📚 AS Cambridge Statistics: Answering Techniques and Marking Criteria | 剑桥AS统计：答题技巧与评分标准
Understanding how examiners allocate marks is just as important as knowing the formulas. This guide breaks down the most common question types in the Cambridge AS Statistics paper (Paper 5 for the 9709 Mathematics syllabus), explains what earns method marks, accuracy marks, and answer marks, and provides proven strategies to avoid losing points on representation, probability, and hypothesis tests. Every section presents the key insight in English first, followed by its Chinese equivalent, so you can reinforce your learning in both languages.
了解考官如何分配分数与掌握公式同等重要。本指南拆解了剑桥AS统计试卷（9709数学大纲中的Paper 5）中最常见的题型，解释如何获得方法分、准确度分和答案分，并提供行之有效的策略以避免在图表表示、概率和假设检验中丢分。每个要点先以英文呈现，再以中文呈现，帮助你用双语巩固所学内容。
1. Understanding the Paper Structure and Mark Allocation | 理解试卷结构与分值分配
The Statistics 1 paper (S1) lasts 1 hour 15 minutes and carries 50 marks, typically comprising 6 to 7 questions. Marks are awarded for method (M), accuracy (A), and answer (B) without working shown. M marks require a clear correct process even if an arithmetic slip happens later. A marks are only given if the preceding M mark has been earned and the numerical result is accurate. Many students lose A marks by not preserving intermediate values to at least four significant figures. Knowing this helps you decide when to show full working and when a rounded final answer is acceptable.
统计1（S1）试卷考试时长为1小时15分钟，满分50分，通常包括6至7道题。分数分为方法分（M分）、准确度分（A分）和无需展示步骤的答案分（B分）。M分要求清晰的正确过程，即使后续出现计算错误也能获得。A分只有在已获得前置M分且数值结果准确时才给予。许多学生因没有将中间值保留至少四位有效数字而丢失A分。了解这一点有助于你判断何时展示完整步骤，以及何时可以接受舍入的最终答案。
2. Representing Data: Box-and-Whisker Plots and Histograms | 数据表示：箱线图与直方图
When drawing a box-and-whisker plot, the examiner looks for a linear scale and five key values: minimum, lower quartile (Q₁), median (Q₂), upper quartile (Q₃), and maximum. The median line must be drawn inside the box, not on its edge. A common mistake is plotting outliers as whisker ends; instead, mark outliers with crosses (×) and end whiskers at the nearest value that is not an outlier. Outliers are defined as more than 1.5 × IQR beyond Q₁ or Q₃. The Chinese version: 绘制箱线图时，考官要求线性的标度和五个关键值：最小值、下四分位数(Q₁)、中位数(Q₂)、上四分位数(Q₃)和最大值。中位线必须画在箱体内部，而不是边缘。常见错误是将异常值当作须的端点；应用叉号(×)标记异常值，而须端点在最近的非异常值处。异常值定义为超出Q₁或Q₃的距离超过1.5倍四分位距(IQR)。
For histograms, the frequency density = frequency ÷ class width. Always label the vertical axis as ‘Frequency density’. The area of each bar is proportional to the frequency, not the height. If classes are unequal, a common pitfall is to plot frequency against height, which gives a misleading picture. Use a ruler and draw sharp, clear lines. Scales must be uniform and correctly labelled with units. A mark is often reserved for correct labelling of both axes. 对于直方图，频率密度 = 频率 ÷ 组距。纵轴务必标注为“频率密度”。每个矩形的面积（而非高度）与频率成正比。若各组距不相等，常见陷阱是直接用频率作为高度作图，这会误导结果。务必使用直尺，画出清晰尖锐的线条。标度必须均匀且正确标注单位。正确标注两条坐标轴通常能获得1分。
3. Measures of Central Tendency and Spread | 集中趋势与离差的度量
