📚 Three Common Misuses of Rankings as a Decision-Making Basis | 把排名当决策依据的三种常见误用
Rankings are seductive because they look like evidence. A league table arrives as a clean numbered list, published on a fixed date, produced by an organisation that has clearly done a great deal of counting. The number feels objective, and objectivity is exactly what a seventeen-year-old choosing a university – or a parent paying for one – wants most.
排名之所以有吸引力，是因为它看起来像证据。排行榜以一份干净的数字清单出现，在固定日期发布，由一个显然做过大量统计的机构制作。这个数字显得客观，而客观恰恰是十七岁选校的学生，或是为其付费的家长最想要的东西。
But a ranking is not evidence about your decision. It is evidence about somebody else’s decision – specifically, about the weighting scheme that the publisher chose. Treating it as a verdict on quality is the origin of almost every expensive mistake in university choice. This article sets out the three most common misuses, why each one is logically flawed, and what to do instead.
但排名并不是关于你的决定的证据。它是关于别人决定的证据——具体说，是关于发布机构所选择的权重方案的证据。把它当作对办学质量的裁决，几乎是大学选择中所有昂贵错误的源头。本文梳理最常见的三种误用、它们各自的逻辑缺陷，以及应当如何替代。
1. Why Rankings Feel Like Evidence | 为什么排名看起来像证据
A ranking is a compression algorithm. It takes hundreds of noisy, partly incompatible variables – citation counts, staff-student ratios, survey responses, graduate salaries, international student proportions – and squeezes them into a single ordinal position. Compression is genuinely useful: it lets you compare 1,200 institutions in ten seconds.
排名是一种压缩算法。它把数百个充满噪声、部分互不兼容的变量——引用量、师生比、问卷回答、毕业生薪资、国际学生比例——压缩成一个序数位次。压缩确实有用：它让你在十秒内比较 1200 所院校。
The problem is that compression is lossy by design. To produce one number you must discard information, and the discarded information is usually precisely the part that matters to a specific applicant: class size in your subject, assessment format, cost net of scholarships, the strength of the alumni network in the city where you intend to work.
问题在于，压缩在本质上是有损的。要得出一个数字，就必须丢弃信息，而被丢弃的往往正是对某个具体申请者最重要的部分：你所在专业的班级规模、考核形式、扣除奖学金后的实际成本、以及你打算工作的那座城市里的校友网络强度。
There is also a measurement problem. A ranking claims to measure “quality”, but quality of what? Teaching quality, research quality and student experience are three distinct constructs. A single position number cannot be a valid measure of all three simultaneously, because the three do not move together. A research-intensive department can be a poor teaching environment for a student who needs structure and frequent feedback.
还有测量学上的问题。排名声称衡量”质量”，但衡量的是什么质量？教学质量、科研质量和就读体验是三种不同的构念。单一的名次无法同时有效测量这三者，因为它们并不同向变动。一个科研密集型院系，对需要明确结构和频繁反馈的学生来说，可能恰恰是糟糕的教学环境。
Composite Score = w₁S₁ + w₂S₂ + … + wₙSₙ , where Σwᵢ = 1
Everything that follows depends on that formula. The scores Sᵢ are contested measurements, and the weights wᵢ are value judgements wearing the costume of technical neutrality.
接下来的一切都取决于这个公式。各项得分 Sᵢ 是有争议的测量值，而权重 wᵢ 是披着技术中立外衣的价值判断。
2. The Three Misuses in One Table | 三种误用一览表
Before analysing each misuse in detail, it helps to see them side by side. Each one has a distinctive signature, and each one produces a different class of bad decision.
在逐一分析之前，先把三种误用并列对照会很有帮助。每一种都有独特的表现形式，也各自导致不同类型的错误决策。
| Misuse 误用 | What it looks like 表现 | What it hides 掩盖了什么 | Diagnostic question 诊断问题 |
|---|---|---|---|
| 1. Treating one number as quality 把单一数字当作质量 | “It is ranked 12th, so it is better.” | The publisher’s weights and the loss of multi-dimensional information 发布机构的权重与多维信息的丢失 | Would the order change if the weights changed? 权重一变，次序会变吗？ |
| 2. Wrong unit of analysis 分析单位错误 | “The university is top 50, therefore my course is top 50.” | Within-institution variation between departments 校内各院系之间的巨大差异 | What is the rank of the department, not the institution? 排的是院系还是学校？ |
| 3. Confusing rank with fit and outcome 把排名与匹配度、结果混同 | “Higher rank means better salary and a better fit for me.” | Selection effects, confounding by intake, and personal fit 选择效应、生源混杂因素与个人匹配度 | Is this measuring the institution or the students it admits? 这衡量的是学校还是它招到的学生？ |
The three misuses compound. A student who commits all three ends up optimising a number that was never a measure of what they wanted, computed at the wrong level of aggregation, for an outcome that was never caused by the institution in the first place.
