📚 CAIE AS Statistics: Exam Writing Framework & Model Answers | CAIE AS 统计：考试写作框架与范文
In CAIE AS Statistics (Paper 5), marks are not only awarded for correct numerical answers, but also for clear reasoning, precise notation, and effective communication. Developing a robust written framework can help you secure all method, accuracy, and communication marks, even when working under time pressure. This article breaks down structured writing techniques and provides full model answers for typical AS topics.
在CAIE AS统计学（试卷5）中，分数不仅授予正确的数值答案，还授予清晰的推理、精确的符号和有效的表达。即使时间紧张，建立一个稳固的写作框架也能帮助你拿到所有方法分、准确度分和交流分。本文将拆解结构化写作技巧，并提供典型AS专题的完整范文。
1. Understanding the Mark Scheme | 理解评分标准
CAIE Statistics mark schemes allocate M marks for correct method, A marks for accuracy, and B marks for independent statements such as interpreting a p-value or comparing a measure of spread. Examiners look for logical flow: you must show the steps, not just the answer. Every time you define a random variable, state hypotheses, write a probability statement, or end with a conclusion in context, you are earning marks.
CAIE统计评分标准将M分授予正确方法，A分授予准确性，B分授予独立陈述，例如解释p值或比较离散程度。考官看重逻辑流程：你必须展示步骤，而不仅仅是答案。每次你定义随机变量、陈述假设、写出概率表达式或在上下文中给出结论，都在得分。
Always annotate your work clearly. For example, write ‘P(X ≥ 15) = 0.0207’ rather than just ‘0.0207’. Include units where applicable and avoid vague language. The mark scheme often awards a final B mark for a non-technical conclusion that refers back to the original problem.
始终清晰地标注你的解答。例如，写出’P(X ≥ 15) = 0.0207’，而不只是’0.0207’。在适当处写单位，避免模糊用语。评分方案常在给出结合原问题的非技术性结论时奖励最后的B分。
2. The 5-Step Response Structure | 五步答题结构
Almost every AS Statistics question can be approached with a common five-step structure: Identify, Model, Solve, Check, Conclude. Identify the type of problem and the given data. Model it with an appropriate distribution or statistical measure. Solve mathematically, showing all calculations and substitution steps. Check conditions (e.g. np > 5 for normal approximation, or variance known). Conclude by answering the original question with a non-technical sentence.
几乎每道AS统计题都可以用通用的五步结构处理：识别、建模、求解、检查、结论。识别问题类型与给定数据。用适当的分布或统计量建模。数学求解，展示所有计算和代入步骤。检查条件（例如正态近似需np > 5，或方差已知）。通过非技术性语句回答原问题给出结论。
This framework ensures you never skip a condition check or forget the contextual conclusion. In hypothesis testing, the ‘model’ step includes hypotheses; in probability, it includes defining the variable and its parameters. Write each step on a new line or clearly label it so the examiner can follow your logic instantly.
此框架确保你不会跳过条件检查或忘记上下文结论。在假设检验中，“建模”步骤包含假设；在概率中，包含定义变量及其参数。将每一步写在新行或清晰地标记，使考官能立即跟上你的逻辑。
3. Hypothesis Testing: A Step-by-Step Framework | 假设检验：逐步框架
Step 1: Define the test statistic or random variable. Example: Let X be the number of defective items in a sample of n. Then state the distribution under H₀: X ~ B(n, p₀).
步骤1：定义检验统计量或随机变量。例如：设X为样本中缺陷品数，然后在H₀下写出分布：X ~ B(n, p₀)。
Step 2: State the null and alternative hypotheses clearly. H₀: p = 0.3, H₁: p > 0.3 (one‑tailed) or H₁: p ≠ 0.3 (two‑tailed). Always specify the parameter being tested.
步骤2：清晰陈述原假设和备择假设。H₀: p = 0.3, H₁: p > 0.3（单尾）或 H₁: p ≠ 0.3（双尾）。务必指明被检验的参数。
Step 3: State the significance level α, e.g. 5%. Decide the method: p‑value or critical region. For p‑value, calculate the probability of the observed result (or more extreme) assuming H₀ is true.
步骤3：写出显著性水平α，如5%。决定方法：p值法或临界区域法。对于p值，计算在H₀为真下观察到该结果（或更极端）的概率。
Step 4: Compare the p‑value with α, or the test statistic with critical value. If p‑value < α, reject H₀. Otherwise, do not reject H₀.
