Introduction

The precarious working conditions in academia have attained increased attention in the scientific literature and political debate (e.g., Afonso, 2016; Stephan, 2012; Ullrich, 2019). While the vulnerable phases of academic career trajectories vary in different national contexts, an increase in “outsiders”, academics working on a series of fixed-term employment contracts with little to no opportunities of future stable employment, is observed in many countries (Afonso, 2016). This is in most cases, a problem for post PhD researchers [hereinafter “postdocs”Footnote 1 (see ibid.)]. In many European countries and the US, the situation for postdocs has worsened with a “movement of casualization”, i.e. a drastic increase in fixed-term employment (e.g., Afonso, 2016; Stephan, 2012). These employment characteristics of academia are more of a traditional situation in the German context that has only exacerbated (Afonso, 2016).

Already in 1919, Max Weber described the characteristics of an academic career in Germany as a highly insecure path. The only certain way to achieve secure employment is scavenging one of the few positions as a full professor, who constitute 10 percent of academic positions (Kreckel, 2017). Special characteristics of the German system are: (1) a second thesis (Habilitation), which is often still required to attain professorship. (2) The chair-system, which means that university chairs are held by professors, who then represent the respective discipline at their university (Kreckel, 2017). Financial means and personal infrastructure are attached to the chairs, but individually negotiated and set to the individual professor holding the chair. (3) The third peculiarity concerns the “legal ban on internal appointments and promotions” (Kreckel, 2017, p. 5). This means that postdocs cannot become professors at the university of their doctoral studies (Afonso, 2016; Kreckel, 2017). As a result, there are mostly “two categories of academic staff” at German universities: “tenured professors recruited from outside and assisting staff in transitory employment whose task is further qualification, with no career prospects in their own university” (Kreckel, 2017, p. 6). Compared to many other European countries, but also to the United States, Germany stands out by the fact that there are only few permanent positions for PhD graduates (Afonso, 2016; Kreckel, 2016, 2017). Austria and Switzerland are similar to Germany in this respect (Afonso, 2016; Kreckel, 2016). Hence, postdocs have to endure a long period of insecurity with consecutive career decisions – staying or leaving. Postdocs may find it critical to estimate their subjective success probabilities, which are part of decision theories (e.g., Opp, 2019; Kroneberg, 2006, Breen & Goldthorpe, 1997), in order to decide whether to race for one of the scarce positions as professor.

Empirical evidence supports the importance of social capital for a successful academic career and indicates that talent alone may not be sufficient (e.g., Godechot, 2016). We argue in this article, that postdocs with higher professional social capital—postdocs who are well integrated into their scientific communities and have a good relationship with their superior professors—rate their success probability for an academic career higher and are therefore more likely to keep pursuing an academic career. We justify the central role of the professor in this context as follows: As indicated, the terms of employment and privileges of full professors in Germany stand in steep contrast to the insecurities faced by postdocs. They decide whom to hire, promote and recommend (Gallas, 2018; Ullrich, 2019). Qualitative data suggest that full professors are mainly responsible for promoting the careers of their employees, yet perform this task at their own discretion (Richter & Reul, 2016). Professors are not only employers and mentors, but they also facilitate their postdocs’ access to their network and function as gatekeepers to their scientific communities (Jungbauer-Gans & Gross, 2013).

While, overall, empirical studies support that social capital has an impact on various career outcomes of postdocs at different career stages to some extent, there is also conflicting evidence (see section “Social Capital and Postdocs’ Career Intentions (Success Probability)” for more details). It also remains unclear how social capital shapes the ambitions, career intentions and decisions of early career scientists and postdocs.

Against this backdrop, we pursue the questions how integration in their respective scientific community and the relationship to their superior professors affects postdocs in terms of their career intentions: to pursue an academic career and professorship in the long-term. We analyze these questions using data from a postdoc survey in the disciplines of medicine and basic life sciences. With these two disciplines, we are also able to compare two groups who face different labour market circumstances, i.e. different “exit possibilities”. Whereas the basic life sciences are rather “oversupplied” (Stephan, 2012) and face more difficulties in the non-academic labour market (Plasa, 2014), physicians face excellent labour market circumstances as clinicians (Demary & Koppel, 2013).

Postdoc’s career intentions in the light of subjective expected utility (SEU)

In the light of the long career insecurity postdocs have to face the extreme uncertainty in regards to whether they will find permanent employment in academia at all (e.g., Stringer et al., 2018), it is likely that the question of “staying or leaving” is not a onetime question (Wöhrer, 2014). Specifically, when opting out of academia is not perceived as easy and career opportunities outside seem to diminish with time spent in the “ivory tower” (Enders, 1996; Grönlund, 2020).

