Pre-U CAIE Statistics: High-Frequency Topics and Common Mistakes Analysis | Pre-U CAIE 统计：高频考点与易错题分析

📚 Pre-U CAIE Statistics: High-Frequency Topics and Common Mistakes Analysis | Pre-U CAIE 统计：高频考点与易错题分析

Cambridge Pre-U Statistics (9795) challenges students with a blend of pure statistical theory and applied data analysis. Success demands not only fluency in probability distributions and inference but also vigilance against subtle conceptual errors. This article dissects the most frequently examined topics and the mistakes that repeatedly cost candidates marks, equipping you with the clarity needed for top-tier performance.

剑桥 Pre-U 统计 (9795) 融合了纯统计理论与应用数据分析，对考生提出了很高的要求。要想取得好成绩，不仅要熟练掌握概率分布和推断方法，还要警惕那些细微的概念性错误。本文拆解最高频的考点以及导致考生反复失分的错误，帮助你理清思路，达到顶尖水平。

---

1. Core Probability Distributions | 核心概率分布

The binomial distribution B(n, p) models the number of successes in n independent trials with constant probability p. Its probability mass function is P(X = r) = ⁿCᵣ pʳ (1-p)ⁿ⁻ʳ. A common pitfall is failing to verify the independence and constant probability conditions before applying the binomial model; exam questions often contain subtle dependencies that invalidate it.

二项分布 B(n, p) 用于描述 n 次独立重复试验中成功次数，每次成功概率 p 恒定。其概率质量函数为 P(X = r) = ⁿCᵣ pʳ (1-p)ⁿ⁻ʳ。常见错误是在未验证独立性和恒定概率条件时就贸然套用二项模型；考题中常包含微妙的关联性，使二项分布不再适用。

Poisson distributions Po(λ) model rare events occurring randomly and independently at a constant average rate λ. A typical mistake is using Poisson when events cluster or when the rate is not constant over time or space. Students also frequently misread units, for example using a rate per hour when the question works in minutes.

泊松分布 Po(λ) 用于模拟以恒定平均发生率 λ 随机且独立发生的稀有事件。典型错误是当事件存在聚集现象，或发生率随时间/空间变化时仍然使用泊松分布。此外，考生经常误读单位，比如题目以分钟为单位，却使用了以小时为单位的率。

For a normal distribution N(μ, σ²), standardisation Z = (X − μ)/σ is fundamental. Many candidates lose marks by misreading the continuity of the variable: P(X > 120) is not the same as P(X ≥ 120) for a continuous distribution, but confusing this with discrete settings leads to unnecessary continuity corrections or omission of them when needed.

对于正态分布 N(μ, σ²)，标准化 Z = (X − μ)/σ 是基础操作。许多考生因对变量连续性理解不清而失分：对于连续分布，P(X > 120) 与 P(X ≥ 120) 等价，但将其与离散情形混淆会导致错误地引入或省略连续性校正。

---

2. Approximations and the Continuity Correction | 近似方法与连续性校正

Binomial to Poisson approximation works well when n is large and p is small, with λ = np. Many students apply it without checking that n > 50 and np < 5, or that p is indeed small. Similarly, normal approximations to binomial or Poisson require continuity corrections, but candidates often forget to adjust the boundary by 0.5.

二项分布近似为泊松分布时，要求 n 大 p 小，且 λ = np。许多学生未检查 n > 50 且 np < 5 的条件，或者并未确认 p 是否真的足够小。类似地，二项或泊松分布的正态近似需要连续性校正，但考生常忘记将边界值调整 0.5。

For a binomial random variable X ~ B(n, p) approximated by a normal Y ~ N(np, np(1-p)), the correct approximation is P(X ≥ a) ≈ P(Y > a − 0.5). A frequent mistake is using P(Y ≥ a) directly, which can lead to inaccurate p-values in hypothesis testing and lost accuracy marks.

对于二项随机变量 X ~ B(n, p) 用正态 Y ~ N(np, np(1-p)) 近似时，正确的近似是 P(X ≥ a) ≈ P(Y > a − 0.5)。常见错误是直接使用 P(Y ≥ a)，这会在假设检验中导致不准确的 p 值，损失精度分。

Poisson to normal approximation requires λ > 10. When applying continuity correction here, use P(X ≤ k) ≈ P(Y < k + 0.5). Candidates who skip this step often fall into the trap of unadjusted tail probabilities, especially when the test statistic sits near a critical value.

