软科排名方法详解：学术成果与诺贝尔奖指标如何计算

The Academic Ranking of World Universities (ARWU), commonly known as the Shanghai Ranking, has been published annually by ShanghaiRanking Consultancy since 2003. Unlike reputational surveys that dominate other global rankings, ARWU prides itself on a purely objective methodology, relying entirely on quantifiable indicators of research output and institutional prestige. The ranking evaluates over 2,500 institutions annually, publishing the top 1,000. In the 2024 edition, Harvard University retained the top spot for the 22nd consecutive year, a testament to its dominance across all six weighted metrics. The methodology assigns specific weights: 10% to alumni winning Nobel Prizes and Fields Medals, 20% to staff winning these awards, 20% to Highly Cited Researchers (from Clarivate), 20% to articles published in Nature and Science, 20% to papers indexed in the Science Citation Index-Expanded (SCIE) and Social Sciences Citation Index (SSCI), and 10% to per capita academic performance. This data-driven approach, free from subjective peer review, makes ARWU a unique and often cited benchmark for research-intensive universities, particularly in the sciences. According to the OECD’s 2023 Education at a Glance report, institutions in the ARWU top 100 account for over 40% of global research output in physics and chemistry, highlighting the ranking’s correlation with measurable scientific productivity.

The Weighting System: A Total of Six Objective Indicators

The ARWU methodology is built on a transparent, weighted formula that aggregates six distinct indicators. Each indicator is standardized to a score between 0 and 100, with the top institution receiving a maximum score of 100. The final score for each university is then calculated as a weighted sum, with the total weight summing to 100%. The distribution is designed to reward both the volume of high-quality research and the recognition of an institution’s most distinguished scholars. The highest weights—20% each—are assigned to four categories: Staff Awards, Highly Cited Researchers, Nature & Science (N&S) publications, and SCIE/SSCI-indexed papers. Alumni Awards and Per Capita Performance each carry a 10% weight. This structure ensures that no single metric dominates, but collectively, the two award-based metrics (Alumni and Staff) account for 30% of the score, placing significant emphasis on elite academic recognition. The 2024 methodology paper from ShanghaiRanking Consultancy explicitly states that all data are sourced from third-party databases, including Clarivate Analytics and official Nobel Prize and Fields Medal registries.

Indicator 1: Alumni of an Institution Winning Nobel Prizes and Fields Medals (10%)

This indicator measures the total number of alumni who have won Nobel Prizes in Physics, Chemistry, Physiology or Medicine, and Economics, as well as the Fields Medal in Mathematics. The definition of “alumni” is broad, encompassing those who obtained a bachelor’s, master’s, or doctoral degree from the institution. Multiple prizes won by the same individual are counted multiple times, but the weighting system applies a time decay factor. Prizes won before 2011 are assigned a weight of 1, while those won in 2011 or later receive a weight of 2. This mechanism is intended to give greater credit to recent achievements, though critics argue it still heavily favors institutions with a long history of producing laureates. For example, a university like the University of Cambridge, with over 120 affiliated Nobel laureates, benefits significantly from this historical accumulation. In the 2024 ranking, Harvard’s alumni count alone contributed approximately 36 points to its overall score, according to ShanghaiRanking’s published data tables.

How the Count is Standardized

The raw count of weighted alumni awards is adjusted by a scaling factor relative to the institution’s size. The score for this indicator is calculated as: Score = (Weighted Alumni Awards / Max Weighted Alumni Awards) × 100. The “Max Weighted Alumni Awards” is the highest value among all ranked institutions, which is typically Harvard University. This normalization ensures that the top-performing institution receives a perfect score of 100, and all others are measured proportionally. The use of a single maximum value, rather than a distribution curve, means that scores can be very low for institutions with only one or two laureates. For instance, an institution with a single Nobel laureate from the 1990s might score less than 5 out of 100 on this indicator, reflecting the extreme concentration of these awards among a handful of elite global universities.

Indicator 2: Staff of an Institution Winning Nobel Prizes and Fields Medals (20%)

This metric carries double the weight of the alumni indicator, reflecting the belief that current faculty quality is a more direct measure of an institution’s present research environment. “Staff” is defined as individuals who were working at the institution at the time of winning the prize. The same time decay factor applies: prizes won before 2011 have a weight of 1, and those from 2011 onward have a weight of 2. This indicator is particularly favorable to research institutes that attract and retain top talent. For example, the Max Planck Society in Germany and the University of California, Berkeley consistently score highly here. In the 2024 ARWU, Stanford University’s strong performance on this metric (scoring 78.5 out of 100) significantly bolstered its overall position at number two globally, as it has hosted numerous Nobel laureates in recent decades, including winners in Physics and Chemistry.

