📚 How a 20% Citations-per-Faculty Weight Reveals a University's Research Strength | 师均引用率 20%：如何反映一所大学的科研强度
When a ranking table gives "Citations per Faculty" a 20% weighting, it is making a bold claim: that the citation impact generated by each academic staff member is one fifth of what makes a university good. Understanding how that number is built, what it rewards, and where it distorts reality is essential for anyone reading rankings seriously - whether you are a student choosing a department, a parent comparing offers, or a teacher explaining research culture to A-level applicants.
当一份排名表把"师均引用率"（Citations per Faculty）赋予 20% 的权重时，它其实在做一个大胆的断言：每位学术人员所产生的引用影响力，占一所大学"好坏"的五分之一。理解这个数字是如何计算出来的、它奖励什么、又在哪里扭曲了现实，对任何认真解读排名的人都至关重要——无论你是正在挑专业的学生、比较 offer 的家长，还是向 A-level 申请者解释科研文化的老师。
1. What the Metric Actually Counts | 这个指标究竟在统计什么
Citations per Faculty, often shortened to CpF, is exactly what its name suggests: the total number of citations earned by a university's research output over a defined period, divided by the number of academic staff. In the QS World University Rankings that period is five years, and the citation records are drawn from Elsevier's Scopus database.
师均引用率（常缩写为 CpF）正如其名：一所大学的科研产出在某一固定时期内获得的总引用次数，除以该校学术人员的数量。在 QS 世界大学排名中，这个时期是五年，引用记录来自爱思唯尔的 Scopus 数据库。
The numerator counts citations, not papers. A university that publishes fewer but more influential articles can outscore one that publishes far more. The metric rewards impact per paper and per researcher, not output volume by itself.
分子统计的是引用次数，不是论文篇数。一所发文较少但论文影响力更高的大学，可能胜过发文多得多的大学。这个指标奖励的是"每篇论文、每位研究者的影响力"，而不是单纯的产出规模。
The denominator is full-time-equivalent (FTE) academic staff, not students. This is crucial: the metric asks how much citation impact each researcher generates, so a large student body neither helps nor hurts the score.
分母是全时当量（FTE）学术人员，而不是学生。这一点至关重要：该指标问的是"每位研究者产生了多少引用影响力"，因此学生规模庞大既不加分也不减分。
CpF = Total citations received over 5 years ÷ FTE academic staff
2. Why the Weight Is 20% | 为什么权重定在 20%
Citations are quantitative, reproducible and independent of opinion surveys. Reputation indicators, by contrast, depend on who happens to be asked. Giving CpF 20% is a compromise: enough weight to reward genuine research intensity, but not so much that a single database quirk could reorder the entire table.
引用数据是可量化、可复现、不依赖问卷样本的。相比之下，声誉类指标取决于"问到了谁"。把 CpF 定在 20% 是一种折中：权重足够大到能奖励真正的科研强度，又不至于让某个数据库的统计偏差颠覆整张排名表。
To see how that 20% sits inside the whole picture, look at the full QS weighting scheme:
要看这 20% 处在什么位置，可以看 QS 的完整权重结构：
| Indicator | 指标 | Weight | 权重 |
|---|---|
| Academic Reputation | 学术声誉 | 30% |
| Citations per Faculty | 师均引用率 | 20% |
| Employer Reputation | 雇主声誉 | 15% |
| Faculty Student Ratio | 师生比 | 10% |
| International Faculty Ratio | 国际教师比例 | 5% |
| International Students Ratio | 国际学生比例 | 5% |
| International Research Network | 国际研究网络 | 5% |
| Employment Outcomes | 就业成果 | 5% |
| Sustainability | 可持续发展 | 5% |
Notice what this means in practice: CpF plus Academic Reputation together carry half the total weight. A university cannot buy its way up the table with citations alone, but it also cannot hide weak research output behind a famous name.
