📚 Year 13 WJEC Statistics: In-Depth Analysis of Past Papers | Year 13 WJEC 统计：历年真题深度解析
The WJEC Year 13 Statistics course pushes your understanding of data, probability, and inference to a sophisticated level. A study of past papers reveals not only the recurring topics but also the examiner's expectations for precision, interpretation, and method clarity. This article dissects key areas through the lens of actual exam questions, highlighting common pitfalls and effective strategies to boost your grade.
WJEC 13 年级统计课程将你对数据、概率和推断的理解提升至精密层次。对历年真题的研习不仅能揭示反复出现的主题，也能洞察考官对精确性、解释力和方法清晰度的要求。本文通过实际考题的视角深度剖析核心领域，点明常见陷阱与高效提分策略。
1. Exam Structure and Question Types | 考试结构与题型分析
WJEC A2 Statistics comprises two units: Unit 4 (Statistical Investigations) and Unit 5 (Statistical Inference). Past papers consistently show that Unit 4 emphasises probability distributions, correlation, and sampling, while Unit 5 is heavily weighted towards hypothesis testing, confidence intervals, and chi-squared tests. A typical paper mixes short structured questions with multi-part scenarios, often carrying a final "interpret your result in context" demand.
WJEC A2 统计包含两个单元：单元 4（统计调查）和单元 5（统计推断）。历年试卷一致显示单元 4 侧重概率分布、相关性和抽样，而单元 5 重点考查假设检验、置信区间和卡方检验。典型试卷混合了短小的结构题和多环节的情景题，通常最后一个要求是“在具体情景中解释你的结果”。
Command words such as "State", "Calculate", "Interpret" and "Comment" appear regularly. Examiners expect precise numerical answers to a given degree of accuracy, often three significant figures. A common weakness is failing to provide a contextual conclusion after a statistical test; in many 2022 scripts, candidates correctly computed a test statistic but lost marks by not linking it back to the scenario.
指令词如“陈述”、“计算”、“解释”和“评论”频繁出现。考官希望数值答案精确到指定有效数字，通常是三位有效数字。常见失分点是完成统计检验后未提供情境性结论；在 2022 年多份答卷中，考生正确计算出检验统计量，却因未将其与情境关联而丢分。
2. Discrete Random Variables and Expectation | 离散随机变量与期望
er parts, E(3X-5) or the standard deviation. In a June 2021 paper, a very common error was forgetting that Var(aX+b) = a²Var(X), not aVar(X).
许多真题以离散变量 X 的概率质量函数表开局。常要求计算 E(X)、Var(X)，以及后续部分求 E(3X-5) 或标准差。在 2021 年 6 月试卷中，一个非常普遍的错误是忘记 Var(aX+b) = a²Var(X) 而非 aVar(X)。
Examiners also test the concept of a 'fair game' using expectation. A question in 2020 required candidates to set the price of a ticket so that the expected profit was zero. The algebraic manipulation of E(winnings – cost) often tripped students; the key is always to equate the expected winnings to the cost per game.
考官还通过期望值考查“公平游戏”概念。2020 年一题要求考生为彩票定价，使期望利润为零。对 E(获胜金额 – 成本) 的代数处理常让学生出错；关键始终是令期望获胜金额等于单次游戏成本。
3. Binomial and Poisson Distributions: Key Differences | 二项与泊松分布的关键区别
Candidates often confuse when to apply the binomial versus the Poisson model. Binomial conditions are a fixed number of trials n, constant probability p, and independence. Poisson requires events occurring singly, at a constant average rate λ, and independently in a continuous interval. A 2019 question described flaws in a roll of fabric and asked for an appropriate model. Many incorrectly chose binomial because they thought of each metre as a trial, whereas the continuous length and small defect rate clearly suited a Poisson distribution.