Following the well accepted notion that human behaviour is goal oriented (Opp, 2019), but acknowledging that careers’ utility is not only bound to monetary rewards, but may also serve other purposes as, e.g., self-fulfilment (e.g., Gubler et al., 2014; Hall, 2004), we frame our research question within a wide version of rational choice/ subjective expected utility theory. We assume that postdocs aspire to an academic career/professorship, if they subjectively perceive this goal as attainable (success probability) and desirable (utility) in the face of alternatives and the associated costs (e.g., Breen & Goldthorpe, 1997; Kroneberg, 2006; Opp, 1999, 2019).

While other behavioural models that focus on attitudes (Ajzen, 1991; Ajzen & Fishbein, 1980) may be combined with wide rational choice approaches (Opp, 1999, 2019), we see a difficulty of separating attitudes from preferences/utilities, e.g., a positive attitude towards flexible work and a preference for flexible work will probably be highly confounded. Existing research on most prevalent perceived utilities and costs within academia, gives us the opportunity to include relevant control variables while focussing on the impact of the professional social environment of postdocs, i.e., integration into the scientific community and relationship with the superior professor. However, since postdocs cannot decide whether they will achieve a professor position in academia, but can only aspire to that position, we will use intentions as dependent variables instead of actual career decisions (Ajzen, 1991; Ajzen & Fishbein, 1980).

In the empirical implementation, we follow an indirect approach of operationalizing the rational choice model. This means that we do not measure utilities, costs and subjective success probabilities directly, but derive them from theory and empirical research (“bridge assumptions”). While this approach does not allow for a direct empirical test of the theory, it is preferrable in the context of surveys, e.g., due to problems of retroactive rationalization tendencies (summarized in Brüderl, 2004).

Social capital and postdocs’ career intentions (success probability)

Although there are many different theoretical approaches to social capital, theorists agree that social capital describes resources that can only be accessed through social ties: “The social capital metaphor is that people who do better are somehow better connected” (Burt, 2001, p. 32). Nahapiet and Ghoshal (1998) define social capital as “the sum of the actual and potential resources embedded within, available through, and derived from the network of relationships possessed by an individual or social unit” (p. 243). Social ties are therefore crucial for scientific careers, as ties to other scientists can translate into social capital and thus provide access to career-enhancing resources, such as domain-specific knowledge, career-relevant information or scientific reputation.

In this section, we discuss two types of social capital that we consider essential for postdocs’ career intentions, as they affect postdocs’ productivity and career success (cf. e.g. Lee & Bozeman, 2005; Wieczorek et al., 2020, Godechot, 2016): The integration into the scientific community and the relationship to their superior professors. In the next two chapters (“Scientific community integration” and “Postdocs’ relationship with their superior professor”), we argue that both forms of social capital not only directly affect career success, but they also have an impact on postdocs’ career intentions by affecting their subjectively perceived success probabilities of attaining a long-term academic career/professorship.

Scientific community integration

Integration into the scientific community is essential for postdocs’ career success in academia, as academic careers evolve within scientific communities who attach prestige and recognition to individual scientists by, e.g., peer review and citations (Laudel & Gläser, 2007). Being referenced by other scientists leads to a higher visibility and increases the impact of the scientific work (Bikard et al., 2015). Moreover, research is produced by teams rather than single scientists and the average number of authors per paper has been increasing in every discipline (Wuchty et al., 2007; Schmidt et al., 2017). Publishing with other scientists increases the visibility and impact of one’s research, since more scientists can share their work with their contacts (Bikard et al., 2015). Working with other scientists also increases productivity, since social ties can give access to human capital, i.e., domain-specific knowledge and skills and/or access to needed technical resources, and encompasses more opportunities for co-authorship (Lee & Bozeman, 2005; Wieczorek et al., 2020). Consequently, the “lone scientist” may have little chance of surviving the academic pipeline (Leahey, 2016). Considering that the scientific community may increase postdocs’ productivity and visibility, scientific community integration may have an impact on postdocs’ career intentions and decisions by affecting their actual and perceived success probabilities.

Numerous studies find that social capital is related to career outcomes in academia by positively affecting research productivity (e.g., Davis, 2009; Epstein & Lachmann, 2018; Lee & Bozeman, 2005; Scaffidi & Berman, 2011; Wieczorek et al., 2020; Zubieta, 2009). Given the growing importance of collaboration in modern research (e.g., Gross & Jungbauer-Gans, 2007; Leahey, 2016), it is no surprise that social capital plays a major role in scienstists’ productivity. Social capital in general has been linked to finding employment (Hadani et al., 2012; Walker & Yoon, 2016), also specifically in academic research: Walker and Yoon (2016) find that social capital is particularly important for PhD-graduates holding permanent research and teaching positions. In their study, postdocs with more professional contacts got more job opportunities. However, other studies suggest a lower relevance of social capital. Schröder et al. (2021) and Plümper and Schimmelfennig (2007) do not support that social capital impacts chances of being appointed as a professor in political science in Germany. However, Schröder et al. (2021) measure social capital indirectly via mobility, co-authors and positions as interim professors. These variables may not fully reflect the actual social capital of their respondents, nor the qualitative differences of their contacts. In Plümper and Schimmelfennig’s study, social capital was nonsignificant when accounting for publications, however, they established a link between social capital and productivity. This makes it difficult to clearly distinguish between the two.

