QS排名 指标权重 年度审议 未加权指标
QS 世界大学排名每年更新一次，榜单里的名次由一组带权重的指标合成。权重怎么定、谁来决定、多久审一次，QS 官网在方法论页面里写得直接：权重每年审议。这句话背后是一套固定流程，从顾问委员会的评估到新指标的试跑期，都有公开说明。很多学生以为排名是固定公式换一批数据，实际上公式本身也在动，只是动得很克制。看懂这一层，你才分得清名次变化里哪些来自学校，哪些来自方法。
Every June QS publishes a fresh edition of its World University Rankings, and every edition scores universities against a set of weighted indicators. QS states plainly on its official methodology page that those weightings are reviewed on an annual basis, and its FAQ describes how a proposed change is assessed before anything is altered. Many applicants treat the formula as fixed and assume only the data moves, when in practice the formula moves too, slowly and deliberately. Reading that process correctly is what separates a rank that reflects a university from a rank that reflects a revision.
QS 官网写明的年度审议机制
在 QS 官网的世界大学排名方法论页面底部，有一句容易被忽略的说明：各项权重每年审议一次。同一句话出现在每一项指标的单独页面上，措辞完全一致。官网 FAQ 补充了流程，QS 每年审议方法论，并设有一套正式的调整与新增程序。评估意见来自独立的全球排名顾问委员会，QS 会先看证据，再看改动可能带来的影响。审议的结果不一定是改权重，2027 版与 2028 版的权重表就写明继续沿用现有设置。这条机制真正的意义在于，权重表是当前版本，不是恒定常数。
At the foot of the QS methodology page sits a line most readers skip: the weightings are reviewed on an annual basis, and the same sentence, worded identically, appears on each individual indicator page, which makes it a deliberate policy rather than a footnote. The QS FAQ describes a formal process for adjustments and additions, with the independent Global Rankings Advisory Board asked to weigh the evidence and the likely impact first. Reviews do not have to end in change, and the 2027 and 2028 editions are already confirmed to keep the present weightings. Treat the published table as the current version, not a permanent constant.
2027 版的五组权重与十项指标
2027 版把十项指标归进五组权重。研究与发现占 50%，其中学术声誉 30%、师均引用 20%。就业与成果占 20%，其中雇主声誉 15%、就业成果 5%。学习体验占 10%，全部来自师生比。全球参与占 15%，拆成国际教师 5%、国际研究网络 5%、国际学生 5%。可持续发展单列 5%。十项指标里九项参与计分，国际学生多样性一项的权重写着 0%，只公布分数、不进总分。这组数字来自 QS 官网的权重表，可以逐项对照。
The 2027 edition groups ten indicators into five weighted lenses. Research and Discovery carries half the total, split between Academic Reputation at 30 per cent and Citations per Faculty at 20 per cent. Employability and Outcomes accounts for a fifth through Employer Reputation and Employment Outcomes, while Learning Experience rests on a single Faculty Student Ratio worth 10 per cent. Global Engagement adds 15 per cent across three scored indicators, Sustainability stands alone at 5 per cent, and International Student Diversity is published with a weighting of zero. Nine of the ten shape your position; one is displayed for information only.
未加权指标先试跑再计分
新指标不会一上线就计分。QS 官网 FAQ 写明，一项新指标经过论证后，通常先以未加权形式进入榜单，用来观察它对结果的实际影响。国际学生多样性正处在这个状态：2026 版在 2025 年 6 月 19 日发布时作为未加权指标引入，2027 版继续不计分，官网还写明 2028 版仍然使用国际学生比例。在 QS 欧洲区域排名里，这项指标已经计入 5%，说明它还在等待全球口径的验证。这里的节奏值得记住，先公开、后计分，中间隔着至少两三个版次。
New indicators do not enter the score immediately: the QS FAQ explains that a proposal accepted by the process is normally introduced first as an unweighted indicator, so its effect can be studied before it carries any weight. International Student Diversity is exactly that case, published unweighted from the 2026 edition released on 19 June 2025, still unscored in 2027, with the FAQ stating that the 2028 edition will again use International Student Ratio. The same indicator already counts for 5 per cent in the regional Europe table, which tells you the global version is still under test, so publication comes first and scoring comes later.
2024 版那次改权重与三项新增
权重最近一次大改发生在 2024 版，那一版在 2023 年 6 月发布。QS 官网当版公告写明新增三项指标：可持续发展、就业成果、国际研究网络。腾出权重的做法是把学术声誉从 40% 降到 30%，师生比从 20% 降到 10%，并把雇主声誉从 10% 提到 15%。三项新指标合计 15%，恰好来自学术声誉与师生比让出的那部分空间。这次调整之后，2025 版与 2026 版都沿用同一套权重，没有再动。你今天在榜单上看到的分项分，用的就是这套结构。
The last substantial reweighting arrived with the 2024 edition, published in June 2023, when QS announced three new indicators: Sustainability, Employment Outcomes and International Research Network. Room was made by cutting Academic Reputation from 40 to 30 per cent and Faculty Student Ratio from 20 to 10 per cent, while Employer Reputation rose from 10 to 15 per cent. The three newcomers together hold 15 per cent, most of it drawn from the two indicators that shrank, and the 2025 and 2026 editions then kept that structure untouched. Nothing in a current indicator score predates it.
