The relationship between coauthorship and the research impact of medical doctoral students: A social capital perspective

Abstract

Research impact is an important manifestation of research competence and the focus of medical education. This study is dedicated to exploring the relationships between coauthorship networks and the research impact of Chinese medical doctoral students from a social capital perspective. A total of 16291 scientific papers from 237 doctoral students and 126 mentors at Chinese universities were selected from databases, and a study dataset including 19 variables was constructed. Nine independent variables were defined and obtained through coauthorship network analysis, and the doctoral students’ research impact, as the only dependent variable, was used to test the hypothesized relationships among the variables. The results show that the betweenness centrality, student-mentor coauthorship count and the partnership ability index significantly affect the h-index. Specifically, the coauthorship unit count plays the most important role in developing centrality, which, in turn, produces a higher h-index. In addition, betweenness centrality, student–mentor coauthorship count and the partnership ability index are good predictors of the likelihood of doctoral students entering the greater research impact group and especially improving their betweenness centrality. Specifically, doctoral students whose partnership ability index is greater than 2, student-mentor coauthorship count is greater than 4, coauthorship unit count is greater than 6, and betweenness centrality is greater than 0.02 are considered to have greater research impact. These findings suggest that the important roles of cooperation in the development of research competence and good mentorship in the acquisition of social capital by doctoral students should be emphasized. Several strategies are advised for harnessing social capital rooted in doctoral students’ coauthorship network for relevant organizations, mentors and doctoral students who want to increase medical doctoral students’ research impact.

Introduction

Research competence is regarded as the core competence of medical practitioners to meet growing global health care needs, and the educational community has called for medical students to develop and acquire research skills and attributes as early as possible (Laidlaw et al. 2012). Publications in journals are seen as important outcomes of scientific research. Universities in many countries, including China, have adopted a policy in which publications are required for medical doctoral students to obtain their degrees (Cargill et al. 2018; Corsini et al. 2022). This has led to China’s rapid ascent to second place globally in the number of Science Citation Index (SCI) articles, and it may also affect the quality of publications to some extent (Zhu et al. 2014). The research impact of publications represents the intellectual contribution in the academic field, which is seen as an indicator to evaluate the intrinsic quality of the research (Penfield et al. 2014). Therefore, it is very important to enhance the research impact of doctoral students to promote the development of medical education.

Social capital theory (SCT) has been widely applied to improve the relationships among educational groups, thereby enhancing learning outcomes and teaching effectiveness (Keung and Cheung, 2023). Social capital is often defined as all potential resources rooted in social networks formed by interpersonal interactions (Adler and Kwon, 2002). It includes three dimensions: structural capital, cognitive capital, and relational capital (Claridge, 2018). Structural capital describes the patterns of connections among people or units in social network structures (Li et al. 2013). Cognitive capital generally refers to individuals’ knowledge, skills, professional discourse, and practical learning norms from interactions with others within a collective (Wasko and Faraj, 2005). Relational capital, which focuses on how people’s special relationships, such as respect and friendship, affect their behavior, is the capital created and leveraged through interpersonal relationships (Nahapiet and Ghoshal, 1998). Scientific cooperation can bring abundant resources, a reasonable social division of labor and channels for knowledge diffusion, which has been proven to be conducive to increasing research productivity (Abramo et al. 2017; Yue et al. 2019). Coauthorship represents an important form of cooperation, which occurs when researchers collaborate and publish an article together (Kumar, 2015), and multiple coauthorship relationships are combined to construct the coauthorship network. Coauthorship networks are valuable social networks, and social capital from these networks has been proven to have a positive effect on research impact (Li et al. 2013; Xu and Chang, 2020).

As most doctoral students are novices in academia, they do not have the ability to independently publish papers with others to establish a coauthorship relationship at the beginning of their studies, mentorship is adopted in medical graduate education to help them develop their abilities (Witry et al. 2013). Mentors provide students with research and teaching guidance to help them develop their skills and expand their influence in the professional world, provide them with emotional and psychological support, and motivate them to overcome difficulties and move forward toward their goals (Carey and Weissman, 2010). These interactions help doctoral students develop their cognitive capital. The more professional knowledge and skills a mentor possesses, the greater the likelihood is that he or she will contribute cognitive capital to his or her doctoral students (Wasko and Faraj, 2005). As cognitive capital also includes mastering the application of expertise, mentors with more experience may better understand the relevance of their expertise and have cumulative advantages that enable them to provide their doctoral students with more helpful guidance (Malmgren et al. 2010; Wasko and Faraj, 2005).

However, the volume of this flow of social capital from mentors to doctoral students can vary greatly, and creating a more satisfactory relationship may promote the transfer of capital (Ahmad et al. 2023). When doctoral students are coauthors with their mentors, these relationships are good demonstrations of mutual trust, commitment and reciprocity (Moorman et al. 1993). Relational capital exists under these conditions and plays a role in providing more resources to doctoral students. For example, more collaborators provide new knowledge and perspectives to generate creativity (Perretti and Negro, 2006), and diverse cooperative teams provide their own specific expertise, enabling students to overcome their individual cognitive limitations and generate new ideas (Oh et al. 2005) and maintain stable cooperative relationships with others to share resources and maximize cooperative benefits (Li et al. 2013).

As doctoral students mature and gradually establish their individual coauthorship networks, their position in these networks brings structural capital. Centrality, an important structural attribute, is often used to represent the formal power or prominence of a target in a network relative to others, including closeness centrality, betweenness centrality and degree centrality (Kumar, 2015). Degree centrality is defined as the number of direct connections between the target doctoral student and his or her coauthors regardless of connection strength, closeness centrality is defined as the mean shortest distance from the target doctoral student to all other coauthors, and betweenness centrality is defined as the proportion of the shortest paths between all pairs of nodes that pass through the target doctoral student (Borgatti, 2005). Doctoral students with high centrality in coauthorship networks are likely to have access to more resources to enhance their research impact.

Previous studies have confirmed that a good relationship between a mentor and mentee is more helpful for realizing the objectives of mentorship (Straus et al. 2013; Straus et al. 2009; Yu et al. 2022). However, these studies mainly make recommendations on the basis of feedback from interviews and questionnaires, and few studies have used coauthorship network analysis to further mentorship development in the medical field. The purpose of this study is to explore the relationship between coauthorship and the research impact of doctoral students from a social capital perspective. This study attempts to answer the following three questions: 1) What social capital from the coauthorship network affects the research impact of doctoral students? 2) What is the path through which these types of social capital affect the research impact? 3) What special effect pattern does this social capital have on doctoral students’ research impact? The findings of this study can identify different types of social capital presented in coauthorship that might contribute to doctoral students’ research impact and help us better understand and implement mentorship in medicine.

Methodology

Samples

A total of 250 medical doctoral students who graduated from 2016 to 2021 from two major medical universities, Army Medical University and Chongqing Medical University, in Chongqing, China, and their mentors were selected as the research objects. This represented 14.5% of all the doctoral students who graduated from the two universities during this period. We chose these universities for the following reasons: (1) They have consistently performed well in scientific research among Chinese medical universities. (2) Both universities are located in Chongqing, and they have a long history of extensive cooperation in postgraduate training and scientific research, which enables us to explore their coauthorship networks and collaboration. (3) We could easily verify and refine the author’s information from the two universities through interpersonal relationships to ensure the accuracy of the data as much as possible. (4) Both universities require doctoral students to publish academic works with certain impact factors in foreign journals included in the SCI before earning a degree. (5) All the graduate papers from the two universities can be retrieved from the Chinese National Knowledge Infrastructure (CNKI) database. We sampled doctoral graduates according to their disciplines, including clinical medicine and basic medicine. There were 125 samples from each university. We used the following exclusion criteria to filter the raw sample and obtain the effective sample: (1) Doctoral students who had not published SCI papers during their studies; (2) Incomplete data or extensive missing data; (3) Cases in which it could not be determined whether the data were related to the research object. Ultimately, 237 valid samples were obtained, including 117 from Army Medical University and 120 from Chongqing Medical University.