要注意这在实践中意味着什么：CpF 加上学术声誉合计占了一半权重。一所大学无法仅靠引用量冲榜，但同样也无法用一个响亮的名号掩盖薄弱的科研产出。
3. The Formula and the Logarithmic Transformation | 公式与对数变换
The raw ratio is easy to compute, but raw ratios are badly skewed. A handful of institutions post CpF values many times higher than the median, and if scores were scaled linearly the whole ranking would collapse into a short tail of giants. Ranking organisations therefore apply a logarithmic transformation before rescaling to 0-100.
原始比值很容易计算，但原始比值分布极度偏斜。少数院校的 CpF 值是中位数的许多倍，如果按线性方式换算分数，整张排名会塌缩成少数"巨无霸"的长尾。因此排名机构会先做对数变换，再重新标定到 0-100 分。
Score = 100 × [ ln(CpF) − ln(CpF_min) ] ÷ [ ln(CpF_max) − ln(CpF_min) ]
The practical consequence is important. Because the scale is logarithmic, moving from 200 to 400 citations per faculty does not double your score - it adds a fixed increment. Conversely, institutions at the bottom of the range can climb quickly with modest improvements.
其实践后果很重要。由于采用对数刻度，师均引用从 200 提升到 400，并不会让分数翻倍，而只是加上一个固定的增量。反过来说，处于分数区间底部的院校只要小幅改善，就能较快爬升。
It also means that small differences in published scores should not be over-interpreted. Two universities 3 points apart may be statistically indistinguishable in research terms.
这也意味着，公布的分数之间的小幅差距不应被过度解读。相差 3 分的两所大学，在科研层面可能根本没有统计意义。
4. The Five-Year Window and the Lag Problem | 五年窗口与滞后问题
Citations accumulate slowly. A paper published last month may have one or two citations; the same paper in five years may have fifty. Rankings therefore count citations to work published inside a rolling five-year window, which means the metric always describes the recent past, not the present.
引用是缓慢累积的。上个月发表的论文可能只有一两次引用，五年后同样这篇论文可能有五十次。因此排名统计的是"滚动的五年窗口内所发表论文"的引用，这意味着该指标描述的永远是近期过去，而不是当下。
This lag has three practical consequences worth remembering:
这种滞后有三个值得记住的实际后果：
A university investing heavily in new research staff will not see a payoff for several years. | 一所大举引进科研人员的大学，要过几年才能看到回报。
A department that has just lost its star professors will keep a high score for a while on the strength of past work. | 一个刚刚失去明星教授的院系，会靠着过去的成果在一段时间内维持高分。
Fields with fast citation cultures (molecular biology, computer science) look "current" much sooner than fields with slow ones (history, classics, law). | 引用节奏快的学科（分子生物学、计算机科学）比引用节奏慢的学科（历史、古典学、法学）更快显得"当下"。
5. Field Normalisation: Why Raw Counts Mislead | 学科归一化：为什么原始计数会误导
Different disciplines simply cite at different rates. A typical medical paper may collect dozens of citations, while a landmark monograph in philosophy may collect a handful. Comparing them directly is like comparing sprint times with marathon times.
不同学科的引用习惯本就不同。一篇普通医学论文可能获得几十次引用，而一部哲学里程碑式的专著可能只有几次。直接比较二者，就像拿百米成绩和马拉松成绩相比。
To handle this, bibliometricians use field-normalised indicators. The most common are Field-Weighted Citation Impact (FWCI) and Category Normalised Citation Impact (CNCI). A value of 1.00 means "cited exactly as often as the world average for this field, this year, this document type".
为解决这个问题，文献计量学界使用学科归一化指标，最常见的是"学科加权引用影响力"（FWCI）和"学科归一化引用影响力"（CNCI）。数值为 1.00 意味着"与该领域、该年份、该文献类型的全球平均水平完全相同"。
FWCI = 1.00 → exactly world average | 恰好等于世界平均
This matters because a university strong in humanities may post a modest CpF yet have excellent FWCI within its own disciplines. Rankings that mix all fields into one CpF figure can therefore penalise depth in book-based subjects.
