📚 Test with Category: A Complete Guide to Categorised A-Level Testing | 按类别测试：A-Level 分类测试完全指南
Every A-Level student meets the same wall sooner or later. You have worked through hundreds of past-paper questions, you know the content, and yet a mock exam still finds the same two or three topics and takes marks off you in exactly the same way. The problem is almost never effort. It is structure. A random test measures your whole syllabus at once and tells you almost nothing about which category of question actually failed.
每个 A-Level 学生迟早都会撞上同一堵墙。你刷过上百道真题，知识点也背过，但模拟考仍然精准地找到那两三个老问题，用同样的方式扣分。问题几乎从来不是努力不够，而是结构不对。随机测试一次性测量整个考纲，却几乎无法告诉你究竟是哪一类题目真的出了问题。
The method described in this article - testing with a category label attached to every question - fixes that. You attach a named category to each question before you attempt it, mark the paper honestly, and then calculate a separate score for every category. Within a few weeks you own a ranked list of your own weak areas instead of a vague overall percentage. It works for AQA, Edexcel, OCR, WJEC and Cambridge International students alike, in Year 12 and Year 13.
本文要讲的这套方法——给每道题贴上类别标签之后再做测试——正好解决这个问题。你在动笔之前给每道题标注一个明确类别，认真批改，然后为每个类别单独计算分数。几周之内，你得到的不是模糊的总分，而是一份属于你自己的薄弱项排名。无论你学的是 AQA、Edexcel、OCR、WJEC 还是剑桥国际考纲，无论你是 Year 12 还是 Year 13，这套方法都适用。
1. What "Test with Category" Really Means | "按类别测试"到底指什么
A categorised test is any test in which every question carries a label taken from a fixed, pre-agreed list of topics, and in which the results are recorded at the level of those labels rather than as a single total. The label is decided before you start writing, never afterwards. That single rule is what turns a pile of past-paper questions into diagnostic data.
所谓"按类别测试"，是指每一道题都带有一个来自固定清单的主题标签，而成绩是按这些标签分别记录的，不是只记一个总分。标签必须在动笔之前就定好，绝不能事后补。正是这一条规则，把一堆零散的真题变成了可用于诊断的数据。
The core calculation is simple and should be repeated for every category you test:
核心计算非常简单，每一个被测试的类别都要重复一次：
Category Score = (Marks Earned in Category ÷ Marks Available in Category) × 100%
Three components make the system work: a category map (the fixed list of topics drawn from your specification), a question bank tagged by category, and a results sheet that records marks per category rather than per paper. Remove any one of the three and the method collapses back into ordinary revision.
这套系统由三个部件支撑：分类地图（从考纲整理出的固定主题清单）、按类别打标签的题库，以及一份按类别而非按整卷记录分数的成绩表。三者缺一，方法就会退化成普通复习。
2. Why Categorised Testing Beats Random Practice | 为什么分类测试优于随机刷题
A full past paper answers one question: "how did I do today?" A categorised test answers a much more useful question: "which type of question costs me the most marks per attempt?" The first is a snapshot, the second is a diagnosis. Diagnosis is what changes a grade.
整套真题只回答一个问题："我今天考得怎么样？"分类测试回答的则是一个有用得多的问题："哪一类题目，平均每次让我失分最多？"前者是快照，后者是诊断。真正能改变成绩的是诊断。
There is also a memory effect. When questions from several categories are mixed inside one test, your brain has to decide which method to use as well as how to use it. That decision step is exactly what an exam demands, and it is exactly what topic-by-topic textbook exercises never train.
此外还有记忆层面的效应。当一份测试里混入多个类别的题目时，你的大脑不仅要执行方法，还要先判断该用哪个方法。这个"判断步骤"正是考试真正要求的能力，也正是按章节顺序做的课本习题永远训练不到的部分。
| Aspect | 对比项 | Random practice | 随机刷题 | Categorised testing | 分类测试 |
|---|---|---|
| Feedback | 反馈 | A single percentage | 只有一个百分数 | A score per topic area | 每个主题各有分数 |
| Revision target | 复习目标 | Guesswork | 靠感觉猜 | Ranked, evidence-based | 有排名、有证据 |
| Time cost | 时间成本 | 2 hours per paper | 每套两小时 | 45-75 minutes per set | 每套 45-75 分钟 |
| Motivation | 动力 | Slow, invisible | 缓慢且不可见 | Visible per-category progress | 每个类别的进步看得见 |
3. Building Your Category Map from the Specification | 依据考纲建立你的分类地图
Download the specification for your exact subject and exam board. Read the subject content section and the assessment objectives. Your categories should be finer than the top-level headings but coarser than individual bullet points. As a rule of thumb, most A-Level subjects break into 12 to 25 usable categories.
先下载你所学科目与考试局的官方考纲（specification），认真读"学科内容"部分与"考核目标"（assessment objectives）。你的类别要比一级大标题更细，但比每个小知识点更粗。一个经验法则是：大多数 A-Level 科目可以拆成 12 到 25 个可用类别。
Then weight each category by the number of marks it typically attracts. A category worth 4% of the paper deserves four times less testing time than one worth 16%. Mark weightings can be estimated by counting marks in three or four recent past papers.
