📚 Exam Techniques and Marking Criteria for Pre-U CAIE Statistics | Pre-U CAIE统计：答题技巧与评分标准
Mastering exam technique is just as important as understanding statistical concepts for Pre-U CAIE Statistics. Two candidates may know exactly the same formulae and still finish the paper ten marks apart, simply because one of them writes statistics the way an examiner expects to read it. This article explains the marking criteria used by examiners and provides practical strategies to maximise your marks by presenting your knowledge effectively.
对于Pre-U CAIE统计学而言，掌握考试技巧与理解统计概念同等重要。两名考生可能掌握完全相同的公式，最终分数却相差十分，原因仅仅在于其中一人按照考官期望的方式来书写统计答案。本文解释考官使用的评分标准，并提供实用策略，通过有效展示你的知识来最大化你的得分。
1. Why Exam Technique Matters | 为什么答题技巧重要
Understanding a concept and earning marks for that concept are two separate skills. You may know precisely why a two-tailed test is needed, but if you never write the word 'two-tailed' and never halve the significance level, the examiner cannot award the mark. Marks live on the page, not in your head.
理解一个概念与因为这个概念而得分，是两项不同的技能。你可能非常清楚为什么需要使用双尾检验，但如果你从未写下"双尾"这个词，也没有把显著性水平除以二，考官就无法给你分。分数存在于纸面上，而不是存在于你的脑海中。
Statistical questions also differ from pure mathematics questions: they usually contain a real-world context, and that context is part of the marks. A correct number with no context can lose one or two marks out of five, which is a large proportional loss on a short question.
统计题目与纯数学题目还有一个区别：它们通常带有真实情境，而情境本身就是评分的一部分。一个正确的数字如果脱离情境，可能在一道五分题中失掉一到两分，这在短题中占比很大。
2. Understanding the Mark Scheme | 理解评分方案
In Pre-U CAIE Statistics, marks are classified into method marks (M), accuracy marks (A), and independent marks (B). Understanding the difference is crucial. Method marks are awarded for a correct approach, even if numerical mistakes occur later. Accuracy marks require the final answer to be correct, usually following a correct method. Independent marks are given for statements, definitions, or correct use of a formula regardless of method.
在Pre-U CAIE统计学中，分数分为方法分(M)、准确分(A)和独立分(B)。理解它们的区别至关重要。方法分是针对正确解题思路而给出的，即使后续出现数字错误也能获得。准确分要求最终答案正确，通常基于正确的方法。独立分是对表述、定义或正确使用公式给出的，与方法无关。
The mark scheme also uses a set of abbreviations that tell you how strictly an answer is judged. Learning them lets you predict where marks are available before you write anything.
评分方案还使用一组缩写，说明答案被评判的严格程度。掌握这些缩写，你就能在动笔之前预判分数藏在哪里。
| Mark Type / 分数类型 | Explanation / 解释 |
|---|---|
| M (Method / 方法) | Awarded for a correct approach or correct formula application, even if the arithmetic later fails. 针对正确思路或公式应用给分，即使后续运算出错。 |
| A (Accuracy / 准确) | Awarded for a correct value; normally depends on the corresponding M mark. 针对正确数值给分，通常依赖相应的M分。 |
| B (Independent / 独立) | Awarded for a statement, definition, hypothesis or interpretation, independent of any method. 针对陈述、定义、假设或解释给分，与方法无关。 |
| cao | Correct answer only - no method mark available. 只看最终答案，无方法分。 |
| ft | Follow through - marks awarded from your earlier (possibly wrong) answer. 后续分可依据你前面（可能错误）的答案给予。 |
| oe | Or equivalent - any mathematically correct form is accepted. 或等价形式，任何数学上正确的形式都接受。 |
| isw | Ignore subsequent working - once the correct answer appears, later mess is ignored. 忽略后续过程。 |
| ag | Answer given - you must show full working to justify the printed result. 题目已给答案，必须写出完整过程。 |
3. How Marks Are Allocated in a Hypothesis Test | 假设检验中的分数如何分配
A typical allocation: in a hypothesis test, stating the hypotheses earns B marks, calculating the test statistic earns M and A marks, and writing the conclusion in context earns B marks. Always demonstrate the method clearly - you can secure M marks even if the final answer is wrong.