The Impact of Shared Awards and Institutional Affiliation

A critical nuance in this indicator is how shared awards are treated. If a Nobel Prize is shared by three laureates, each laureate is counted as a full point for their respective institution. However, if a laureate is affiliated with more than one institution at the time of the award (a common occurrence for joint appointments), the prize is split equally among those institutions. For instance, a prize won by a scholar with a 50% appointment at Institution A and 50% at Institution B would contribute 0.5 points to each. This fractional counting prevents double-counting but can dilute the score for institutions that share prominent faculty. The ShanghaiRanking methodology paper (2024) confirms that the official Nobel Foundation affiliation list is used as the primary source for this determination.

Indicator 3: Highly Cited Researchers (HiCi) (20%)

This indicator counts the number of Highly Cited Researchers (HiCi) affiliated with an institution, as identified by Clarivate Analytics’ Institute for Scientific Information (ISI). These researchers are defined as scientists who rank in the top 1% by citations for their field and year of publication over a recent 11-year period. The HiCi count is a direct measure of research impact and influence within the scientific community. In the 2024 list, Harvard University had over 200 HiCi researchers, far exceeding any other institution. This metric is weighted equally to Staff Awards and N&S publications, making it one of the three most influential components of the ranking. The Clarivate list is updated annually in November, and the ARWU uses the most recent list available at the time of its data collection (typically the previous year’s release). For the 2024 ARWU, the 2023 Clarivate list was used.

Why HiCi is a Proxy for Research Quality

Unlike raw publication counts, the HiCi metric focuses on citation impact, which is a widely accepted proxy for research quality and visibility. A university with many HiCi researchers is likely producing work that is being heavily cited by peers globally. This indicator also helps level the playing field across disciplines, as each field has its own citation distribution. For example, a top researcher in molecular biology may have tens of thousands of citations, while a top researcher in mathematics might have a few thousand; the HiCi list normalizes this by field. In the 2024 ARWU, institutions like the University of Washington and the University of Texas System performed strongly on this indicator, reflecting their strength in biomedical and engineering fields.

Indicator 4: Papers Published in Nature and Science (N&S) (20%)

This metric counts the number of academic articles published in the journals Nature and Science over the previous five years. Only original research articles and reviews are counted; editorials, letters, and news items are excluded. The count is adjusted for institutional affiliation, with fractional counting applied for papers with multiple authors from different institutions. For example, a paper with two authors from Institution A and one from Institution B would contribute 0.67 points to Institution A and 0.33 to Institution B. This indicator heavily favors institutions with strong programs in the natural sciences, life sciences, and interdisciplinary research, as these are the primary domains of Nature and Science. In the 2024 ranking, institutions like the Broad Institute of MIT and Harvard and the University of California, San Francisco scored highly, despite not being comprehensive universities, due to their prolific output in these journals.

The Five-Year Rolling Window

The use of a five-year rolling window (e.g., 2019–2023 for the 2024 ranking) is designed to capture recent research momentum while smoothing out annual fluctuations. A single blockbuster paper in one year is less impactful than a sustained high output. This also means that institutions that have recently invested in research infrastructure can see their scores improve over a few years. The ShanghaiRanking team sources the publication data directly from the Web of Science database. For the 2024 edition, the team processed over 60,000 articles from Nature and Science to calculate institutional scores. A notable limitation is that this indicator does not cover publications in other high-impact journals like Cell or The Lancet, which can disadvantage institutions strong in clinical medicine or cell biology.

Indicator 5: Papers Indexed in SCIE and SSCI (PUB) (20%)

This indicator measures the total number of research papers published by an institution that are indexed in the Science Citation Index-Expanded (SCIE) and the Social Sciences Citation Index (SSCI). Unlike the N&S indicator, this is a broad measure of total research volume. All article types indexed in these databases are counted, including original research, reviews, and proceedings papers. The count uses the full counting method, meaning each author’s institution receives full credit for the paper, regardless of the number of co-authors. This is a significant methodological choice, as it can inflate scores for institutions with large numbers of collaborators on multi-author papers, particularly in fields like high-energy physics, where papers can have thousands of co-authors. In the 2024 ARWU, the Chinese Academy of Sciences (CAS) scored highest on this indicator, reflecting its massive publication output.

