A Qualitative Study on the Relationship between Faculty Mobility and Scientific Impact: Toward the Sustainable Development of Higher Education

Abstract

Faculty mobility is one of the most important research issues in the field of higher education. Reasonable faculty mobility can actively promote the fair, coordinated, balanced, healthy, and sustainable development of higher education. Scientific impact is the best proof of faculty members’ research abilities and is often represented by the quality of their articles. In particular, the gradual increase in citations of high-quality papers is undoubtedly an important reflection of healthy development in the academic field. This paper aims to explore the influence of faculty mobility on scientific impact, while comparative analysis is used to investigate whether there are disciplinary differences in the relationship between faculty mobility and scientific impact. Four major disciplines—sociology, mathematics, mechanical engineering, and philosophy—are selected as the scope of this study. Articles in these four major disciplines from 2000 to 2020 are obtained from the Web of Science, and Spearman’s rank correlation coefficient and the Wilcoxon signed-rank test are used to analyze the collected data. The results indicate the following: (1) faculty mobility has increased, with differences across disciplines; (2) mobility leads to a decrease in the number of citations, which decreases significantly with increased mobility frequency; and (3) the impact of mobility has disciplinary differences, with a relatively obvious decrease in mechanical engineering.

1. Introduction

Faculty is the primary resource of universities, serving as dynamic and vital agents in educational activities, and is a crucial social human resource [1,2,3]. Faculty mobility is a longstanding and ever-relevant issue in the field of higher education, which not only fulfills social functions and roles but also actively promotes the sustainable development of higher education [4,5]. Its impact on higher education is multifaceted and cannot be underestimated. Faculty mobility can attract outstanding talent and distinguished experts from various disciplines to universities, thereby increasing career options and ensuring that talents are utilized to their fullest potential. It helps reduce the uneven distribution of educational resources, balance the differences in teachers’ performance between regions and countries, and achieve educational equity. Therefore, faculty mobility has become a significant research topic for the sustainable development of higher education, and our primary concern is exploring the relationship between faculty mobility and scientific impact [6].

Currently, there are studies analyzing the patterns of faculty mobility within and across countries [7,8,9], its impact on academic careers [10,11,12,13], and reasons for relocation [14,15]. Among these studies, an increasing number focus on the relationship between faculty mobility and scientific impact [16,17,18]. In [16], the authors compared the research influence of publications in a notable Global Western repository by foreign academics in China and by their Chinese counterparts in 15 double-first-class universities. The study found that foreign academics’ publications attract more citations. Moreover, existing works [17,18] mainly use publicly available CV data from official websites, resulting in the research subjects being limited to a small group, such as elite scientists or tenures. Therefore, current research suffers from small sample sizes and limited representativeness in terms of subjects. Inspired by these studies, our work expands the research subject to the disciplinary level, which includes global researchers within the same discipline. Since the academic community comprises not only leading scholars but also ordinary researchers, comprehensive research by disciplines can therefore provide a solid foundation for further understanding the impact of faculty mobility on scientific impact. Furthermore, we select four representative disciplines to explore whether there are disciplinary differences in the relationship between faculty mobility and scientific impact.

This study aims to explore the impact of mobility on the academic development of ordinary teachers from a macro perspective, rather than focusing on leading scholars. We select the disciplines of sociology, mathematics, mechanical engineering, and philosophy as research subjects. Then, we extract faculty mobility data from articles published over 20 years (2000–2020) in these fields, using information obtained from the Web of Science (WOS) database. We examine the relationship between faculty mobility and changes in citation counts of their papers before and after the mobility. In addition, we observe the differences across disciplines. The development of diverse disciplines is a crucial criterion for university competitiveness and evaluation, and exploring disciplinary differences is of great importance.

2. Literature Review

2.1. Faculty Mobility

The mobility of scientific talent is complex and multi-dimensional [19]. As an essential component of scientific talent, faculty mobility is equally intricate. Based on varying research contents and emphases, studies on this topic can be broadly categorized into global mobility, cross-regional and cross-disciplinary mobility, and factors influencing faculty mobility.

Global Mobility. Altbach posited that “globalization refers to the broad trends in economic, political, social, technological, and scientific domains that directly and inevitably impact higher education worldwide” [20]. Therefore, globalization forms the backdrop for current faculty mobility, with higher education inescapably influenced by global factors. As economic globalization deepens, barriers to talent mobility have been broken, enhancing mobility across various regions and significantly increasing transnational mobility rates [21,22,23]. Consequently, research on global talent mobility has emerged, primarily comprising two components: theoretical analysis of mobility phenomena to identify patterns, and empirical analysis through international surveys to examine mobility characteristics and trends [24].

From a theoretical perspective, the push-pull model explains the forces driving talent mobility, suggesting that both the source and destination regions have push (“−”) and pull (“+”) factors [25,26]. The push factors in the source region include insufficient economic development and imperfect labor market mechanisms, while the pull factors in the destination region are characterized by a favorable academic environment and attractive salaries and benefits, giving the destination a strong competitive edge. Empirical analysis through international surveys also highlights the characteristics and trends of mobility. The Carnegie Foundation for the Advancement of Teaching took the various crises facing the American university teaching profession seriously and conducted comparative studies on the profession across countries and regions. This survey empirically revealed issues faced by university teachers in different parts of the world [27].