典型的分配：在假设检验中，陈述假设可获得B分，计算检验统计量可获得M和A分，写出情境下的结论可获得B分。务必清晰地展示解题过程——这样即使最终答案错误，你也能确保获得方法分。
almost every paper. A random sample of 25 observations from a normal population has mean x̄ = 51.2, and the population standard deviation is known to be σ = 3.0. Test at the 5% level whether the population mean exceeds 50.
考虑一个几乎每份试卷都会出现的解题范式。取自正态总体的25个观测值，其均值 x̄ = 51.2，已知总体标准差 σ = 3.0。在5%显著性水平下检验总体均值是否大于50。
H₀: μ = 50 H₁: μ > 50
z = (x̄ − μ₀) ÷ (σ ÷ √n) = (51.2 − 50) ÷ (3.0 ÷ √25) = 2.00
nd reaching z = 2.00 is A1. Comparing 2.00 with the critical value 1.645 is M1. Concluding 'there is sufficient evidence at the 5% level that the mean exceeds 50' orrect answer with no hypotheses and no conclusion can still lose three of the five marks.
两条假设合计可得B1。写出标准化公式得M1。正确代入并得到 z = 2.00 得A1。将 2.00 与临界值 1.645 比较得M1。写出"在5%水平下有充分证据表明均值大于50"得B1。一个计算完全正确、却没有写假设也没有写结论的答案，仍可能失掉五分中的三分。
4. Showing Method: The Golden Rule | 展示过程：黄金法则
The golden rule of the mark scheme is that an examiner can only reward what is visible. Write the formula before you substitute, substitute before you evaluate, and evaluate before you conclude. Every one of those four steps is usually a separate mark.
评分方案的黄金法则是：考官只能奖励他看得见的内容。先写公式再代入，先代入再计算，先计算再下结论。这四个步骤中的每一步通常都是一分。
This matters most on a calculator paper, where a candidate can press the correct buttons and write only the answer. Even when the answer is right, an unsupported final value often scores fewer marks than a fully written solution with a small arithmetic slip.
这一点在可使用计算器的考试中尤其重要：考生可能按对了所有按键，却只写下答案。即使答案正确，一个没有过程支撑的结果，往往比一个写得很完整但有小计算失误的解答得分更低。
5. Accuracy, Rounding and Significant Figures | 准确度、四舍五入与有效数字
Accuracy marks are lost more often through rounding than through algebra. Use the convention that answers are given to three significant figures unless the question states otherwise, and never round intermediate values that you will use again.
准确分的丢失，往往源于四舍五入而不是代数运算。请遵循以下惯例：除非题目另有说明，答案保留三位有效数字；对于后面还要继续使用的中间量，绝不要提前取整。
Final answers → 3 s.f. | Intermediate values → keep full calculator accuracy
ility should not be rounded to 0.000 - write it as 3.14 × 10⁻⁵ or as a stated value such as 0.0000314. If a question says 'give your answer correct to 4 decimal places', that instruction overrides the general rule.
概率是特殊情况。二项分布或泊松分布的概率通常保留三位有效数字；一个非常小的概率不应被写成 0.000，而应写成 3.14 × 10⁻⁵ 或 0.0000314 这类明确数值。如果题目要求"保留四位小数"，则该要求优先于一般规则。
Also check whether the quantity needs a unit or a label. Writing '51.2' when the answer is '51.2 kg' or '51.2 seconds' can cost the accuracy mark even when the arithmetic is perfect.
同时要检查该量是否需要单位或标签。当答案应是"51.2 kg"或"51.2 秒"时只写"51.2"，即使运算完美也可能失掉准确分。
6. Writing Conclusions in Context | 结合情境写出结论
Examiners consistently report that context marks are the most commonly dropped marks on the whole paper. A conclusion must refer to the real-world quantity named in the question, not to 'the mean' or 'the claim' in the abstract.
考官反复指出，情境分是整份试卷中最常被丢掉的分数。结论必须指向题目中提到的现实量，而不能抽象地说"均值"或"该说法"。
Compare these two answers to the test in Section 3. 'Reject H₀' earns only part of the credit. 'There is sufficient evidence at the 5% level to support the claim that the mean weight of the packets exceeds 50 g' earns full credit, because it names the significance level, the direction of the evidence and the variable.
