📚 QS vs THE vs ARWU vs US News: How the Big Four University Rankings Really Differ | QS、THE、软科、US News 四大排名口径差异全解析
Every autumn, four different organisations publish four different “world university rankings”, and every autumn students and parents ask the same question: which one is right? The honest answer is that all four are right within their own definitions, and all four are wrong if you treat a single number as a measure of educational quality. This guide unpacks the methodology behind QS, THE, ARWU (ShanghaiRanking) and US News, so you can read a ranking table the way a researcher reads a dataset – with attention to what was measured, what was weighted, and what was quietly left out.
每年秋天，四个不同机构会发布四份不同的”世界大学排名”，而每年秋天学生和家长都会问同一个问题：到底哪个排名是对的？诚实的答案是：在各自的定义体系内，四个排名都是对的；但如果你把一个数字当成教育质量的衡量标准，四个排名又都是错的。本文拆解 QS、THE、软科（ARWU）和 US News 的方法论，让你像研究者看数据集一样看排名表——关注它测了什么、权重给了谁、又悄悄遗漏了什么。
1. Why the Big Four Never Agree | 四大排名为何永远不一致
Put the top ten of QS, THE, ARWU and US News side by side and you get four visibly different lists. Harvard sits at or near number one in ARWU and US News Global, yet ranks around fourth or fifth in QS. Imperial College London can beat Cambridge and Oxford in QS while sitting well below them in ARWU. That disagreement is not a scandal; it is arithmetic. Each ranking chooses a different proxy for “excellence”, then normalises and weights those proxies differently.
把 QS、THE、软科和 US News 的前十名并排放，你会看到四份明显不同的名单。哈佛在软科和 US News 全球榜上稳居第一或接近第一，在 QS 上却排在第四、第五左右。帝国理工在 QS 上能超过剑桥和牛津，在软科上却明显落后。这种分歧不是丑闻，而是算术结果：每个排名对”卓越”选用了不同的代理变量，再用不同的方式标准化和加权。
A ranking is a model, not a measurement of truth. Once you accept that, the useful question shifts from “which ranking is correct?” to “which ranking’s definition of quality matches the decision I am about to make?”
排名是一个模型，不是真理的测量。接受这一点之后，有用的问题就从”哪个排名是对的”变成”哪个排名对质量的界定，与我即将做的这个决定相匹配”。
2. A Short History of the Big Four | 四大排名简史与发布机构
QS World University Rankings is produced by Quacquarelli Symonds, a London-based education consultancy. From 2004 to 2009 it was published jointly with Times Higher Education as the “THE-QS World University Rankings”; the partnership broke up, and since 2010 the two have been separate products. QS has always been the most reputation- and employability-driven of the four.
QS 世界大学排名由总部在伦敦的教育咨询公司 Quacquarelli Symonds 发布。2004 到 2009 年间，它与《泰晤士高等教育》联合发布”THE-QS 世界大学排名”；合作结束后，自 2010 年起两者各自成为独立产品。QS 一直是四大排名中最重视声誉与就业表现的。
THE World University Rankings is published by Times Higher Education, a UK magazine. Its methodology is designed around the idea of a balanced research university, with teaching, research environment, research quality, international outlook and industry income treated as separate pillars.
THE 世界大学排名由英国杂志《泰晤士高等教育》发布。它的方法论围绕”均衡的研究型大学”这一理念设计，把教学、研究环境、研究质量、国际化视野和产业收入作为相互独立的支柱。
ARWU, commonly called the Shanghai Ranking, was launched in 2003 by Shanghai Jiao Tong University and is now run by ShanghaiRanking Consultancy. It is the most purely bibliometric of the four: there is no reputation survey and no teaching data at all. US News is different again, because it publishes two entirely separate products – Best Colleges, a domestic undergraduate ranking, and Best Global Universities, a research-output ranking launched in 2014.