Cross-Regional and Cross-Disciplinary Mobility. The unequal development across regions significantly influences faculty mobility. Economic development levels, academic platforms, and living environments in developed regions are catalysts for changes in faculty members’ career development [28]. By examining data samples to observe mobility trends, empirical research has described the layout and patterns of university faculty mobility, detailing the history and characteristics of this mobility. A notable example of empirical research in this area is the work conducted by Chinese scholars Zhang and Chen, who analyzed faculty mobility patterns across several major Chinese universities [29]. Their study utilized a comprehensive dataset spanning multiple institutions and regions, providing a broader perspective on the factors driving faculty mobility within the country. The research highlighted significant regional disparities and identified key factors influencing faculty decisions to move, thus offering a more representative view of national trends.

Factors Influencing Faculty Mobility. Understanding these factors is fundamental to educational management, as they determine how stakeholders in higher education allocate resources and manage funding, including the development strategies of universities. This section includes two main aspects: the intentions and influencing factors of faculty mobility, and descriptive analysis based on disciplinary differences. Numerous studies have concentrated on the intention or influencing factors of faculty mobility. University teachers’ turnover intentions are mainly influenced by their current work environment, job satisfaction, and the lure of non-academic labor markets to varying degrees. Work-related and personal factors, such as time spent on teaching and research, participation in institutional management, gender, marital status, and salary, also influence mobility [30]. Descriptive analysis based on disciplinary differences divides university teachers into four major categories: economics and management, engineering, sciences, and humanities and social sciences [31,32]. Analyzing online resume samples reveals that humanities teachers have higher intra-provincial mobility rates, while science and engineering teachers have higher inter-provincial mobility rates, with teachers in advantageous disciplines exhibiting greater mobility rates [33,34].

2.2. Scientific Impact

From a macro perspective, research has shown that talent introduction policies and research funding significantly impact university teachers’ scientific influence. During the quasi-tenure period, the number of publications by teachers generally increases steadily but often declines sharply after they gain tenure [35,36,37]. Additionally, with increased research funding, the quality of teachers’ papers and their academic contributions also improve. By examining the causal effects of institutional factors on faculty mobility, it becomes clear how talent programs and funding influence the career movements of scientists, partially demystifying the relationship between these factors and career mobility. At the meso level, studies on the age structure and composition of faculty reveal that the optimal scientific impact of universities can be theoretically achieved when the proportions of young, mid-career, and senior faculty are 51.2%, 43.0%, and 5%, respectively [38,39,40]. Increasing the number and quality of research assistants within a fixed number of full-time faculty members can further enhance scientific impact. However, hiring non-tenure-track faculty does not improve scientific impact as expected. Changes in the proportion of traditionally disadvantaged groups among faculty have not systematically affected the overall scientific output and influence of universities. At the micro level, research shows that individual characteristics such as gender, institution, qualifications, academic rank, discipline, and years of service influence teachers’ scientific impact [41,42]. Some studies indicate a gender disparity in research output, with many suggesting that male scientists tend to produce more [43,44]. However, when redefining the nature of research output, findings show that female scientists often have higher output and tend to emphasize the quality of their research more [45].

The globalization of science and the mobility of scientists have had a profound impact on the progress of science and technology, as well as scholars’ academic achievements [1], particularly in the context of China. Chinese returnees, with their unique blend of international experience, expertise, and social capital, have opportunities to win academic titles or funding, which can guarantee them tenure, a higher salary, and other academic resources [46]. More researchers have focused on assessing the effectiveness of these talent-recruitment policies launched by the Chinese government and investigating whether the research performance of Chinese returnees is better than that of their local counterparts [47,48]. Studies have indicated that Chinese returnees publish more papers and more corresponding-author papers than their local counterparts, have higher research caliber, and tend to make outstanding contributions to the link between China and global networks [49,50,51].

2.3. Relationship between Faculty Mobility and Scientific Impact

One aspect of the research focus is the overall descriptive analysis of faculty after they have moved. Studies have indicated that the level of academic productivity is related to the prestige and level of the institution to which the teacher moves [52,53,54]. Moving to higher-level institutions can further enhance scientists’ research output and impact, while downward mobility tends to have increasingly adverse effects over time. Although mobility enhances the efficiency of research and innovation in universities, its effects vary significantly across different groups. Research has shown that upward mobility can improve researchers’ output and citation rates, whereas downward mobility decreases research productivity [55,56]. Compared to non-mobile teachers, mobile teachers have significantly higher research impact. Furthermore, mobile researchers not only experience an increase in citation counts but also diversify their research topics and expand their academic collaboration networks. Another aspect involves analyzing the correlation and differences in scientific impact based on disciplinary differences. Studies have found that the relationship between mobility, productivity, and impact varies across disciplines [57,58]. Using methods such as inverse probability weighting, research has shown that the positive effects of researcher mobility differ by discipline. For example, in the fields of science, engineering, agriculture, and medicine, the quantity of papers has increased significantly, but the quality of these papers has generally declined [59,60]. According to Peter’s principle, employees’ performance decreases after promotion until they reach a level where they cannot handle critical responsibilities. Although mobility positively influences academic career development, frequent mobility can delay academic promotion.