泊松分布正态近似要求 λ > 10。此处应用连续性校正时，应使用 P(X ≤ k) ≈ P(Y < k + 0.5)。省略这一步的考生常常掉入未调整尾部概率的陷阱，特别是当检验统计量接近临界值时。

---

3. Hypothesis Testing Framework | 假设检验框架

A clear statement of the null hypothesis H₀ and alternative hypothesis H₁ is essential. A common error is writing H₁ with a strict inequality when a two-tailed test is required, or vice versa. Always match the direction of the test to the wording of the question; “increased” suggests an upper-tail test, while “changed” indicates two-tailed.

清晰地陈述原假设 H₀ 和备择假设 H₁ 至关重要。常见错误是在需要双尾检验时，H₁ 却写成了严格不等式，或反之。始终将检验方向与题目措辞匹配；”increased” 提示上尾检验，而 “changed” 则应使用双尾检验。

When using p-values, reject H₀ if p < α. Many candidates compare the p-value directly to the test statistic or confuse the rejection region with the p-value. In Pre-U, showing both the critical region and the p-value is good practice, but they must be consistent.

使用 p 值时，若 p < α 则拒绝原假设。许多考生将 p 值直接与检验统计量比较，或者混淆拒绝域与 p 值。在 Pre-U 考试中，同时展示临界区域和 p 值是良好习惯，但两者必须保持一致。

For binomial and Poisson tests, finding the exact p-value can require summing probabilities in the appropriate tail. A frequent mistake is summing only the probability of the observed value rather than all values at least as extreme. In two-tailed tests, candidates often mistakenly double the one-tailed p-value without checking the symmetry of the distribution.

对于二项和泊松检验，计算精确 p 值需要对相应尾部概率求和。常见错误是只对观测值的概率求和，而忽略了所有更极端的情形。在双尾检验中，考生经常不检查分布的对称性就直接将单尾 p 值乘以 2。

---

4. Type I and Type II Errors | 第一类和第二类错误

A Type I error occurs when H₀ is true but rejected; its probability is the significance level α. A Type II error happens when H₀ is false but not rejected, with probability β. Many candidates struggle to articulate these concepts in context and mistakenly think that decreasing α always improves the test, without realising that it increases β for a fixed sample size.

第一类错误发生在原假设为真却被拒绝时，其概率为显著性水平 α。第二类错误发生在原假设为假却未被拒绝时，概率为 β。许多考生难以在具体情境中解释这两个概念，并错误地认为减小 α 总能使检验更优，却没有意识到在固定样本量下这会增大 β。

The power of a test is 1 − β, the probability of correctly rejecting a false H₀. Pre-U questions frequently ask about the impact of sample size on power. A larger sample size reduces standard error, making it easier to detect a given effect, thus increasing power. Candidates often forget to mention that it also narrows the confidence interval.

检验的功效为 1 − β，即正确拒绝错误原假设的概率。Pre-U 试题常问及样本量对功效的影响。更大的样本量会减小标准误，更容易检测到给定效应，从而提升功效。考生常常忘记提及样本量同时也会缩窄置信区间。

---

5. Confidence Intervals | 置信区间

A 95% confidence interval for the population mean μ when σ is known is x̄ ± z₀.₀₂₅ × σ/√n. When σ is unknown and estimated by s, the t distribution with n−1 degrees of freedom must be used. Students frequently apply z intervals in all situations, ignoring the t distribution entirely, which causes inaccuracies in small samples.

当总体标准差 σ 已知时，μ 的 95% 置信区间为 x̄ ± z₀.₀₂₅ × σ/√n。当 σ 未知用 s 估计时，必须使用自由度为 n−1 的 t 分布。学生经常在所有情况下都用 z 区间，完全忽略 t 分布，这在小样本下会产生严重误差。

A common misinterpretation is that a 95% confidence interval means there is a 95% probability that the true parameter lies within the computed interval. Instead, 95% refers to the long-run proportion of such intervals that capture the parameter. Candidates often repeat this probabilistic phrasing in exam answers and lose marks.

一个常见的错误理解是认为 95% 置信区间表示真参数落在该计算区间内的概率为 95%。而正确的理解是：95% 指的是长期重复抽样下，由该方法构造的区间能捕获该参数的比例。考生常在答案中重复这种概率表述而失分。

For proportions, the confidence interval is p̂ ± z* × √[p̂(1−p̂)/n]. A careless mistake is forgetting the condition that np̂ > 5 and n(1−p̂) > 5 for the normal approximation to be valid, or using p̂ in the standard error formula without checking if it is based on a sufficiently large sample.