Since achieving a position is not only a result of individual decisions and goals, it is also important to consider how social capital shapes the intentions of postdocs. We look at intentions both as a proxy for decisions (Ajzen, 1991; Ajzen & Fishbein, 1980) and an opportunity to analyze what postdocs want, with a lower impact of what opportunity structures allow them to achieve.

H1

Positive ratings of scientific community integration are significantly and positively related to (a) postdocs’ intentions to pursue an academic research career, and (b) negatively related to the intention to leave one’s current job.

Postdocs’ relationship with their superior professor

The extent to which resources can be mobilized depends on both access to networks and one’s degree of centrality within networks (Nahapiet & Ghoshal, 1998). Professors are likely to hold more central positions (Burt, 2001) within their scientific community. Actors who hold central positions “carry more valued resources and exercise greater power” (Lin, 1999, p. 31). Specifically, superior professors may have a meaningful impact on the career trajectories of their employed postdocs since they combine several potentially career-enhancing roles: They hold a twofold powerful position as superiors with a quasi-employer function and as supervisors who evaluate scientific work. They decide about the hiring, promotion and recommendation of their postdocs (Gallas, 2018; Ullrich, 2019). In addition, superior professors influence their postdocs’ careers by assigning them administrative tasks that keep them from doing research (summarized in Dorenkamp & Weiss, 2018). “[The postdoc’s] career [is] highly dependent on a single professor who is responsible for work tasks, evaluations, and contract renewal and whose recommendation is often critical when the postdoc applies for a permanent position elsewhere” (Dorenkamp & Weiss, 2018, p. 750). Moreover, professors also function as gatekeepers to their scientific communities and can provide postdocs with access to their networks (Jungbauer-Gans & Gross, 2013). Therefore, professors can have a tremendous impact on their postdocs’ careers – not only by providing or denying access to resources/networks and employment opportunities, but also due to the impact of their perception of their postdocs’ potential and resulting encouragement or discouragement to pursue an academic career. By their degree of support, professors may affect postdocs’ motivation and their perceived success probabilities to obtain tenure with one of the scarcely available professor positions.

Studies indicate a career-enhancing effect of a supportive relationships between postdocs and experienced scientists/professors: Davis (2009) finds that postdocs, whose supervisors write down a plan with them, submit and publish more articles and are more successful with grant proposals. Social contact to experienced scientists has further been linked to finding employment as a postdoc (Fuchs et al., 2001; Lang & Neyer, 2004; Schubert & Engelage, 2011). Lang and Neyer (2004), for instance, discover that the supervisor’s productivity increases a PhD student’s chances of finding a postdoc position. Social capital has also been linked to reaching tenure (Combes et al., 2008; Godechot, 2016; Lutter & Schröder, 2016). Combes et al. (2008) find that the presence of a PhD advisor in a hiring committee increases chances of success as much as having published five additional papers. Similarly, Godechot (2016) finds that the (random) presence of a PhD advisor doubles an applicant’s chances of being short-listed. While effects of social capital on tenure are mixed in the German context (Jungbauer-Gans & Gross, 2013; Plümper & Schimmelfennig, 2007), differing operationalization of social capital and survivorship bias needs to be considered. Against this backdrop, we propose the following hypotheses:

H2

Positive ratings of professor relationship are significantly and positively related to (a) postdocs’ intentions to pursue an academic research career, and (b) negatively related to the intention to leave one’s current job.

Other influencing factors on postdocs’ career intentions in the context of SEU

Social capital is not the only factor influencing postdocs’ career intentions. Therefore, we present other relevant factors and subordinate them into the Subjective Expected Utility Model. The factors presented herein will be introduced as control variables in our statistic models.

Productivity/human capital (success probability)

It can be assumed that scientists perceive an academic career as attainable if they have a high level of productivity and hence human capital. While credentials do not perfectly depict the human capital of an individual, publications have become the most important signal of human capital in academia. They are the strongest predictor for achieving professorship (Plümper & Schimmelfenning, 2007; Jungbauer-Gans & Gross, 2013; Lutter & Schröder, 2016). Furthermore, PhD graduates with more publications aspire more strongly for an academic career (Epstein & Fischer, 2017). A study on postdocs in the life sciences finds that postdocs are mostly concerned about publishing as a first author for their career progress (Müller, 2012). Interviewees in the study use “drastic language” to describe the importance of publications, such as “you’re dead if you don’t publish well” (Müller, 2012, p. 25). In addition, the acquisition of third-party funds is an indicator for high academic ability. It indicates that one’s research idea was rated positively and prevailed against competing proposals, signaling the ability to acquire third-party funds in the future which is crucial considering their important role in funding academic research (e.g., Afonso, 2016; Gross & Jungabauer-Gans, 2007). The acquisition of third-party funds has also been associated with being appointed as professor (Schröder et al., 2021).