3. Data and Research Methodology

3.1. Dataset

The dataset used for this study is extracted from the WOS database. WOS is a comprehensive, interdisciplinary academic information resource that includes over 8000 peer-reviewed, high-quality, and influential journals worldwide. Our aim is to examine the relationship between faculty mobility and scientific impact across different disciplines. This study selects four primary disciplines: mathematics, philosophy, mechanical engineering, and sociology. We first utilize the “Topic” field in the WOS Core Collection search page to search within the four disciplines for documents published from 2000 to 2020. The document type is specified as “Article”. The “All records on page” are then exported to extract information about the papers. Using Python, the data are processed to extract the corresponding authors’ addresses and citation details from each paper. Moreover, in case there is a name disambiguation problem in the subset we used for our experiment, we apply the method described in [61] to solve it.

Table 1. Statistical information about the dataset.

Discipline	People	Papers
Mathematics	255,755	920,885
Philosophy	30,367	42,548
Mechanical Engineering	439,285	1,048,575
Sociology	145,491	185,686

provides statistical information about the dataset used in this study.

3.2. Research Methodology

3.2.1. Descriptive Statistical Analysis

The articles’ data acquired from the WOS database include relevant information about the scholars. In this paper, we use the citation counts of articles to represent the scientific impact of faculty since it is one of the most widely used indicators. After processing the information, we count the authors whose correspondence addresses have changed and record the annual citation counts of their papers before and after the address change. This provides the total citation counts and correspondence addresses for all authors each year. From the perspectives of data availability and reliability, this study uses the change in authors’ correspondence addresses as an indicator of mobility frequency and the citation counts of papers as an important metric to represent the scientific impact of faculty. The following definitions clarify the key metrics used in this study:

• Mobility Frequency: This study uses the change in the authors’ correspondence addresses as an indicator of mobility frequency. Samples with abnormal data and excessive mobility experiences were excluded, and only samples with 1–5 instances of mobility were included in the subsequent analysis.
• Citation Count: The number of times a paper is cited by other papers after publication is called the citation count. This reflects the referential value and importance of an original paper for subsequent research. Highly cited papers represent the frontier and hot issues in the field.
• Difference in Citation Count ($\delta$): This refers to the difference in citation counts of papers by faculty members after mobility compared to the citation counts before mobility.

The total number of faculty in this study is 870,898. Of these, 324,320 individuals experienced mobility, accounting for 37.23% of the total, while 657,378 individuals did not experience mobility, accounting for 75.48%. In terms of mobility frequency, there are 135,438 individuals who moved once, accounting for 41.76% of the total number of individuals who experienced mobility; 60,656 individuals who moved twice, accounting for 18.70%; 33,985 individuals who moved three times, accounting for 10.47%; 21,431 individuals who moved four times, accounting for 6.60%; and 14,905 individuals who moved five times, accounting for 5.00%.

Regarding disciplinary differences, in the mathematics discipline, mobile scholars account for 4.83%, while non-mobile scholars account for 95.17%. In the philosophy discipline, mobile scholars account for 21.82%, while non-mobile scholars account for 78.18%. In the mechanical engineering discipline, mobile scholars account for 37.58%, while non-mobile scholars account for 62.42%. In the sociology discipline, mobile scholars account for 20.27%, while non-mobile scholars account for 79.73%. From the overall descriptive data, the number of mobile scholars exceeds one-third of the total scholars, with mobility proportions across disciplines as follows: mechanical engineering > philosophy > sociology > mathematics.

Table 2. Descriptive statistics of faculty mobility across the four major disciplines.

Subject	Mathematics	Philosophy	Mechanical Engineering	Sociology	Total
Total number	255,755	30,367	439,285	145,491	870,898
Non-mobile individuals	243,413	23,742	274,221	116,002	657,378
Mobile individuals	12,342	6625	165,064	29,489	324,320
Individuals with 1 mobility	42,102	4254	73,308	15,774	135,438
Individuals with 2 mobilities	21,338	1363	31,951	5954	60,656
Individuals with 3 mobilities	13,288	511	17,352	2834	33,985
Individuals with 4 mobilities	9111	234	10,487	1599	21,431
Individuals with 5 mobilities	6777	110	7032	986	14,905

shows the overall statistics of the mobile population across the four disciplines in this study.

3.2.2. Normality Test

The normality test is a specialized hypothesis test used to determine whether a set of observations (or data transformed by some function) or a set of random numbers comes from a normal population and follows a normal distribution. The most commonly used normality test is the Kolmogorov–Smirnov ($\mathcal{K}-\mathcal{S}$) test.

To explore whether there is a statistically significant difference in the citation counts of papers before and after the mobility of university teachers, this study conducted a non-parametric single-sample $\mathcal{K}-\mathcal{S}$ normality test on the difference in citation counts d before and after mobility using SPSS 26.0 software. The results indicate that the data for all four major disciplines do not follow a normal distribution (p-value = 0.000, all less than 0.05), as shown in

Table 3. Normality test of the difference in citation counts before and after mobility across four major disciplines.

	Discipline	Statistical Magnitude	df	Sig.
	Philosophy	0.301	6472	0.000
$\delta$	Mathematics	0.280	92,666	0.000
	Sociology	0.281	27,147	0.000
	Mechanical Engineering	0.246	133,098	0.000

. Therefore, non-parametric tests can be used for further analysis.

By examining the group of teachers with mobility frequencies of 1–5 times, the one-sample $\mathcal{K}-\mathcal{S}$ normal distribution test results for the difference $\delta$ in citation counts before and after mobility show that the data do not follow a normal distribution ($p=0.000$, all less than 0.05; see

Table 4. Normality test of the difference between the overall mobility frequency and citation counts.

