We invited all enrolled third–year undergraduate students at the target medical school to participate via class channels (hyperlink/QR code). Eligibility required current third–year status and completion of all questionnaire items; participation was voluntary and anonymous. Relevant studies suggest that for networks with fewer than 20 nodes, a minimum sample size should generally exceed 20 times the number of nodes or comprise at least 250–350 cases (Constantin and Cramer, 2018). Moreover, whether a given sample affords sufficient statistical power in network analysis should be judged alongside stability parameters (Constantin et al., 2023). We distributed 540 questionnaires and analyzed 499 valid responses (see Table 1 for gender, place of origin, academic rank, and parental education), meeting the minimum sample requirements. We focused on third–year students because their career planning and learning goals are typically more defined at this stage, which reduces heterogeneity in motivational orientations and allows a clearer depiction of engagement. This study was approved by the Ethics Committee of Jining Medical University. Clinical trial number: not applicable.

This study used the Chinese version of the learning motivation scale developed by Amabile et al. and adapted by Chi and Xin (2006). The scale comprises 30 items distributed across six dimensions: challenge (8 items), enthusiasm (6 items), dependence on others' evaluations (6 items), choosing simple tasks (4 items), focusing on interpersonal competition (4 items), and pursuit of rewards (2 items). The “challenge” and “involvement” dimensions constitute the intrinsic motivation subscale, whereas “reliance on others” “evaluations,” “preference for simple tasks,” “focusing on interpersonal competition,” and “pursuit of rewards” constitute the extrinsic motivation subscale. Each item is rated on a 4-point Likert scale, from 1 (“completely inconsistent”) to 4 (“completely consistent”), with higher scores indicating stronger learning motivation. In this study, the overall Cronbach's α of the scale was 0.874.

This study employed a modified version of the Utrecht Work Engagement Scale–Student (UWES-S) developed by Dong (2021). The scale comprises 15 items across three dimensions: vitality (3 items), dedication (3 items), and behavior (9 items). Each item is rated on a 5-point Likert scale, from 1 (“completely inconsistent”) to 5 (“completely consistent”), with higher scores indicating higher levels of learning engagement. In this study, the overall Cronbach's α of the scale was 0.906.

Data analysis involved descriptive statistics and network analysis. First, SPSS 24.0 was used for descriptive statistics to explore participants' basic information and relevant demographic variables. Second, R 4.4.2 was employed for network analysis to investigate the structural relationships among the dimensions of learning motivation and learning engagement. The network analysis included the following five components: network estimation, network visualization, centrality estimation, network comparison, and assessments of network accuracy and stability.

Table 1 presents the general demographic characteristics of the participants. The results indicate that among government-funded medical students, the proportion of males (52.8%) exceeded that of females (47.2%). Moreover, the majority of students (54.1%) came from towns or villages, with rural-origin students comprising a significantly larger share of the government-funded group (65.3%) compared to those from urban areas (43.0%). In terms of academic performance, 60.5% ranked in the top 50% of their cohort. Additionally, most fathers and mothers had a junior high school education or below, at 40.9% and 52.5%, respectively.

As depicted in Figure 1, the overall sample network—comprising nine nodes and 23 edges—had a mean edge weight of 0.087 and a density of 0.639. The learning motivation nodes and learning engagement nodes formed two distinct clusters. The primary link connecting learning motivation and learning engagement was between challenge and learning dedication, showing the strongest correlation (weight = 0.22). Within the learning motivation cluster, challenge was closely associated with both enthusiasm and focusing on interpersonal competition (weights = 0.35 and 0.34, respectively). Within the learning engagement cluster, learning dedication was strongly linked to learning vitality (weights = 0.43).

Figures 2, 3 illustrate the network structures of learning motivation and engagement for government-funded and self-funded medical students, respectively. In terms of learning motivation, both groups exhibited a strong association between focusing on interpersonal competition and challenge. In learning engagement, learning dedication and learning vitality also showed a robust connection in both groups. Further analysis revealed differences in the network linkages between motivation and engagement: among government-funded students, focus on interpersonal competition was most tightly connected to learning dedication (weight = 0.21), whereas among self-funded students, the most prominent link was between challenge and learning behavior (weight = 0.23).