Preferences (utility and costs)

When choosing a job, personal preferences determine whether certain aspects of the job are perceived as utility or cost. Empirical findings support that a strong preference for research activities is an important reason to pursue an academic career (e.g., Åkerlind, 2005; Roach & Sauermann, 2010; Wöhrer, 2014). With a sample of postdocs, we can assume that the majority of postdocs like the tasks associated with an academic career. Moreover, work autonomy is one main contributor to job satisfaction in academic research (e.g., Amarasena et al., 2015; Bellamy et al., 2003; Cano & Castillo, 2004; Teelken & van der Weijden, 2018). The freedom and autonomy/flexibility that academia provides in comparison to a company job may be perceived favourably. Unsurprisingly, insecure employment is one major source of job dissatisfaction in academia (e.g., Castellacci & Viñas-Bardolet, 2020; Felisberti & Sear, 2014; Teelken & van der Weijden, 2018). In addition, the duration of the postdoc might increase perceived costs of leaving, while perceived chances of achieving tenure might decrease.

Gender (success probability and costs)

In most disciplines, women are underrepresented at the professor level. In fields with a majority of female students, like biology and medicine, the drop out from PhD to professorship is specifically high (Destatis, 2020).Footnote 2 Women in both disciplines publish fewer articles during their PhD period (Epstein & Fischer, 2017; Feldon et al., 2017), although only in medicine recent female PhDs are less interested in an academic career (Epstein & Fischer, 2017). In addition, female physicians start their specialization less often at a university hospital and base their career decisions more on family planning (Busche, 2017). Given that our sample consists of postdocs, (self-)selection processes may already have taken place at that point, since previous studies indicate that after the second thesis, women do not seem to be disadvantaged in terms of being appointed as a professor (Brodesser & Samjeske, 2015; Jungbauer-Gans & Gross, 2013; Lind & Löther, 2007; Lutter & Schröder, 2016). In theory, women may perceive lower success probabilities, and possibly higher costs when considering sacrifices such as postponing family planning.

SEU of career alternatives

In disciplines with fewer non-academic employment options, postdocs tend to stay longer in academia (Stephan & Ma, 2005). Moreover, postdocs may perceive themselves as less suitable for the non-academic job-market in comparison to PhD students (Müller, 2012). A qualitative interview study with postdocs in the social sciences supports that lacking an alternative career plan is inhibiting postdocs from leaving academia (Wöhrer, 2014).

In addition, postdocs are lacking information and career guidance in respect to the non-academic labour market. Supervisors strongly encourage a career in academic research while not being able to give advice for non-academic careers (Hayter & Parker, 2019; Puljak & Sharif, 2009; Sauermann & Roach, 2012; Teelken & van der Weijden, 2018). Leaving the academic career context and moving to another, possibly extremely different, work environment and culture might also be frightening (Blackford, 2018). While we have no information of postdocs’ perceptions on alternative career paths in our dataset, we have two groups with different career chances outside of academia. While the “basic life sciences” represent a variety of disciplines and specializations, one can generally say that (PhD) graduates from those fields face problematic labour market conditions both inside and outside academia (e.g., Klöck, 2010; Jaksztat et al., 2010; Stephan, 2013; Plasa, 2014). However, in medicine, clinical practitioners are in high demand (Gerst & Hibbeler, 2012; Stallmach et al., 2011). Empirical evidence suggests that recent medical PhD graduatesFootnote 3 in Germany are less interested in academic careers in comparison to those from the basic life sciences (Epstein & Fischer, 2017). Based on previous findings and theoretical considerations, we assume that postdocs in medicine are less interested in a long-term research career, but perceive less pressure to leave academia.

Conceptual framework

To give a brief overview of the discussed aspects that presumably affect postdocs’ career intentions, we present our Conceptual Framework in Figure 1. Since our respondents were not asked directly about success probabilities, utilities and costs and their perceived career alternatives, and our data is cross-sectional, we would like to stress, that we cannot test the theoretical model empirically and thus we do not claim causality for our results.

Figure 1
figure 1

Conceptual framework

Full size image

The present study

The present study aims to investigate the impact that professional social capital has on postdocs’ career intentions with cross-sectional survey data including postdocs from medicine and the basic life sciences. Professional social capital is operationalized by postdocs’ ratings of their relationship with their superior professor and their scientific community integration. In addition, co-authorships are assessed, as they are not only an indicator of productivity but also an indicator of scientific community integration. Furthermore, they can serve as a more objective indicator of integration, i.e., indicating the quality of the postdocs’ professional ties. Our hypotheses and control variables are derived from the perspective of a wide rational choice approach. We expect a positive and significant relationship between our measures of professional social capital and postdocs career intentions in our multivariate models.