ARWU 通常被称为”软科排名”或”上海排名”，2003 年由上海交通大学发起，现由上海软科教育信息咨询公司运营。它是四大中最纯粹的文献计量型排名：完全没有声誉调查，也没有任何教学数据。US News 则更为特殊，因为它发布两个完全不同的产品——Best Colleges（美国国内本科排名）和 Best Global Universities（2014 年推出的科研产出型全球排名）。
3. The Core Logic: Indicators, Weights, Normalisation | 核心逻辑：指标、权重与标准化
All four rankings follow the same three-step recipe. First, choose indicators – measurable quantities such as citation counts, staff-student ratios, or survey scores. Second, normalise them so that numbers with different units can be added together. Third, weight them so that the total sums to one. Two rankings built on identical indicators can still produce different orders if their weights differ.
四个排名遵循同一套三步配方。第一步，选择指标——可测量的量，例如引用次数、师生比或调查得分。第二步，标准化，使单位不同的数字可以相加。第三步，加权，使总权重之和为 1。即使两个排名使用完全相同的指标，只要权重不同，最终顺序就可能不同。
The most common normalisation in ARWU is proportional scaling, where the top-performing institution becomes the reference point:
软科最常用的标准化方式是比例缩放，表现最好的机构成为参照点：
Scoreᵢ = 100 × (Valueᵢ ÷ Max Value in the cohort)
QS and THE lean more towards z-score standardisation, which expresses each value in units of standard deviations from the cohort mean:
QS 和 THE 则更倾向于使用 Z 分数标准化，把每个数值表示为距离全体均值若干个标准差：
z = (x − μ) ÷ σ
The final composite is a weighted sum:
最终的综合得分是加权求和：
Total = Σ (wᵢ × zᵢ), where Σ wᵢ = 1
This matters because z-scores are sensitive to the distribution of the whole cohort. If a small group of universities has an extremely high citation-per-faculty figure, the standard deviation σ becomes large, and everyone else’s z-score is compressed towards zero.
这一点很关键，因为 Z 分数对整个群体的分布非常敏感。如果少数高校的”师均引用”数值极高，标准差 σ 就会变大，其他所有学校的 Z 分数都会被压向零。
4. QS: Reputation, Employability, Internationalisation | QS：声誉、就业与国际化
QS gives the largest single weight in the entire sector to subjective reputation. Academic reputation alone has recently accounted for around 30% of the total, with employer reputation adding roughly another 15%. Together, that is nearly half the score decided by surveys rather than by documents or data.
QS 在整个行业中给主观声誉的权重最大。近年来，仅学术声誉一项就约占总分的 30%，雇主声誉再加约 15%。合计接近一半的分数由问卷调查而非文献或数据决定。
Recent QS editions have used roughly the following nine indicators:
近几版 QS 大致使用以下九项指标：
- Academic Reputation ≈ 30% | 学术声誉 约 30%
- Employer Reputation ≈ 15% | 雇主声誉 约 15%
- Citations per Faculty ≈ 20% | 师均引用 约 20%
- Faculty Student Ratio ≈ 10% | 师生比 约 10%
- International Faculty Ratio ≈ 5% | 国际教师比例 约 5%
- International Students Ratio ≈ 5% | 国际学生比例 约 5%
- International Research Network ≈ 5% | 国际研究网络 约 5%
- Employment Outcomes ≈ 5% | 就业成果 约 5%
- Sustainability ≈ 5% | 可持续发展 约 5%
Two consequences follow directly. First, institutions with strong global brands and large international student bodies are rewarded, which explains the consistently high placement of UK and Australian universities. Second, the citations-per-faculty indicator is a ratio, so a small, citation-dense institution can outrank a large comprehensive university on that component.
