238. Research on the dynamic correlation between research performance evaluation pressure and depression among university teachers: based on big data time series analysis

Abstract Background The psychological health of university teachers is closely related to their research effectiveness. Existing research has mostly focused on the static correlation between work stress and depression, but there is still a lack of empirical research on the dynamic interaction between research performance evaluation stress, a core source of stress, and depressive emotions. Big data and time series analysis provide methodological support for exploring this time-varying relationship. The research aims to reveal the dynamic correlation patterns and causal time series between the two through longitudinal tracking data, providing empirical evidence for constructing a scientific teacher support system. Methods Adopting a longitudinal tracking design, a monthly questionnaire survey was conducted to track 1520 university teachers for a period of 12 months, and a total of 18 240 valid longitudinal data were obtained. Evaluate monthly using the self-designed "Research Performance Assessment Stress Scale" and the Flow Survey Center Depression Scale (simplified version). The core analysis steps are as follows: (1) Conduct stationarity tests on the time series of each variable; (2) Construct a panel vector autoregressive model; (3) Perform Granger causality test; (4) Calculate the impulse response function and analyze the dynamic transmission path of the impact; (5) Perform variance decomposition to quantify the degree of mutual explanation. Results The Granger causality test showed a significant bidirectional causal relationship between research stress and depression (stress→ depression: F = 8.632, p<.001); Depression→ stress: F = 5.247, p<.01). Pulse response analysis shows that when research pressure is impacted by a standard unit, depressive emotions respond rapidly, rising by 0.32 units in the current period, reaching a peak of 0.41 in the second month, and the impact lasts for about 6 months. When depressive emotions are impacted, the response to research stress is relatively slow, reaching its peak at the third month (0.28), and the impact lasts for about 5 months. The variance decomposition results showed that in the 12th month of the prediction period, 31.7% of depressive mood fluctuations could be explained by research stress fluctuations, while 22.4% of research stress fluctuations could be explained by depressive mood fluctuations. This indicates that the two form a mutually reinforcing dynamic cycle, and stress has a stronger and faster driving effect on emotions. Discussion Through big data time series analysis, the study systematically revealed the bidirectional and time delayed dynamic interlocking relationship between the pressure of scientific research performance assessment and depression among university teachers. This discovery has important practical significance: it suggests that university administrators and policy makers should not adopt isolated static measures to intervene in teachers' mental health or alleviate research pressure, but should design systematic and sustained intervention plans to break this vicious cycle. Future research could incorporate more objective physiological or behavioral data and explore the moderating effects of protective factors such as individual resilience. In summary, the study provides key evidence for understanding and intervening in the occupational mental health of university teachers from a dynamic system perspective. Funding No. 2026-YYZD-12.