Methods

Description of survey data

To analyze the research question, we used data from a German research project that investigated academic research careers in medicine and the basic life sciences (Epstein et al., 2020; Meuleners et al., 2020). Within the project, recent PhD graduates and postdocs from 13 medical and biological faculties from three German federal statesFootnote 4 were surveyed. To explore our research questions, we combined two datasets from the project, both collected in 2018 and comprising postdocs at different career stages. The larger data set originated from a cross-sectional survey conducted with postdocs and professors, who were recruited from the websites of their universities (N = 447). The second dataset represents the last wave of a multicohort-panel survey with PhDs who graduated in 2013, 2014 and 2015 (N = 395). Both surveys were undertaken in the same year with the same target group, entailing the same questions, only the way of recruitment differed, therefore we see their combination as unproblematic. To account for the possibility of unobserved heterogeneity, i.e. systematic differences not captured by our control variables, we included a dummy variable for sample association (panel = 0 and internet recruitment = 1).

We excluded respondents who indicated to have left academia (N = 184). Moreover, we excluded professors (N = 54) and postdocs who had completed their PhD 20 years ago and more (N = 92). In the final sample, 53 percent of postdocs were men, 46 percent women and 1 percent diverse, 37.5 came from medicine and 63.5 percent from the basic life sciences. Years as a postdoc varied between 0 and 19.5 years with a mean of 6.5 years and a median of 5 years.

Measures

The surveys were conducted in German language and are only translated for the readers of this article.

Dependent Variables: We assessed career intentions with three single itemsFootnote 5: Academic Career Intention, Professorship Intention and Job Leave Intention. For the variables Academic Career Intention and Professorship Intention respondents indicated on 5-point-Likert scales to which extent they agreed or disagreed with the statements “In the long term, I would like to pursue a research career at a university or university hospital” and “In the long term, I would like to pursue professorship” (1 = completely disagree, 5 = completely agree) (cf. Table 7 in Appendix). The variable job leave intention included several “positive”-categories signifying different stages of changing jobs: believing that a change in jobs is worth considering, browsing through job advertisements, actively applying, already having a new job offer, and working on becoming self-employed. The “negative” category stated that respondents were not currently considering a job change. We created a dummy variable with the value “0”, the negative category, and “1” for all positive categories (cf. Table 8 in Appendix).

Influencing Variables: The central influencing variable professor relationship, was operationalized by 13 items (5-point Likert scaled, 1 = completely disagree, 5 = completely agree). The scale was conceptualized by the researchers involved in the research project (Epstein et al., 2020; Meuleners et al., 2020). It included the disciplinary fit between postdoc and professor (“my professor is a specialist in my research field”), since a professor with a specialization too far from the postdoc can likely not be as good a mentor as someone from a more closely related field and may not be able to provide as much professional advice as someone from the same scientific community. Moreover, the scale encompasses the professors’ professional support of the postdoc (e.g., “my professor is a great support in publishing), career support (e.g., “my professor talks to me about my career progression”) and contact frequency (e.g., “I have regular employee interviews with my professor”), (cf. Table 5 in Appendix). An exploratory, unrotated factor analysis (minimum eigenvalue > 1), a screeplot (cf. Table 5, Figure 2 in Appendix), and Velicer’s Minimal Average Partial Correlation test (MAP-test) (Velicer, 1976; Velicer et al., 2000) suggested a one-dimensional scale. We excluded items that referred to career advice outside academia (items 9 and 13), due to low factor loadings and overall approval, supporting that non-academic careers are not discussed with supervisors/professors (Åkerlind, 2005; Scaffidi & Berman, 2011). Since professors probably do not have enough expertise to give guidance for non-academic careers, these items designate neither a negative nor a positive relationship. We also excluded the item “I have regular performance reviews with my professor” (item 10). The relatively low fit of item 10 suggests that informal contact with one’s professor (item 11) and the contents discussed (e.g., items 5 and 7) are more important than formal performance reviews. After excluding the described items, we summarized the remaining items into an additive index. Scientific community integration was operationalized with five items (5-point Likert scaled, 1 = completely disagree, 5 = completely agree) that addressed the quantity of contacts (e.g., “I know many members of my scientific community personally”) but also quality of contacts (e.g., “I feel that I can always ask questions to members of my scientific community”) (cf. Table 6 in Appendix) (Epstein & Lachmann, 2018). An exploratory, unrotated factor analysis (minimum eigenvalue of factors > 1), a screeplot (cf. Table 6, Figure 3 in Appendix) and Velicer’s MAP-test suggested a one-dimensional scale. Items were summarized into an additive index. The variables “Scientific community integration” and “professor relationship” were barely correlated with r = 0.19.