考生常混淆二项与泊松的适用情景。二项条件为：固定的试验次数 n、恒定的概率 p 且试验独立。泊松要求事件单个发生、以恒定平均速率 λ 且在一段连续区域内独立发生。2019 年一题描述一卷布匹上的瑕疵，要求选择合适的模型。许多人错误选择二项，因为他们把每米视为一次试验，而连续的长度与低缺陷率明显适合泊松分布。
n is expected. A typical pitfall is misreading binomial cumulative tables: for P(X ≤ 5). In the 2018 paper, candidates sometimes read the value for p=0.6 instead of p=0.4, highlighting the need for careful navigation of statistical tables.
计算 P(X > k) 常需用补集 P(X ≤ k-1)。在大 n 的二项问题中，需要使用表格或累积分布函数。典型陷阱是误读二项累积表：如求 P(X ≤ 5)。2018 年试卷中，考生偶尔将 p=0.6 的值当成 p=0.4，这凸显了仔细查阅统计表格的必要性。
4. Normal Distribution and Standardisation | 正态分布与标准化
hat appears in nearly every session. Exam questions frequently provide a worded scenario, and candidates must extract μ and σ correctly. In the 2023 Unit 5 paper, a "probability of exceeding 50 kg" ndardisation and then using the symmetry of the normal curve to find the tail probability.
利用 Z = (X - μ) / σ 对正态变量进行标准化是一项几乎每场考试都出现的基本技能。考题常提供文字情景，考生必须正确提取 μ 与 σ。2023 年单元 5 试卷中，一道“超过 50 kg 的概率”题要求标准化，再利用正态曲线的对称性求得尾部概率。
Many past papers also test inverse normal calculations, where you are given a probability and must find the corresponding X-value or a parameter. Here, setting up the correct equation with the standard normal percentage point is crucial. A recurring mistake is forgetting to adjust the sign when the probability is less than 0.5 but on the left side of the mean, leading to an incorrect Z-value.
历年真题还常考查逆正态计算，即给出概率求对应的 X 值或某个参数。此时，建立包含标准正态百分位点的正确方程至关重要。反复出现的错误是：当概率小于 0.5 但位于均值左侧时，忘记调整符号，从而导致错误的 Z 值。
5. Confidence Intervals: Means and Proportions | 置信区间：均值和比例
Constructing a 95% confidence interval for a population mean, typically using the formula x̄ ± z × (σ / √n), is a staple of Unit 5. WJEC often provides the sample mean and standard deviation. In a 2017 question, candidates were asked to calculate the interval and then interpret it. Many scored full marks for computation but wrote "there is a 95% chance the mean lies in this interval" — a misinterpretation. The correct interpretation is that we are 95% confident the interval contains the true mean, or that if we repeated the sampling many times, 95% of such intervals would capture μ.
构建总体均值的 95% 置信区间，通常用公式 x̄ ± z × (σ / √n)，是单元 5 的重头戏。WJEC 常提供样本均值和标准差。2017 年一题中，考生计算区间后需解释其含义。许多人计算满分，却写道“均值有 95% 的可能落在此区间内”——这是误解。正确的解释是：我们有 95% 的信心认为该区间包含了真实均值，或者说如果重复多次抽样，则 95% 的此类区间会捕获 μ。
For proportions, the confidence interval uses p̂ ± z × √[p̂(1-p̂)/n]. Prior to 2020, examiners occasionally asked for Wilson' nt syllabuses focus on the standard Wald interval with the condition that n is large and p̂ not too extreme. Always check that np̂ > 5 and n(1-p̂) > 5 before applying the normal approximation.
对于比例，置信区间用 p̂ ± z × √[p̂(1-p̂)/n]。2020 年前，考官偶尔要求采用威尔逊置信区间，但近年大纲聚焦于标准沃尔德区间，条件是 n 较大且 p̂ 不太极端。在使用正态近似之前，务必检查 np̂ > 5 且 n(1-p̂) > 5。
6. Hypothesis Testing: Critical Values and p-Values | 假设检验：临界值与 p 值
Hypothesis tests in WJEC past papers follow a precise structure: state H₀ and H₁, identify the test statistic, calculate its value, and compare to a critical value or determine the p-value. A 2022 question on a binomial test required finding the critical region for X ~ B(20, 0.3) at a 5% significance level. The most common mistake was constructing a one-tailed region when the question specified a two-tailed alternative.
