Kingsbridge 君桥 选院数据库College Explorer
君桥官网Main site 选院咨询Consultation
选院数据库College Explorer / 牛津Oxford / 口径与出处Methods & sources

牛津 · 口径与出处Oxford · Methods & sources

Provenance

这些数从哪里来、比率怎么算、哪些地方有边界。Where these numbers come from, how the rates are calculated, and where their limits are.

数据源

Data sources

SOURCES
  • 官方统计Official stats牛津大学官方公开的本科招生统计看板（Admissions statistics）：分专业的申请、面邀、offer、入学（按英国 / 欧盟 / 非欧盟常住地拆分），分学院的申请、offer、入学，学院成功率，国家和地区，直接申请与 Open 申请。覆盖 2007–2025 年入学。本站整理后核对过：各专业加总等于全校合计，各学院直接申请数等于各专业直接申请数。The University of Oxford’s official undergraduate admissions statistics dashboards (Admissions statistics): applications, shortlisting, offers and acceptances by subject (split into UK / EU / non-EU domicile), applications, offers and acceptances by college, college success rates, countries and territories, and direct vs open applications, for 2007–2025 entry. After compiling them we checked that the subjects add up to the university totals and that each college’s direct applications add up to each subject’s direct applications.
  • 院系报告Departments数学、计算机、物理、工程、经济与管理、PPE、法律、医学 / 生物医学、化学九个院系每年公开的招生报告，取其中的笔试均分（MAT、PAT、TSA、LNAT、UCAT）。有的年份用的是学院网站转载的院系报告原件；只有 Mansfield 学院转述版给了的数标 †。每个数都在原文里找到过出处。The annual admissions reports published by nine departments – Mathematics, Computer Science, Physics, Engineering, Economics and Management, PPE, Law, Medicine / Biomedical Sciences and Chemistry. We use their average admissions test scores (MAT, PAT, TSA, LNAT, UCAT). Some years come from copies of the department report posted on college websites; figures given only in Mansfield College’s own summary are marked †. Every number was located in its source text.
  • 信息公开FOI replies牛津大学依英国《信息自由法》（FOI）公开答复的数据：分学院、分专业的申请、面邀、offer 人数（2023–2025 年三年合计），以及中国大陆常住地申请人分专业、分学院的人数（2018–2025）。Data the University of Oxford has released in public replies under the UK Freedom of Information Act (FOI): applications, shortlisting and offers by college and subject (2023–2025 combined), and mainland China-domiciled applicants by subject and by college (2018–2025).
  • 中学榜Schools牛津官方「UCAS Apply Centre」看板，2021–2025 年每个报考中心的申请、offer、入学。哪些中心在中国大陆，是本站逐个核定的。Oxford’s official “UCAS Apply Centre” dashboard: applications, offers and acceptances for every centre, 2021–2025. We checked one by one which centres are in mainland China.