Table 1 Gender differences in dependent, influencing and control variables (scales)
Full size table
Table 2 Disciplinary differences in dependent, influencing and control variables (scales)
Full size table
Figure 2
figure 2

Postestimation Screeplot, Professor Relationship Scale

Full size image
Figure 3
figure 3

Postestimation Screeplot, Scientific Community Integration Scale

Full size image

Control Variables: We controlled for self-reported publications as lead, co, and last author (a key position in the life sciences, cf. Wren et al., 2007) per postdoc yearFootnote 6 and controlled for the total amounts of years as a postdoc. Moreover, we controlled for the successful acquisition of third-party funds (self-reported, 0 = no, 1 = yes). In addition, we measured personal preferences in terms of importance of work autonomy, importance of career prospects and importance of work-life balance. Herby, we asked the respondents to rate the importance of the aspects “independent scheduling and organization of work”, “long-term employment prospects”, and “compatibility with family/private life” for their professional future (all assessed with 5-Point Likert scales, 1 = not important at all, 5 = very important). Because work-life-balance needs may not only vary between genders but also affect career decisions differently by gender, we included an interaction term of gender and importance of work-life-balance. Moreover, we controlled for gender (women vs. men/diverse) and discipline (medicine vs. basic life sciences) as well as for permanent vs. fixed-term employment. Since we combined two different datasets, we included a dummy variable to control for sample association (panel = 0 and internet recruitment = 1).

Variables not measured: The dataset did not include preferences on a task-level. Moreover, our dataset only comprised rough details on the respondents’ family situation; we did not include further measures in addition to work-life-balance.

Data analysis

We carried out descriptive and bivariate analyses to investigate variable distributions and possible differences between gender and disciplines. To determine group differences, we conducted t-tests for the scale variables and chi-square tests for dummy variables. We report the results of the chi-square tests in the text only. To test our hypotheses, we carried out linear multivariate regression analyses, and logistic multivariate regression analyses when the dependent variable was dummy coded. We achieved a sample size of 512 respondents within the linear multivariate regression models in which missing values were imputed. Missing values were imputed with Stata’s “sem command”, specifying for a full information likelihood approach (FIML) (Medeiros, 2016). By imputing missing values, 316 cases were added to the final regression model. On the level of individual variables, the maximum amount of missing values was 50 percent for the variable “professor intention” and less than 33 percent for the rest of variables. For descriptive, bivariate and logistic regression analysis, only complete cases were analyzed.

Results

Descriptive and bivariate results

In Table 1, we depict the results of the t-tests by gender and in Table 2 we show the results by discipline.

In Table 1, there are no significant differences between female and male respondents with the exception of co-authorships. Since the difference is so small, we argue it is negligible. Women and men wanted to leave their current position at comparable rates (64 and 65 percent, chi2 = 0.032, df = 1, p = 0.860). There were also no significant differences in the number of fixed-term contracts (72 percent of men and 71 percent of women, chi2 = 0.056, df = 1, p = 0.813). Women had acquired slightly more third-party funding (50 percent of men and 56 percent of women, chi2 = 1.26, df = 1, p = 0.261).

Postdocs from the life sciences have a higher intention to pursue an academic research career, this difference is apparent in both variables assessing the intention to pursue an academic research career (cf. Table 2). Moreover, postdocs from the life sciences were better integrated into their scientific communities. Life sciences postdocs intended to leave their current position to a significantly higher extent than respondents from medicine (71 percent vs. 54 percent, chi2 = 11.63, df = 1, p = 0.001). 72 percent in the basic life sciences and 68 percent in medicine were employed on fixed-term contracts (chi2 = 0.62, df = 1, p = 0.431).

Respondents from the life sciences published significantly more lead and co-author articles and had more often acquired third-party funding (66 percent vs. 35 percent, chi2 = 34.56, df = 1, p = 0.000). Postdocs from the basic life sciences indicated a higher importance of work autonomy and career prospects.

Multivariate results

In Table 3, we report the results of the logistic regression analyses with odds ratios (ORs) concerning the dependent variables “job leave intention”. Respondents who wanted to leave their job to attain professorship were not included in this analysis. Following the recommendations of Best and Wolf (2012), we interpret the direction (OR > 1 positive effect, OR < 1 negative effect) and significance of the ORs but not effect sizes.

Table 3 Logistic regression analyses, job leave intention
Full size table

Professor relationship is negatively related to the intention to leave jobs (support of Hypothesis 2b) scientific community integration is not (against Hypothesis 1b). In addition, postdocs in the fields of medicine are less inclined to leave jobs. The assessed performance variables, publications and having acquired third-party funding, are nonsignificant. Autonomy and career prospects are not related to job leave intention; however, a fixed-term contract is significantly and positively so. There are no significant differences between genders concerning job leave intention, including the interaction term between gender and the importance of work-life-balance. The amount of postdoc years and integration into the scientific community are not significantly related to job leave intention.