Data collection and preprocessing

All the data for this study were gathered from the CNKI database and Web of Science (WoS) database. The CNKI is the largest Chinese academic literature database, with a complete collection and continuously updated graduate dissertations and related information (Zuo et al. 2021), and the WoS contains publications in almost all major scientific fields and is considered an ideal database for literature research (Zuo et al. 2021).

The specific process of data collection was as follows: We obtained the doctoral student’s name, department, major, doctoral study periods and mentor information from the CNKI database. Then, we obtained the publications and h-index of the samples through the WoS. Publications were retrieved from the WoS through a set of developed search terms, and the time span was set for doctoral students as their training period and for mentors as the mentors’ tenure. Publications, which encompassed all papers in which the target appeared in any position in the authorship, included articles, reviews, meeting abstracts, and short essays. For each publication, we extracted the title, publication date, coauthors and addresses. The data collection from the CNKI database was completed in May 2022, and the publication collection was completed in December 2022. The raw data consisted of 21686 publications by 250 doctoral students and their mentors.

The preprocessing of publication data is key to obtaining reliable analysis results. In this study, the original data were processed as follows: (1) Publications were checked manually by author names, units and research areas to disambiguate authors with the same name (Zeng et al. 2017). First, we checked whether the author’s byline name was the same. For example, the results of a search for papers by Li Xuliang, an author from Chongqing Medical University showed that Li Xuliang, Li Xu Liang, Li, Xuliang, Li Xu-liang and Li, Xianliang might or might not be the same author. Then, we further checked whether the above authors were from Chongqing Medical University, and papers not from this school were excluded. Finally, we checked whether the research topics of the remaining papers were in line with the target author’s major and research field from the CNKI database. If not, the authors were judged not to be the target author. If the author’s name in the article was abbreviated in the WoS, it was reverified with the DOI (digital object identifier), title and other information through other databases, such as PubMed and Sci-Hub, and if it was still not confirmed, it was deleted. (2) One researcher labeled the papers of nonsubjects and eliminated the duplicate papers under one author, and another researcher crosschecked the data until the two reached a consensus. (3) We then manually standardized the writing formats of each author’s name in all publications. Ultimately, 5395 papers were excluded, and the remaining 16291 papers, by 237 doctoral students and 126 mentors, were retained. Among them, 1441 papers were published by authors who were then doctoral students. We linked all the information retrieved from the above database with the student-mentor pairs in a new dataset so that we obtained a research sample consisting of 237 such pairs (see Dataset 1).

Variables

Dependent variable

We used the h-index of the doctoral students’ published papers as the only dependent variable. The well-known h-index, proposed by Hirsch in 2005 (Hirsch, 2005), is used to evaluate an individual’s research impact objectively (Wendl, 2007). This index is calculated on the basis of publications and citations (Abbas, 2012) as follows: h=max(i):C_i ≥ i. (Suppose the papers are arranged in descending order of the number of citations, and let C_i be the number of citations of a paper numbered i.). The h-index in this study was obtained through the WoS database on the basis of published papers and citations.

Independent variables

We constructed nine variables from cognitive capital, structural capital and relational capital in doctoral students’ coauthorship networks.

1. Cognitive capital included mentors’ h-index and fecundity, which are common measures of mentors’ academic success (Ma et al. 2020; Malmgren et al. 2010). Fecundity was defined as the number of postgraduates supervised by the mentor from the enrollment year of the first postgraduate supervised by the mentor to the graduation year of the target doctoral student, excluding the target doctoral student.

2. Structural capital was evaluated by doctoral students’ closeness centrality, betweenness centrality and degree centrality (Kumar, 2015). The centralities were calculated via UCINET software (Borgatti et al. 2002) based on the coauthorship networks, which were constructed via Co-Occurrence (COOC) software (Qin et al. 2022).

3. Relational capital was measured by the doctoral students’ coauthor count, coauthorship unit count, student-mentor coauthorship count and the partnership ability index (Schubert, 2012). The doctoral students’ coauthor count and doctoral students’ coauthorship unit count were obtained by manual calculation on the basis of the students’ publications, and the duplicate coauthors and coauthor units were calculated only once. The student-mentor coauthorship count was the number of publications in which the student appeared with his or her mentor in the coauthorship of a publication. The partnership ability index, proposed by Schubert to assess authors’ cooperation behaviors, comprehensively reflects the number of collaborators and number of repeats of cooperation (Schubert, 2012). Its calculation method is shown in Fig. S1 (see the Supplementary file).

Control variables

The characteristics of the doctoral students and mentors were used as control variables, such as the doctoral students’ subject, doctoral students’ training period, mentors’ title and mentors’ tenure. The doctoral students’ subject and training period were controlled for heterogeneity across the thesis research fields and for research time effects. The title and tenure of the mentor were controlled for his or her experience and time influence in research and mentoring activities. These variables were not relevant to the study’s aims but were controlled because they might have influenced the outcomes (Corsini et al. 2022). All the variables included in our analysis are listed in Table 1 with a short description and sources.

Table 1. List of variables used in the analysis.

Variables	Description	Source
Dependent variable
Doctoral students’ h-index	The h-index was calculated according to the number and citations of publications between Yg and Yea.	WoS database
Independent variables
Cognitive capital
Mentors’ h-index	The h-index was calculated according to the number and citations of publications between Yg and Yf.	WoS database
Mentors’ fecundity	The number of all postgraduate students supervised by the mentor from Yf to Yg minus 1.	Data accumulation through CNKI database
Structural capital
Degree centrality	The number of direct connections between the target doctoral student and other coauthors regardless of the connection strength in the coauthorship network.	UCINET software
Closeness centrality	The mean shortest distance from the target doctoral student to all the other coauthors in the coauthorship network.	UCINET software
Betweenness centrality	The proportion of the shortest paths between all the pairs of nodes that pass through the target doctoral student in the coauthorship network.	UCINET software
Relational capital
Doctoral students’ coauthor count	The number of coauthors in the doctoral student’s publications excluding herself or himself, and the count is still 1 if the same coauthor appears more than once in the publications.	Data calculation based on publications analysis
Doctoral students’ coauthorship unit count	The number of coauthorship units in the doctoral student’s publications excluding her or his own unit, and the count is still 1 if the same unit appears more than once in the publications.	Data calculation based on publications analysis
Student-mentor coauthorship count	The number of publications in which the student coappeared with his or her mentor in the coauthorship of a publication.	Data calculation based on publications analysis
Doctoral students’ partnership ability index	The collaborators of a certain author are listed in descending order by the frequency of their cooperation with the author. When the author serial number is greater than the corresponding cooperation frequency for the first time, the serial number minus 1 is the author’s partnership ability index.	Data calculation based on publications analysis
Control variables
Doctoral students’ subject	Dummy variables that equal one if the doctoral student’s subjects are in clinical medicine.	CNKI database
Doctoral students’ training period	The number of years elapsed from Ye to Yg.	Calculation by data from CNKI database
Mentors’ title	A dummy variable that equals one if the mentor is a professor.	CNKI database
Mentors’ tenure	The number of years elapsed from Yf to Yg.	Calculation by data from CNKI database

Note: Y_e is the target doctoral student’s enrollment year; Y_g is the target doctoral student’s graduation year; Y_f is the enrollment year of the first postgraduate supervised by the mentor.

Hypotheses

A predesigned model (see Fig. 1) on the basis of SCT was proposed to facilitate answering the first two questions, that is, what social capital from coauthorship networks influences the research impact of doctoral students and its action paths? To examine the influence of cognitive capital on research impact, we predicted that doctoral students who are supervised by more academically successful mentors will have greater research impact (H1a, H1b), that is:

Fig. 1 [Images not available. See PDF.]

Conceptual framework.

H1a: A higher mentor’ s h-index is associated with a higher doctoral students’ h-index.

H1b: Higher mentor’ s fecundity is associated with a higher doctoral students’ h-index.