步骤4：比较p值与α，或检验统计量与临界值。若p值 < α，拒绝H₀。否则，不拒绝H₀。
Step 5: Write a conclusion in context. E.g., ‘There is sufficient evidence at the 5% significance level to suggest that the proportion of defective items has increased.’ Avoid saying ‘accept H₀’ – always use ‘do not reject’.
步骤5：写出上下文结论。例如：“在5%显著性水平下，有充分证据表明缺陷品比例已增加。” 避免说“接受H₀”——始终用“不拒绝”。
4. Writing Probability & Normal Distribution Solutions | 撰写概率与正态分布解答
For normal distribution problems, always begin by defining the variable: X ~ N(μ, σ²). Mention if the variance or standard deviation is given. Then state the required probability, standardise, and use the standard normal table. Never jump from X directly to Φ(z) without showing the z‑score calculation. Write: Z = (X − μ)/σ, then P(X > k) = P(Z > (k−μ)/σ).
对于正态分布问题，始终先定义变量：X ~ N(μ, σ²)。注明所给的是方差还是标准差。然后陈述所需概率，标准化，并使用标准正态表。切勿不展示z值计算就由X直接跳到Φ(z)。应写出：Z = (X − μ)/σ，然后 P(X > k) = P(Z > (k−μ)/σ)。
Explicitly show symmetry operations: P(Z > 1.2) = 1 − Φ(1.2). When finding an unknown mean or standard deviation, set up an equation involving Φ⁻¹. For discrete distributions (Binomial, Poisson), write the probability in full, e.g. P(X = 3) = ⁿC₃ p³ (1−p)ⁿ⁻³, and then substitute values. Where tables are used, cite the table value clearly.
明确展示对称性操作：P(Z > 1.2) = 1 − Φ(1.2)。在求未知均值或标准差时，建立含Φ⁻¹的方程。对于离散分布（二项、泊松），完整写出概率，如 P(X = 3) = ⁿC₃ p³ (1−p)ⁿ⁻³，然后代入数值。使用表格时，清楚注明查表值。
5. Permutations and Combinations: Logic and Justification | 排列与组合：逻辑与论证
Marks in permutations and combinations questions depend heavily on clear reasoning. Start by identifying whether order matters (arrangement → permutations) or not (selection → combinations). State whether objects are distinct or include repeats. Break complex scenarios into stages, using the multiplication principle. Write the number of ways for each stage as a separate expression before multiplying.
排列与组合题目的得分很大程度上依赖于清晰的推理。首先判断是否有序（排列）或无序（组合）。说明对象是否互异或含重复。将复杂情境分阶段，使用乘法原理。在相乘前，将每阶段的方法数分别写出表达式。
For arrangements with restrictions, say: ‘Treat the block of items that must be together as a single object, then arrange internally.’ When using factorials, cancel systematically and show the simplified expression. If you use the term ‘number of ways’, write n(E) for the event space. Never skip justification – a bald answer without an explanation may lose method marks.
对于带限制的排列，可以说：“将必须相邻的项目块视为一个对象，再内部排列。”使用阶乘时，系统约分并展示简化后的表达式。如果使用术语“方法数”，写出事件空间的n(E)。切勿跳过论证——无解释的裸答案可能失去方法分。
6. Data Interpretation and Modeling Questions | 数据解释与建模题
When asked to compare data sets using stem‑and‑leaf diagrams, box plots, or histograms, always discuss three aspects: location (median, mean), spread (interquartile range, standard deviation), and shape (symmetry, skewness). Begin with a specific comparison: ‘The median mark of Class A (67) is higher than that of Class B (58), suggesting better typical performance.’
当要求使用茎叶图、箱线图或直方图比较数据集时，始终讨论三个方面：位置（中位数、均值）、离散程度（四分位距、标准差）和形状（对称性、偏态）。以具体的比较开始：“A班中位数（67）高于B班（58），表明典型成绩更好。”
Then discuss spread: ‘Class B has a larger interquartile range (20) compared to Class A (12), indicating greater variability.’ Finally, comment on skewness: ‘Class A’s box plot shows a negative skew, with the median closer to the upper quartile.’ Each statement must be backed by data read from the graph, and you should use statistical terms accurately.
