Subject Rankings vs World University Rankings: What the Indicators Really Measure | 学科排名与世界大学排名的指标差异

📚 Subject Rankings vs World University Rankings: What the Indicators Really Measure | 学科排名与世界大学排名的指标差异

Every year students, parents and counsellors place league tables side by side, yet rarely notice an obvious paradox: a university ranked 80th in the world can sit inside the global top 20 for one subject, while another institution with a strong overall rank barely appears in that same subject table. The explanation is not inconsistency or bias but design. World university rankings and subject rankings answer different questions, and therefore rely on different indicator sets, different data scopes and different weighting logics.

每年学生、家长和升学顾问都会把各类排行榜并排比较，却很少注意到一个显而易见的悖论：一所世界排名第80位的大学，可能在某个学科上进入全球前20；而另一所综合排名很高的大学，在同一份学科榜单上却几乎看不到。原因并非榜单自相矛盾或存在偏见，而是设计使然：世界大学排名与学科排名回答的是不同的问题，因而采用不同的指标体系、不同的数据范围与不同的权重逻辑。

---

1. Why the Same University Gets Two Different Ranks | 同一所大学为何有两个不同名次

The first thing to understand is the unit of analysis. A world university ranking treats the institution as a single object: one score, one position. It aggregates medicine, law, engineering, history and music into a composite number. A subject ranking narrows the lens to one discipline and asks how strong the institution is in that field alone.

首先要理解的是”分析单位”的差异。世界大学排名把整所大学当作一个对象：一个总分、一个名次，它把医学、法学、工程、历史、音乐等全部学科汇总成一个综合数字。而学科排名则把镜头收窄到某一个学科，只问”这所大学在这个领域有多强”。

Because the unit of analysis differs, the two rankings are not competing versions of the same measurement. They are complementary instruments, and the gap between them is often where the most useful information for applicants hides.

由于分析单位不同，两类排名并不是同一种测量的两个竞争版本，而是互为补充的两套工具。二者之间的落差，往往正隐藏着对申请者最有价值的信息。

---

2. Institutional Data vs Subject-Level Data | 院校层面数据与学科层面数据

World rankings typically draw on institution-wide data: total research income, total citations across all fields, the overall faculty-to-student ratio, the overall international staff share, and a global reputation survey covering every discipline. Subject rankings instead draw on a restricted journal set, a discipline-specific citation window, and a reputation panel made up only of academics working in that field.

世界大学排名通常使用全校层面的数据：科研总经费、覆盖所有领域的引用总量、整体师生比、整体国际教师比例，以及涵盖全部学科的国际声誉调查。学科排名则使用受限的数据：特定学科的期刊集合、符合该学科引用习惯的统计窗口，以及只由该领域学者构成的声誉评审团。

This distinction produces predictable effects. Large comprehensive universities with medical schools accumulate enormous citation totals and therefore gain an advantage in world rankings, while small specialist institutions can dominate a subject table without ever entering the global top 200.

这种差异会带来可预期的结果：拥有医学院的大型综合性大学能积累巨大的引用总量，从而在世界大学排名中占据优势；而规模不大但高度专精的院校，可能长期雄踞某个学科榜单，却根本进不了世界前200。

The composite logic behind both types of table is the same weighted sum:

两类榜单背后的合成逻辑都是同一个加权求和公式：

Composite Score = w₁I₁ + w₂I₂ + w₃I₃ + … + wₙIₙ , where Σwᵢ = 1

The difference lies entirely in which indicators Iᵢ are included and how large each weight wᵢ is.

差别完全在于：纳入哪些指标 Iᵢ，以及每个权重 wᵢ 有多大。

---

3. QS World University Rankings: Indicator Architecture | QS世界大学排名：指标架构

QS is the most reputation-heavy of the major systems. Academic reputation alone carries roughly 30 per cent of the total, and employer reputation adds a further 15 per cent, meaning that subjective surveys account for nearly half of a university’s final position.