We further hypothesized that doctoral students with greater centrality in their coauthorship networks will have greater research impact (H2a, H2b, H2c), as follows:

H2a: Higher doctoral students’ closeness centrality is associated with a higher h-index.

H2b: Higher doctoral students’ betweenness centrality is associated with a higher h-index.

H2c: Higher doctoral students’ degree centrality is associated with a higher h-index.

Building on the previous hypotheses, H2a, H2b and H2c sought to verify the relationship between structural capital and research impact, and the effects of relational capital on this relationship were also tested by the following hypothesis. Doctoral students who have established good cooperative relationships with others (units) will have greater research impact (H3a, H3b, H3c, H3d). Therefore, we postulated the following:

H3a: A higher doctoral students’ coauthor count is associated with a higher h-index.

H3b: A higher doctoral students’ coauthorship unit count is associated with a higher h-index.

H3c: A higher doctoral students’ student-mentor coauthorship count is associated with a higher h-index.

H3d: A higher doctoral students’ partnership ability index is associated with a higher h-index.

Moreover, the associations among the three dimensions of social capital and among the three centralities of structural capital were used to refine the model. Doctoral students with insufficient structural capital in the early stages of their studies are likely to expand their social networks by investing in their relational and cognitive capital through cognitive efforts and proactive behaviors (Li et al. 2013; Smith, 2007). Mentors with a high h-index may work with more collaborators, which, in turn, may bring more close collaborators to students (McCarty et al. 2013). Thus, we predicted that doctoral students supervised by mentors with a higher h-index will have more structural capital and will especially improve their closeness centrality and degree centrality (H4a, H4b). The specific hypotheses were as follows:

H4a: A higher mentors’ h-index is associated with higher doctoral students’ closeness centrality.

H4b: A higher mentors’ h-index is associated with higher doctoral students’ degree centrality.

Similarly, we hypothesized that doctoral students with more relational capital will bring in more structural capital (H5a, H5b, H5c, and H5d), and H5a covers the three assumptions about the effects of one indicator of relationship capital on closeness centrality, betweenness centrality and degree centrality, as do H5b, H5c and H5d. Thus, the following hypotheses were proposed:

H5a (1,2,3): A higher doctoral students’ coauthor count is associated with higher closeness centrality/betweenness centrality/degree centrality.

H5b (1,2,3): A higher doctoral students’ coauthorship unit count is associated with higher closeness centrality/betweenness centrality/degree centrality.

H5c (1,2,3): A higher doctoral students’ student-mentor coauthorship count is associated with higher closeness centrality/betweenness centrality/degree centrality.

H5d (1,2,3): A higher doctoral students’ partnership ability index is associated with higher closeness centrality/betweenness centrality/degree centrality.

Additionally, an actor should first establish direct contact and then develop close relationships with her or his collaborators so that she or he can become the key controller of effective communication in the network. Therefore, we predict that the higher the degree centrality or closeness centrality is, the greater the actor’s betweenness centrality is likely to be (H6a, H6b).

H6a: Higher doctoral students’ degree centrality is associated with higher betweenness centrality.

H6b: Higher doctoral students’ closeness centrality is associated with higher betweenness centrality.

Measurement

In this study, Co-Occurrence12.8 (COOC) software was used to construct and analyze the co-occurrence matrix. COOC software is bibliometric software that uses accurate character segment recognition algorithms to ensure the quality of data analysis (Qin et al. 2022). The author list from each student’s papers was input into the software to construct a co-occurrence matrix, with each number in the matrix indicating the number of times any two people were coauthors, which was the basis for calculating the doctoral students’ coauthor count, the partnership ability index and the student-mentor coauthorship count. Additionally, the co-occurrence matrix is the data form analyzed by UCINET software, which is a common tool for social network analysis that can compute the centralities for each node in a matrix (Borgatti et al. 2002). All the authors listed in the papers from 237 doctoral students were imported into the COOC software, from which 4380 coauthors were extracted to construct a co-occurrence matrix (4380×4380), which was then imported into the UCINET software to calculate the closeness centrality, betweenness centrality and degree centrality of the doctoral students. The software operation process is shown in the Supplementary File.

Finally, the data were analyzed using SPSSAU online application software (version 23.0), which was retrieved from https://www.spssau.com. Nonparametric tests were selected because of the nonnormal distribution of the data. Descriptive statistics, Spearman correlation analysis and path analysis were used to answer the first two research questions: what social capital affects the research impact of doctoral students, and what is the action path of this effect? Descriptive statistics and Spearman correlation analysis were used to analyze the basic characteristics of the relationships among all the indicators. Path analysis in structural equation modeling (SEM) was adopted to test the hypothesized associations and the predesigned model (see Fig. 1), which can simultaneously evaluate relationships among multiple variables and provide model fitting data to verify the degree of match between the predesigned model and the empirical data (Davvetas et al. 2020). Quantile regression, logistic regression and receiver operating characteristic (ROC) curves were further used to answer the last research question: what is the special effect pattern of this social capital on the research impact of doctoral students? We focused on the influencing trends and predictive effects among the variables, which are valuable for improving the doctoral students’ learning process. Quantile regression was performed to explore the influence of independent variables on the h-index when the dependent variables were divided into 9 parts at a 10% interval (Das et al. 2019). Subsequently, the doctoral students at the top and bottom 30% of the h-index were divided into an excellent group and a poor group. The Wilcoxon rank-sum test was used to test differences in the indicators between the groups. Logistic regression was used to identify the independent variables that could predict the probability of a doctoral student entering the excellent group. The single-variable ROC curves were chosen instead of the comprehensive multivariable ROC curve because they could further evaluate the predictive efficacy of each independent variable on the students’ h-index and determine its optimal cutoff value, which provides a specific standard reference in practice (Søreide et al. 2011). Finally, a robustness check was used to ensure the reliability of the results, and two-tailed p < 0.05 was considered statistically significant.

Results

Descriptive statistics, correlation analysis, path analysis and quantile regression analysis were first performed on the total sample (n = 237). The logistic analysis and ROC curve analysis were then performed on the subsample (n = 175). Finally, a robustness check was used to explore the stability of the results across the whole sample (n = 237).

Descriptive statistics

The sample consisted of 237 doctoral students, 79% of whom were clinical medicine students (n = 188); the rest were basic medicine students. The mean doctoral training time of the students was 3.6 years (SD = 0.99). They had published an average of 6.1 papers during their training, with the publication years ranging from 2012–2021. The mentors were almost all professors, with an average tenure of 15.8 years (SD = 4.71), an average of 62.7 publications, and the highest h-index of 38 (see Table 2). The h-index of the doctoral students had a positive relationship with all the independent variables (r_s= 0.26 ~ 0.80, p < 0.01) except mentors’ fecundity (see Table S1).

Table 2. Descriptive statistics for the 237 doctoral student-mentor pairs.