To calculate mean and variance from grouped data, always use the midpoints of intervals. The formula for the mean is Σfx / Σf, and for variance use (Σfx² / Σf) − (mean)². The examiner awards M1 for substituting into any correct formula and A1 for the correct numerical result. Never round midpoints prematurely. If class boundaries are given as, for example, 10–, 20–, 30–, the midpoints are 15, 25, 35 unless the data are continuous in a different way. Check carefully the wording: ’10–19′ implies a class width of 10 with midpoint 14.5 or 15 depending on the convention clarified in the question. 对于分组数据，计算均值与方差时务必使用组中值。均值公式为 Σfx / Σf，方差公式为 (Σfx² / Σf) − (均值)²。考官对代入任何正确公式给予M1分，对正确的数值结果给予A1分。切勿过早舍入组中值。若组边界以10–、20–、30–等形式给出，除非数据在别处有不同说明，组中值应为15、25、35。仔细检查措辞：“10–19”表示组距为10，组中值可能是14.5或15，需根据题目说明确定惯例。
When using coded data, you must uncode the mean and standard deviation correctly. If coding was Y = (X − a)/b, then X mean = a + b × Y mean, and Sx = |b| × Sy. A very common loss of A marks arises from forgetting to multiply the standard deviation by b. 使用编码数据时，必须正确反编码均值和标准差。若编码为 Y = (X − a)/b，则 X 均值 = a + b × Y 均值，Sx = |b| × Sy。一个极其常见的丢分点是忘记将标准差乘以 b。
4. Probability and Set Notation | 概率与集合符号
Always translate the question into precise set notation before solving. For ‘and’ use ∩ (intersection), for ‘or’ use ∪ (union). The addition rule P(A ∪ B) = P(A) + P(B) − P(A ∩ B) is examined frequently. Examiner expects the formula to be stated or clearly applied; method mark M1 is given for correct structure. If the question gives a Venn diagram with unknown regions, label them algebraically (x, y, z) and sum to 1 to form an equation. State the equation and solve. 在解题前，务必将题目转化为精确的集合符号。“且”使用 ∩（交集），“或”使用 ∪（并集）。加法公式 P(A ∪ B) = P(A) + P(B) − P(A ∩ B) 经常考查。考官期望明确写出该公式或清晰应用；正确结构给予方法分M1。若题目给出带有未知区域的维恩图，用代数符号（x, y, z）标记各个区域，并使其总和为1以建立方程。明确写出方程并求解。
Conditional probability P(A | B) = P(A ∩ B) / P(B). The denominator is the probability of the given event B, not the whole sample space. Many candidates use P(A ∩ B) / P(A) by mistake. To avoid this, highlight the ‘given that’ event B and write down P(B) first. For independent events, P(A ∩ B) = P(A)P(B), but be careful: independence is often tested in reverse, asking you to decide whether two events are independent by checking if P(A ∩ B) = P(A) × P(B). 条件概率 P(A | B) = P(A ∩ B) / P(B)。分母是给定事件B的概率，而非整个样本空间。许多考生误用了 P(A ∩ B) / P(A)。为避免此错误，先高亮“给定”事件B，并写下 P(B)。对于独立事件，P(A ∩ B) = P(A)P(B)，但需注意：独立性经常反向考查，要求你通过验证 P(A ∩ B) = P(A) × P(B) 来判断两个事件是否独立。
5. Discrete Random Variables and Expectation Algebra | 离散随机变量与期望代数
For a discrete random variable X, the sum of all probabilities must equal 1. This is often the first step in finding an unknown constant k. Always write Σ P(X = x) = 1 and solve. Marks are given for forming the equation (M1) and for the correct k (A1). E(X) is Σ x·p, and Var(X) = E(X²) − [E(X)]², where E(X²) = Σ x²·p. Keep full precision for E(X²) to avoid losing accuracy in variance. 对于离散随机变量 X，所有概率之和必须等于1。这通常是求未知常数 k 的第一步。务必写出 Σ P(X = x) = 1 并求解。建立方程得M1分，正确的 k 值得A1分。E(X) = Σ x·p，Var(X) = E(X²) − [E(X)]²，其中 E(X²) = Σ x²·p。在使用 E(X²) 时保持全精度，以避免方差计算中的精度损失。