QS 是主流排名体系中最倚重声誉的一个。仅学术声誉就约占30%，雇主声誉再占15%，也就是说主观调查几乎决定了最终名次的半壁江山。

• Academic Reputation (about 30%) – a global survey of academics asked to name the strongest institutions in their field. | 学术声誉（约30%）——向全球学者调查其所在领域最强的院校。

• Employer Reputation (about 15%) – a survey of graduate employers. | 雇主声誉（约15%）——面向毕业生雇主的调查。

• Citations per Faculty (about 20%) – total citations divided by academic staff headcount. | 师均引用（约20%）——引用总量除以教师人数。

• Faculty-Student Ratio (about 10%) – a proxy for teaching resource. | 师生比（约10%）——教学资源的替代指标。

• International Faculty, International Students, International Research Network, Employment Outcomes and Sustainability (about 5% each). | 国际教师比例、国际学生比例、国际研究网络、就业成果、可持续性（各约5%）。

Note the phrase ‘citations per faculty’. The denominator is the whole institution’s academic staff, not the staff of any one department, which is precisely why the indicator cannot transfer to a subject-level table.

请注意”师均引用”这一指标：分母是全校教师人数，而不是某个院系的教师人数——这正是它无法直接搬到学科榜单上的原因。

---

4. QS Subject Rankings: A Reputation-Led Design | QS学科排名：声誉主导的设计

QS subject tables keep reputation dominant but replace the faculty-normalised citation measure with two paper-level indicators. In most subject areas the four core components are academic reputation at about 40 per cent, employer reputation at about 10 per cent, citations per paper at about 20 per cent and the H-index at about 20 per cent, with an international research network component of roughly 10 per cent in recent editions.

QS 学科榜单仍以声誉为绝对核心，但把”师均引用”换成了两个论文层面的指标。在大多数学科领域中，四个核心部分约为：学术声誉40%、雇主声誉10%、篇均引用20%、H指数20%，近几版还加入了约10%的国际研究网络指标。

The shift from ‘citations per faculty’ to ‘citations per paper’ matters enormously. Dividing by papers rather than by people removes the size penalty faced by large departments and the size bonus enjoyed by institutions that simply employ more staff.

从”师均引用”改为”篇均引用”意义重大：以论文数而非人数作分母，消除了大院系因规模大而吃亏、或仅因教师多而占便宜的规模效应。

The H-index component further rewards sustained output, because a high H-index requires many papers each attracting many citations, not one lucky highly cited article.

H指数部分则进一步奖励持续产出：高H指数要求大量论文各自获得大量引用，而不是靠某一篇偶然的高被引论文。

---

5. THE World University Rankings: Five Balanced Pillars | THE世界大学排名：五大支柱

Times Higher Education organises its world ranking around five pillars rather than a long list of separate indicators. The current framework gives roughly 29.5 per cent to Teaching, 29 per cent to Research Environment, 30 per cent to Research Quality, 7.5 per cent to International Outlook and 4 per cent to Industry.

泰晤士高等教育（THE）的世界大学排名以五大支柱组织，而不是罗列一长串独立指标。现行框架大约为：教学29.5%、研究环境29%、研究质量30%、国际视野7.5%、产业收入4%。

Pillar | 支柱	Weight | 权重	Typical sub-indicators | 典型子指标
Teaching | 教学	≈ 29.5%	Reputation survey, staff-to-student ratio, doctorates awarded, institutional income | 声誉调查、师生比、授予博士学位数、机构收入
Research Environment | 研究环境	≈ 29%	Reputation survey, research income, research productivity | 声誉调查、科研经费、科研产出量
Research Quality | 研究质量	≈ 30%	Citation impact, research strength, research excellence, research influence | 引用影响力、研究实力、卓越研究、研究影响
International Outlook | 国际视野	≈ 7.5%	International staff, students and co-authorship | 国际教师、国际学生、国际合作论文
Industry | 产业	≈ 4%	Research income from industry, patents | 来自产业的科研收入、专利

Crucially, even the world-ranking framework uses field-normalised citation metrics, because a paper in molecular biology is cited far more often than a paper in mathematics. This normalisation principle becomes the organising idea of the subject tables.