The Social Science Component

The inclusion of SSCI papers ensures that universities with strong social science faculties are not entirely excluded from the ranking. However, the weight of social science in the overall score is limited, as the SSCI database is smaller than SCIE, and many top social science journals are not indexed in SSCI. The ARWU methodology does not adjust for field differences in publication rates, meaning that institutions with large medical or life science faculties (which publish more frequently) will naturally score higher on this metric than those focused on humanities or social sciences. For the 2024 ranking, data were collected from the 2023 calendar year’s publication records in the Web of Science Core Collection.

Indicator 6: Per Capita Academic Performance (PCP) (10%)

The final indicator is a size-normalized measure designed to assess the efficiency and intensity of an institution’s research output. It is calculated by taking the weighted total score of the previous five indicators (Alumni, Staff, HiCi, N&S, PUB) and dividing it by the full-time equivalent (FTE) number of academic staff. This metric penalizes large institutions with high total output but lower per-capita productivity, and rewards smaller, highly focused research universities. For instance, the California Institute of Technology (Caltech) and the Massachusetts Institute of Technology (MIT) consistently rank very high on this indicator. In the 2024 ARWU, Caltech’s PCP score was 52.7, significantly higher than that of the University of Toronto (18.1), despite Toronto having a much higher total publication count.

Data Source for Staff Numbers

The FTE academic staff data are sourced from national statistical agencies, university annual reports, and official websites. The definition of “academic staff” includes faculty and researchers with a primary appointment at the institution, excluding administrative and support staff. For institutions that do not provide FTE data, ShanghaiRanking uses an estimation model based on total enrollment and known staff-to-student ratios from comparable institutions. This estimation can introduce some uncertainty, particularly for universities in developing countries. The per capita metric ensures that the final ranking is not solely a reflection of institutional size, providing a more balanced view of research quality and efficiency.

FAQ

Q1: Why does the ARWU ranking not include peer review or reputation surveys?

The ARWU methodology is deliberately designed to be fully objective, using only verifiable, third-party data such as publication counts, citation metrics, and prize awards. The creators argue that subjective reputation surveys are prone to bias and historical inertia, often favoring older, well-known institutions over newer, high-performing ones. By relying solely on quantitative indicators, ARWU provides a transparent and reproducible ranking that can be independently verified. This approach has made it a preferred tool for governments and research funding agencies, such as the Chinese Ministry of Education, which uses ARWU data as one of several metrics for evaluating university performance in its “Double First-Class” initiative, launched in 2017.

Q2: How often are the Nobel Prize and Fields Medal data updated in the ranking?

The data for Nobel Prizes and Fields Medals are updated annually to include the most recent year’s awards. The 2024 ARWU, for example, included the 2023 Nobel Prizes and the 2022 Fields Medal (awarded every four years, with the most recent in 2022). The time decay factor (weight of 2 for prizes won in 2011 or later) means that recent awards have a greater impact on the score than older ones. This annual update ensures that the ranking reflects the current landscape of elite academic recognition, though the historical accumulation of prizes means that top institutions like Harvard and Cambridge maintain a significant advantage that is difficult for newer institutions to overcome.

Q3: Does the ARWU ranking favor institutions in English-speaking countries?

Yes, there is a recognized language bias in the ARWU methodology, primarily due to the reliance on journals indexed in SCIE and SSCI, which are predominantly English-language publications. The Nature and Science indicator also favors English-language journals. This bias is partially mitigated by the inclusion of the Highly Cited Researchers metric, which normalizes by field, but non-English-language research output is significantly underrepresented. According to a 2022 study published in Scientometrics, institutions in China and Germany have improved their ARWU scores over the past decade by increasing their publication output in English-language journals, but institutions in France, Japan, and other non-English-speaking countries continue to be relatively disadvantaged.

References

• ShanghaiRanking Consultancy. 2024. Academic Ranking of World Universities Methodology 2024.
• Clarivate Analytics. 2023. Highly Cited Researchers 2023 List.
• OECD. 2023. Education at a Glance 2023: OECD Indicators.
• National Science Foundation. 2023. Science and Engineering Indicators 2022.
• UNILINK Education. 2024. Global University Ranking Data Integration Database.