	Mobility Frequency	Statistical Magnitude	df	Sig.
	1	0.279	135,438	0.000
	2	0.264	60,656	0.000
$\delta$	3	0.257	33,985	0.000
	4	0.270	21,431	0.000
	5	0.273	14,905	0.000

). Therefore, a non-parametric test can be used to analyze the relationship between mobility frequency and the difference in citation counts before and after mobility.

The one-sample $\mathcal{K}-\mathcal{S}$ normal distribution test results for the difference $\delta$ in the number of citations before and after mobility for scholars in the mathematics, philosophy, mechanical engineering, and sociology disciplines, with mobility frequencies of 1–5 times, show that the data do not follow a normal distribution ($p=0.000$, all less than 0.05; see

Table 5. Normality test of difference between mobility frequency and paper citations across four disciplines.

Discipline		Mobility Frequency	Statistical Magnitude	df	Sig.
		1	0.291	42,102	0.000
		2	0.277	21,388	0.000
Mathematics	$\delta$	3	0.259	13,288	0.000
		4	0.301	9111	0.000
		5	0.217	6777	0.000
		1	0.322	4254	0.000
		2	0.272	1363	0.000
Philosophy	$\delta$	3	0.263	511	0.000
		4	0.239	234	0.000
		5	0.261	110	0.000
		1	0.284	15,774	0.000
		2	0.274	5954	0.000
Mechanical Engineering	$\delta$	3	0.252	2834	0.000
		4	0.257	1599	0.000
		5	0.331	986	0.000
		1	0.252	73,308	0.000
		2	0.238	31,951	0.000
Sociology	$\delta$	3	0.236	17,352	0.000
		4	0.241	10,487	0.000
		5	0.209	7032	0.000

). Therefore, non-parametric tests can be used to analyze the correlation and differences.

3.2.3. Spearman’s Rank Correlation Coefficient

This study investigates the relationship between faculty mobility and scientific impact using Spearman’s rank correlation coefficient. Spearman’s rank correlation coefficient calculates the differences in ranks for each pair to measure the correlation between two sets of continuous variables that are not normally distributed (or contain outliers that cannot be eliminated). When using Spearman’s rank correlation analysis, two hypotheses need to be considered:

Hypothesis 1.

The observed variables are continuous variables that are not normally distributed (or contain outliers that cannot be eliminated). In this study, the citation frequency of academic papers among university teachers exhibits a non-normal distribution. Although the overall mobility frequency across the four major disciplines follows a normal distribution, when separated by discipline, the mobility frequency does not.

Hypothesis 2.

There is a monotonic relationship between the variables. Analysis shows that the mobility frequency of university teachers in this study and the difference in citation counts before and after their mobility exhibit a monotonic relationship. When calculated separately for each discipline, the mobility frequency and the difference in citation counts before and after mobility still exhibit a monotonic relationship.

Based on the above analysis, the data utilized in this study fulfill Hypotheses 1 and 2, thus satisfying the criteria for conducting Spearman’s rank correlation analysis.

3.2.4. Wilcoxon Signed-Rank Test

When the same experimental unit from the same population undergoes a “treatment” and produces two corresponding results, both parametric and non-parametric tests can be conducted. In the empirical study on the relationship between faculty mobility and scientific impact, the sample’s scientific impact is based on the output at different time points from the same sample. Therefore, it is appropriate to consider both parametric and non-parametric tests, such as the Wilcoxon signed-rank test. In this study, when analyzing the normality of the difference variable $\delta$, the null hypothesis of non-normal distribution is satisfied. Therefore, the Wilcoxon signed-rank test is used as one of the methods for analyzing the data in this study. When using the Wilcoxon signed-rank test, the following three hypotheses need to be satisfied:

Hypothesis 3.

The observed variables are continuous or ordered categorical variables. In this study, the scientific impact of university teachers is a continuous variable, while the mobility frequency is an ordered categorical variable.

Hypothesis 4.

In this study, the data can be divided into two groups: before faculty mobility ($Pre\text{\_}TC$) and after faculty mobility ($Post\text{\_}TC$).

Hypothesis 5.

The data in this study on “faculty mobility and scientific impact” are in the form of pairs of the research subjects themselves.

Based on the above analysis, the data applied in this study meet Hypotheses 3–5, thus satisfying the criteria for conducting the Wilcoxon signed-rank test.

4. Results

4.1. Spearman’s Rank Correlation Analysis of Faculty Mobility and Citations

4.1.1. Correlation Analysis of Overall Faculty Mobility Frequency and Citations

After conducting a basic analysis of the descriptive statistics of the mobility numbers across four disciplines, this study explores the correlation between faculty mobility frequency and the number of citations before and after mobility using Spearman’s rank correlation analysis. Calculating the Spearman correlation coefficient between faculty mobility frequency and the difference in paper citations before and after mobility yields a p-value less than 0.05, indicating a significant difference between them. The results in

Table 6. Spearman correlation coefficients between faculty mobility frequency and citation counts across four disciplines.

		Faculty Mobility Frequency	Difference in Paper Citations $\mathit{\delta }$
Faculty Mobility Frequency	Correlation Coefficient $\rho$	1.000	−0.042 **
	Sig. (2-tailed)	.	0.000
	N	266,415	266,415
Difference in Paper Citations $\delta$	Correlation Coefficient $\rho$	−0.042 **	1.000
	Sig. (2-tailed)	0.000	.
	N	266,415	266,415

**. Correlation is significant at the 0.05 level (2-tailed).

show that $\rho =-0.042$, suggesting a slight negative correlation between faculty mobility frequency and the difference in paper citations before and after mobility. In other words, without considering specific disciplines, the more frequently faculty move, the lower the number of citations they tend to receive. It is found that the higher the frequency of faculty mobility, the lower the number of citations. To further observe and differentiate the differences between disciplines, it is necessary to compare the correlation between faculty mobility frequency and the difference in paper citations among different disciplines.