In our network, challenge (intrinsic) and focusing on interpersonal competition (extrinsic) exhibit the highest Expected Influence (EI) (Figure 4). Practically, small, intentional adjustments to these two dimensions are most likely to cascade to multiple engagement behaviors (e.g., dedication, vitality). For example, in a diagnostic-reasoning course, instructors can deploy staged case-based teaching to elicit engagement. Students begin with a patient presenting with acute abdominal pain and must sequentially analyze history, electrocardiogram, and imaging findings. As information unfolds and conflicting data appear (e.g., discrepant test results), the level of challenge escalates. This design fosters intense focus, prompts autonomous study after class, and sustains motivation derived from resolving difficult problems. Dependence on others' evaluations displayed the lowest EI, suggesting a relatively peripheral position in the learning motivation–engagement network with weak associations to other behavioral indicators.

Figure 5 illustrates the EI indices for government-funded and self-funded students. In the government-funded sample, focusing on interpersonal competition ranked the highest, whereas choosing simple tasks showed the lowest EI. Among self-funded students, challenge was the most influential dimension, while dependence on others' evaluations had the lowest EI.

Based on the stability estimates for the centrality metrics (Figure 6) and the bootstrap results of the edge weights (Figure 7), all three networks (overall sample, government-funded students, self-funded students) exhibited high accuracy. The centrality stability (CS) coefficients exceeded 0.75, indicating good overall stability.

Using 95% confidence intervals (CIs) for edge weights to assess network accuracy, the bootstrapped differences in edge weights are shown in Figure 7. The red line indicates edge weights from the original sample, and the black line represents estimates from nonparametric bootstrap methods; the gray area denotes the 95% CI for edge weights. In Figure 7, the red and black lines almost overlap, and the 95% CIs are relatively narrow, suggesting that the estimated edge weights are highly accurate.

Three tests were conducted to compare differences in the learning motivation–engagement networks between government-funded and self-funded medical students. The Network Structure Invariance test showed no significant difference (p = 0.10), and the Global Strength Invariance test indicated that the two groups did not differ significantly in overall network strength (p= 0.74). Additionally, none of the edges in the Local Invariance test reached statistical significance (p > 0.05). These findings suggest a high degree of stability in learning motivation and engagement dimensions across different training models.

Our findings indicate that challenge and focusing on interpersonal competition function as central hub nodes in the network, exerting significant influence on the overall structure. When these nodes are activated, their effects spread throughout the network by way of interconnected nodes, thereby shaping the dynamic evolution of the symptom network (Hofmann et al., 2016). As a sub-dimension of intrinsic motivation, challenge emphasizes individual growth and perseverance. Network analyses revealed a strong link between challenge and learning dedication—a dimension of learning engagement—suggesting that these dimensions bridge the learning motivation and engagement sub-networks. This connection may arise because challenging tasks strongly evoke intrinsic motivation, prompting higher levels of engagement among students (Larson and Rusk, 2011). Additionally, such tasks can bolster self-efficacy and a sense of accomplishment, further reinforcing behaviors tied to learning engagement (Shang and Ma, 2024). Within the learning motivation network, challenge is closely linked to both involvement (weight = 0.35) and focusing on interpersonal competition (weight = 0.34). This finding implies that challenge may activate motivation via two pathways: interest-driven involvement and external competitive stimuli.

Prior research has identified an inverted U-shaped relationship between challenge and intrinsic motivation (Ma et al., 2017): motivation increases as challenge level rises until an optimal point, after which motivation begins to decline. Appropriate levels of challenge maximize intrinsic motivation, manifesting in enhanced focus, perseverance, and determination—factors known to significantly improve task performance (Aubé et al., 2014). This has important implications for medical education: when designing learning tasks, educators should account for both the optimal degree of challenge and individual differences. Such precise task design can maximize intrinsic motivation and foster more effective learning and professional growth.

