We tried to overcome this problem by allowing the students to choose whether or not to participate in the surveys and one did not respond to survey 4. Another limitation of our study is that these network maps are snapshots of particular moments along the PhD trajectory. Some connections might disappear from the knowledge networks but do not disappear in real life. Our research does not include emotional bonds forged among individuals which might persist over the long term. These problems might be overcome by capturing live dynamic networks which allow additions of new contacts that can be linked to a specific moment in the PhD trajectory or a specific decision made by the PhD student or supervisor.

Our study has revealed the differences in the access to and use of knowledge networks among PhD students, and the influence of these knowledge networks on researcher role identity formation. Differences in PhD students’ networks lead to distinct researcher role identities depending on the characteristics of the respective individual in the knowledge network and their individual agency, i.e. level of self-monitoring. These results add to the theory on researcher role identity formation by PhD students and knowledge networks. In terms of practical contributions, our findings highlight the importance of networks for PhD students to access knowledge and develop a researcher role identity that is aligned with the PhD students’ expectations. We also show the importance of supervisory support at all stages during the PhD trajectory.

Appendix

See Table

Table 4 Profiles of the interviewees

4.

Table

Table 5 Structural characteristics of the networks

5 presents an overview of the evolution of the PhD students’ knowledge networks in terms of size and connectedness.

We observe only a slight increase in the number of nodes (individuals) in the knowledge network over the project’s lifetime. This observation indicates that the basic structure of the network was already set before we launched the first survey. Through regular project meetings, all project members got to know each other and thus form a rather closed network from the start of the project. Looking at the density of the network, we observe a low level of connectedness throughout the study period. Networks with a low density are sparse and non-redundant. In our context, it shows that the surveyed PhD students rarely name each other as important contacts but rather mention more experienced people in their direct working environment as crucial sources of knowledge. This finding reflects the spread of the PhD positions among eight different institutions around Europe. The diameter and average path length are measures to describe the interconnectedness within the network. The diameter represents the longest possible distance between 2 nodes. As the network grows from survey 1 to survey 3, the diameter grows. In survey 3, the network reaches the highest number of nodes and edges (connections). The slight drop in these numbers from surveys 3 to 4 can be explained by the fact that one PhD did not participate in survey 4. Taking that into consideration, the difference in nodes and edges between these surveys is minimal.

In short, we observe an increased connectedness in the network around the time of the third and fourth surveys, when most of the PhD students were in the middle of their PhD trajectory and engaged in mandatory mobility at one of the consortium institutes or companies. Interestingly, we observe that as soon as research stays ended and a collaborative paper was written, the relationships established during that period of mobility became weaker or even disappeared. Strong ties in the knowledge network seem to consist of colleagues and supervisors from the same working environment. Surprisingly, the number of edges in survey 5 even dropped below the number of edges of survey 1 and showed a slightly lower density. Survey 5 was launched during the COVID-19 pandemic when all mobility was cancelled; hence, collaborative work became more difficult to explain this observation. Additionally, some PhD students were approaching the end of their PhD trajectories and were focused on finishing their thesis rather than setting up or maintaining collaboration.