4.1.2. Correlation Analysis of Faculty Mobility Frequency and Paper Citations by Discipline

Calculating the Spearman correlation coefficient between the mobility frequency of faculty in the mathematics discipline and the difference in paper citations before and after mobility yields a p-value less than 0.05, indicating a significant correlation. The results in

Table 7. Spearman correlation coefficient between faculty mobility frequency and citation counts in the discipline of mathematics.

		Faculty Mobility Frequency in Mathematics	Difference in Paper Citations $\mathit{\delta }$
Faculty Mobility Frequency in Mathematics	Correlation Coefficient $\rho$	1.000	−0.045 **
	Sig. (2-tailed)	.	0.000
	N	92,666	92,666
Difference in Paper Citations $\delta$	Correlation Coefficient $\rho$	−0.045 **	1.000
	Sig. (2-tailed)	0.000	.
	N	92,666	92,666

**. Correlation is significant at the 0.05 level (2-tailed).

show that $\rho =-0.045$, suggesting a slight negative correlation between the mobility frequency of faculty in the mathematics discipline and the difference in paper citations before and after mobility. In other words, within this discipline, the more frequently faculty move, the lower the number of citations they tend to receive.

For philosophy, the p-value is less than 0.05, indicating a significant correlation. The results in

Table 8. Spearman correlation coefficient between faculty mobility frequency and citation counts in the discipline of philosophy.

		Faculty Mobility Frequency in Philosophy	Difference in Paper Citations $\mathit{\delta }$
Faculty Mobility Frequency in Philosophy	Correlation Coefficient $\rho$	1.000	−0.055 **
	Sig. (2-tailed)	.	0.000
	N	6472	6472
Difference in Paper Citations $\delta$	Correlation Coefficient $\rho$	−0.055 **	1.000
	Sig. (2-tailed)	0.000	.
	N	6472	6472

**. Correlation is significant at the 0.05 level (2-tailed).

show that $\rho =-0.055$, suggesting a moderate negative correlation between the mobility frequency of faculty in the philosophy discipline and the difference in paper citations before and after mobility. This means that within this discipline, the more frequently faculty move, the lower the number of citations they tend to receive.

For mechanical engineering, the p-value is less than 0.05, indicating a significant correlation. The results in

Table 9. Spearman correlation coefficient between faculty mobility frequency and citation counts in the discipline of mechanical engineering.

		Faculty Mobility Frequency in Mechanical Engineering	Difference in Paper Citations $\mathit{\delta }$
Faculty Mobility Frequency in Mechanical Engineering	Correlation Coefficient $\rho$	1.000	−0.052 **
	Sig. (2-tailed)	.	0.000
	N	140,130	140,130
Difference in Paper Citations $\delta$	Correlation Coefficient $\rho$	−0.052 **	1.000
	Sig. (2-tailed)	0.000	.
	N	140,130	140,130

**. Correlation is significant at the 0.05 level (2-tailed).

show that $\rho =-0.052$, suggesting a slight negative correlation between the mobility frequency of faculty in the mechanical engineering discipline and the difference in paper citations before and after mobility. Specifically, within this discipline, the more frequently faculty move, the lower the number of citations they tend to receive.

As shown in

Table 10. Spearman correlation coefficient between faculty mobility frequency and citation counts in the discipline of sociology.

		Faculty Mobility Frequency in Sociology	Difference in Paper Citations $\mathit{\delta }$
Faculty Mobility Frequency in Sociology	Correlation Coefficient $\rho$	1.000	−0.097 **
	Sig. (2-tailed)	.	0.000
	N	27,147	27,147
Difference in Paper Citations $\delta$	Correlation Coefficient $\rho$	−0.097 **	1.000
	Sig. (2-tailed)	0.000	.
	N	27,147	27,147

**. Correlation is significant at the 0.05 level (2-tailed).

, for sociology, the p-value is less than 0.05, indicating a significant negative correlation. In addition, $\rho =-0.097$, suggesting a strong negative correlation between the mobility frequency of faculty in the sociology discipline and the difference in paper citations before and after mobility. In other words, within this discipline, the more frequently faculty move, the lower the number of citations they tend to receive.

4.2. Wilcoxon Signed-Rank Test Analysis of Faculty Mobility and Paper Citations

In addition to examining the mean number of citations before and after faculty mobility, this study also focuses on the median values of citations. Based on the statistical analysis, the median and mean citation counts of papers before and after faculty mobility are presented in

Table 11. Descriptive statistics of overall paper citations across the four major disciplines.

	Mathematics		Philosophy		Mechanical Engineering		Sociology
	Median	Mean	Median	Mean	Median	Mean	Median	Mean
$Pre\text{\_}TC$	2	7.24	10	20.87	12	30.93	4	8.88
$Post\text{\_}TC$	0	3.63	5	13.22	4	16.45	1	5.44
$\delta$	0	3.61	2	7.65	4	14.48	1	3.44

.

The analysis results (

Table 12. Wilcoxon test results for citations before and after the mobility of four subjects.

