Perceived supervisor guidance was measured through five areas: the supervisor’s academic competence, guidance style, supervisory philosophy, timeliness of feedback and access to resources such as funding, equipment, data, platforms and academic exchange opportunities. Perceived graduate affairs management support covered the implementation of training regulations, management of academic procedures, life-related services and psychological care. The outcome combined students’ assessments of research engagement, perceived research competence, research output and research–life balance. All three measures were reported on five-point scales, with higher scores indicating more favorable perceptions. The questionnaire was adapted from relevant research and adapted to the institutional context, reviewed by five higher-education experts and piloted with 50 students. Its internal consistency was high, with Cronbach’s alpha values of 0.850, 0.841 and 0.843 for supervisor guidance, management support and research development, respectively.

In regression models adjusted for student gender and grade, supervisor age and title, supervisory load, and university fixed effects, perceived supervisor guidance was associated with research development when entered alone, with an unstandardized coefficient of 0.321 and a standardized coefficient of 0.306. Management support also had a strong independent association when entered alone, producing an unstandardized coefficient of 0.432 and a standardized coefficient of 0.418. When both support measures appeared in the same model, supervisor guidance retained a positive association with research development at B = 0.165 and p = 0.003, while management support remained more strongly associated at B = 0.364 and p < 0.001. The complete model explained 23.0% of the variance in self-reported research development, with an adjusted R-squared of 0.200. These estimates describe average associations across the sample and should not be read as estimates of what would happen if a particular university changed its support arrangements.

The statistical analysis also found that students who rated their supervisor guidance more favorably tended to rate management support more favorably, with a coefficient of 0.429. A bootstrap analysis identified a cross-sectional indirect association between supervisor guidance and research development through management support, estimated at B = 0.156 with a 95% bootstrap confidence interval from 0.103 to 0.217. This result does not establish a temporal chain or causal mediation because all variables were measured at the same time. An exploratory test found no evidence that management support amplified the supervisor-guidance association in a multiplicative interaction: the interaction coefficient was 0.005, its 95% confidence interval ranged from −0.107 to 0.117, and p was 0.933. In other words, the data did not show that the relationship between supervisor guidance and research development became systematically stronger only at particular levels of management support.

Latent profile analysis identified four distinct configurations among the nine item-level indicators of supervisor guidance and management support. The “fully supported” profile contained 129 students, or 34.1% of the sample, and had estimated mean scores of about 4.25 for supervisor guidance and 4.19 for management support. The “management-support-dominant” profile included 94 students, or 24.9%, with comparatively low supervisor-guidance scores of about 2.89 and high management-support scores of about 3.97. The “supervisor-support-dominant” profile comprised 103 students, or 27.2%, showing the reverse configuration: supervisor guidance was about 3.55, while management support was about 2.84. The remaining 52 students, or 13.8%, formed an “insufficiently supported” profile with estimated scores of about 2.26 and 2.50 in the two domains. The four-profile solution was retained after the five-profile model was rejected because of a near-singular class-specific variance estimate.

Comparisons adjusted for uncertainty in profile classification showed a consistent ordering in students’ self-reported outcomes. The total research-development mean was highest in the fully supported group at 3.74, followed by the management-support-dominant group at 3.38, the supervisor-support-dominant group at 2.98 and the insufficiently supported group at 2.80. The same general pattern appeared for engagement, perceived competence, output and research–life balance. Fully supported students reported scores of 3.78 for engagement, 3.66 for competence, 3.66 for output and 3.87 for balance, whereas the insufficiently supported group reported corresponding values of 2.70, 2.75, 2.84 and 2.89. These differences support the conclusion that the combination of support mattered: students with one comparatively strong source of support did not report the same outcomes as students reporting strong support across both domains.

Interviews with 18 graduate students, five supervisors and six graduate affairs administrators clarified the distinct functions represented by the two measures. Supervisors were described as providing substantive academic support, including topic selection, methodological instruction, experimental guidance, feedback on theses and manuscripts, academic judgment and help in interpreting failed experiments. Graduate affairs administrators were described as managing formal milestones, interpreting training plans and regulations, arranging proposal and assessment procedures, organizing thesis review and graduation requirements, providing financial and life-related assistance, offering psychological care and identifying students at risk of delayed progress. Students’ accounts suggested that procedural explanations could reduce uncertainty even when they did not solve technical research problems. One student said that a clear explanation of the proposal-defense process meant that many unknowns became known, even though the required work remained.

The qualitative material also identified sources of support gaps. Some students described uncertainty over how to begin a research project, how to break a proposal into daily tasks or whether a proposed technical route was feasible. Others described anxiety associated with experimental setbacks, difficult equipment access, manuscript submission, thesis review, employment decisions and overlapping graduation deadlines. Administrators noted that late-stage pressure could contribute to insomnia, avoidance of submitted materials or reluctance to read review comments. These accounts did not suggest that administrative support could replace academic supervision. Instead, participants consistently separated the roles: supervisors remained responsible for research direction, methodological judgment and thesis quality, while graduate affairs units clarified procedures, coordinated services and tracked risks. The data therefore support a “support ecology” in which multiple actors contribute connected but non-identical resources.

The study has important boundaries. It was cross-sectional, used a single self-report source and measured research development through students’ own assessments rather than verified publications, patents or supervisor ratings. The sample was limited to three science and engineering universities in one Chinese province, so generalizability to other regions, institutions and disciplines remains uncertain. The researchers also could not directly model clustering of students within supervisors because identifiable supervisor information was not retained. Although the interview sample was drawn from survey participants, de-identification prevented reliable linkage between individual interviews and profile membership. The findings therefore do not show that strengthening one support domain will automatically improve the other or that institutional support produces better research. They do show that students’ perceptions of both academic guidance and process-based support were independently connected with their reported engagement, competence, output and ability to balance research with personal life.

Subject of Research: How perceived academic and institutional support relate to master’s students’ self-reported research development

Article Title: Support ecology in master’s research development: perceived supervisor guidance and graduate affairs management support

Article References: Zhu, M., & Zhou, S. (2026). Support ecology in master’s research development: perceived supervisor guidance and graduate affairs management support. Frontiers in Psychology, 17, Article 1927035. https://doi.org/10.3389/fpsyg.2026.1927035

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1927035

Keywords: graduate education, research development, supervisor guidance, institutional support, master’s students, science and engineering, latent profile analysis, mixed-methods research, Support, ecology, master, research