接着讨论离散程度：“B班四分位距（20）大于A班（12），表明变异性更大。”最后评论偏态：“A班的箱线图呈现负偏态，中位数更靠近上四分位数。”每个陈述必须有从图中读取的数据支持，并准确使用统计术语。
7. Common Pitfalls in Written Communication | 书面表达中的常见错误
Avoid writing ‘accept H₀’ – the correct phrasing is ‘do not reject H₀’. Never state that a hypothesis test proves the null hypothesis; it only assesses evidence against it. Omission of the random variable definition and its distribution loses the first mark in almost every hypothesis test.
避免写“接受H₀”——正确的表述是“不拒绝H₀”。永远不要说假设检验证明了原假设；它只是评估反对原假设的证据。漏掉随机变量定义及其分布会丢失大多数假设检验中的第一个分数。
Using non‑standard abbreviations or ambiguous notation (e.g. writing just ‘p’ instead of P(X > 3)) confuses examiners. Remember to attach ‘in context’ to your conclusion; a statistical statement without real‑world meaning will not earn the final communication mark. Also check that you have used the correct variance or standard deviation – confusing σ² with σ is a frequent slip.
使用非标准缩写或模糊符号（例如只写“p”而不是P(X > 3)）会使考官迷惑。记得在结论中加入“上下文”；没有实际意义的统计陈述将无法得到最后的交流分。还要检查是否使用了正确的方差或标准差——混淆σ²和σ是常见疏忽。
8. Model Answer 1: Binomial Hypothesis Test | 范文1：二项假设检验
Question: A die is rolled 30 times and the number ‘6’ occurs 9 times. Test at the 10% significance level whether the die is biased in favour of ‘6’.
问题：一个骰子掷30次，数字“6”出现了9次。在10%显著性水平下检验骰子是否偏向出现“6”。
Step 1: Let X be the number of sixes in 30 rolls. Under H₀, X ~ B(30, 1/6).
步骤1：设X为30次投掷中6的次数。在H₀下，X ~ B(30, 1/6)。
Step 2: H₀: p = 1/6, H₁: p > 1/6 (one‑tailed, bias in favour of 6).
步骤2：H₀: p = 1/6，H₁: p > 1/6（单尾，偏向6）。
Step 3: Significance level α = 0.10. Observed x = 9. Compute p‑value: P(X ≥ 9 | p = 1/6). Using B(30, 1/6) tables or calculator, P(X ≥ 9) = 1 − P(X ≤ 8) = 1 − 0.9496 = 0.0504.
步骤3：显著性水平α = 0.10。观测值x = 9。计算p值：P(X ≥ 9 | p = 1/6)。查B(30, 1/6)表或计算器，P(X ≥ 9) = 1 − P(X ≤ 8) = 1 − 0.9496 = 0.0504。
Step 4: Since p‑value = 0.0504 < 0.10, we reject H₀.
步骤4：由于p值 = 0.0504 < 0.10，拒绝H₀。
Step 5: There is sufficient evidence at the 10% significance level to suggest that the die is biased in favour of the number 6.
步骤5：在10%显著性水平下，有足够证据表明骰子偏向出现数字6。
9. Model Answer 2: Normal Distribution Probability | 范文2：正态分布概率
Question: The mass of cereal in a box is normally distributed with mean 500 g and variance 36 g². Boxes weighing less than 492 g are rejected. Find the probability that a randomly selected box is rejected.
问题：一盒谷物质量服从正态分布，均值为500 g，方差为36 g²。质量低于492 g的盒子被拒收。求随机选取一盒被拒收的概率。
Solution: Let X be the mass of a box, X ~ N(500, 36). We need P(X < 492). Standardise: Z = (X − 500)/√36 = (X − 500)/6. Then P(X < 492) = P(Z < (492−500)/6) = P(Z < −1.333...). Using symmetry: P(Z < −1.333) = 1 − Φ(1.333) ≈ 1 − 0.9088 = 0.0912 (from normal table). Therefore, the probability a box is rejected is approximately 0.0912.
解答：设X为盒子质量，X ~ N(500, 36)。需P(X < 492)。标准化：Z = (X − 500)/√36 = (X − 500)/6。则P(X < 492) = P(Z < (492−500)/6) = P(Z < −1.333...)。利用对称性：P(Z < −1.333) = 1 − Φ(1.333) ≈ 1 − 0.9088 = 0.0912（查正态表）。因此，盒子被拒收的概率约为0.0912。
10. Model Answer 3: Permutations with Restrictions | 范文3：带限制的排列
Question: Find the number of different arrangements of the letters of the word ‘STATISTICS’ in
Published by TutorHao | Year 12 统计 Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导