Subject		Difference in Paper Citations $\mathit{\delta }$
Mathematics	Z	−102.524 b
	Asymptotic Sig. (2-tailed)	0.000
Philosophy	Z	−30.444 b
	Asymptotic Sig. (2-tailed)	0.000
Mechanical Engineering	Z	−127.505 b
	Asymptotic Sig. (2-tailed)	0.000
Sociology	Z	−70.285 b
	Asymptotic Sig. (2-tailed)	0.000

^b Based on positive ranks.

) indicate significant differences in the citation counts before and after mobility across various disciplines, with $p=0.000$. Specifically, for the mathematics discipline, the Wilcoxon signed-rank test yielded $Z=-102.524$ with an asymptotic significance $p=0.000$; for the philosophy discipline, the Wilcoxon signed-rank test yielded $Z=-30.444$ with $p=0.000$; for the mechanical engineering discipline, the Wilcoxon signed-rank test yielded $Z=-127.505$ with $p=0.000$; and for the sociology discipline, the Wilcoxon signed-rank test yielded $Z=-70.285$ with $p=0.000$. Combining these results with the median citation counts before and after mobility (Table 11), it is evident that the citation counts of papers in all disciplines significantly decreased after faculty mobility ($p<0.05$).

4.3. Wilcoxon Signed-Rank Test Analysis of Faculty Mobility Frequency and Paper Citations

4.3.1. Analysis of Differences in Paper Citations by Overall Faculty Mobility Frequency

We found that the citation count of scholars in all disciplines significantly decreased after mobility. Analyzing academic productivity across different mobility frequencies helps explore the impact of mobility frequency on academic productivity. The previous section analyzed the relationship between changes in faculty paper citations without differentiating between mobility frequencies. To ensure data accuracy, this section excludes invalid data and outliers, focusing on mobility frequencies of 1–5 times only.

Table 13. Descriptive statistics of overall mobility frequency and paper citations.

	Mobility Frequency
	1		2		3		4		5
	Median	Mean	Median	Mean	Median	Mean	Median	Mean	Median	Mean
$Pre\text{\_}TC$	6	16.03	8	18.07	8	19.25	9	20.68	9	20.72
$Post\text{\_}TC$	4	11.35	3	10.39	3	9.71	3	8.98	2	8.11
$\delta$	0	4.67	2	7.68	4	9.54	4	11.70	5	12.61

provides descriptive statistics for paper citations at different mobility frequencies.

From the analysis results (

Table 14. Wilcoxon test results for overall mobility frequency and paper citations.

Mobility Frequency		Difference in Paper Citations $\mathit{\delta }$
1	Z	−103.517 b
	Asymptotic Sig. (2-tailed)	0.000
2	Z	−92.806 b
	Asymptotic Sig. (2-tailed)	0.000
3	Z	−80.284 b
	Asymptotic Sig. (2-tailed)	0.000
4	Z	−71.573 b
	Asymptotic Sig. (2-tailed)	0.000
5	Z	−62.879 b
	Asymptotic Sig. (2-tailed)	0.000

^b Based on positive ranks.

), it can be seen that the test results of differences in paper citations for each mobility frequency are significant with $p=0.000$. The two-sided test results are as follows: for a mobility frequency of 1, $Z=-103.517$ and $p=0.000<0.05$; for a mobility frequency of 2, $Z=-92.806$ and $p=0.000<0.05$; for a mobility frequency of 3, $Z=-80.284$ and $p=0.000<0.05$; for a mobility frequency of 4, $Z=-71.573$ and $p=0.000<0.05$; and for a mobility frequency of 5, $Z=-62.879$ and $p=0.000<0.05$.

4.3.2. Analysis of Differences in Citations by Discipline

Mathematics. From the analysis results (

Table 15. Results of Wilcoxon test for mobility frequency and paper citations in mathematics.

Mobility Frequency		Difference in Paper Citations $\mathit{\delta }$
1	Z	−51.859 b
	Asymptotic Sig. (2-tailed)	0.000
2	Z	−50.296 b
	Asymptotic Sig. (2-tailed)	0.000
3	Z	−46.726 b
	Asymptotic Sig. (2-tailed)	0.000
4	Z	−42.788 b
	Asymptotic Sig. (2-tailed)	0.000
5	Z	−38.501 b
	Asymptotic Sig. (2-tailed)	0.000

^b Based on positive ranks.

), it can be seen that the test results of differences in paper citations for each mobility frequency in the mathematics discipline are significant with $p=0.000$. The two-sided test results are as follows: for a mobility frequency of 1, $Z=-51.859$ and $p=0.000<0.05$; for a mobility frequency of 2, $Z=-50.296$ and $p=0.000<0.05$; for a mobility frequency of 3, $Z=-46.726$ and $p=0.000<0.05$; for a mobility frequency of 4, $Z=-42.788$ and $p=0.000<0.05$; and for a mobility frequency of 5, $Z=-38.501$ and $p=0.000<0.05$.

Based on the median citation results before and after mobility (

Table 16. Descriptive statistics of mobility frequency and paper citations in mathematics.

	Mobility Frequency
	1		2		3		4		5
	Median	Mean	Median	Mean	Median	Mean	Median	Mean	Median	Mean
$Pre\text{\_}TC$	3	7.70	4	9.15	4	10.14	5	10.29	5	10.9
$Post\text{\_}TC$	2	5.66	1	5.30	1	5.35	1	5.29	1	4.83
$\delta$	0	2.05	1	3.84	2	4.80	3	5.00	3	6.06

), it can be concluded that the citation count of papers from mathematics faculty significantly decreased for mobility frequencies of 1–5 times ($p<0.05$). Moreover, a higher mobility frequency is associated with a greater decrease in citation counts, as indicated by the change in the Z value.