Expectation algebra is frequently tested: E(aX + b) = aE(X) + b, Var(aX + b) = a² Var(X). Notice that adding b does not affect variance. For a sum or difference of independent variables, E(X ± Y) = E(X) ± E(Y) and Var(X ± Y) = Var(X) + Var(Y). The sign of the difference does not change the variance: it’s always a plus when variables are independent. Examiners set common traps by asking for Var(X − Y) and students incorrectly subtract the variances. 期望代数经常考查：E(aX + b) = aE(X) + b，Var(aX + b) = a² Var(X)。注意增加常数 b 不影响方差。对于独立变量的和或差，E(X ± Y) = E(X) ± E(Y)，以及 Var(X ± Y) = Var(X) + Var(Y)。差异的符号不改变方差：只要变量独立，方差始终相加。考官常设计陷阱，要求学生计算 Var(X − Y)，而学生错误地减去方差。
6. Binomial Distribution: Conditions and Calculations | 二项分布：条件与计算
A binomial distribution is valid if there are a fixed number n of independent trials, each with two outcomes (success/failure), with constant probability of success p. Always state these conditions briefly but clearly in words: ‘fixed number of trials’, ‘independent’, ‘constant probability’, ‘two outcomes’. A typical question asks to explain why a variable can be modelled by a binomial distribution; the marks are for stating at least two conditions correctly. 二项分布有效的条件是：固定试验次数 n，每次试验独立，每次只有两种结果（成功/失败），且成功的概率 p 恒定。务必用语言简要而清晰地陈述：“固定试验次数”、“独立”、“恒定概率”、“两种结果”。典型题目会要求解释为什么某个变量可用二项分布建模；分数通常来自正确陈述至少两个条件。
For probabilities, use the formula P(X = r) = ⁿCᵣ pʳ qⁿ⁻ʳ where q = 1 − p. In the exam, you may use the calculator’s binomial probability function, but you must show the parameters n and p, and the value of r. Simply writing an answer from a calculator with no indication of the distribution may lose method marks. It is good practice to write X ~ B(n, p) then state P(X = r) or P(X ≤ r). For cumulative probabilities, clearly indicate ‘using Bcd’ or ‘from calculator’. 对于概率，使用公式 P(X = r) = ⁿCᵣ pʳ qⁿ⁻ʳ，其中 q = 1 − p。考试中可以使用计算器的二项概率功能，但必须展示参数 n 和 p，以及 r 的值。直接从计算器得出答案而无任何分布说明，可能会失去方法分。良好的做法是写出 X ~ B(n, p)，然后陈述 P(X = r) 或 P(X ≤ r)。对于累积概率，明确注明“使用 Bcd”或“由计算器得”。
7. Normal Distribution: Standardisation and Inverse | 正态分布：标准化与逆运算
When approximating a binomial with a normal, continuity correction is necessary. If X ~ B(n, p) is approximated by Y ~ N(np, npq), then P(X ≥ a) is approximated by P(Y > a − 0.5). Similarly, P(X ≤ a) uses P(Y < a + 0.5). Write the correction explicitly; one mark is specifically for the proper continuity correction. Always check that np and nq are both > 5 before approximating. 当用正态分布近似二项分布时，必须进行连续性校正。若 X ~ B(n, p) 用 Y ~ N(np, npq) 近似，那么 P(X ≥ a) 近似为 P(Y > a − 0.5)。类似地，P(X ≤ a) 使用 P(Y < a + 0.5)。明确写出校正项；通常有1分专门用于正确的连续性校正。在近似之前，务必检查 np 与 nq 是否都大于5。
Standardisation formula: z = (x − μ) / σ. In reverse normal problems where you are given a probability and asked to find the mean or standard deviation, always set up an equation using the inverse normal. For example, P(X > 12) = 0.2 becomes P(Z > (12 − μ)/σ) = 0.2. Then find the z-value, usually z = 0.8416 for a right-tail of 0.2. The examiner expects to see the equation linking z, μ, and σ, not just a calculator command. Standardisation is the most heavily weighted skill in the normal distribution topic. 标准化公式：z = (x − μ) / σ。在已知概率、需逆向求均值或标准差的问题中，务必利用逆正态建立方程。例如，P(X > 12) = 0.2 转化为 P(Z > (12 − μ)/σ) = 0.2。然后找出 z 值，对于右尾面积0.2，通常 z = 0.8416。考官期望看到连接 z、μ 和 σ 的方程，而不仅仅是计算器指令。标准化是正态分布主题中权重最高的技能。
8. Hypothesis Testing: Correct Structure Wins Marks | 假设检验：正确结构赢取分数