比较一下针对第3节检验的两种答案。"拒绝H₀"只能拿到部分分。而"在5%水平下有充分证据支持包装平均重量超过50克的结论"可获满分，因为它写明了显著性水平、证据方向以及所涉及的变量。
When the test does not reject the null hypothesis, the correct phrasing is 'there is insufficient evidence to suggest that...', never 'this proves the mean is equal to...'. Failing to reach significance is not proof of the null hypothesis.
当检验未拒绝原假设时，正确的表述是"没有充分证据表明……"，绝不能写成"这证明均值等于……"。未达到显著并不等于证明了原假设。
7. Command Words and Question Decoding | 指令词与审题
Every command word signals how much working is expected and where the marks sit. Spending thirty seconds decoding the instruction is usually worth several marks.
每个指令词都提示了需要写多少过程以及分数在哪里。花三十秒解读指令，通常能换来好几分。
| Command Word / 指令词 | What It Demands / 要求 |
|---|---|
| State / Write down / 写出 | No working required; a single correct value or word. 无需过程，一个正确数值或词语即可。 |
| Calculate / Find / 计算、求 | Full method must be shown for M and A marks. 必须写出完整方法以获取M分和A分。 |
| Show that / 证明（给出结论） | Answer is printed; every algebraic step must be justified. 答案已给出，每一步都要有依据。 |
| Hence / 由此 | You must use the previous part's result, not start again. 必须使用上一问的结果，不能从头另算。 |
| Verify / 验证 | Substitute the given value and confirm it works. 代入给定值并确认其成立。 |
| Comment / Interpret / 评论、解释 | A sentence in context, referring to the data or the model. 一句结合情境的话，须指向数据或模型。 |
| Justify / 说明理由 | Give statistical reasoning, not just a numerical comparison. 给出统计学理由，而非仅数值比较。 |
| Estimate / 估计 | An approximate value is acceptable; read from a graph or table if needed. 允许近似值，可读图或查表。 |
8. Presenting Calculations: Layout and Notation | 计算呈现：书写与符号
Clear layout is not a cosmetic issue; it is how you make each mark visible. Start each part on a new line, label the parts (a), (b), (c) exactly as printed, and never bury a final answer inside a block of algebra.
书写清晰不是美观问题，而是让每一分都看得见的方式。每一小问另起一行，按题目原样标注 (a)、(b)、(c)，绝不要把最终答案埋在一大段代数式中间。
Use correct statistical notation throughout. Write x̄ X ≤ r) for a cumulative probability. Using the wrong symbol can make a correct calculation look like a conceptual error.
全程使用正确的统计符号。样本均值写 x̄，总体均值写 μ；样本标准差写 s，总体标准差写 σ；样本容量写 n；累积概率写 P(X ≤ r)。用错符号会让正确的计算看起来像概念错误。
Binomial: P(X = r) = ₙCᵣ pʳ (1 − p)ⁿ⁻ʳ | Poisson: P(X = r) = e⁻λ λʳ ÷ r!
χ² = Σ (O − E)² ÷ E | r = Sₓᵧ ÷ √Sₓₓ Sᵧᵧ
Finally, finish with a short 'Therefore ...' sentence. It takes five seconds to write and it makes the conclusion mark impossible to miss.
最后，用一句简短的"因此……"收尾。写这句话只需五秒，却能让结论分不可能被漏掉。
9. Common Mark-Losing Errors | 常见失分错误
Certain errors appear in almost every examiner report. Check your own answers against this list while you practise, and the same pattern will not survive into the exam room.