Philosophy. From the analysis results (

Table 17. Results of Wilcoxon test for mobility frequency and paper citations in philosophy.

Mobility Frequency		Difference in Paper Citations $\mathit{\delta }$
1	Z	−20.058 b
	Asymptotic Sig. (2-tailed)	0.000
2	Z	−16.820 b
	Asymptotic Sig. (2-tailed)	0.000
3	Z	−12.118 b
	Asymptotic Sig. (2-tailed)	0.000
4	Z	−8.279 b
	Asymptotic Sig. (2-tailed)	0.000
5	Z	−6.893 b
	Asymptotic Sig. (2-tailed)	0.000

^b Based on positive ranks.

), it can be seen that the test results of differences in paper citations across each mobility frequency in the philosophy discipline are significant with $p=0.000$. The two-sided test results are as follows: for a mobility frequency of 1, $Z=-20.058$ and $p=0.000<0.05$; for a mobility frequency of 2, $Z=-16.820$ and $p=0.000<0.05$; for a mobility frequency of 3, $Z=-12.118$ and $p=0.000<0.05$; for a mobility frequency of 4, $Z=-8.279$ and $p=0.000<0.05$; and for a mobility frequency of 5, $Z=-6.893$ and $p=0.000<0.05$.

Based on the median citation results before and after mobility (

Table 18. Descriptive statistics of mobility frequency and paper citations in philosophy.

	Mobility Frequency
	1		2		3		4		5
	Median	Mean	Median	Mean	Median	Mean	Median	Mean	Median	Mean
$Pre\text{\_}TC$	1	6.28	2	8.61	3	9.72	3	8.99	5	12.28
$Post\text{\_}TC$	0	3.99	0	3.27	0	2.41	0	2.42	1	2.57
$\delta$	0	2.29	1	5.34	1	7.31	1.50	6.56	3	9.71

), it can be concluded that the citation count of papers from philosophy faculty significantly decreased for mobility frequencies of 1–5 times ($p<0.05$). Moreover, a higher mobility frequency is associated with a greater decrease in citation counts, as indicated by the change in the Z value.

Mechanical Engineering. From the analysis results (

Table 19. Results of Wilcoxon test for mobility frequency and paper citations in mechanical engineering.

Mobility Frequency		Difference in Paper Citations $\mathit{\delta }$
1	Z	−76.797 b
	Asymptotic Sig. (2-tailed)	0.000
2	Z	−68.009 b
	Asymptotic Sig. (2-tailed)	0.000
3	Z	−58.012 b
	Asymptotic Sig. (2-tailed)	0.000
4	Z	−51.113 b
	Asymptotic Sig. (2-tailed)	0.000
5	Z	−44.285 b
	Asymptotic Sig. (2-tailed)	0.000

^b Based on positive ranks.

), it can be seen that the test results for differences in paper citations across each mobility frequency in the mechanical engineering discipline are significant with $p=0.000$. The two-sided test results are as follows: for a mobility frequency of 1, $Z=-76.797$ and $p=0.000<0.05$; for a mobility frequency of 2, $Z=-68.009$ and $p=0.000<0.05$; for a mobility frequency of 3, $Z=-58.012$ and $p=0.000<0.05$; for a mobility frequency of 4, $Z=-51.113$ and $p=0.000<0.05$; and for a mobility frequency of 5, $Z=-44.285$ and $p=0.000<0.05$.

Based on the median citation results before and after mobility (

Table 20. Descriptive statistics of mobility frequency and paper citations in mechanical engineering.

	Mobility Frequency
	1		2		3		4		5
	Median	Mean	Median	Mean	Median	Mean	Median	Mean	Median	Mean
$Pre\text{\_}TC$	9	19.03	11	21.93	12	23.64	13	26.01	15	26.39
$Post\text{\_}TC$	6	13.80	5	13.00	5	12.30	4	11.41	4	11.04
$\delta$	1	5.23	4	8.94	5	11.34	7	14.60	8	15.34

), it can be concluded that the citation count of papers from mechanical engineering faculty significantly decreased for mobility frequencies of 1–5 times ($p<0.05$). Moreover, a higher mobility frequency is associated with a greater decrease in citation counts, as indicated by the change in the Z value.

Sociology. From the analysis results (

Table 21. Results of Wilcoxon test for mobility frequency and paper citations in sociology.

Mobility Frequency		Difference in Paper Citations $\mathit{\delta }$
1	Z	−42.631 b
	Asymptotic Sig. (2-tailed)	0.000
2	Z	−35.590 b
	Asymptotic Sig. (2-tailed)	0.000
3	Z	−28.616 b
	Asymptotic Sig. (2-tailed)	0.000
4	Z	−25.429 b
	Asymptotic Sig. (2-tailed)	0.000
5	Z	−21.708 b
	Asymptotic Sig. (2-tailed)	0.000

^b Based on positive ranks.