Variables	Mean (SD)/no. (%)	Min	Max
Doctoral students’ h-index	4.17 (2.96)	0.00	17.00
Doctoral students’ publications	6.08 (4.46)	1.00	22.00
Doctoral students’ yearly citation count	16.37 (17.93)	0.00	128.78
Mentors’ h-index	18.41 (7.28)	0.00	38.00
Mentors’ publications	62.66 (39.96)	1.00	202.00
Mentors’ yearly citation count	119.13 (109.76)	0.00	731.10
Mentors’ fecundity	44.87 (29.58)	1.00	113.00
Degree centrality	0.64 (0.49)	0.02	3.56
Closeness centrality	1.00 (0.18)	0.02	1.04
Betweenness centrality	0.22 (0.46)	0.00	4.38
Doctoral students’ coauthor count	16.50 (17.22)	1.00	123.00
Doctoral students’ coauthorship unit count	6.86 (6.26)	0.00	43.00
Student-mentor coauthorship count	4.64 (3.91)	0.00	18.00
Doctoral students’ partnership ability index	3.15 (1.81)	0.00	10.00
Doctoral students’ training period	3.64 (0.99)	3.00	10.00
Mentors’ tenure	15.80 (4.71)	3.00	25.00
Doctoral students’ subject		0.00	1.00
Clinical medicine	188 (79.3%)
Basic medicine	49 (20.7%)
Doctoral students’ university
Army Medical University	117 (49.4%)
Chongqing Medical University	120 (50.6%)
Mentors’ title		0.00	1.00
Professor	233 (98.3%)
Associate professor	4 (1.7%)

Path analysis

The path analysis results (see Fig. 2) revealed that betweenness centrality (H2b: β = 0.29, p < 0.001), the partnership ability index (H3d: β = 0.38, p < 0.001), and the student-mentor coauthorship count (H3c: β = 0.33, p < 0.001) had a significant impact on the doctoral students’ h-index. The values of the variance inflation factor (VIF) of the variables were less than 5 except for degree centrality (VIF = 7.23), which was below the common threshold of 10.0, indicating that multicollinearity among these variables was acceptable (see Table S1). This finding supported the important hypothesis that the structural capital and relational capital of doctoral students positively affect their h-index. The results regarding the influence of cognitive capital and relational capital on structural capital revealed that mentors’ h-index, doctoral students’ coauthor count, coauthorship unit count, the partnership ability index and student-mentor coauthorship count all had significant effects on the three centralities of structural capital. Moreover, degree centrality had a significant positive effect on betweenness centrality (H6a: β = 1.01, p < 0.001), and closeness centrality did not have this effect. In addition, mentors’ tenure, as a control variable, had a significant effect on doctoral students’ h-index (β = −0.14, p < 0.01). The independent variables explained 72% of the h-index, 76% of the betweenness centrality, 9% of the closeness centrality and 64% of the degree centrality. Therefore, nine hypotheses proposed in this study were accepted, and the others were rejected (see Table 3).

Fig. 2 [Images not available. See PDF.]

Structural equation modeling (SEM) of the relationships between the variables.

The coauthor count represents doctoral students’ coauthor count; the coauthorship unit count represents doctoral students’ coauthorship unit count; the SM coauthorship count represents the student-mentor coauthorship count; and the partnership ability index represents doctoral students’ partnership ability index.

Table 3. Hypothesis testing.

Hypothesis	Relationship			UnStd.	S.E.	C.R.	p	Std.(β)	Results
H2b	Betweenness centrality	→	Doctoral students’ h-index	1.83	0.45	4.10	<0.001	0.29	Supported
H3c	Student-mentor coauthorship count	→	Doctoral students’ h-index	0.25	0.05	4.69	<0.001	0.33	Supported
H3d	Doctoral students’ partnership ability index	→	Doctoral students’ h-index	0.62	0.12	5.31	<0.001	0.38	Supported
H4a	Mentors’ h-index	→	Closeness centrality	0.01	0.00	3.03	<0.01	0.19	Supported
H4b	Mentors’ h-index	→	Degree centrality	0.01	0.00	2.40	<0.05	0.10	Supported
H5a (3)	Doctoral students’ coauthor count	→	Degree centrality	0.01	0.00	7.05	<0.001	0.33	Supported
H5b (3)	Doctoral students’ coauthorship unit count	→	Degree centrality	0.04	0.00	11.95	<0.001	0.55	Supported
H5d (3)	Doctoral students’ partnership ability index	→	Degree centrality	0.07	0.02	3.34	<0.01	0.25	Supported
H6a	Degree centrality	→	Betweenness centrality	0.96	0.05	19.08	<0.001	1.01	Supported
H1a	Mentors’ h-index	→	Doctoral students’ h-index	0.02	0.02	1.22	0.22	0.05	Not Supported
H1b	Mentors’ fecundity	→	Doctoral students’ h-index	0.01	0.01	1.63	0.10	0.09	Not Supported
H2a	Closeness centrality	→	Doctoral students’ h-index	−0.35	0.59	−0.59	0.56	−0.02	Not Supported
H2c	Degree centrality	→	Doctoral students’ h-index	0.54	0.55	0.97	0.33	0.09	Not Supported
H3a	Doctoral students’ coauthor count	→	Doctoral students’ h-index	−0.01	0.01	−0.60	0.55	−0.03	Not Supported
H3b	Doctoral students’ coauthorship unit count	→	Doctoral students’ h-index	0.01	0.03	0.51	0.61	0.03	Not Supported
H5a (1)	Doctoral students’ coauthor count	→	Closeness centrality	0.00	0.00	0.81	0.42	0.06	Not Supported
H5a (2)	Doctoral students’ coauthor count	→	Betweenness centrality	0.00	0.00	1.50	0.13	0.06	Not Supported
H5b (1)	Doctoral students’ coauthorship unit count	→	Closeness centrality	0.00	0.00	0.33	0.74	0.03	Not Supported
H5b (2)	Doctoral students’ coauthorship unit count	→	Betweenness centrality	−0.01	0.00	−1.98	<0.05	−0.10	Not Supported
H5c (1)	Student-mentor coauthorship count	→	Closeness centrality	0.00	0.01	0.27	0.79	0.03	Not Supported
H5c (2)	Student-mentor coauthorship count	→	Betweenness centrality	−0.02	0.01	−2.74	<0.01	−0.17	Not Supported
H5c (3)	Student-mentor coauthorship count	→	Degree centrality	−0.03	0.01	−2.78	<0.01	−0.21	Not Supported
H5d (1)	Doctoral students’ partnership ability index	→	Closeness centrality	0.01	0.01	0.94	0.35	0.11	Not Supported
H5d (2)	Doctoral students’ partnership ability index	→	Betweenness centrality	−0.04	0.02	−2.67	<0.01	−0.17	Not Supported
H6b	Closeness centrality	→	Betweenness centrality	−0.09	0.08	−1.06	0.29	−0.04	Not Supported

Note: UnStd. = unstandardized path coefficients, S.E. = standard error, C.R. = critical ratio, and Std. (β) = standardized path coefficients.

We assessed the goodness of fit of the model via several commonly used measures: the chi-square degree of freedom ratio (χ²/df), the comparative fit index (CFI), the standardized root mean square residual (SRMR), the root mean square error of approximation (RMSEA), and the nonnormed fit index (NNFI) (Hooper et al. 2008; McDonald and Ho, 2002). The widely used empirical rules state that the model fit is excellent when the value of χ²/df is less than 3 (χ²/df=1.12, p > 0.05), the CFI and NNFI exceed 0.95, the SRMR is less than 0.05 (SRMR = 0.02) and the RMSEA is less than 0.05 (RMSEA = 0.02) (Hooper et al. 2008; Keinänen et al. 2018).

Quantile regression analysis

In the quantile regression analysis, we focused on the four variables that had a direct significant effect on the h-index in the path analysis: betweenness centrality, student-mentor coauthorship count, the partnership ability index and mentors’ tenure. The results revealed different influence trends of these variables (see Fig. 3). The strength of betweenness centrality increased incrementally on the h-index from 0.15 (B: β = 1.25, p < 0.001) to 0.65 (B: β = 2.81, p < 0.001), then decreased to 0.75 (B: β = 2.40, p < 0.001) and finally began to increase rapidly to 0.95 (B: β = 3.82, p < 0.001). The significant effect of the student-mentor coauthorship count on the h-index increased incrementally from quantiles 0.05 (A: β = −0.14, p < 0.001) to 0.85 (A: β = 0.40, p < 0.001) and then decreased to quantiles 0.95 (A: β = 0.31, p < 0.05). The effect of the partnership ability index on the h-index fluctuated before the 0.55 quantile, and the effect gradually increased from the 0.55 quantile (B: β = 0.50, p < 0.001), especially from the 0.85 quantile (B: β = 0.74, p < 0.001) to the 0.95 quantile (B: β = 1.13, p < 0.001). Mentors’ tenure had a significant negative effect only on the h-index at the 0.95 quantile (B: β = −0.05, p < 0.05) and had no effect on the other quantiles.

Fig. 3 [Images not available. See PDF.]