对于比例，置信区间为 p̂ ± z* × √[p̂(1−p̂)/n]。一个粗心错误是忘记正态近似有效的条件 np̂ > 5 和 n(1−p̂) > 5，或者在没有检验样本量是否足够的情况下就使用 p̂ 计算标准误。

---

6. Chi-Squared Tests | 卡方检验

In goodness-of-fit tests, the test statistic is χ² = Σ (O − E)² / E, where O are observed frequencies and E are expected frequencies under H₀. A crucial condition is that all expected frequencies should be at least 5; if not, categories must be combined. Candidates often overlook this, leading to distorted p-values and invalid conclusions.

在拟合优度检验中，检验统计量为 χ² = Σ (O − E)² / E，其中 O 为观测频数，E 为在原假设下的期望频数。一个关键条件是所有期望频数至少为 5；否则必须合并类别。考生常忽略这一点，导致 p 值扭曲、结论无效。

For tests of association in contingency tables, the expected values are computed as (row total × column total) / grand total. A widespread mistake is calculating degrees of freedom incorrectly: for an r × c table, df = (r−1)(c−1). Using the wrong df can lead to selecting an incorrect critical value from the chi-squared table.

对于列联表独立性检验，期望值按 (行合计 × 列合计) / 总计 计算。一个普遍错误是自由度计算错误：对于 r × c 表格，df = (r−1)(c−1)。使用错误的自由度会导致从卡方表中选取不正确的临界值。

Students sometimes treat percentages or proportions as frequencies in these tests. Chi-squared tests require actual counts, not relative frequencies. Entering percentages into the formula will produce a grossly deflated statistic and an erroneous acceptance of H₀.

学生有时会把百分比或比例当作频数代入检验。卡方检验需要的是实际计数，而非相对频率。将百分比代入公式会得出严重缩小的统计量，并错误地接受原假设。

---

7. Correlation and Linear Regression | 相关与线性回归

Pearson’s product-moment correlation coefficient r measures linear association. A high |r| close to 1 indicates strong linear correlation, but a low r does not necessarily mean no relationship; the variables could be strongly related in a non-linear way. This is a classic exam pitfall.

皮尔逊积矩相关系数 r 衡量线性关联的强弱。接近 1 的 |r| 表示强线性相关，但低 r 值并不一定意味着没有关系；变量间可能存在强烈的非线性关系。这是典型的考试陷阱。

The least-squares regression line of y on x is y = a + bx, where b = Sxy / Sxx and a = ȳ − b x̄. Candidates often mix up the dependent and independent variables, fitting x on y when y on x is required. Misidentification leads to a completely different line and invalid predictions.

最小二乘回归线 (y 对 x) 为 y = a + bx，其中 b = Sxy / Sxx，a = ȳ − b x̄。考生常混淆因变量和自变量，在需要 y 对 x 的回归时却拟合了 x 对 y 的回归，导致完全不同的直线和无效的预测。

Extrapolation involves using the regression line to predict y for an x-value outside the range of the original data. This practice is unreliable because the linear relationship may not hold beyond the observed span. Pre-U examiners frequently penalise candidates who blanketly extend predictions without acknowledging this risk.

外推是指用回归线预测原始数据范围之外的 x 所对应的 y 值。这种做法不可靠，因为线性关系在观测范围之外可能不再成立。Pre-U 考官经常处罚那些不加说明就随意外推的考生。

---

8. Non-Parametric Tests | 非参数检验

The Wilcoxon signed-rank test is used for paired data or a single sample when the assumption of normality is questionable. The test statistic T is the smaller sum of the ranks of positive or negative differences. A frequent error is mishandling zero differences: these pairs should be discarded, and the sample size reduced accordingly, but many students keep them and assign arbitrary ranks.

威尔科克森符号秩检验用于配对数据或单样本，当正态性假设不满足时尤为适用。检验统计量 T 是正差或负差秩和中的较小者。常见错误是零差值处理不当：这些配对应被剔除并相应减少样本量，但许多学生保留它们并随意赋秩。

When performing the test, the differences are ranked in order of absolute value. Using raw values rather than absolute values for ranking is a typical slip. Also, after ranking, the signs must be reattached to the ranks before summing; omitting this step invalidates the test.

执行检验时，先按差值的绝对值排序。典型失误是用原始值而非绝对值来排序。此外，排完秩后必须将正负号重新赋予秩值再求和；省去这一步将使检验无效。

Pre-U also examines the Wilcoxon rank-sum (Mann-Whitney) test for two independent samples. Candidates often confuse which test is appropriate, applying the signed-rank test to independent data. Paying attention to the study design is crucial.

Pre-U 还考察两独立样本的威尔科克森秩和 (曼-惠特尼) 检验。考生经常混淆哪个检验适用，将符号秩检验用于独立数据。留意研究设计至关重要。

---

9. Central Limit Theorem and Sampling Distributions | 中心极限定理与抽样分布

The Central Limit Theorem (CLT) states that, for sufficiently large sample size n, the sampling distribution of the sample mean x̄ is approximately N(μ, σ²/n), regardless of the population distribution. Many candidates invoke the CLT but then fail to check that n is large enough (typically n ≥ 30) or incorrectly claim that the original population becomes normally distributed.