Although the majority of research emphasizes the positive effects of intrinsic motivation on learning engagement, extrinsic motivation also exerts a certain degree of positive influence on learning engagement (Goswami and Urminsky, 2017; Liu et al., 2022). In our study, the focusing on interpersonal competition serves as a pivotal node, significantly impacting the overall network as well. Existing research has explored the relationship between competition and learning engagement from various perspectives. For instance, Xu et al., 2018 empirically found that a healthy competitive attitude can enhance students‘ level of learning engagement, whereas an excessively competitive attitude may diminish their motivation to engage in learning. (Meng et al., 2024) demonstrated a positive correlation between competition and learning engagement. From a positive standpoint, competition can stimulate individuals' intrinsic motivation, prompting learners to focus more intently on learning tasks to gain an advantage in competitive scenarios. Moreover, within a competitive atmosphere, learners are likely to set higher goals and adopt more proactive learning strategies, such as in-depth study of materials, actively seeking feedback, and continuous self-reflection, all of which contribute to improving their academic performance.

According to Maslow's hierarchy of needs (Maslow, 1943), competition may help students recognize their own value and gain respect from others. External feedback is thus internalized into intrinsic motivation, generating sustained positive reinforcement that further promotes competitively oriented learning behaviors. Nevertheless, excessive competition may induce anxiety and stress, ultimately diminishing motivation. Therefore, teaching methods such as problem-based learning (PBL) and case-based learning (CBL) can be adopted to encourage mutual support and teamwork, striking a balance between cooperation and competition.

Comparing networks for publicly funded and self-funded medical students revealed notable differences. Among publicly funded students, interpersonal competition is the most influential node and shows the strongest association with absorption. Instructional design should therefore channel competitive pressure into mastery-oriented collaborative environments. For instance, a redesigned leaderboard can rank not only by position but also recognize accuracy and improvement, and award points for helping peers learn. This preserves the motivational benefits of competition while fostering mutual support and collective progress. During clinical placements, a “buddy-pair” system—two students jointly recording and checking each other's skill gains—can retain the energizing aspects of comparison while curbing unhealthy competition. These strategies maintain the motivational edge of competition yet redirect it toward collective growth, thereby supporting sustained focus and engagement.

For self-funded students, challenge is the central node and is most tightly associated with absorption. Curriculum and assessment can therefore prioritize calibrated difficulty and autonomy-supportive choice. In coursework, progressive-complexity assignments (worked example → partial example → independent case) and choice-based “stretch tasks” invite deeper problem-solving while avoiding overload. In clinical training, short, mentored intensives (e.g., a “mini-fellowship week” on airway management or ultrasound-guided procedures) can consolidate difficult skills, which—per the challenge-dedication- vitality pathways—should propagate to sustained engagement. Reflective difficulty-calibration notes (what was hard, why, how I'll approach it next time) help students internalize adaptive challenge rather than avoid it.

Preference for simple tasks generally reflects a tendency, driven by extrinsic motivation, to choose tasks with clear goals and predictable processes. Our findings show that among government-funded students, preference for simple tasks had the lowest expected influence index, suggesting that it exerts a relatively weak effect on learning behavior in this group. Faced with high levels of competitive pressure, these students may actively opt for more difficult tasks to improve their capabilities and competitiveness, rendering simple tasks insufficiently challenging to spark motivation. Meanwhile, among self-funded students, reliance on others' evaluations exhibited the lowest expected influence index, potentially for two reasons: first, such students tend to be more autonomous and independent, preferring to evaluate their own performance and set personal standards; second, many self-funded students chose the medical profession out of personal interest, resulting in a stronger intrinsic drive and less dependence on external validation. From an intervention perspective, therefore, educators of government-funded students might focus on stimulating intrinsic motivation by providing more challenging, practice-oriented tasks that meet their needs for skill enhancement. For self-funded students, strategies should emphasize fostering autonomy and self-reflection, encouraging learning plans driven by personal interests and internal goals rather than external feedback.