Moving to Table 4, professor relationship is positively related to academic career intentions and professorship intention, which further supports Hypothesis 2a. Scientific community integration is also significantly related to both dependent variables (support of Hypothesis 1a). There is a small significant effect of postdoc years on the intention to pursue an academic research career. Further, there is a positive significant effect of publications as lead author on the intention to pursue an academic career, and a small positive effect of co-authorships on professorship intention. We understand the latter as a network effect, with co-authorships being an objective marker for scientific community integration. There is no significant effect of publications as last author. Having received third-party funds and having a high need for autonomy are only positively related to professorship intention. The importance of work-life-balance has a significant, negative effect on the intention to pursue an academic career and the intention to pursue professorship. Postdocs with a background in medicine are more inclined to pursue professorship. However, there are no significant differences between the disciplines when it comes to the intention to pursue an academic career in the multivariate model. The control variable for sample affiliation shows that the respondents who were recruited from the websites of their university are significantly more inclined to pursue an academic research career.

Table 4 Multivariate regression analyses, academic career intention and professorship intention
Full size table

Discussion

The results of our study suggest, that social capital does not only directly impact postdocs career success (e.g., Davis, 2009; Hadani et al., 2012; Lee & Bozeman, 2005; Scaffidi & Berman, 2011; Zubieta, 2009), but has the potential to alter their career trajectories by shaping their ambitions. Studies who solely focus on the impact of social capital on being appointed as a professor (e.g., Plümper & Schimmelpfennig, 2007) do not capture the effects of social capital at different career stages and the indirect effect it has on career trajectories by affecting career intentions and presumably decisions. Due to the special status of the professor, the primary goal of our study was to investigate the link between postdocs’ relationship with their superior professor and their career intentions. As we expected, postdocs with higher professor relationship ratings reported higher academic career and professorship intentions. They were also less inclined to leave their current position. Since the data analyzed are cross-sectional, we cannot rule out that postdocs with high aspirations build a better relationship with their professors. Theoretically, a reciprocal relationship is conceivable. Longitudinal data on postdocs may reveal more about the direction of those correlations, but also qualitative data may be helpful to understand the mechanisms behind them. Following the rationale of SEU theories, we assumed that this would be the case because a positive relationship with the professor would affect postdocs' subjective probabilities of success with respect to an academic career. We were not able to test this theoretical mechanism with the present data. Apart from the professor relationship, the integration into the scientific community was also significantly related to academic career and professorship intention. Further, co-authorships were significantly related to the intention to pursue a professorship. Co-authorships may be a better parameter for the quality of scientific community integration. Integration into the scientific community that does pay off in terms of a longer publication record may be specifically relevant to postdocs’ career intentions and specifically important if postdocs aspire to professorship.

With respect to utilities and costs, we supported the preference of work autonomy for academic scientists (Amarasena et al., 2015; Bellamy et al., 2003; Cano & Castillo, 2004; Teelken & van der Weijden, 2018). A high need for autonomy was significantly related the intention to pursue an academic career and professorship. Our results further show that postdocs from both disciplines highly agree that work-life balance is important to them. The negative association between the importance of work-life-balance and intentions to pursue an academic research career and professorship suggest that postdocs perceive this aspect rather as a cost of the academic career path. This is no surprise, considering the “publish or perish mentality” in academic research and the competition for the few permanent positions the system has to offer (e.g., Afonso, 2016). While the importance of career prospects received high approval, it did not have the expected negative effects on career intentions. These results may reflect perceived career alternatives of postdocs: while academia lacks long-term employment opportunities/career prospects, postdocs may fear a lack of alternatives (Müller, 2012). Moreover, postdocs on fixed-term contracts, approx. 70 percent, had higher job leave intentions.

Our results suggest a small positive association between the postdoc duration and academic career intention, possibly because postdocs with these career goals will work for them more persistently but may also perceive limited career alternatives over time.

Looking at objective performance variables, publications were significantly related to academic career and professorship intentions. Other measures of publication quality and visibility (e.g., citations) might be even better measures of academic productivity and were not available here. The reception of third-party funding was also positively related to the intention to pursue professorship. Postdocs who have successfully acquired third-party funding probably have higher chances to apply successfully for a position as professor, as it has already been shown in political sciences (Lutter and Schröder, 2021).

We found no meaningful gender differences in our sample. Since our data comprised postdocs at different career stages, we cannot rule out that gender differences exist, e.g., in the early postdoc stage as previously shown (e.g., Epstein & Fischer, 2017). We further did not analyze gender effects within the different disciplines in order to keep our sample size big enough. Different disciplinary cultures could affect career decisions differently by gender. Our data also did not comprise details about the respondent’s family situation, it is possible that our sample is not representative in this respect and that our female respondents are less burdened by family duties than the average. As we measured intentions rather than behaviour, we do not know how the assessed variables will affect the “final” career decisions and outcomes for women and men. This is a question for future research.