Quantile regression of variables on the h-index.

a. Betweenness centrality; b. Student-mentor coauthorship count; c. Doctoral students’ partnership ability index; d. Mentors’ tenure; e. Results of quantile regression analysis.

Logistic regression and ROC curve analysis

The subsample consisted of 175 doctoral students and was divided into an excellent group (n = 88) and a poor group (n = 87). Wilcoxon rank-sum tests were conducted to compare the numbers of publications and yearly citation counts between the two groups, revealing significant differences (p < 0.001), but there was no significant difference in the doctoral students’ training period (Fig. 4c). The three variables of betweenness centrality, student-mentor coauthorship count and the partnership ability index were included in the subsequent analysis, while mentors’ tenure had only a weak negative effect on the h-index for specific doctoral students in the quantile regression; thus, it was excluded from the variables. The logistic regression model was statistically significant (χ² = 194.47, df = 3, p < 0.001) and explained 80% of the reasons for students entering different groups. The independent variables of student-mentor coauthorship count (z = 2.65, p < 0.01), the partnership ability index (z = 3.17, p < 0.01) and betweenness centrality (z = 3.82, p < 0.001) had significant predictive effects on the h-index. By coauthoring publications with their mentors, doctoral students increased the likelihood of their h-index being in the top 30% by 87.6% (OR 1.88, 95% CI 1.18 ~ 2.99). For each one-unit increase in the partnership ability index (OR 7.73, 95% CI 2.19 ~ 27.32) and betweenness centrality (OR 2476.18, 95% CI 44.97 ~ 136339.17), doctoral students were nearly 8-fold and more than 2000-fold more likely to enter the excellent group, respectively. The overall prediction accuracy of the model was 94.29%, and the Hosmer–Lemeshow test (p > 0.05) revealed that the model fit was acceptable. The formula of the model is expressed as follows:$$\begin{array}{c}\ln \left(\frac{p}{1-p}\right)=-9.91+0.63\times SMcoauthored+2.05\times Partnership\hfill \\ +7.81\times B\mathrm{etweenness}\mathrm{centrality}\hfill \end{array}$$Note: p represents the probability of being in the excellent group, SMcoauthored represents the student-mentor coauthorship count, and Partnership represents the partnership ability index.

Fig. 4 [Images not available. See PDF.]

Differences in the variables between the excellent and poor groups.

a. Doctoral students’ publications; b. Doctoral students’ yearly citation count; c. Doctoral students’ training period.

ROC curve analysis was performed on the important predictors of the above results, including betweenness centrality, student-mentor coauthorship count and the partnership ability index. However, since betweenness centrality could only be obtained through social network analysis software, it would be difficult to formulate improvement measures for management. Therefore, doctoral students’ coauthorship unit count, which had a significant effect on betweenness centrality in path analysis, was also included in the analysis. The area under the ROC curve (AUC) values of the student-mentor coauthorship count, partnership ability index, betweenness centrality and coauthorship unit count were 0.92 (95% CI 0.87 ~ 0.96, p < 0.001), 0.96 (95% CI 0.93 ~ 0.98, p < 0.001), 0.86 (95% CI 0.80 ~ 0.91, p < 0.001), and 0.88 (95% CI 0.84 ~ 0.93, p < 0.001), respectively. The Youden index values were 0.77, 0.77, 0.62 and 0.60, respectively. The results revealed that the student-mentor coauthorship count and the partnership ability index were highly valuable in predicting whether doctoral students would enter the excellent group, with optimal cutoff values of 4 and 2, respectively. The betweenness centrality and coauthorship unit count also had certain values, with optimal cutoff values of 0.02 and 6, respectively (see Fig. 5).

Fig. 5 [Images not available. See PDF.]

ROC curves of the variables.

Robustness check

To make the results more likely to reflect the current situation, we chose doctoral students who had graduated in recent years as the samples. We considered the time lag effect of citations after publication because article citations may be low during the first few years after publication. Although a reasonable citation time window is still under debate, some studies suggest that citations reach a maximum at two years after an article is published (Schreiber, 2015). However, of all the doctoral students’ publications, 6% of the papers (n = 85) were published in 2021, which may have affected the h-index. As a further robustness check, we ran a regression selecting the number of doctoral students’ publications as the dependent variable. The results were consistent with the main findings of this study, which confirms the important role of betweenness centrality, the partnership ability index, and student-mentor coauthorship count in the dependent variable (see Table S2).

Discussion

In this study, we investigate the relationship between coauthorship networks and the research impact of medical doctoral students from a social capital perspective. Research impact has complex connotations, including the changes it brings to academia and beyond, such as social, economic and environmental benefits (Bærøe et al. 2022; Belcher and Halliwell, 2021). In academia, research impact demonstrates the contribution of quality research to academic advances in theory and application, and publications as research outcomes reflect this impact (Greenhalgh et al. 2016). This study focused on publications by doctoral students, as these are almost universal research outputs that students generate during their studies, thus facilitating comparisons with each other, but it may have ignored whether their research produced more value in fields outside academia. Although publications and impact indicators do not represent the full research competence of a person or institution, they are easy to quantify and obtain and not easily affected by personal subjective judgment, which makes them widely accepted and used for comparison and ranking between universities or individuals in the context of increasingly fierce quality competition (Musselin, 2018). This also leads to a focus on the published papers and impact index of doctoral students in their training without guiding the students to develop needed competencies in the implementation process. In this study, publications and the h-index were all obtained from the WoS database, which includes almost all global scientific documents and provides high-quality and continuously updated data to meet the needs of scientific research (Birkle et al. 2020) and as a result was a reliable data source. The results confirmed the positive influence of coauthorship on the research impact of doctoral students, which reflects the importance of cooperation in the development of doctoral students’ research competence. This is a necessary supplement to the core competence that medical students should possess under a competency-based medical education model and should receive increased attention from the government or at the university level (Leiphrakpam and Are, 2023).

Deepening the understanding of the role of mentorship, this study provides evidence that mentorship enhances the research impact of doctoral students by furnishing favorable social capital through their coauthorship networks. Previous studies have shown that the greater a mentor’s research impact is, the better the research performance of her or his postgraduates is. For example, mentors who are themselves Nobel Prize winners often cultivate more future Nobel Prize winners (Chariker et al. 2017). This may lead doctoral students to choose mentors with high academic performance, resulting in the Matthew effect of the “rich get richer” in science (Perc, 2014). However, our research suggests that the student-mentor coauthorship count and the partnership ability index of relational capital and the betweenness centrality of structural capital are the main factors that have a positive effect on the h-index. We did not find a direct influence of mentors’ h-index on doctoral students’ h-index, which is in contrast to Jing Shang’s research (Shang et al. 2022). This finding indicates that it is very important for doctoral students to maintain trusting and committed relationships with their mentors to increase their relational capital first. In addition to providing professional and emotional support to doctoral students, mentors should guide them in scientific cooperation. In addition, this suggests that doctoral students should consider the selection of mentors more comprehensively rather than being determined by mentors’ research impact, which may be more conducive to the establishment of good mentorship and also bring more equitable opportunities for young menors or those in disadvantaged disciplines to recruit students.

To increase their coauthorship, the quantile regression indicated that the top 5% of doctoral students with an h-index above 9 should focus on strengthening cooperation with different researchers and increasing their control over resources in the coauthorship network to achieve more innovation breakthroughs rather than relying on cooperation with their mentors, which is similar to the advice for scholars in Xu and Chang’s study (Xu and Chang, 2020). Close relationships with their mentors can lead to exclusivity and leave no space for others to collaborate, which is not conducive to innovation (Liu-Lastres and Cahyanto, 2023). Furthermore, the partnership ability index reflects strong ties between coauthors rather than simply increasing the number of collaborators, enabling better research performance. This is also consistent with the findings of (Abbasi et al. 2011). In addition, we confirmed the negative effect of mentors’ tenure on the h-index of doctoral students, which is consistent with previous research on citations (Corsini et al. 2022). Specifically, students in the top 5% of the h-index may experience some negative effects, and the impact on the remaining students can be ignored.