In AS Statistics, hypothesis tests are usually one-tailed or two-tailed binomial tests. The five-step structure is essential: (1) Define the null hypothesis H₀: p = [value], and alternative H₁: p < or p > or p ≠ [value]. (2) State the significance level, typically 5% or 10%. (3) Calculate the test statistic, usually P(X ≤ observed) or P(X ≥ observed) under H₀. (4) Compare this probability with the significance level. (5) Conclude in context: either ‘reject H₀’ if p-value < significance level, else 'do not reject H₀'. The conclusion must be written in words referring to the original claim, not just 'reject H₀'. 在AS统计中，假设检验通常是单尾或双尾的二项分布检验。五步结构至关重要：(1) 定义原假设 H₀: p = [值]，以及备择假设 H₁: p < 或 p > 或 p ≠ [值]。(2) 陈述显著性水平，通常为5%或10%。(3) 计算检验统计量，通常是在 H₀ 下计算 P(X ≤ 观测值) 或 P(X ≥ 观测值)。(4) 将此概率与显著性水平比较。(5) 在上下文中得出结论：若 p 值 < 显著性水平，则“拒绝 H₀”，否则“不拒绝 H₀”。结论必须用文字针对原声明进行表述，而不只是“拒绝 H₀”。
For two-tailed tests, compare the probability of the observed extreme with half the significance level, or double the one-tail probability and compare with the full significance level. Many students forget to halve or double and lose the final A mark. A clear statement such as ‘2 × P(X ≤ 3) = 0.048 < 0.05, so reject H₀' shows the method correctly. Also, always define the probability p in words before starting the test, e.g., 'p is the proportion of defective items'. 对于双尾检验，将观测到的极端概率与显著性水平的一半比较，或将单尾概率加倍后与完整的显著性水平比较。许多学生忘记折半或加倍，因此丢失最后的A分。清晰陈述例如“2 × P(X ≤ 3) = 0.048 < 0.05，因此拒绝 H₀”能正确展示方法。此外，在进行检验之前，务必用文字定义概率 p，例如“p 为缺陷品的比例”。
9. Permutations and Combinations: Selecting the Right Tool | 排列与组合：选择正确的工具
The majority of marks in permutation and combination questions are method marks for recognising whether order matters. If order matters, use permutations (ⁿPᵣ); if only selection matters, use combinations (ⁿCᵣ). For arrangements with identical items, use n!/(a!b!…). Always write the formula in numbers before calculating; this shows the examiner your reasoning. A phrasing like ‘number of ways to choose a committee of 3 from 10’ usually requires combination, while ‘arranging 5 books on a shelf’ is permutation. 排列组合题目中，绝大多数分数是方法分，用于识别顺序是否重要。若顺序重要，使用排列（ⁿPᵣ）；若仅与选择有关，则使用组合（ⁿCᵣ）。对于含有相同物品的排列，使用 n!/(a!b!…)。在计算前，务必用数字写出公式；这向考官展示了你的推理过程。诸如“从10人中选出3人委员会的方法数”通常需要组合，而“将5本书排列在书架上”则是排列。
Examiners set tricky multi-stage problems: e.g., selecting a team with restrictions such as ‘at least one woman’. The best approach is to split into mutually exclusive cases, compute each case with combinations, and then add the results. Avoid using complement unless you are sure about the total unrestricted selections. Show the cases clearly: Case 1: 1 woman and 2 men, Case 2: 2 women and 1 man, etc. This structured working secures M1 for method and A1 for each correct case, even if the final total is slightly wrong. 考官常设计多阶段难题：例如，选择团队并带有“至少有一名女性”的限制。最佳方法是拆分成互斥的情况，用组合分别计算每种情况，然后相加。除非你对无限制的总选择数非常确定，避免使用补集。清晰地展示各种情况：情况1：1女2男，情况2：2女1男，等等。这种结构化的步骤为每种正确情况赢得M1和A1分，即使最终总计数略有错误。
10. Geometric Distribution (if included) | 几何分布（若包含）
Some AS syllabi include the geometric distribution. If X ~ Geo(p) models the number of trials up to and including the first success, then P(X = x) = qˣ⁻¹ p, where q = 1 − p. Expected value E(X) = 1/p. The mode is always 1. A common exam question asks for P(X > x) which equals qˣ. This elegant shortcut is often worth a B mark. Justify it simply: ‘probability that first x trials are all failures’. 某些AS大纲包含几何分布。若 X ~ Geo(p) 模拟直至并包含首次成功的试验次数，则 P(X = x) = qˣ⁻¹ p，其中 q = 1 − p。期望值 E(X) = 1/p。众数总是1。常见考题要求计算 P(X > x)，它等于 qˣ。这个简洁的快捷公式通常值一个B分。只需简单说明：“前 x 次试验均为失败的概率”。