某些错误几乎出现在每一份考官报告中。练习时就对照这份清单检查自己的答案，这些毛病就不会被带进考场。
Using the wrong tail or failing to halve the significance level for a two-tailed test. 双尾检验时用错尾侧，或忘记将显著性水平除以二。
Writing a z-value of 2.00 when the question used the sample standard deviation and should have used t. 当题目使用样本标准差、本应使用 t 分布时，却用了 z。
obability that contradicts the question - for example P(X = r) when P(X ≤ r) was required. 计算了与题目要求相反的概率，例如题目要 P(X ≤ r) 却算了 P(X = r)。
Quoting a critical value with the wrong degrees of freedom, such as n instead of n − 1. 引用临界值时自由度错误，例如用 n 而不是 n − 1。
Rounding an intermediate result and carrying the error into the final answer. 对中间结果提前取整，把误差带入最终答案。
Giving a bare 'accept H₀' with no reference to the context. 只写"接受H₀"，完全不结合情境。
Skipping the definition of a random variable before using a distribution. 使用分布前未定义随机变量。
10. Diagrams, Distribution Tables and Sketch Marks | 图表、分布表与草图分
Sketch marks are among the easiest marks on the paper, and they are also among the most frequently missed. Whenever you use a normal distribution, draw a small bell curve, mark the mean at the centre, shade the required tail or region, and label the boundary value.
草图分是整份试卷中最容易拿到的分数之一，也是最常被忽略的分数之一。只要用到正态分布，就画一个小钟形曲线，在中心标出均值，把需要的尾侧或区域涂上阴影，并标注边界值。
In correlation and regression questions, always sketch or describe the scatter pattern before quoting r. A correlation coefficient of 0.8 means very different things for a linear pattern and for a strongly curved one, and a comment mark often depends on stating whether a linear model is appropriate.
在相关与回归题目中，引用 r 之前一定要先描绘散点形态。相关系数 0.8 对线性形态和明显弯曲的形态含义完全不同，而评论分往往取决于你是否说明了线性模型是否合适。
When using statistical tables, copy the row and column headings you used. Examiners frequently award a method mark for a correctly identified critical value even when the final comparison is wrong.
使用统计表时，把你所用的行标题与列标题抄下来。即使最终比较写错，考官也常因临界值识别正确而给出方法分。
11. Time Management and Checking Strategy | 时间管理与检查策略
Divide the total time by the total marks to get your minutes per mark, then add a five-minute buffer at the end for checking. If a part is taking more than twice its allotted time, write down the method you have so far, leave it, and return later.
用总时间除以总分，得出每题应花的时间，最后再留出五分钟检查缓冲。如果某一小问耗时超过应分配时间的两倍，就把已写出的方法写下来，先跳过，之后再回来。
Check in a fixed order rather than randomly. First confirm that you answered the question that was asked; second check the direction of your inequality and the number of tails; third verify significant figures and units; fourth confirm that every conclusion contains a context sentence.
检查要按固定顺序，而不是随意翻看。第一，确认你回答了题目所问；第二，检查不等号方向和尾侧数；第三，核对有效数字与单位；第四，确认每个结论都包含一句情境表述。
For long questions worth eight or more marks, read the whole question before starting. Later parts often tell you which distribution or which variable earlier parts should have used, and that saves re-writing.
对于八分以上的长题，先通读整题再动笔。后面的小问常常会提示前面该用哪个分布或哪个变量，这能省下大量重写时间。
12. Final Checklist and Exam-Day Routine | 考前清单与考试当天流程
Before the exam, build a one-page summary containing the distributions you must recognise, the test statistics with their conditions, and the exact wording you will use for conclusions. Practise past papers under timed conditions and mark them yourself using the published mark scheme, awarding marks only for what you literally wrote.
考前制作一页总结，内容包括必须识别的分布、各检验统计量及其适用条件，以及你将在结论中使用的固定句式。在限时条件下练习真题，并用官方评分方案自评，只为你真正写在纸上的内容给分。
On the day, read the rubric first, note the total number of questions and the marks, and write your hypotheses before you touch the calculator. In the final ten minutes, review every 'comment', 'interpret' and 'conclude' question - those are the marks most often left on the table.
考试当天，先读答题说明，记下题目总数与分值，在碰计算器之前先写下假设。最后十分钟，专门复查每一道"评论""解释""下结论"的题目——这些正是最常被白白丢掉的分数。
With a clear understanding of M, A and B marks, disciplined rounding, and conclusions that speak the language of the context, your exam technique will stop being the gap between your ability and your grade.
当你清楚理解M分、A分与B分，养成严谨的取整习惯，并写出使用情境语言的结论时，答题技巧将不再是你的能力与成绩之间的那道鸿沟。
Published by TutorHao | Pre-U Statistics Revision Series | aleveler.com
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