由此直接产生两个后果。第一，全球品牌强、国际学生规模大的院校会获得奖励，这解释了英国和澳大利亚高校在 QS 上长期靠前。第二，”师均引用”是一个比值指标，因此一所规模小但引用密度高的院校，在这一项上可以超过规模庞大的综合性大学。
5. THE: A Balanced Research Scorecard | THE：研究型大学平衡计分卡
THE deliberately spreads its weight across five pillars instead of concentrating on reputation. In recent editions the structure has been approximately 29.5% teaching, 29% research environment, 30% research quality, 7.5% international outlook and 4% industry.
THE 刻意把权重分散到五个支柱，而不是集中在声誉上。近几版的结构大致是：教学 29.5%、研究环境 29%、研究质量 30%、国际化视野 7.5%、产业收入 4%。
The teaching pillar is not based on student satisfaction surveys; it uses proxies such as the staff-to-student ratio, the ratio of doctoral to bachelor’s degrees awarded, the number of doctorates awarded per academic staff member, and institutional income per academic staff member.
教学支柱并不基于学生满意度调查，而是使用代理变量，例如师生比、授予博士学位与学士学位的比例、师均授予博士数，以及师均机构收入。
The research quality pillar, introduced as a distinct pillar in the 2024 edition, is built from field-normalised citation impact, research strength, research excellence and research influence. Field normalisation means a paper in mathematics is compared with other mathematics papers rather than with medicine, which corrects a well-known distortion in raw citation counts.
研究质量支柱在 2024 版中被独立出来，由领域归一化引用影响力、研究实力、研究卓越度和研究影响力构成。领域归一化意味着数学论文与数学论文比较，而不是与医学论文比较，这修正了原始引用次数中一个众所周知的偏差。
A practical caveat: THE requires institutions to submit data for several indicators. Universities that decline to submit may be excluded from the main table or listed without a rank, which means the absence of a university from THE is not evidence of low quality.
一个实用提醒：THE 要求院校为若干指标提交数据。拒绝提交的大学可能被排除在主榜之外，或只列出而不给排名。因此，一所大学没有出现在 THE 榜上，并不等于它质量低。
6. ARWU: Pure Research Output, No Surveys | 软科：纯科研产出，没有声誉调查
ARWU makes no attempt to measure teaching, student experience or employer opinion. It counts six things, all of them research or prestige related:
软科完全不去测量教学、学生体验或雇主评价。它只统计六项内容，且都与科研或学术声望相关：
- Alumni winning Nobel Prizes or Fields Medals ≈ 10% | 校友获诺贝尔奖或菲尔兹奖 约 10%
- Staff winning Nobel Prizes or Fields Medals ≈ 20% | 教师获诺贝尔奖或菲尔兹奖 约 20%
- Highly Cited Researchers ≈ 20% | 高被引科学家 约 20%
- Papers in Nature and Science ≈ 20% | 《自然》与《科学》论文 约 20%
- Papers indexed in SCIE and SSCI ≈ 20% | SCIE 与 SSCI 收录论文 约 20%
- Per capita academic performance ≈ 10% | 师均学术表现 约 10%
Because awards and citations accumulate over decades, ARWU changes extremely slowly. Harvard has topped it every year since the ranking began. That stability is a strength for long-term reputation research and a weakness for anyone trying to judge a young, fast-improving university.
由于奖项和引用是数十年累积的结果，软科变化极其缓慢。自该排名创立以来，哈佛每年都位居第一。这种稳定性对长期声誉研究是优点，但对想判断一所年轻、快速进步大学的人来说则是缺点。
ARWU also has a structural bias towards science and medicine. Institutions famous for law, economics, business or the arts – the London School of Economics being the standard example – perform far worse here than in QS or THE, because Nobel Prizes and Nature papers are not the currency of those fields.
软科还存在对理工医的结构性偏向。以法律、经济、商科或艺术闻名的院校——伦敦政治经济学院是最典型的例子——在这里的表现远差于在 QS 或 THE 上的表现，因为诺贝尔奖和《自然》论文并非这些领域的通用货币。
7. US News
Published by TutorHao | Exam Prep Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导