), it can be seen that the test results for differences in paper citations across each mobility frequency in the sociology discipline are significant with $p=0.000$. The two-sided test results are as follows: for a mobility frequency of 1, $Z=-42.631$ and $p=0.000<0.05$; for a mobility frequency of 2, $Z=-35.590$ and $p=0.000<0.05$; for a mobility frequency of 3, $Z=-28.616$ and $p=0.000<0.05$; for a mobility frequency of 4, $Z=-25.429$ and $p=0.000<0.05$; and for a mobility frequency of 5, $Z=-21.708$ and $p=0.000<0.05$.

Based on the median citation results before and after mobility (

Table 22. Descriptive statistics of mobility frequency and paper citations in sociology.

	Mobility Frequency
	1		2		3		4		5
	Median	Mean	Median	Mean	Median	Mean	Median	Mean	Median	Mean
$Pre\text{\_}TC$	10	26.93	13	31.54	16	36.78	20	46.65	21	48.75
$Post\text{\_}TC$	5	17.17	4	16.31	4	15.61	3	15.05	3	10.36
$\delta$	1	9.76	5	15.231	8	21.17	13	31.60	14.5	38.39

), it can be concluded that the citation count of papers from sociology faculty significantly decreased for mobility frequencies of 1–5 times ($p<0.05$). Moreover, a higher mobility frequency is associated with a greater decrease in citation counts, as indicated by the change in the Z value.

In conclusion, through Spearman’s rank correlation coefficient analysis, we can draw two conclusions. Firstly, regardless of the discipline, there is a low negative correlation between the frequency of faculty mobility and the change in citation counts before and after mobility. Secondly, within specific disciplines, this correlation varies: it shows a low negative correlation, a moderate negative correlation, a low negative correlation, and a high negative correlation in mathematics, philosophy, mechanical engineering, and sociology, respectively. Through the Wilcoxon signed-rank test analysis, we find that across all disciplines, regardless of mobility frequency, there is a significant decrease in citation counts after mobility compared to before. When analyzed by discipline and mobility frequency, significant changes are observed in citation counts before and after mobility.

5. Discussion and Conclusions

Faculty mobility has increased, with differences across disciplines. The rationale for faculty mobility across universities includes optimizing the allocation of human resources, promoting scientific collaborations within or across disciplines, accelerating the development of science and technology, and improving education quality in higher education. Faculty mobility should become a norm, with orderly “flow” and effective “movement” [62]. Mobility or migration can also be seen as a positive selection mechanism, where only scientists with high impact have the ability to migrate and gain more career opportunities [63]. Firstly, from the analysis of faculty mobility frequency, the overall effective sample in this study consists of 870,898 individuals, with about 37.23% being mobile, indicating that more than one-third of the scholars have undertaken mobility actions, making faculty mobility a norm in universities. Secondly, the proportion of mobile scholars is 4.83% in mathematics, 21.82% in philosophy, 37.58% in mechanical engineering, and 20.27% in sociology, showing significant differences in mobility among different disciplines.

Mobility leads to a decrease in the number of citations, with the decrease becoming more significant as mobility frequency increases. Research has shown that faculty mobility has a significant negative impact on citation influence [7,64]. Based on data availability, this study uses citation counts as an indicator of scientific impact. Spearman’s rank correlation analysis reveals that, regardless of the discipline, the relationship between faculty mobility frequency and the difference in paper citations before and after mobility shows a high, medium, or low degree of negative correlation. In other words, faculty mobility leads to a decrease in paper citations. The results of the Wilcoxon signed-rank test $p=0.000$, all less than 0.05, also indicate that faculty mobility has a significant negative impact on paper citations across all disciplines.

The impact of mobility varies by discipline, with a relatively obvious decrease in mechanical engineering. Recent research results indicate that while mobility may temporarily reduce scientific output, it can benefit scientists’ long-term career development. Moreover, the size and cohesion of research teams can also affect the quantity of research results produced by university faculty. For university faculty, differences in academic networks, funding constraints, and the quality of academic groups often lead to significant restrictions on their scientific impact after mobility. In this study, when analyzing mobility frequencies of 1–5 times without distinguishing disciplines, Spearman’s correlation analysis r shows that the impact of mobility frequency on the change in paper citations before and after mobility is most significant for faculty in mechanical engineering, followed by philosophy, sociology, and mathematics.

When examining the data by discipline, Spearman’s correlation analysis r confirms that the change in paper citations before and after mobility is most significant for faculty in mechanical engineering. Through the Wilcoxon signed-rank test analysis of the overall data, the Z value and its changes also show disciplinary differences. Further comparison of the Z values confirms that the change in paper citations before and after mobility for faculty in mechanical engineering is more significant compared to other disciplines. However, across all data, comparing the Z values reveals that more frequent mobility is associated with a greater decline in paper citations for faculty in both mechanical engineering and mathematics.

Furthermore, there is still room for further modifications. Firstly, due to data processing constraints and the timeframe of this study, the analysis of scientific impact was limited to paper citations. However, other critical variables, such as the number of papers published by faculty and their sociological and educational backgrounds, are equally important for a comprehensive analysis. Secondly, based on data availability, this study did not account for the timing of faculty departures and arrivals when analyzing paper citations. The lack of comparison over equivalent time periods before and after mobility may lead to discrepancies in conclusions about research quality. Future studies should incorporate factors such as the timing of mobility events. Lastly, this study is confined to data correlation analysis and difference testing and lacks qualitative insights, such as interviews or group surveys on faculty mobility intentions. Future research should integrate qualitative methods for a more in-depth exploration, which would be beneficial for thoroughly explaining the relationship between faculty mobility and research impact.