Postdocs in medicine had significantly lower ratings on many of the independent variables: i.e., importance of career prospects and autonomy, integration into the scientific community and publications as lead and co-author. These results are not surprising, considering that postdocs in medicine are usually very engaged with patient care. Since physicians in university hospitals are embedded in a larger health care team and must rely on the specific expertise of other health professionals, autonomy may be less crucial to them. In addition, they may for example cooperate with basic life scientists who conduct the laboratory work for their research.

Postdocs from medicine have no reason to fear for their future employment, hence, it is clear why they rated the importance of career prospects lower. The same interpretation applies to their lower job leave intention. While postdocs from medicine had lower academic career and professorship intentions on the bivariate level, this was not the case in the multivariate models. Their intention to pursue professorship was even higher here. This result may support that postdocs in medicine do not pursue a traditional professorship but an extraordinary professorship, which is equivalent to an honorary title but not a professor position at the university. In line with our results, a study conducted by Sorg et al. (2016) finds that physicians mostly pursue a Habilitation (second thesis) to advance their clinical careers and to a lesser extent to achieve a position as professor (Sorg et al., 2016). Acquiring societal prestige and climbing up the clinical career ladder may be a more important aspect here than being a scientist. However, more data is needed to support this claim.

Limitations

As already discussed, the data of this study is cross-sectional and therefore is limited with respect to the prediction of actual future behaviour and with respect to the direction and causality of correlations. Moreover, actual behaviour is not only influenced by intentions but also the opportunity structures of the respondent, i.e. career alternatives, career opportunities in academic research, career decisions based on the family situation, the willingness to move etc. While our study incorporated aspects of the quality of social capital, there is potential in capturing these aspects more deeply. Qualitative research for instance could help to get a deeper insight into how superior professors support their postdocs and how it affects them.

Further, we only investigated the impact of professional social capital. Social capital outside the workplace, e.g., resources and support by the family, may also be important for career development. For instance, if the postdoc’s close family/spouse expects her or him to provide a stable and secure income, this could push postdocs away from academia.

With respect to human capital, our study included publications and self-acquired third-party funds. In addition, other human capital, such as specific knowledge and skills, may also affect postdocs’ career chances but were not included in the survey. Further, our results may not be directly transferable to other disciplines; however, we argue that the structure of academia is similar in most disciplines. Differences may occur e.g., due to different labour market structures and opportunities. In addition, our results apply to a context of high job insecurity of postdocs. Postdocs who work in contexts in which a permanent position is more attainable and does not require switching universities, such as for instance in France or Spain (Afonso, 2016), may not be as dependent on their professional social ties to secure a permanent position. In this context, comparative research could yield interesting results.

As mentioned in the section entitled “Postdoc’s Career Intentions in the Light of Subjective Expected Utility (SEU)”, we used an indirect approach to operationalize our theoretical frame, a wide rational choice approach. Our results are not a direct test of the theory but rather a test of our theoretically and empirically derived bridge hypotheses (cf. Brüderl, 2004). Experimental studies/vignette studies may be a possibility for a direct test of the theory in the setting of an academic career, including our postulated link between social capital and subjective success probabilities.

Practical implications

Our results have implications for career strategy of postdocs on the individual level: to choose professional ties that are beneficial to their productivity. Moreover, when choosing a position, postdocs should take into consideration the supportive nature of their superior professor. Building a network that entails other professors in one’s research area may also be beneficial, to reduce dependency on one person. In order to increase their integration into the scientific community postdocs should actively attend congresses and approach experienced researchers from their community and initiate collaboration.

However, postdocs should not be left alone with this task. More institutionalized, structured and meaningful career support for postdocs and early career scientists that is not bound to their superior seems necessary. This support should entail help to build a professional network supportive of one’s career goals. In addition, professors may not always be aware of the impact they have on their postdocs’ career trajectories. In addition, even if they are aware, they may not be prepared to support their postdocs optimally. In this sense, it would be desirable to better support and train professors in their role as mentors.

Since work-life-balance was an aspect that received high approval among the study participants, the culture of academia might have to change in order to attract young generations who are not willing to sacrifice large amounts of their private life. Our results also stress the need for better long-term career prospects in (German) academia. As other studies have stressed, insecure employment is an important factor for job dissatisfaction in academia (e.g., Castellacci & Viñas-Bardolet, 2020; Felisberti & Sear, 2014; Teelken & van der Weijden, 2018). The respondents in our sample rated long-term career prospects as very important for their career. However, these are currently lacking for the vast majority. Better and securer long-term prospects may further diminish the impact of superiors on postdocs’ career intentions and may give postdocs more independence with respect to their career planning.