WJEC 历年真题中的假设检验遵循精确的框架：陈述 H₀ 与 H₁，明确检验统计量，计算其数值，并与临界值比较或确定 p 值。2022 年一道关于二项检验的题目要求找出 X ~ B(20, 0.3) 在 5% 显著性水平下的临界域。最常见的错误是当问题明确要求双尾备择时却构造了单尾域。
and α = 0.05. Although the conclusion was obvious, some candidates failed to state the conclusion in context: "There is sufficient evidence to suggest that the new drug is more effective." Simply writing "reject H₀" lost marks.
p 值考得越来越多。若 p 值小于显著性水平 α，则拒绝 H₀。2021 年一道题给出 p 值 0.041 与 α = 0.05。尽管结论显而易见，部分考生未能在情境中陈述结论：“有足够证据表明新药更有效。”仅写“拒绝 H₀”而丢分。
7. Chi-Squared Tests: Common Pitfalls | 卡方检验：常见陷阱
ust be accurate. Examiners frequently deduct marks for missing the Yates continuity correction in a 2×2 table when total frequency is moderate; however, WJEC guidance now varies, so check the latest specification.
卡方拟合优度与独立性检验定期出现。期望频数的计算和检验统计量 χ² = Σ (O - E)² / E 必须精确。考官常因 2×2 表中总频数中等时遗漏耶茨连续性校正而扣分；不过当前 WJEC 指引有所变化，故务必核对最新规格。
hen used the wrong critical value because they misapplied the significance level. Another typical error is combining rows or columns to avoid small expected frequencies without documenting it, resulting in an unexplained change in degrees of freedom.
2019 年一题考查列联表的自由度。很多人正确写下 (r-1)×(c-1)，却因误用显著性水平而用了错误的临界值。另一典型错误是为避免较小的期望频数而合并行或列却不加说明，导致自由度发生未解释的变化。
8. Correlation and Regression: Interpretation and Limitations | 相关与回归：解释与局限
tistics or calculator. A 2020 question gave Σx, Σy, Σx², Σy², Σxy and required interpreting r = 0.87 in context. Strong positive correlation was noted, but many failed to add that correlation does not imply causation — an essential comment for full marks.
散点图与积矩相关系数 r 高频出现。WJEC 考试期待考生利用汇总统计或计算器算出 r。2020 年一题给出 Σx, Σy, Σx², Σy², Σxy，要求在情境中解释 r = 0.87。大家注意到了强正相关，但许多人未补充“相关不代表因果”——这是满分必需的评论。
. A subtle trap in past papers involves predicting outside the range of the data (extrapolation). In a 2018 paper, predicting the yield for a fertiliser amount well beyond the sampled range was requested, and candidates had to state that the prediction would be unreliable because the relationship might change.
最小二乘回归直线 y = a + bx 也常被要求。历年试卷中的一个微妙陷阱涉及在数据范围之外进行预测（外推）。2018 年试卷中，要求对远超采样范围的肥料用量预测产量，考生必须说明预测不可靠，因为关系可能发生改变。
9. Probability Generating Functions in Exam Questions | 概率生成函数在考题中的应用
PGFs are a distinctive WJEC topic. G(t) = E(tˣ) = Σ tˣ ndependent variables. The 2019 paper had a PGF G(t) = (2+t)²/9, and candidates needed to expand to retrieve the probability distribution — those who remembered that coefficients of tˣ divided by G(1) give probabilities were most successful.
概率生成函数是 WJEC 的一个特色主题。G(t) = E(tˣ) = Σ tˣ P(X=x)。典型题目要求证明给定函数是 PGF，通过求导找出 E(X) 和 Var(X)，或推导独立变量之和的分布。2019 年试卷出现 PGF G(t) = (2+t)²/9，考生需展开以还原概率分布——记住 tˣ 的系数除以 G(1) 即得概率的考生最为成功。
Using G'(1) = E(X) and G'' (X-1)) to find variance is a standard routine. A common slip is forgetting that Var(X) = G''(1) + G'(1) - [G'(1)]², not simply G'' past mark schemes, the final A mark was awarded only if the variance was fully simplified.