计算规则

Calculation rules

RULES
  • 年份Years一律按入学年：2025 ＝ 2024 年 10 月递交申请、2025 年入学。院系报告写「2024-25」的已换成入学年 2025。All years are years of entry: 2025 = applied in October 2024, starting in 2025. Department reports labelled “2024-25” have been converted to entry year 2025.
  • 主指标Main measure最终进牛津% ＝ 直接申请这所学院、最后在牛津任一学院入学的人 ÷ 直接申请这所学院的人，来自牛津官方的学院成功率。被别的学院调剂录取也算进了牛津。它按入学算，不按 offer 算；只算直接申请，不含 Open 申请。Admitted to Oxford % = people who applied directly to this college and were admitted to any Oxford college ÷ people who applied directly to this college, from Oxford’s official college success rates. Being admitted by another college still counts. It is based on acceptances, not offers, and covers direct applications only, not open applications.
  • 次指标Other measures在本院入学% ＝ 直接申请并最后在这所学院入学的人 ÷ 直接申请的人，永远不大于主指标。外院调入占比 ＝ 这所学院录取的直接申请人里，原本直接申请别的学院的人所占比例（不算 Open 申请入学的人）。Admitted to this college % = people who applied directly and were admitted to this college ÷ direct applicants; it is never higher than the main measure. Intake from other colleges’ applicants = among the direct applicants this college admitted, the share who had applied directly to a different college (open applicants aren’t counted).
  • 稳定性Consistency学院之间的差别常常只有一两个人。所以专业版表格另给「稳定性」：把最近 12 年切成 4 个三年段（2014–16、2017–19、2020–22、2023–25），看这所学院每一段的「最终进牛津%」是不是高于这个专业全部直接申请人的平均。一段申请不足 10 人的不判断。表格的高亮和简单版的「多年稳定」榜，只给能判断的段数不少于 3、而且最多只有 1 段不高于平均的学院。Differences between colleges are often just one or two people. So the Pro table adds “Consistency”: the last 12 years are split into four three-year periods (2014–16, 2017–19, 2020–22, 2023–25) and we check whether the college’s “Admitted to Oxford %” was above the average for all direct applicants in the subject in each one. Periods with fewer than 10 applicants aren’t judged. The table highlight and the Simple view’s “consistently high” list only include colleges with at least 3 periods judged and no more than one of them at or below the average.
  • 不用的算法Not used学院层不用「offer ÷ 申请」：牛津的申请记在第一志愿学院名下，面邀记在做决定的学院，offer 记在发 offer 的学院；2025 年约三成 offer 不是第一志愿学院发的。三个数不是同一批人，相除不是这所学院的录取率。At college level we don’t use “offers ÷ applications”: Oxford counts applications under the first-choice college, shortlisting under the college that made the decision and offers under the college that made the offer; in 2025 about three in ten offers were not made by the first-choice college. The three numbers aren’t the same people, so dividing them doesn’t give the college’s admission rate.
  • 全专业层Subject level专业层的「面邀率」「offer 率」＝ 面邀人数 ÷ 申请、offer ÷ 申请，按常住地分英国、欧盟、非欧盟；中国大陆申请人在「非欧盟」里，看板不单列。At subject level, “shortlist rate” and “offer rate” = shortlisted ÷ applications and offers ÷ applications, split into UK, EU and non-EU domicile; mainland China applicants are part of “non-EU” – the dashboard doesn’t list them separately.
  • 大陆申请占比Mainland share对比、打分、向导里的「中国大陆申请占比」＝这所学院 2023–2025 三年全部专业的申请人里，中国大陆（常住地）申请人所占的比例，来自牛津的信息公开答复和官方分学院看板。它反映的是申请人的构成，不是录取率；分专业、分学院的大陆人数太少，官方多数没给具体数字，所以用全部专业合计。“Mainland China share of applicants” in the comparison, scoring and finder = among all applicants to the college across all subjects in 2023–2025, the share domiciled in mainland China, from Oxford’s FOI replies and its official college dashboard. It describes who applies, not an admission rate; figures by subject and college are too small to be published, so all subjects are combined.
  • 样本门槛Sample size所选时间段里直接申请不足 20 人的学院，比率灰显并标 *，排在最后，不参与高亮，也不进结论卡的前 3 名。Colleges with fewer than 20 direct applicants in the chosen period have their rates greyed out and marked *; they are listed last, never highlighted and never in the top 3 of the summary card.