这一点很关键：一所以人文学科见长的大学，CpF 可能不高，但在本学科内的 FWCI 可能非常优秀。把所有学科混在一起算一个 CpF 的排名，因此可能惩罚那些以专著为主的学科强校。
6. The Citation Multiplier: Medicine and Life Sciences | 引用乘数：医学与生命科学
Institution-level citation totals are dominated by a small number of fields. Medical, biomedical and life-science research routinely accounts for a disproportionate share of all citations a large university receives, even when those staff are a minority of the faculty.
在院校层面，引用总量往往被少数几个学科主导。医学、生物医学与生命科学研究通常占一所大型大学所获全部引用的极大比例，哪怕这些教师只占全体人员的一小部分。
Indicative patterns - not exact figures, but the general shape of the data - look roughly like this:
以下是示意性的模式（并非精确数值，但反映了数据的大致形态）：
| Field | 学科 | Relative citation intensity | 相对引用强度 |
|---|---|
| Clinical Medicine | 临床医学 | Very high | 极高 |
| Biomedical and Life Sciences | 生物医学与生命科学 | Very high | 极高 |
| Physical Sciences and Engineering | 物理科学与工程 | High | 高 |
| Social Sciences | 社会科学 | Moderate | 中等 |
| Arts and Humanities | 艺术与人文学科 | Low | 低 |
The practical lesson for applicants is straightforward: a strong CpF may say more about a university's medical school than about the department you actually want to join.
对申请者而言，实际的启示很直接：一个漂亮的 CpF，可能更多反映的是这所大学的医学院，而不是你真正想去的那个院系。
7. The Denominator Effect: Big Versus Small | 分母效应：大校与小校之别
Because CpF is a ratio, it can be gamed by shrinking the denominator. A small, highly specialised institution with 500 research-active staff can post spectacular figures without generating anything like the total research output of a comprehensive university with 6,000 staff.
由于 CpF 是一个比值，它可以通过缩小分母来"优化"。一所只有 500 名科研人员的小型专业化院校，可以交出了不起的数字，但其科研总产出远不及一所有 6,000 名教职员工的综合性大学。
There is a second, subtler issue. Staff counts include teaching-only and clinical academics who publish little. How those are classified varies between institutions, which introduces a definitional wobble that no amount of mathematical elegance can remove.
还有第二个更微妙的问题：教师统计中包含以教学为主、以及几乎不发表论文的临床教师。这些人的归类方式各校不一，由此产生的定义性波动，是任何数学上的精巧都无法消除的。
A useful sanity check is to read CpF alongside absolute research output. A university with high CpF and high total output is genuinely research-intensive; high CpF with low total output suggests a small, focused institution.
一个有用的核查方法是把 CpF 与绝对科研产出并列来看：CpF 高且总产出也高的大学，是真正的科研密集型；CpF 高但总产出低，则往往意味着这是一所规模小、方向集中的院校。
8. Self-Citation, Citation Clubs and Gaming Risk | 自引、互引圈子与博弈风险
Every metric attracts behaviour. The best-documented distortions in citation-based measures are self-citation, reciprocal citation networks among small groups of institutions, and salami-slicing - splitting one study into several minimal publishable units.
任何指标都会引来博弈行为。引用类指标中被记录得最清楚的扭曲包括：自引、少数院校之间的小圈子互引，以及"切片发表"——把一项研究拆成若干最低可发表单元。
Ranking providers push back in several ways:
排名机构主要从几个方面进行反制：
Excluding self-citations beyond a threshold. | 超过一定阈值的自引予以剔除。
Using five-year rather than one-year windows, which dilutes short-term manipulation. | 采用五年窗口而非一年窗口，稀释短期操纵的效果。
Field and document-type normalisation, which removes the easiest inflation routes. | 进行学科与文献类型归一化，堵住最容易被注水的路径。
Auditing sudden, unexplained jumps in an institution's citation profile. | 审查某院校引用画像中突然出现且无法解释的跃升。
None of these is perfect. The reasonable conclusion is not that citations are worthless, but that they should never be read alone.
