*4.5. Principal Component Analysis*

A principal component analysis (PCA) of the 20 impact factors was performed to group similar items into subgroups for data reduction, the results are shown in Table 6. It can be seen that the measured value of the sample data KMO (Kaiser-Meyer-Olkin Measure of Sampling Adequacy) was 0.802, The Barlett sphericity test was 1936.29 (*p* < 0.01), indicating the suitability for PCA. The PCA resulted in five factors with Eigen-values greater than 1.0 and explained 62% of the total variation. The Cronbach's alpha reliability scores for the five components ranged from 0.725 to 0.809.

The five components were as follows:

Component 1 (departmental influence): four items—atmosphere of academic freedom, educational ideas and culture, academic systems and policies, and quality of student source.

Component 2 (research environment): five items—total research funds, prestige of colleges and departments, research equipment and library facilities, cooperation and relationship with colleagues, and extent of connection with academic circle.

Component 3 (geographic location and social atmosphere): five items—university geographic location, climate of university geographic location, social atmosphere, recreational and leisure facilities in the community, and educational facilities in the community.

Component 4 (income): three items—total personal income, potential income and insurance benefits, and housing size.

Component 5 (spouse): three items—spouse's career development opportunities, spouse's workplace, and spouse's total income.


**Table 6.** Principal component analysis.


Because of its strong correlation with income, we chose *mobility frequency* as the dependent variable for the regression model. Since housing size was not correlated with moving or not or with mobility frequency, it was not included in the model. The independent variables in the model included the income variables as well as academic titles and prestige of colleges and departments.

As noted in Table 7, a stepwise analysis was employed to assess the additive effects of the variables. The final model has good explanatory power (R-square = 0.698), indicating that the selected variables explain faculty mobility. Overall, the factors related to performance, income, institutional prestige and academic titles contribute to predicting faculty mobility.

Specifically, considering all publications at home and abroad in the past five years as an indicator of academic performance can demonstrate scholars' academic capability. In the regression model, the relationship between publication quantity and mobility frequency passed the significance level test. The income indicators also had strong explanatory power in this model, especially the direct income indicator, with a value reaching 13.2%, while the total research funds indicator for indirect income was over 10%. Another indirect income indicator, spouse's total income, had weak explanatory power, only 2.4%. This result reveals that in the faculty mobility process at Chinese research universities, income continues to play a very important role, which is reflected not only in scholars' personal income but also in their research funds. It is also partially related to the possession of research resources. As an indicator of the prestige of colleges and departments, types of doctoral degrees granted by the institution contributed significantly to this model. When we ranked the prestige of institutions from

high to low (based on a scale of 1–4), we found that the lower an institution's rank was, the less often one moved.


**Table 7.** Regression models of the impact factors at Chinese research universities.

*Levels:* \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05.

This study included cumulative academic work time abroad in the regression model. The results indicate that there was a significant positive correlation between the experience of working abroad at academic institutions (regarded as "international reputation") and mobility frequency. The more cumulative time that was spent working abroad in academic institutions, the more times the faculty member moved. Additionally, academic titles also had an important influence on the regression model. With titles ranked from high to low (1 for professor, 5 for teaching assistant), we found that the higher one's title was, the more frequently he or she moved.

The regression models presented in Table 7 indicate that the effects of income on faculty mobility are weaker than the effects of academic performance. It should be noted that types of doctoral degrees granted by the institution not only contributed to the model but also reduced the explanatory power. This model intuitively demonstrates that among faculty members at Chinese research universities, neither direct nor indirect income is the core impact factor, although they play crucial roles in the mobility process.

#### **5. Discussion**

The findings of this study demonstrate that although income contributes to faculty mobility at major Chinese research universities, a number of additional factors remain important for faculty and their work, particularly for those who move to a different institution. The desire for academic performance and institutional prestige are equal to if not more important than income. In fact, Chinese scholars frequently have discussions about the impact factors for faculty mobility at the theoretical level, but most of the impact factors have not yet been tested, and non-monetary factors are considered key for influencing faculty mobility. The first of these is educational background and "guanxi". The Chinese academic labor market is not yet completely marketized [19,54,55], so positive "guanxi" is needed for some aspects, such as obtaining new job information [56], acquiring help in the mobility process [36] and career development following a move [6,57–59], and these close and even private academic contacts are based primarily on the relationship between supervisors and students as well as self-established relationship networks in academic circles. Another of our studies analyzed the curriculum vitae of faculty members from "211 Project" universities and found that the rate of academic inbreeding at a few research universities (even well-known universities at the top of the Shanghai JiaoTong University Ranking list) is still higher than 60%, so the low mobility rate of Chinese academics is related to academic inbreeding. In contrast, the collapse of or tension in "guanxi" at faculty members' original academic institutions is also important for faculty mobility. For example, an interview about faculty mobility among Chinese scholars suggested that academic organizational culture is a significant factor. Academic organizational culture will undoubtedly generate centrifugal forces if the team disagrees, if there are strains in interpersonal relationships and if team members defame each other [56]. The second factor is academic titles. Some studies [60,61] have asserted that

the pursuit of senior academic titles provides motivation to transition to a new academic position, but this notion is controversial and has not undergone comprehensive empirical testing. For instance, some studies have found that only a few faculty members transfer to lower-ranked universities because of the pressure of academic titles, while fewer studies have reported that the issue of academic titles is resolved by transferring to lower-ranked universities [62].