关键在于，即便是世界排名框架也使用了按学科归一化的引用指标，因为分子生物学论文的被引次数远高于数学论文。这一归一化原则，正是学科榜单的组织核心。

---

6. THE Subject Rankings: Weight Reallocation | THE学科排名：权重重分配

THE subject rankings use the same five pillars but reallocate the weights so that they match the research culture of each discipline. In the arts and humanities, where journal citation is a weak signal of quality, Research Quality drops to roughly 15 per cent while Teaching and Research Environment each rise above 37 per cent.

THE 学科排名使用同样的五大支柱，但重新分配权重，使其贴合各学科的研究文化。在艺术与人文学科中，期刊引用是质量信号较弱，研究质量权重降至约15%，而教学与研究环境各升至37%以上。

Subject area | 学科领域	Teaching | 教学	Research Environment | 研究环境	Research Quality | 研究质量	International | 国际	Industry | 产业
Arts and Humanities | 艺术与人文	≈ 37.4%	≈ 37.6%	≈ 15.0%	7.5%	2.5%
Social Sciences | 社会科学	≈ 30.4%	≈ 30.4%	≈ 25.0%	7.5%	4.2%
Engineering | 工程	≈ 27.5%	≈ 27.5%	≈ 32.5%	7.5%	5.0%
Clinical and Health | 临床与健康	≈ 26.8%	≈ 26.5%	≈ 35.0%	7.5%	4.2%

These values are approximate and are revised in each published cycle, so always read the methodology note attached to the specific edition you are using.

上述数值为近似值，且每一发布周期都会修订，因此使用时应查阅该版本自带的官方方法说明。

---

7. Shanghai ARWU: Output-Driven Objectivity | 软科ARWU：产出驱动的客观性

The Shanghai Academic Ranking of World Universities takes the opposite approach to QS: it deliberately excludes reputation surveys and relies on countable, verifiable outputs. Alumni and staff winning Nobel Prizes or Fields Medals, highly cited researchers, papers in Nature and Science, and total publications indexed in major citation databases form the backbone.

上海软科世界大学学术排名（ARWU）与 QS 的路线相反：它刻意排除声誉调查，完全依赖可清点、可核验的产出——校友与教师获诺贝尔奖或菲尔兹奖、高被引科学家、在《自然》与《科学》上发表的论文，以及被主要引文数据库收录的论文总数。

Indicator | 指标	Weight | 权重	Dimension | 维度
Alumni awards | 校友获奖	10%	Quality of education | 教育质量
Staff awards | 教师获奖	20%	Quality of faculty | 教师质量
Highly cited researchers | 高被引科学家	20%	Quality of faculty | 教师质量
Papers in Nature and Science | 《自然》《科学》论文	20%	Research output | 科研产出
Total indexed publications | 论文总数	20%	Research output | 科研产出
Per capita performance | 师均表现	10%	Per capita | 人均

This design explains why ARWU tends to reward large, science-heavy research universities and why humanities-focused institutions underperform relative to their teaching reputation.

这一设计解释了为什么 ARWU 往往利好规模大、理工医为主的研究型大学，而人文见长的院校在此榜上的表现常低于其教学声誉。

---

8. ARWU GRAS: Subject-Specific Benchmarks | 软科世界一流学科排名：学科专属基准

For subject-level comparison, ARWU publishes the Global Ranking of Academic Subjects, which uses a completely different indicator family from the main ARWU. The five components are research output measured by Q1 journal papers, research influence measured by category-normalised citation impact, international collaboration, research quality measured by papers in top journals of the field, and major international awards.

在学科层面，软科发布”世界一流学科排名”（GRAS），其指标与主榜完全不同，由五部分构成：以 Q1 期刊论文衡量的科研产出、以学科归一化引用影响力（CNCI）衡量的研究影响、国际合作、以本领域顶级期刊论文衡量的研究质量，以及国际重大奖项。

The weights are not uniform across disciplines.

Published by TutorHao | Exam Prep Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）