This study revealed that the betweenness centrality positively affected the h-index, but degree and closeness centralities did not have any significant effect, which is similar to Eldon Y. Lia’s research on scholars’ citation counts (Li et al. 2013). The greater the betweenness centrality of a doctoral student is, the stronger his or her ability to control information and resources beyond other persons in the network is (Xu and Chang, 2020). There were significant positive relationships between betweenness centrality and both degree centrality (r_s= 0.81, p < 0.01) and closeness centrality (r_s= 0.65, p < 0.01). This result is consistent with those of previous studies, but these studies did not explain the influence relationship between them (Abbasi et al. 2011; Li et al. 2013; Xu and Chang, 2020). We further confirmed that only degree centrality had a significant positive effect on betweenness centrality (β = 1.01, p < 0.001), whereas closeness centrality had no significant effect. Moreover, degree centrality was significantly positively affected by the coauthor count, coauthorship unit count, the partnership ability index of a doctoral student, and mentors’ h-index, especially the coauthorship unit count (β = 0.55, p < 0.001), which had the greatest effect. This is because degree centrality is a measure of the number of collaborators (Lu and Feng, 2009). Therefore, increasing the number of coauthors and coauthorship units of doctoral students not only directly affects degree centrality but also indirectly enhances betweenness centrality to increase their research impact.

Additionally, we found that the student-mentor coauthorship count, the partnership ability index and betweenness centrality had significant predictive effects on the h-index. In particular, increasing the partnership ability index and betweenness centrality greatly increased a student’s likelihood of having high research impact. Specifically, doctoral students whose partnership ability index is greater than 2, whose student-mentor coauthorship count is greater than 4, whose coauthorship unit count is greater than 6, and whose betweenness centrality is greater than 0.02 have high productivity. The results have some practical implications for mentors, doctoral students, and educational institutions. As role models for doctoral students, mentors should be good at cooperating with others; building teams; and creating an atmosphere of fairness, trust and respect so that doctoral students can integrate into the teams quickly and establish good cooperative relationships with team members. Additionally, mentors should assume the responsibility for developing students’ collaborative capacity and help them expand their cooperation networks with individuals, teams and relevant industries through different research projects (Gisbert, 2017). For doctoral students, establishing a win-win mode of thinking, viewing colleagues as potential partners rather than competitors, and learning to coexist and develop harmoniously with others should be a recommended value orientation for their growth (Zhang et al. 2021). Educational institutions should promote the scientific cooperation of doctoral students at the organizational level, providing policy and financial support to address barriers to intersectional and multidisciplinary cooperation (Leenaars et al. 2015). Moreover, these cooperation indicators should be included in the evaluation system to track doctoral students’ research performance to identify and cultivate future stars in academia earlier.

Limitations of the study

This study has several limitations. First, we used the h-index as the only dependent variable representing the research impact, and more indicators from funds, patents and technology transfer may be further used to confirm these results. Meanwhile, nine independent variables were explored via SCT. Although structural capital is well explained by the three centrality measures, other measures may better reflect relational capital and cognitive capital. For example, relational capital may be further measured by students and their mentors’ ranking on the author list of an article (Xie et al. 2022). The research theme similarity between students and mentors can also be considered a direction for exploring the relationship between cognitive capital and research impact in the future (Liénard et al. 2018). Moreover, the results reveal the influence of degree centrality and betweenness centrality in structural capital on the h-index, but the effect of closeness centrality should be considered in further research. Second, this study is based on 237 medical doctoral students who graduated from two Chinese universities from 2016–2021 and their mentors, which may hinder the generalizability of the results to other periods or disciplines. Future studies could generalize these findings by replicating studies in multiple disciplines using larger datasets and longer study periods. Third, there is a growing need for interdisciplinary research in academia. This study examined the impact of increasing the number of cooperative units on the research impact of doctoral students, while different cooperative units may indicate the diversity of the subject majors of the collaborators. It is necessary to further verify the differences in social capital resulting from interdisciplinary and multiprofessional cooperation and their effects on students’ research impacts in future studies.

Conclusion

This study defines nine independent variables from doctoral students’ coauthorship network based on three dimensions of social capital theory and examines how these indicators interact, influence, and predict doctoral students’ h-index. The results show that utilizing social capital from coauthorship networks, especially betweenness centrality, student-mentor coauthorship count and the partnership ability index, is an effective way to increase the research impact of medical doctoral students. The results show the important role of collaboration in improving doctoral research competence and deepen the understanding of good mentorship in doctoral students’ development. Most importantly, they provide several strategies for harnessing social capital for relevant organizations, mentors and doctoral students who want to increase their research impact. First, cooperation in cultivating doctoral students should be considered, and the shaping of cooperative spirit, norms and skills in education should be strengthened. Second, more abundant cooperation resources for doctoral students should be provided, and a smoother cooperation platform through policy orientation and administrative intervention should be built to promote the cross-unit cooperation mode among multiple scholars with strong connections. Third, a concrete and feasible development plan should be created according to the judgment criteria and predictive value of the indicators. For example, a partnership ability index greater than 2 inspired us to believe that doctoral students should have a team of at least 3 mentors from different institutions at the beginning of their schooling to accelerate collaboration and promote research impact. Meanwhile, doctoral students should strengthen communication with their mentors, maintain good teacher‒student relationships, and enhance their personal ability to obtain information and resources in coauthorship networks. Doctoral students with high h-index levels should not rely on mentors but should expand their scope of cooperation, actively cooperate with different scholars from different institutions, and strengthen their cooperative relationships.

Acknowledgements

This work was supported by the Postgraduate Education Reform Foundation of Chongqing [grant number yjg212042]; the Research Project of Graduate Education and Teaching Reform in Army Medical University [grant number 2022yjgA02]; and the Project of Association of Chinese Graduate Schools [grant number ACGS05-2024040]. We express our thanks to the Xueshudiandi Team for the development of the Co-Occurrence 12.8 (COOC) software.

Author contributions

GC: Investigation; Data curation; Formal analysis; Writing - original draft. W-WY: Investigation; Data curation; Formal analysis. X-YW: Investigation; Data curation. QN: Formal analysis. YX: Data curation. XM: Conceptualization; Supervision. J-JY: Conceptualization; Formal analysis; Funding acquisition; Project administration; Writing - review & editing. All authors approved the final manuscript.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethical approval

This article does not contain any studies with human participants performed by any of the authors.

Informed consent

This article does not contain any studies with human participants performed by any of the authors.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1057/s41599-024-03813-9.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

Abbas, AM. Bounds and Inequalities Relating h-Index, g-Index, e-Index and Generalized Impact Factor: An Improvement over Existing Models. PLoS ONE; 2012; 7, 4 e33699.2012PLoSO..733699A[COI: 1:CAS:528:DC%2BC38Xls1Ghs7o%3D] [DOI: https://dx.doi.org/10.1371/journal.pone.0033699] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22496760][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3319552]

Abbasi, A; Altmann, J; Hossain, L. Identifying the effects of co-authorship networks on the performance of scholars: A correlation and regression analysis of performance measures and social network analysis measures. J. Informetr.; 2011; 5, pp. 594-607. [DOI: https://dx.doi.org/10.1016/j.joi.2011.05.007]

Abramo, G; D’Angelo, AC; Murgia, G. The relationship among research productivity, research collaboration, and their determinants. J. Informetr.; 2017; 11, 4 pp. 1016-1030. [DOI: https://dx.doi.org/10.1016/j.joi.2017.09.007]

Adler, PS; Kwon, S-W. Social capital: prospects for a new concept. Acad. Manag. Rev.; 2002; 27, 1 pp. 17-40. [DOI: https://dx.doi.org/10.5465/amr.2002.5922314]

Ahmad, R; Nawaz, MR; Ishaq, MI; Khan, MM; Ashraf, HA. Social exchange theory: Systematic review and future directions [Systematic Review]. Front. Psychol.; 2023; 13, [DOI: https://dx.doi.org/10.3389/fpsyg.2022.1015921] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36710813][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9878386]