声誉调查为什么计入五年答复
声誉类指标靠问卷，但问卷不是只看当年。QS 官网写明，学术声誉与雇主声誉的调查每年进行，计入的是近五年的答复，其中第四年与第五年的答复权重更低。学术声誉虽然从 40% 降到 30%，仍是单项权重最高的指标。2027 版的数据口径里，这两项声誉分来自 121,024 位学者与 69,432 位雇主的答复。问卷规则里有一条防作弊设计，学者不能给自己所在的院校投票。五年窗口解释了为什么声誉分很少剧烈波动。
Reputation indicators run on surveys, but not on this year's responses alone: QS states that its academic and employer surveys are conducted annually while the score draws on five years of replies, with lower weight given to the fourth and fifth years. Academic Reputation remains the heaviest single indicator even after falling from 40 to 30 per cent, and the 2027 edition drew on replies from 121,024 academics and 69,432 employers. Respondents are barred from nominating their own institution, a rule aimed at removing the most obvious bias, which is why reputation scores rarely swing hard within a single year.
引用数据来自 Scopus 与六年窗口
研究类指标用的是 Elsevier 的 Scopus 数据库，不是院校自己填报的数字。师均引用把一所院校的引用总量除以教师人数，论文看五年、引用看六年。自引会被剔除，五个学科领域先等权再折算成人均。从 2026 版起，纳入研究的机构数扩大约 65%，覆盖 8,000 所以上，这让师均引用与国际研究网络两项出现跨年位移。国际研究网络只统计持续合作，五年窗口里合著三篇以上才算数。数据源固定，院校无法直接修改这一部分。
The research indicators draw on Elsevier's Scopus rather than figures a university supplies about itself, with Citations per Faculty dividing an institution's citation volume by its faculty headcount. That measure uses a five-year paper window and a six-year citation window, removes self-citations and weights the five faculty areas equally before the per-capita step. From the 2026 edition QS widened its research mapping by roughly 65 per cent to more than 8,000 institutions, which shifted both Citations per Faculty and International Research Network, while the network indicator itself counts only sustained partnerships of three or more joint papers inside five years.
归一化调整为何让总分变化
2026 版还改了一件更隐蔽的事，归一化方法。QS 在当版说明里写明，这次细化让分数分布发生变化，除国际研究网络外所有指标都受影响。结果是同一所院校的名次可能只挪一两位，总分却与上一年明显不同。官网 FAQ 也提到，QS 刻意保持方法论稳定，改动通常经过长期咨询，目的是让名次变化来自院校表现而不是公式。这两句话并不矛盾，方法论求稳，归一化属于技术层面，两年版次之间仍会留下痕迹。看总分之前，先看当版说明。
One quieter change arrived with the 2026 edition: the normalisation method, which QS says shifted the distribution of scores and affected every indicator except International Research Network. An institution can therefore hold roughly the same position while its overall score moves noticeably, and the FAQ argues that QS protects stability deliberately, changing indicators only after long consultation, so that a rank shift reflects performance rather than a formula rewrite. Both statements can be true at once. Methodology is guarded, yet the arithmetic behind it still gets tuned, and the edition notes are where you find the trace.
评估规模与上榜资格条件
不是所有大学都能进榜。QS 官网的资格条件包括同时开设本科与研究生学位课程，覆盖至少两个学科领域，学术声誉进入全球前 20%，五年内至少有 100 篇被 Scopus 索引的论文。学生人数少于 5,000 的院校另有三项门槛，需要在其中一项进入前 1,000，或两项进入前 1,200，或全球前 1,300。这份榜单在 2026 年 6 月 18 日发布，评估了 8,808 所院校，最终公布 1,504 所，分布在 106 个教育体系。英国有 93 所院校上榜。评估数与上榜数差着近六倍，筛选规则本身也值得知道。
Not every university can appear in the table, since QS requires degree programmes at both undergraduate and postgraduate level, teaching in at least two faculty areas, an Academic Reputation position inside the global top 20 per cent, and at least 100 Scopus-indexed papers over five years. Institutions with fewer than 5,000 students face extra thresholds, needing a top-1,000 place in one of three indicators, a top-1,200 place in two, or a top-1,300 place overall. The 2027 edition was published on 18 June 2026, evaluating 8,808 institutions and publishing 1,504 of them across 106 higher education systems. That gap between evaluated and published is a reminder that selection, not only scoring, shapes the list.