这三种误用会叠加。同时犯下三种误用的学生，最终优化的是一个从未衡量过其所求之物的数字，在错误的聚合层级上计算，指向一个本来就不由院校导致的结果。
3. Misuse One: One Number for Many Things | 误用一：用一个数字代表许多东西
The first misuse is reification: treating the composite index as if it were a property of the institution, rather than an artefact of a formula. This is the error of mistaking the map for the territory.
第一种误用是”具体化”：把综合指数当成院校本身的属性，而不是一个公式的产物。这是把地图误当领土的错误。
Consider two universities with fixed, uncontested underlying scores. University A scores 9/10 on research and 4/10 on teaching. University B scores 5/10 on research and 8/10 on teaching. Nothing about the two institutions changes. Only the weights change.
设想两所院校，其底层得分固定且无争议。A 校科研 9 分（满分 10）、教学 4 分；B 校科研 5 分、教学 8 分。两所院校本身没有任何变化，变的只是权重。
If w_R = 0.7 and w_T = 0.3 : A = 7.5 , B = 5.9 → A ranks higher
If w_R = 0.3 and w_T = 0.7 : A = 5.5 , B = 7.1 → B ranks higher
The rank order flips with no change in any underlying fact. This is not a hypothetical curiosity; it is exactly what happens when different publishers weight research output differently, which is why the same university can sit at 34th in one table and 91st in another in the same year.
名次顺序在任何底层事实都未改变的情况下发生了翻转。这不是假设中的趣闻，而是不同发布方对科研产出赋予不同权重时的真实结果，也正是同一所大学在同一年可以在一张榜上排 34 名、在另一张榜上排 91 名的原因。
A single number also destroys information that is genuinely decision-relevant:
单一数字还会摧毁真正与决策相关的信息：
- Distribution, not just the mean: an average class size of 20 can mean “always 20” or “10 in tutorials and 60 in lectures”. 分布而非均值：平均班额 20 人，可能是”始终 20 人”，也可能是”小课 10 人、大课 60 人”。
- Variance across departments, which is usually far larger than variance across universities. 院系之间的方差，通常远大于院校之间的方差。
- Cost and funding: a rank says nothing about net price after scholarships. 成本与资助：排名对扣除奖学金后的净支出只字不提。
- Direction of travel: a rising department and a falling one can share a position this year. 变化趋势：一个上升中的院系和一个下滑中的院系，今年可能名次相同。
4. The Aggregation Problem: Weights You Never Chose | 聚合问题：你从未选择的权重
The weights inside a composite index are a value judgement disguised as a technical choice. Somebody decided that citations matter twice as much as student satisfaction. That somebody is not you, and their preferences are not your preferences.
综合指数内部的权重，是披着技术选择外衣的价值判断。有人决定了引用量的重要性是学生满意度的两倍。那个人不是你，而他的偏好也不是你的偏好。
The practical remedy is sensitivity analysis, and it takes about twenty minutes. First, find the published methodology and write down the indicator weights. Second, re-rank your shortlist using your own weights. Third, check whether the top choice survives.
实用的补救办法是做敏感性分析，大约只需二十分钟。第一，找出公布的方法论，记下各指标的权重；第二，用你自己的权重给候选名单重新排序；第三，检查首选是否依然胜出。
A useful shortcut is the dominance test. If option A beats option B on every criterion you care about, and the weights cannot change that, the decision is robust. If the winner changes when you adjust the weights, then the ranking was never doing the work – your preferences were, and you have simply made them explicit.
一个有用的捷径是”优势检验”。如果选项 A 在你关心的每一项标准上都优于 B，且权重无法改变这一结论，那么决策是稳健的。如果调整权重后胜者改变，那么起作用的从来不是排名，而是你的偏好——你只是把它明确表达出来了。
There is a second aggregation problem: the level at which the average is taken. An institutional average blends medicine, engineering, law and history into one figure. Unless your own subject mix matches the publisher’s student mix, the average is being computed over a population you are not part of.
还有第二个聚合问题：取平均的层级。院校层面的平均值把医学、工程、法学和历史混成一个数字。除非你的专业结构与发布方的学生结构相同，否则这个平均值是在一个你并不属于的总体上计算出来的。
5. Misuse Two: The Wrong Unit of Analysis | 误用二：分析单位选错了
The second misuse is a straightforward level error, but it is remarkably persistent. Institutional rank is not departmental rank, and departmental rank is not the quality of the modules you will actually take.
第二种误用是直接的层级错误，却异常顽固。院校排名不等于院系排名，院系排名也不等于你将实际修读的那些课程的质量。
A university sitting 80th overall may host a department that is genuinely top 10 in its field, often because the institution’s overall score is dragged down by faculties it is weak in, or because the department is small and specialised and the publisher’s indicators do not capture it well.
一所综合排名 80 位的大学，可能拥有一个在其领域确实位居前十的院系，原因往往是学校被其薄弱的学部拖累，或该院系规模小、方向专精，发布方的指标无法很好地捕捉它。
The reverse also holds. A university in the global top 20 may have a department in your subject that is mediocre by international standards, thinly staffed, or in the middle of a leadership transition.
反向情况同样存在。一所全球前 20 的大学，其你所学专业的院系可能以国际
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