这些措施都不完美。合理的结论并不是"引用数据毫无价值"，而是"绝不应孤立地看引用数据"。
9. How CpF Relates to Other Research Indicators | CpF 与其他科研指标的关系
No single number captures research strength. It is far more informative to read several indicators together and ask whether they tell a consistent story.
没有任何单一数字能概括科研实力。更有信息量的做法是同时阅读多个指标，并追问它们是否讲出了同一个故事。
| Indicator | 指标 | What it captures | 反映什么 | Main weakness | 主要弱点 |
|---|---|---|
| Citations per Faculty | 师均引用率 | Impact per researcher | 每位研究者的影响力 | Field and denominator bias | 学科与分母偏差 |
| FWCI / CNCI | 学科加权引用影响力 | Citation impact within field | 本学科内的引用影响力 | Still favours journal articles | 仍偏向期刊论文 |
| Research income | 科研经费 | Competitive funding strength | 竞争性经费实力 | Cost-heavy fields look better | 烧钱学科占优 |
| Peer review exercises | 同行评审评估 | Expert judgement of quality | 专家对质量的判断 | Expensive, periodic, subjective | 昂贵、周期性、主观 |
| Doctoral awards | 博士学位授予量 | Research training capacity | 科研人才培养能力 | Says nothing about quality | 与质量无关 |
When CpF, FWCI and peer review all point the same way, you can be reasonably confident the research base is strong. When they diverge, dig into the departmental level before drawing conclusions.
当 CpF、FWCI 与同行评审三者指向一致时，你可以比较有信心地认为其科研基础扎实。当它们互相矛盾时，在得出结论之前，请深入到院系层面去看。
10. What 20% Can and Cannot Tell You | 20% 能说明与不能说明什么
A 20% weight means the metric is influential but not decisive. Roughly four fifths of a university's position in such a table comes from something else - reputation, staffing ratios, international mix, employability and sustainability.
20% 的权重意味着这个指标有影响力，但不具决定性。一所大学在这类榜单中的位置，约五分之四来自其他因素——声誉、师资比例、国际化程度、就业力以及可持续发展。
What it can tell you:
它能告诉你：
Whether the institution is genuinely research-active rather than mainly a teaching operation. | 该校是真正的科研活跃型，还是以教学为主。
Whether its researchers are producing work that other researchers actually use. | 其研究者的成果是否真的被其他研究者所使用。
Whether research intensity is broadly spread or concentrated in one or two faculties. | 科研强度是普遍分布，还是集中在少数一两个学院。
What it cannot tell you:
它不能告诉你：
Whether the teaching you will receive is any good. Citations measure research reception, not lecture quality. | 你将接受的教学质量如何。引用衡量的是研究被接受的程度，不是讲课水平。
Whether your specific subject is strong there. Institution-level figures hide departments with no research culture at all. | 你想读的那个专业在该校是否强。校级数字会掩盖毫无科研文化的院系。
- <
Published by TutorHao | Exam Prep Revision Series | aleveler.com
广告 / Sponsored9618考试剑桥A level计算机科学教材Cambridge International AS and A Leve…¥491到手价 · 以淘宝页面为准计算机 备考教辅 · 正版现货 · 学习备考通用查看详情 / View on Taobao广告 / Sponsored2027中央戏剧学院612戏剧基础801戏剧影视表演理论与初试考研一本通真题笔记知识重点总结技巧题库学复习资料教材模拟…¥6.8到手价 · 以淘宝页面为准戏剧 备考教辅 · 正版现货 · 学习备考通用查看详情 / View on Taobao更多咨询请联系16621398022（同微信）
备考资讯每日更新，关注不错过考纲变化 · 科目分析 · 学习方法
在微信里长按二维码识别关注
或截图后用「扫一扫 → 相册」选择本图公众号：tutorhao