顾问委员会与学生代表的位置
权重改动要经过一道外部评估。QS 的全球排名顾问委员会是独立机构，官网列出的成员约 39 位，来自 20 多个国家，主席是香港科技大学的 Alison Lloyd，秘书是巴斯大学的 Ludovic Highman，成员里还有一名学生代表。官网 FAQ 给了新增指标必须回答的五个问题，从提出理由、对学生的价值、对院校的价值一直问到优势与弱点。这些问题答完之后，才进入映射与分析，再征求排名与数据团队、执行层、顾问委员会以及学生与院校的意见。学生视角在这套流程里有正式席位，不是装饰。
Weight changes pass an external review, and the QS Global Rankings Advisory Board is an independent panel whose published membership runs to about 39 experts from more than 20 countries, chaired by Alison Lloyd of HKUST with Ludovic Highman of the University of Bath as secretary, plus a student representative. The QS FAQ lists five questions any proposed indicator must answer, covering the reason for it, its value to students, its value to institutions, and its strengths and weaknesses. Only then does mapping and analysis begin, with input sought from the rankings and data teams, QS executives, the board, students and institutions, so the student perspective holds a seat in that room rather than a mention on a slide.
按关注点挑分项而不是看总分
权重表真正的用法是按关注点挑分项。在意班级规模与教学资源，看师生比，它占 10%。在意毕业去向，把雇主声誉 15% 与就业成果 5% 放在一起看，合计 20%。打算读研究型学位，师均引用 20% 与国际研究网络 5% 比总名次更有信息量。在意校园国际化程度，国际学生 5% 与国际教师 5% 才是对应指标。同一所院校在这些分项上的名次差距经常是几百位。总分是加权结果，它替你把取舍做完了，你自己也可以重做一遍。
The weighting table is most useful when read against your own priorities: class sizes and staff resources point you to Faculty Student Ratio at 10 per cent, while graduate destinations point to the employer side, which adds up to 20 per cent across Employer Reputation and Employment Outcomes. A research degree makes Citations per Faculty and International Research Network, at 20 and 5 per cent, more informative than an overall position, and an international campus points to the two five per cent indicators instead. One institution's ranks across those columns can sit hundreds of places apart. The headline number has already done your weighting for you.
跨版次比较的三个事实陷阱
跨版次比较有三种常见陷阱。第一类是把名次直接拉成趋势线，忽略了 2024 版新增三项指标、学术声誉与师生比一起被调权。第二类是拿总分做多年对比，归一化调整会让分数在实质没有变化时也发生位移。第三类是把区域排名与总排名混用，国际学生多样性在欧洲区域排名计入 5%，在旗舰排名是 0%。稳妥的做法是把版次年月与分项分一起记下。比较之前交代方法是否变过，结论才站得住。
Three traps appear whenever editions are compared side by side: the first is drawing a trend line through positions without noting that the 2024 edition added three indicators and reweighted Academic Reputation and Faculty Student Ratio at the same time. The second is comparing overall scores across years, where a normalisation tweak can move a number even when nothing substantive changed. The third is mixing a regional table with the global one, since International Student Diversity carries 5 per cent in the Europe regional ranking and nothing in the flagship. Record the edition and the indicator scores together, and state whether the method changed.
排名与官方数据怎么配合使用
排名之外另有官方数据可以对齐。Discover Uni 官网给出学生体验与毕业去向的官方口径，OfS 官网发布监管信息与准入参与计划，两者都不受排名权重变化影响。打算申请具体专业时，QS 的科目排名与总排名同日发布，权重结构与旗舰排名并不相同，更贴近课程层面的比较。把这几类来源放在一起看，比在同一个榜单里挑数字踏实得多。申请季的时间成本有限，先记下当年版次与分项分，再去核对官网的课程信息。很多学生以为排名越高课程就越合适，实际影响就读体验的是班级规模、考核方式与导师安排。常见的错误是只抄一个名次就做决定，完全没看分项分与课程结构。这两件事花的时间一样多，收获完全不同。
Official sources sit alongside the ranking and answer different questions. Discover Uni publishes student experience and destination data, while the OfS website carries regulatory information and access and participation plans, neither of which reacts to a weighting change. When you are applying to a specific course, the QS subject tables are released on the same day as the overall ranking but use a different weight structure, which brings the comparison closer to what you will actually study. Set those sources next to the headline number and the picture is steadier than anything a single column can give you. Many students assume a higher rank means a better fit, yet class sizes, assessment style and supervision arrangements are what shape day-to-day study. A common mistake is copying one number into a shortlist without reading a single indicator score.
牛津高阶英汉双解词典 第10版 商务印书馆¥168📱 去京东购买京东精选需要一对一辅导？Need One-on-One Tutoring?
A-Level / GCSE / IB / KS2-KS3 全科辅导
16621398022 同微信
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图