利用 G'(1) = E(X) 和 G''(1) = E(X(X-1)) 求方差是常规程序。常见疏忽是忘记 Var(X) = G''(1) + G'(1) - [G'(1)]²，而非简单等于 G''(1)。在多次阅卷标准中，最终的 A 分仅在方差完全化简后给出。
10. Sampling and Data Collection Methods | 抽样与数据收集方法
Although often thought of as a lower-year topic, sampling methodology underpins many higher-mark questions. Stratified, cluster, systematic, and simple random sampling each have advantages and biases. A 2021 question presented a scenario of surveying hospital patients: stratified sampling by ward was appropriate, but many suggested convenience sampling, which the mark scheme penalised for bias.
虽然常被视为低年级主题，但抽样方法论是许多高分题的根基。分层、整群、系统与简单随机抽样各有优缺点和偏倚。2021 年一题呈现了调查医院患者的情景：按病房分层抽样合适，但许多人建议便利抽样，评分方案因其偏倚而扣分。
ion: M marked, C caught in second sample, R recaptures. Candidates who confused M and C usually lost all marks. Examiners also expected an assumption that marked individuals mix evenly — a point often omitted.
2020 年出现了用捕获-再捕获法估计种群规模的题目。公式 N = (M × C) / R 的解读需谨慎：M 为标记数，C 为第二次捕获数，R 为再捕获数。混淆 M 与 C 的考生通常全部丢分。考官还期望写出标记个体均匀混合的假设——这一点常被遗漏。
11. Tips from Examiner Reports | 考官报告中的提分技巧
Examiner reports across multiple series repeat the same advice. Always show your working; many marks are awarded for correct method even if the final numeric answer is wrong. In 2022, a candidate obtained 3.16 instead of 3.18 due to premature rounding, but still earned method marks because the steps were clear.
多份考官报告重复着同样的建议。始终展示解答过程；即便最终数值答案错误，许多分数会因正确方法而给予。2022 年，一位考生因过早舍入得到 3.16 而非 3.18，但仍因步骤清晰获得了方法分。
Time management is critical. Past papers indicate that Unit 5 has 75 marks in 1 hour 45 minutes, meaning roughly 1.4 minutes per mark. Practice under timed conditions and learn which statistical tables you are allowed. Several reports highlight candidates wasting time deriving formulae that were given in the formula booklet. Know what is provided.
时间管理至关重要。历年试卷显示单元 5 共 75 分，考试时间 1 小时 45 分钟，意味着每分约 1.4 分钟。在限时条件下练习，并熟知允许携带的统计表格。多份报告强调，考生浪费时间推导公式小册子中已有的公式。务必了解哪些公式已提供。
12. Handling Proof and Derivation Questions | 应对证明与推导题
WJEC occasionally includes short proof-style questions, for instance, showing that Σ(x - x̄)² = Σx² - (Σx)²/n. These are worth 3– rd, but the mark scheme shows that a few lines of correct algebra earn substantial credit.
WJEC 偶尔会包含简短的证明型题目，例如证明 Σ(x - x̄)² = Σx² - (Σx)²/n。这些题目值 3–5 分，奖励代数流畅性。许多考生跳过它们，认为太难，但阅卷标准显示仅需几行正确代数即可获得可观分数。
E(X) = np. The key is to treat q as (1-p) and use the chain rule. Practice these derivations; they are often the discriminator for top grades.
从 PGF 推导二项分布的期望与方差曾出现过。若 G(t) = (q + pt)ⁿ，求导并代入 t=1 得到 E(X) = np。关键是将 q 视为 (1-p) 并使用链式法则。勤练这类推导；它们常是区分顶尖成绩的关键。
Published by TutorHao | 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