接着按每个类别通常占的分值给它加权。占整卷 4% 的类别，应投入的测试时间大约是占 16% 类别的四分之一。分值权重可以通过统计三四套近年真题的分值来估算。
| Example: A-Level Maths | 示例：A-Level 数学 | Approx. weight | 大致权重 |
|---|---|
| Pure - Algebra and Functions | 纯数-代数与函数 | High | 高 |
| Pure - Differentiation | 纯数-微分 | High | 高 |
| Pure - Integration | 纯数-积分 | High | 高 |
| Pure - Trigonometry and Vectors | 纯数-三角与向量 | Medium | 中 |
| Statistics - Probability and Distributions | 统计-概率与分布 | Medium | 中 |
| Statistics - Hypothesis Testing | 统计-假设检验 | Medium | 中 |
| Mechanics - Kinematics and Newton's Laws | 力学-运动学与牛顿定律 | Medium | 中 |
4. Choosing the Right Question Source for Each Category | 为每个类别挑选合适的题源
For each category you need at least 8 to 15 tagged questions that have official mark schemes attached. Topic-sorted past papers are the gold standard because the wording, command words and mark allocation match the real exam. Textbook end-of-chapter questions are useful for early fluency but are usually too gentle for testing.
每个类别至少需要 8 到 15 道带官方评分标准（mark scheme）的标注题目。按主题分类的真题是最优选择，因为它的措辞、指令词（command word）和分值分配都与真实考试一致。课本的章末练习题适合早期的熟练度训练，但作为测试通常过于温和。
- Past papers by topic | 按主题分类的真题 - closest to exam standard, official mark schemes, free from board websites | 最接近考试标准，有官方评分标准，考试局官网可免费下载
Tip: cut and paste into your own test sheet, keeping the category label in a header row. | 建议：剪贴成自己的测试卷，并在页眉保留类别标签。 - Textbook and revision guide questions | 课本与教辅题目 - good for building fluency before timed testing | 适合在计时测试之前建立熟练度
- Teacher-made or tutoring sets | 老师或辅导机构出的题 - useful when a category is thin on past-paper material | 当某类别真题资源稀缺时很有用
- Language reference tools | 语言工具书 - for essay subjects, a strong monolingual dictionary such as the Oxford Advanced Learner's Dictionary (10th edition, typically around ¥169) sharpens precision in extended writing | 对写作类科目，一本好的英英词典（如《牛津高阶英汉双解词典》第 10 版，参考价约 ¥169）能显著提升长篇写作的用词准确性
5. Designing a Test Set: Length, Timing and Mark Weighting | 设计一套测试卷：题量、时间与分值权重
A single categorised test should take 45 to 75 minutes and contain 6 to 12 questions. This is short enough to be repeated twice a week and long enough to produce meaningful per-category data. Anything under four questions per category gives you noise, not evidence.
单套分类测试建议控制在 45 到 75 分钟，包含 6 到 12 道题。这个长度短到可以一周做两次，又长到能产出有意义的分类数据。每个类别少于四道题，你得到的是噪音，不是证据。
Mirror the exam weighting when you build the paper:
组卷时要让分值权重贴近真实考试：
Marks for Category = Total Test Marks × (Category Weight ÷ 100)
Decide before you start whether the test is untimed or timed. Untimed tests isolate knowledge and method errors; timed tests expose speed and decision problems. Alternate between the two rather than always defaulting to one.
开始之前先决定这次是计时还是不计时。不计时测试能分离出知识与方法层面的错误；计时测试暴露的是速度与判断问题。两种交替使用，不要永远只用一种。
- Weeks 1-4 | 第 1-4 周: untimed, 4-6 questions, one category per section | 不计时，4-6 题，每部分一个类别
- Weeks 5-8 | 第 5-8 周: timed at 1.5 minutes per mark, mixed categories | 计时，约每分 1.5 分钟，类别混排
- Weeks 9-12 | 第 9-12 周: full past papers under exam conditions, categories recorded after marking | 全真整卷，考后记录分类成绩
6. Marking Honestly with Mark Schemes and Grade Boundaries | 用评分标准与分数线严格给分
The mark scheme is your measuring instrument, not your comfort blanket. Award method marks exactly as written: in Edexcel and AQA mathematics, M1 rewards a correct method even when the final answer is wrong, A1 rewards accuracy, and B1 is an independent mark not tied to a method. If you give yourself a mark you have not earned, the whole dataset becomes worthless.
评分标准是测量工具，不是安慰剂。要严格按标准给分：在 Edexcel 与 AQA 数学中，M1 奖励正确的方法（即使最终答案错误），A1 奖励准确性，B1 是不依赖方法的独立分。如果你给了自己不该得的分，整份数据就作废了。
Convert raw marks into a percentage for each category, then compare against published grade boundaries for the same paper series. Grade boundaries are set per paper per session and vary widely: an A* on some maths papers sits near 80-85%, while on certain science and humanities papers it can be close to 70%. Never compare raw marks across different papers.
把每个类别的原始分换算成百分数，再与同一考试季公布的分数线对比。分数线是按每次考试单独划定的，波动很大：某些数学卷的 A* 线接近 80-85%，而部分理科与人文学科的 A* 线可能靠近 70%。千万不要跨卷比较原始分。
Overall Test Score = (Total Marks Earned ÷ Total Marks Available) × 100%
7. Error Logging and Category-Level Diagnosis | 错题记录与分类诊断
An error log turns lost marks into instructions. For every question you lose marks on, record five things: the date, the question reference, the category, the number of marks lost, and the error type. Error types are finite and worth standardising.
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图