已知边界

Known limits

LIMITS
  • 少量“Few”信息公开答复里人数很少的格子，牛津没有给出具体数字（多数阈值没写明；能对照的格子都在 5 人以下，也可能是 0）。答复里没有出现的专业或学院，本站也不当 0，显示「—」。本站显示为「少量」，不当 0 加总，也不拿来算比例；合计写成「至少 …」。中学榜里小于 3 的格子官方留空，本站显示「<3」。In FOI replies Oxford doesn’t give exact figures for very small counts (the threshold usually isn’t stated; where we can cross-check, these cells are below 5 and may be 0). Subjects or colleges missing from a reply aren’t treated as 0 either; they show as “—”. We show these as “few”: they are never added up as zero or used to work out a rate, and totals are given as “at least …”. In the schools ranking Oxford leaves counts below 3 blank; we show them as “<3”.
  • 中国口径China常住地是申请人在 UCAS 申请里填报的长期居住地，不是国籍，也不是在哪里读书。「中国大陆」分常住地和国籍两套数，2025 年常住地申请 1,991 份、国籍 2,126 份。本站逐年、分专业、分学院的数一律用常住地（2018–2020 年的数含澳门，人数很少）；国籍只在全校层单列对照。Domicile is the permanent home the applicant gives on their UCAS application – not nationality, and not where they go to school. “Mainland China” has two sets of figures – domicile and nationality: in 2025 there were 1,991 applications by domicile and 2,126 by nationality. Our year-by-year, subject and college figures all use domicile (2018–2020 figures include Macao, a very small number); nationality appears only at whole-university level for comparison.
  • 学院限制College limits哈里斯·曼彻斯特学院（Harris Manchester）只招 21 岁及以上的本科生；几所规模很小的永久私人学堂（Wycliffe Hall、St Stephen's House、Blackfriars，以及 2022 年关闭的 St Benet's Hall）在官方成功率表里合成一行「其他永久私人学堂」，其中有的已经不招本科生。这两行表格默认去掉，可以一键切换。摄政公园学院（Regent's Park）也是私人学堂，但招普通本科生，照常显示并加标记。Harris Manchester only takes undergraduates aged 21 or over; several very small Permanent Private Halls (Wycliffe Hall, St Stephen's House, Blackfriars, and St Benet's Hall, which closed in 2022) are combined into one row, “Other Permanent Private Halls”, in the official success-rate tables; some no longer take undergraduates. Both rows are hidden by default (one tap shows them again). Regent's Park is also a Permanent Private Hall but admits ordinary undergraduates, so it is shown as normal, with a label.
  • 笔试换代Test changesMAT：2018 年的考试起大纲调整，2023 年 10 月起机考，2025 年 10 月是最后一届，之后数学、计算机考 TMUA。PAT：2008 年起由两份卷子合成一份，2015 年起取消选择题，2024 年 10 月起机考，2025 年 10 月是最后一届，之后工程、物理考 ESAT。TSA 两次改分制；医学由 BMAT 改 UCAT，UCAT 满分又从 3600 降到 2700。这些地方在表格里都用分隔线隔开，前后的分数不能连起来比。MAT: the syllabus was updated for the 2018 test, the test moved online in October 2023, and October 2025 was the last MAT (Mathematics and Computer Science now use the TMUA). PAT: two papers were combined into one from 2008, multiple-choice questions were dropped from 2015, it moved online in October 2024, and October 2025 was the last PAT (Engineering and Physics now use the ESAT). TSA scoring changed twice; Medicine moved from the BMAT to the UCAT, whose maximum then fell from 3600 to 2700. All of these points are marked with a line in the tables, and scores on either side can’t be compared.
  • 学院译名College names牛津各学院没有官方中文名，本站采用中文维基百科「牛津大学学院列表」的通行译名，英文是学院官方名称。Oxford colleges have no official Chinese names; we use the common translations in the Chinese Wikipedia “List of colleges of the University of Oxford”, alongside the colleges’ official English names.
  • 停招专业Old courses已停招的课程（如 2013 年停招的 EEM）不单独成页；数学与统计 2024 年起和数学合并招生，2024 年以后的数字算在「数学」里。Courses that have closed (such as EEM, which stopped in 2013) don’t have their own page; Mathematics and Statistics has been admitted jointly with Mathematics since 2024, and from 2024 its figures are counted under Mathematics.