11. Graphical Displays: Cumulative Frequency and Scatter Diagrams | 图形展示：累积频率与散点图
When drawing a cumulative frequency curve, plot the points at the upper class boundary against cumulative frequency. The first point is at the lower boundary of the first class with cumulative frequency 0. Join points with a smooth curve, not straight lines. To estimate median and quartiles, draw horizontal lines from the frequency axis and then drop vertical lines to the data axis. Mark these crossing points clearly. Even if the curve is slightly skewed, marks are awarded for correct construction and reading, not artistic beauty. 绘制累积频率曲线时，在组上限边界绘制累积频率点。起点位于第一组的组下限，累积频率为0。用光滑曲线连接各点，而非直线。为估计中位数和四分位数，从频率轴画水平线，再垂直到达数据轴。清晰地标出这些交叉点。即使曲线略有歪斜，只要构建与读数正确，依然能得到分数，与画得是否精美无关。
For scatter diagrams, plot given data points accurately with small crosses (×). Never join the points. Draw a line of best fit by ‘eye’, ensuring roughly equal numbers of points above and below the line, and pass it through the mean point (x̅, y̅). The equation of the regression line is often obtained from calculator, but you need to write the equation in the form y = a + bx and give a and b to a stated degree of accuracy, usually three significant figures. Using the line for prediction outside the data range (extrapolation) is unreliable; examiners may ask for a comment to test understanding. 对于散点图，用小叉号（×）准确绘制数据点。切勿连接这些点。通过“目测”画出最佳拟合线，确保线上方和线下方的点数大致相等，并使其穿过均值点 (x̅, y̅)。回归线的方程通常由计算器得出，但需要写成 y = a + bx 的形式，并给出 a 和 b 至一定精度，通常是三位有效数字。在数据范围之外使用该线进行预测（外推）是不可靠的；考官可能要求对此发表评论以检验理解。
12. Final Minutes: Checking and Presentation | 最后几分钟：检查与卷面呈现
Allocate the last 5–8 minutes to review. First, verify all probabilities are between 0 and 1, and that discrete probabilities sum to 1. Check that you have defined any variables or probability symbols you introduced. For hypothesis tests, ensure the conclusion matches the decision. If you rejected H₀, the wording must say ‘there is sufficient evidence to suggest that…’ rather than ‘prove’. If you did not reject, say ‘there is insufficient evidence’. The word ‘prove’ is never acceptable in statistical conclusions. 预留最后5–8分钟进行检查。首先，核实所有概率值介于0和1之间，且离散概率总和为1。检查是否定义了你所引入的任何变量或概率符号。对于假设检验，确保结论与决定一致。若你拒绝了 H₀，措辞必须是“有足够证据表明……”，而非“证明”。若未拒绝，则应说“没有足够证据”。在统计结论中，“证明”一词绝不可接受。
Presentation also matters: number your steps, separate parts clearly, and box final answers. When a question says ‘state an assumption’, write a short phrase like ‘the sample is random’ or ‘items are independent’. Never leave an assumption blank; even a reasonable guess can earn the mark. Finally, check your calculator settings: are you in degree or radian? For statistics, the mode should be in normal or stat mode, and clear old data from lists to avoid carry-over errors. 卷面呈现也很重要：为步骤编号，清晰分隔各部分，并用方框标出最终答案。当题目要求“陈述一条假设”时，写出简短短语，如“样本是随机的”或“物品相互独立”。绝不要把假设留空；即使是合理的猜测也可能得分。最后，检查计算器设置：是角度制还是弧度制？对于统计，模式应在常规或统计模式，并清除列表中的旧数据以避免残留错误。
Published by TutorHao | Statistics 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Cancel reply