Bærøe, K; Kerasidou, A; Dunn, M; Teig, IL. Pursuing impact in research: towards an ethical approach. BMC Med. Ethics; 2022; 23, 1 [DOI: https://dx.doi.org/10.1186/s12910-022-00754-3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35387625][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8988365]

Belcher, B; Halliwell, J. Conceptualizing the elements of research impact: towards semantic standards. Humanities Soc. Sci. Commun.; 2021; 8, 1 183. [DOI: https://dx.doi.org/10.1057/s41599-021-00854-2]

Birkle, C; Pendlebury, DA; Schnell, J; Adams, J. Web of Science as a data source for research on scientific and scholarly activity. Quant. Sci. Stud.; 2020; 1, 1 pp. 363-376. [DOI: https://dx.doi.org/10.1162/qss_a_00018]

Borgatti, SP. Centrality and network flow. Soc. Netw.; 2005; 27, 1 pp. 55-71.MathSciNet ID: 951858[DOI: https://dx.doi.org/10.1016/j.socnet.2004.11.008]

Borgatti, SP; Everett, MG; Freeman, LC. UCINET for windows: software for social network analysis; 2002; Harvard, MA, Analytic Technologies:

Carey, EC; Weissman, DE. Understanding and Finding Mentorship: A Review for Junior Faculty. J. Palliat. Med.; 2010; 13, 11 pp. 1373-1379. [DOI: https://dx.doi.org/10.1089/jpm.2010.0091] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21091022][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3000901]

Cargill, M; Gao, X; Wang, X; O’Connor, P. Preparing Chinese graduate students of science facing an international publication requirement for graduation:Adapting an intensive workshop approach for early-candidature use. Engl. Specif. Purp.; 2018; 52, pp. 13-26. [DOI: https://dx.doi.org/10.1016/j.esp.2018.05.002]

Chariker, JH; Zhang, Y; Pani, JR; Rouchka, EC. Identification of successful mentoring communities using network-based analysis of mentor-mentee relationships across Nobel laureates. Scientometrics; 2017; 111, 3 pp. 1733-1749. [DOI: https://dx.doi.org/10.1007/s11192-017-2364-4] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28596628][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5438421]

Claridge, T (2018). Dimensions of Social Capital - structural, cognitive, and relational. Social Capital Research, 1-4. https://www.socialcapitalresearch.com/structural-cognitive-relational-social-capital/

Corsini, A; Pezzoni, M; Visentin, F. What makes a productive Ph.D. student?. Res. Policy; 2022; 51, [DOI: https://dx.doi.org/10.1016/j.respol.2022.104561]

Das, K; Krzywinski, M; Altman, N. Quantile regression. Nat. Methods; 2019; 16, pp. 451-452.[COI: 1:CAS:528:DC%2BC1MXhtVOgu77N] [DOI: https://dx.doi.org/10.1038/s41592-019-0406-y] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31147637]

Davvetas, V; Diamantopoulos, A; Zaefarian, G; Sichtmann, C. Ten basic questions about structural equations modeling you should know the answers to – But perhaps you don’t. Ind. Mark. Manag.; 2020; 90, pp. 252-263. [DOI: https://dx.doi.org/10.1016/j.indmarman.2020.07.016]

Gisbert, JP. Mentor–mentee relationship in medicine. Gastroenterol.ía y. Hepatol.ía (Engl. Ed.); 2017; 40, 1 pp. 48-57. [DOI: https://dx.doi.org/10.1016/j.gastre.2016.02.022]

Greenhalgh, T; Raftery, J; Hanney, S; Glover, M. Research impact: a narrative review. BMC Med.; 2016; 14, 1 [DOI: https://dx.doi.org/10.1186/s12916-016-0620-8] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27211576][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4876557]

Hirsch, JE. An index to quantify an individual’s scientific research output. Proc. Natl Acad. Sci. USA; 2005; 102, 46 pp. 16569-16572.2005PNAS.10216569H[COI: 1:CAS:528:DC%2BD2MXht1Kgs7fL] [DOI: https://dx.doi.org/10.1073/pnas.0507655102] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16275915][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1283832]

Hooper, D; Coughlan, JP; Mullen, MR. Structural equation modelling: guidelines for determining model fit. Electron. J. Bus. Res. Methods; 2008; 6, 1 pp. 53-60. [DOI: https://dx.doi.org/10.0000/PMID35188134]

Keinänen, M; Ursin, J; Nissinen, K. How to measure students’ innovation competences in higher education: Evaluation of an assessment tool in authentic learning environments. Stud. Educ. Evaluation; 2018; 58, pp. 30-36. [DOI: https://dx.doi.org/10.1016/j.stueduc.2018.05.007]

Keung, C; Cheung, A. A family-school-community partnership supporting play-based learning: A social capital perspective. Teach. Teach. Educ.; 2023; 135, [DOI: https://dx.doi.org/10.1016/j.tate.2023.104314]

Kumar, S. Co-authorship networks: a review of the literature. Aslib J. Inf. Manag.; 2015; 67, 1 pp. 55-73. [DOI: https://dx.doi.org/10.1108/AJIM-09-2014-0116]

Laidlaw, A; Aiton, J; Struthers, J; Guild, S. Developing research skills in medical students: AMEE Guide No. 69. Med. Teach.; 2012; 34, 9 pp. 754-771. [DOI: https://dx.doi.org/10.3109/0142159X.2012.704438]

Leenaars, KEF; Smit, E; Wagemakers, A; Molleman, GRM; Koelen, MA. Facilitators and barriers in the collaboration between the primary care and the sport sector in order to promote physical activity: A systematic literature review. Preventive Med.; 2015; 81, pp. 460-478.[COI: 1:STN:280:DC%2BC28znsVejuw%3D%3D] [DOI: https://dx.doi.org/10.1016/j.ypmed.2015.10.010]

Leiphrakpam, PD, & Are, C (2023). Competency-Based Medical Education (CBME): an Overview and Relevance to the Education of Future Surgical Oncologists. Indian Journal of Surgical Oncology. https://doi.org/10.1007/s13193-023-01716-w

Li, EY; Liao, CH; Yen, HR. Co-authorship networks and research impact: A social capital perspective. Res. Policy; 2013; 42, pp. 1515-1530. [DOI: https://dx.doi.org/10.1016/j.respol.2013.06.012]

Liénard, JF; Achakulvisut, T; Acuna, DE; David, SV. Intellectual synthesis in mentorship determines success in academic careers. Nat. Commun.; 2018; 9, 2018NatCo..9.4840L[COI: 1:CAS:528:DC%2BC1cXitlOksbvN] [DOI: https://dx.doi.org/10.1038/s41467-018-07034-y] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30482900][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6258699]

Liu-Lastres, B; Cahyanto, IP. Are we always ready? Examining event professionals approaches to risk and crisis management and resilience. Tour. Manag. Perspect.; 2023; 46, [DOI: https://dx.doi.org/10.1016/j.tmp.2023.101073]

Lu, H; Feng, Y. A measure of authors’ centrality in co-authorship networks based on the distribution of collaborative relationships. Scientometrics; 2009; 81, 2 pp. 499-511. [DOI: https://dx.doi.org/10.1007/s11192-008-2173-x]

Ma, Y; Mukherjee, S; Uzzi, B. Mentorship and protégé success in STEM fields. Proc. Natl Acad. Sci. USA; 2020; 117, 25 pp. 14077-14083.2020PNAS.11714077M[COI: 1:CAS:528:DC%2BB3cXhsVehsr3N] [DOI: https://dx.doi.org/10.1073/pnas.1915516117] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32522881][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7322065]

Malmgren, RD; Ottino, JM; Amaral, LsAN. The role of mentorship in prote´ge´ performance. NATURE; 2010; 465, pp. 622-626.2010Natur.465.622M[COI: 1:CAS:528:DC%2BC3cXmvFWksL4%3D] [DOI: https://dx.doi.org/10.1038/nature09040] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20520715][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6697046]