向导怎么算的

How the college finder works

METHOD
  • 公开可复算Reproducible第三步选哪一条，就按哪一个数排，不掺别的指标，都按近 10 年（2016–2025）算：进牛津的比例高、而且多年稳定——先按 2014–2025 四个三年段里高于本专业平均的段数（能判断的段数不足 3 的排在后面），一样的再按「最终进牛津%」从高到低；中国学生多一些——按这所学院 2023–2025 全部专业申请人里中国大陆申请人的占比；申请人多一些 / 少一些——按直接申请人数；离上课地点近——按到主要上课地点的平均步行时间。第二步只用来去掉申请人不能申的学院。直接申请不足 20 人的学院不参与排序；如果这样剩下不足 5 所，就把小样本学院也算进来，并在结果页提示。几所小型私人学堂合并的那一行不参与。每一所都把原始数字列在旁边，可以自己核。Whichever option you pick in step 3, colleges are ranked by that one number only, with nothing else mixed in, all over the last 10 years (2016–2025): High and consistent share admitted to Oxford – first by how many of the four three-year periods in 2014–2025 were above the subject average (colleges with fewer than 3 judgeable periods come after), then by “Admitted to Oxford %”, highest first; More mainland Chinese students – by the mainland China share of the college’s 2023–2025 applicants across all subjects; More / fewer applicants – by direct applicants; Close to lectures – by average walking time to the main lecture sites. Step 2 is only used to remove colleges you can’t apply to. Colleges with fewer than 20 direct applicants are left out of the ranking; if that leaves fewer than 5, the small-sample colleges are put back in and the results page says so. The combined row for several small private halls isn’t ranked. Every college is listed with its raw numbers alongside, so you can check them yourself.
  • 它不是什么What it isn’t它不预测录取概率。它只把公开的历史数字按你选的那一条排个序。比例高，也可能只是因为申请这所学院的人本身更强。真正的选院还要看课程、面试形式、笔试成绩和个人陈述。It does not predict your chance of admission. It simply sorts public historical numbers by the option you chose. A high rate may simply mean the people applying to that college were stronger. A real college choice also depends on the course, interview format, admissions test results and your personal statement.

智能打分怎么算

How smart scoring works

SCORING
  • 打分Scoring每一项先在本专业的学院之间换算成 0–1（这一项数字最理想的学院记 1，最不理想的记 0；申请人数和步行时间是越少越好），再按你选的看重程度（不看重 0 到很看重 4）加权平均，换成 0–100 分。各项都是本站公开的数据：往年最终进牛津的比例、多年稳定（四个三年段里高于平均的段数 ÷ 能判断的段数）、在本院入学的比例、中国大陆申请占比（全部专业）、竞争（直接申请人数），以及（整理好上课地点的专业）离上课地点的步行时间。所选时间段里直接申请不足 20 人的学院，两个比率按本专业中位数处理并标「样本少」；只招 21+ 的学院按你填的情况去掉；几所小型私人学堂合并的那一行不参与。Each factor is first rescaled to 0–1 across the colleges for the subject (the college with the best figure scores 1 and the worst 0; for applicant numbers and walking time, fewer is better). The factors are then averaged, weighted by how much you said each one matters (from 0, not important, to 4, very important), and converted to a score out of 100. All of them use data published on this site: the past share admitted to Oxford, consistency (periods above the average ÷ periods that could be judged), the share admitted to the college itself, the mainland China share of applicants (all subjects), competition (direct applicants) and, for subjects whose lecture sites have been compiled, walking time to lecture sites. For colleges with fewer than 20 direct applicants in the chosen period, both rates are set to the subject median and marked “small sample”; the 21+ college is removed based on what you entered; the combined row for several small private halls isn’t scored.
  • 它不是什么What it isn’t它不预测录取概率，只是按你在意的方面把历史数字排个序。It does not predict your chance of admission; it just sorts historical numbers by the things you care about.

纠错

Corrections

CORRECTIONS
  • 逐格回源Check at source若发现任一个数与牛津官方统计不符，或你的学校应在中学榜上却没出现，请告知，我们会复核、补入并公开更正记录。If any number doesn’t match Oxford’s official statistics, or your school should be in the China schools ranking but isn’t, please tell us. We will check it, add it if needed and publish a record of the correction.