中心极限定理 (CLT) 指出，当样本量 n 充分大时，样本均值 x̄ 的抽样分布近似为 N(μ, σ²/n)，无论总体服从何种分布。许多考生虽然引用了中心极限定理，却未能检查 n 是否足够大（通常 n ≥ 30），或者错误地宣称原始总体变为正态分布。

Another mistake is confusing the distribution of a sample statistic with the distribution of the raw data. The standard deviation of the sampling distribution is the standard error σ/√n, not σ. When constructing test statistics, students often use σ instead of the standard error, resulting in an inflated or deflated statistic.

另一个错误是混淆样本统计量的分布与原始数据的分布。抽样分布的标准差是标准误 σ/√n，而非 σ。在构建检验统计量时，学生经常使用 σ 而非标准误，导致统计量虚高或偏低。

---

10. Probability Pitfalls | 概率陷阱

Conditional probability P(A|B) is often confused with P(B|A). A classic scenario involves diagnostic testing: the probability of having a disease given a positive test result is not the same as the probability of testing positive given the disease. Candidates need to apply Bayes’ theorem or contingency tables formally rather than relying on intuition.

条件概率 P(A|B) 常与 P(B|A) 混淆。一个经典场景是诊断检验：给定检测结果呈阳性时患病的概率，与患病情况下检测呈阳性的概率并不相同。考生需要正式应用贝叶斯定理或列联表，而不是依赖直觉。

Mutually exclusive events and independent events are fundamentally different. Two events are mutually exclusive if they cannot occur simultaneously, while independence means P(A∩B) = P(A)P(B). Candidates often treat mutually exclusive events as independent, which leads to incorrect probability calculations.

互斥事件与独立事件有着本质区别。若两事件不能同时发生，则它们互斥；而独立意味着 P(A∩B) = P(A)P(B)。考生经常将互斥事件视为独立，导致概率计算错误。

The addition rule P(A∪B) = P(A) + P(B) − P(A∩B) is frequently misapplied when the events are not disjoint. Simply adding probabilities without subtracting the intersection is a stubborn error, especially in problems involving “at least” or “or”.

加法公式 P(A∪B) = P(A) + P(B) − P(A∩B) 在事件不互斥时经常被错误使用。不做交集扣减就直接相加概率，是一个根深蒂固的错误，尤其是在涉及 “至少” 或 “或” 的问题中。

---

11. Exam Technique and Key Takeaways | 考试技巧与核心总结

Reading the question thoroughly and underlining the command words prevents misinterpretation. State hypotheses clearly, define variables, and always check the conditions for any statistical procedure. A complete answer includes a conclusion in context, not just a numerical result.

仔细读题并划出指令词可以避免曲解题意。清晰地陈述假设、定义变量，并且始终检查所用统计方法的前提条件。一份完整的答案应包括情境化结论，而不仅仅是数值结果。

Review the following table of frequent mistakes and the corresponding preventive actions to consolidate your exam readiness.

回顾以下常犯错误及相应预防措施表，以巩固备考状态。

Common Mistake | 常见错误	Preventive Action | 预防措施
Omitting continuity correction in normal approximation | 正态近似时漏掉连续性校正	Always adjust discrete boundaries by ±0.5 | 始终将离散边界调整 ±0.5
Using z instead of t when σ is unknown | σ 未知时使用 z 而非 t	Check whether σ is known or estimated by s | 检查 σ 已知还是由 s 估计
Confusing one-tailed and two-tailed tests | 混淆单尾与双尾检验	Match H₁ to phrasing: ‘increase’ vs ‘change’ | 将 H₁ 与措辞匹配：”增加” 对 “变化”
Misinterpreting confidence interval probability | 错误解释置信区间概率	Use ‘confidence’ not ‘probability’ in context | 在语境中使用 ‘置信水平’ 而非 ‘概率’
Using percentages in chi-squared test | 卡方检验中使用百分比	Ensure all entries are frequency counts | 确保所有项均为频数计数

Mastering these high-frequency topics and sidestepping the documented errors will sharpen your statistical reasoning and elevate your exam performance. Consistent practice with past papers, while fine-tuning your error-checking routine, is the proven route to a top grade.

掌握这些高频考点并避开上述记录的错误，将提升你的统计推理能力，提高考试成绩。通过历年真题持续练习，同时细化纠错流程，是通往高分的可靠路径。

Published by TutorHao | Statistics Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）