McCarty, C; Jawitz, JW; Hopkins, A; Goldman, A. Predicting author h-index using characteristics of the co-author network. Scientometrics; 2013; 96, 2 pp. 467-483. [DOI: https://dx.doi.org/10.1007/s11192-012-0933-0]

McDonald, RP; Ho, M-HR. Principles and practice in reporting structural equation analyses. Psychological Methods; 2002; 7, 1 pp. 64-82. [DOI: https://dx.doi.org/10.1037/1082-989X.7.1.64] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/11928891]10.1037//1082-989X.7.1.64

Moorman, C; Deshpandé, R; Zaltman, G. Factors Affecting Trust in Market Research Relationships. J. Mark.; 1993; 57, 1 pp. 81-101. [DOI: https://dx.doi.org/10.1177/002224299305700106]

Musselin, C. New forms of competition in higher education1. Socio-Economic Rev.; 2018; 16, 3 pp. 657-683. [DOI: https://dx.doi.org/10.1093/ser/mwy033]

Nahapiet, J; Ghoshal, S. Social Capital, Intellectual Capital, and the Organizational Advantage. Acad. Manag. Rev.; 1998; 23, 2 pp. 242-266. [DOI: https://dx.doi.org/10.5465/amr.1998.533225]

Oh, W; Choi, JN; Kim, K. Coauthorship Dynamics and Knowledge Capital: The Patterns of Cross-Disciplinary Collaboration in Information Systems Research. J. Manag. Inf. Syst.; 2005; 22, 3 pp. 266-292. [DOI: https://dx.doi.org/10.2753/MIS0742-1222220309]

Penfield, T; Baker, MJ; Scoble, R; Wykes, MC. Assessment, evaluations, and definitions of research impact: A review. Res. Evaluation; 2014; 23, 1 pp. 21-32. [DOI: https://dx.doi.org/10.1093/reseval/rvt021]

Perc, M. The Matthew effect in empirical data. J. R. Soc. Interface; 2014; 11, 98 [DOI: https://dx.doi.org/10.1098/rsif.2014.0378] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24990288][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4233686]

Perretti, F; Negro, G. Filling Empty Seats: How Status and Organizational Hierarchies Affect Exploration Versus Exploitation in Team Design. Acad. Manag. J.; 2006; 49, 4 pp. 759-777. [DOI: https://dx.doi.org/10.5465/amj.2006.22083032]

Qin, F; Li, J; Zhang, C; Zeng, G; Huang, D; Tan, X; Tan, H. Biochar in the 21st century: A data-driven visualization of collaboration,frontier identification, and future trend. Sci. Total Environ.; 2022; 818,[COI: 1:CAS:528:DC%2BB3MXisFGjtrfO] [DOI: https://dx.doi.org/10.1016/j.scitotenv.2021.151774] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34801502]

Schreiber, M. Restricting the h-index to a publication and citation time window: A case study of a timed Hirsch index. J. Informetr.; 2015; 9, pp. 150-155. [DOI: https://dx.doi.org/10.1016/j.joi.2014.12.005]

Schubert, AS. A Hirsch-type index of co-author partnership ability. Scientometrics; 2012; 91, pp. 303-308. [DOI: https://dx.doi.org/10.1007/s11192-011-0559-7]

Shang, J; Zeng, M; Zhang, G. Investigating the mentorship effect on the academic success of young scientists: An empirical study of the 985 project universities of China. J. Informetr.; 2022; 16, [DOI: https://dx.doi.org/10.1016/j.joi.2022.101285]

Smith, B. Accessing Social Capital through the Academic Mentoring Process. Equity Excell. Educ.; 2007; 40, 1 pp. 36-46. [DOI: https://dx.doi.org/10.1080/10665680601088465]

Søreide, K; Kørner, H; Søreide, JA. Diagnostic accuracy and receiver-operating characteristics curve analysis in surgical research and decision making. Ann. Surg.; 2011; 253, 1 pp. 27-34. [DOI: https://dx.doi.org/10.1097/SLA.0b013e318204a892] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21294285]

Straus, SE; Johnson, MO; Marquez, C; Feldman, MD. Characteristics of Successful and Failed Mentoring Relationships: A Qualitative Study Across Two Academic Health Centers. Academic Med.; 2013; 88, 1 pp. 82-89. [DOI: https://dx.doi.org/10.1097/ACM.0b013e31827647a0]

Straus, SE; Chatur, F; Taylor, MM. Issues in the Mentor–Mentee Relationship in Academic Medicine: A Qualitative Study. Academic Med.; 2009; 84, 1 pp. 135-139. [DOI: https://dx.doi.org/10.1097/ACM.0b013e31819301ab]

Wasko, MM; Faraj, S. Why Should I Share? Examining Social Capital and Knowledge Contribution in Electronic Networks of Practice. MIS Q.; 2005; 29, 1 pp. 35-57. [DOI: https://dx.doi.org/10.2307/25148667]

Wendl, MC. H-index: however ranked, citations need context. NATURE; 2007; 449, 403.2007Natur.449.403W[COI: 1:CAS:528:DC%2BD2sXhtV2iur7P] [DOI: https://dx.doi.org/10.1038/449403b] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/17898746]

Witry, MJ; Patterson, BJ; Sorofman, BA. A qualitative investigation of protégé expectations and proposition of an evaluation model for formal mentoring in pharmacy education. Res. Soc. Adm. Pharm.; 2013; 9, 6 pp. 654-665. [DOI: https://dx.doi.org/10.1016/j.sapharm.2012.08.003]

Xie, Q; Zhang, X; Kim, G; Song, M. Exploring the influence of coauthorship with top scientists on researchers’ affiliation, research topic, productivity, and impact. J. Informetr.; 2022; 16, 3 [DOI: https://dx.doi.org/10.1016/j.joi.2022.101314]

Xu, QA; Chang, V. Co-authorship network and the correlation with academic performance. Internet Things; 2020; 12, [DOI: https://dx.doi.org/10.1016/j.iot.2020.100307]

Yu, D; Herremans, I; Althouse, N; Hendijani, R; Echegaray, AMN; Donohue, F. The mediating role of a good match in achieving mentorship objectives. Int. J. Manag. Educ.; 2022; 20, 3 100695. [DOI: https://dx.doi.org/10.1016/j.ijme.2022.100695]

Yue, Z; Xu, H; Yuan, G; Pang, H. Modeling study of knowledge diffusion in scientific collaboration networks based on differential dynamics: A case study in graphene field. Phys. A: Stat. Mech. its Appl.; 2019; 524, pp. 375-391.2019PhyA.524.375YMathSciNet ID: 3946349[COI: 1:CAS:528:DC%2BC1MXptF2iu7Y%3D] [DOI: https://dx.doi.org/10.1016/j.physa.2019.04.201]

Zeng, A; Shen, Z; Zhou, J; Wu, J; Fan, Y; Wang, Y; Stanley, HE. The science of science: from the perspective of complex systems. Phys. Rep.; 2017; 714–715, 16 pp. 1-73.2017PhR..714..1ZMathSciNet ID: 3730469[DOI: https://dx.doi.org/10.1016/j.physrep.2017.10.001]

Zhang, S, Zang, X, & Zhang, F (2021). Development and Validation of the Win-Win Scale. Frontiers in Psychology, 12. https://doi.org/10.3389/fpsyg.2021.657015

Zhu, Y; Zhang, CJ; Hu, CL. China’s postgraduate education practices and its academic impact on publishing: is it proportional?. J. Zhejiang Univ. Sci. B; 2014; 15, 12 pp. 1088-1092. [DOI: https://dx.doi.org/10.1631/jzus.B14a0331] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25471839][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4265564]

Zuo, Z; Cheng, J; Guo, H; Li, Y. Knowledge mapping of research on strategic mineral resource security: A visual analysis using CiteSpace. Resour. Policy; 2021; 74, [DOI: https://dx.doi.org/10.1016/j.resourpol.2021.102372]