National Tsing Hua University

EXECUTIVE SUMMARY AND STRATEGIC VISION

National Tsing Hua University presents a robust scientific integrity profile, with an overall risk score of -0.027 indicating a performance well-aligned with global best practices. The institution demonstrates exceptional strengths in its publication strategies, showing very low-risk levels in the Rate of Output in Discontinued Journals and the Rate of Output in Institutional Journals, complemented by a low-risk profile for Hyperprolific Authors. These results point to strong governance and a commitment to high-quality, externally validated research dissemination. This solid foundation supports the university's outstanding thematic performance, as evidenced by its top national rankings in critical areas such as Engineering (Top 2), Physics and Astronomy (Top 3), Computer Science (Top 5), and Mathematics (Top 5), according to SCImago Institutions Rankings data. However, a cluster of indicators at a medium-risk level—including Institutional Self-Citation, Hyper-Authored Output, and Redundant Output—suggests emerging vulnerabilities that could, if left unaddressed, subtly undermine the principles of excellence and transparency central to any leading academic mission. Proactively addressing these moderate deviations will be key to ensuring that the university's operational integrity fully matches its celebrated academic and research achievements.

ANALYSIS BY INDICATOR

Rate of Multiple Affiliations

The institution's Z-score of 0.432, while in the medium-risk range, is notably lower than the national average of 1.166. This suggests that while the university operates within a national context where multiple affiliations are common, it exercises more effective management and control over this practice than its peers. While multiple affiliations can be a legitimate outcome of collaboration, disproportionately high rates can signal strategic attempts to inflate institutional credit. The university's ability to moderate this trend indicates a differentiated approach that helps safeguard against the potential for "affiliation shopping," thereby maintaining a clearer attribution of its scientific contributions.

Rate of Retracted Output

With a Z-score of 0.070, the institution's rate of retractions is nearly identical to the national average of 0.051, placing both in the medium-risk category. This alignment suggests that the university's performance reflects a systemic pattern present throughout the country's research ecosystem. Retractions are complex events, but a rate significantly above the global average can alert to a vulnerability in the institution's integrity culture. In this case, the data points not to an isolated institutional issue but to a shared national challenge, indicating that pre-publication quality control mechanisms may be failing systemically across the board, a situation that warrants collective attention.

Rate of Institutional Self-Citation

The university's Z-score of 0.064 for institutional self-citation marks a moderate deviation from the country's low-risk average of -0.204. This indicates that the institution shows a greater sensitivity to this risk factor than its national counterparts. While a certain level of self-citation is natural, disproportionately high rates can signal concerning scientific isolation or 'echo chambers.' This higher value warns of a potential risk of endogamous impact inflation, suggesting that the institution's academic influence may be oversized by internal dynamics. A review is recommended to ensure its work receives sufficient external scrutiny and validation from the global community.

Rate of Output in Discontinued Journals

The institution demonstrates an exemplary performance with a Z-score of -0.378, which is in the very low-risk category and significantly better than the country's already low-risk average of -0.165. This result shows a consistent and well-established process for selecting high-quality publication venues. The virtual absence of risk signals in this area confirms that the university has robust due diligence mechanisms, effectively protecting its research and reputation from the risks associated with 'predatory' or low-quality publishing practices and aligning perfectly with national standards of academic integrity.

Rate of Hyper-Authored Output

With a Z-score of 0.483, the institution shows a medium-risk level that moderately deviates from the low-risk national average of -0.671. This suggests that the university's research culture is more susceptible to practices of author list inflation than its peers. While extensive author lists are legitimate in 'Big Science' fields, a high Z-score outside these contexts can dilute individual accountability and transparency. This signal serves as an important prompt for the institution to review its authorship practices and distinguish between necessary massive collaborations and potentially 'honorary' or political authorship.

Gap between Impact of total output and the impact of output with leadership

The institution's Z-score of 0.193 indicates a medium-risk gap, representing a moderate deviation from the low-risk national average of -0.559. This suggests that the university's scientific prestige may be more dependent on external collaborations where it does not hold intellectual leadership compared to other institutions in the country. A high value in this indicator signals a potential sustainability risk, inviting reflection on whether its high-impact metrics are the result of genuine internal capacity or strategic positioning in partnerships. Strengthening internal research leadership would ensure its long-term scientific autonomy and structural excellence.

Rate of Hyperprolific Authors

The institution's Z-score of -0.439 is in the low-risk range, a positive contrast to the medium-risk national average of 0.005. This demonstrates significant institutional resilience, suggesting that internal control mechanisms are effectively mitigating the systemic risks of hyperprolificity observed elsewhere in the country. By maintaining a low rate of authors with extreme publication volumes, the university successfully avoids potential imbalances between quantity and quality, steering clear of risks like coercive authorship or the assignment of credit without real participation, which appear to be a greater challenge for its national peers.

Rate of Output in Institutional Journals

With a Z-score of -0.268, the institution shows a very low-risk profile, outperforming the low-risk national average of -0.075. This result highlights a strong commitment to seeking external, independent validation for its research. By minimizing its reliance on in-house journals, the university effectively avoids the conflicts of interest and risks of academic endogamy that can arise when an institution acts as both judge and party. This practice ensures its scientific production is vetted through standard competitive review, enhancing its global visibility and credibility in a way that exceeds the national norm.

Rate of Redundant Output

The institution's Z-score of 0.361 places it in the medium-risk category, a notable deviation from the country's low-risk average of -0.176. This suggests a greater tendency within the university's research culture toward data fragmentation or 'salami slicing' to increase publication output. A high value here alerts to the risk of artificially inflating productivity by dividing coherent studies into minimal publishable units. This practice can distort the scientific record and overburden the review system, indicating a need to reinforce policies that prioritize the publication of significant, new knowledge over sheer volume.