Pre-U Cambridge 统计：公式定理速查手册

什么是 Cambridge Pre-U 统计学？

Cambridge Pre-U（预科）是剑桥大学国际考评部（CAIE）为16-19岁学生设计的高级学术资格课程。统计学作为数学体系中的一个独立分支，在Pre-U课程中被开设为一门完整的学科，涵盖概率论、统计推断、回归分析和假设检验等核心内容。这门课程不仅为大学阶段的统计学、数据科学、经济学、心理学和工程学等专业打下坚实基础，更培养学生在现实世界中运用数据做出理性判断的能力。

Cambridge Pre-U Statistics is an advanced academic qualification designed by Cambridge Assessment International Education (CAIE) for students aged 16–19. As a standalone discipline within the mathematical sciences, Pre-U Statistics covers probability theory, statistical inference, regression analysis, and hypothesis testing. The course not only provides a solid foundation for university-level study in statistics, data science, economics, psychology, and engineering, but also cultivates the ability to reason with data in real-world contexts.

Pre-U 统计学与 A-Level 统计学的区别

与A-Level数学中附带的统计学模块不同，Pre-U统计学是一门独立的、深度更高的课程。它引入了A-Level通常不涉及的专题，如贝叶斯推断（Bayesian Inference）、非参数检验（Non-Parametric Tests）和线性模型（Linear Models）的矩阵表示。Pre-U更强调统计思维的本质——理解数据生成过程、评估模型假设的合理性、以及解释结果的统计意义和实际意义之间的区别。

Unlike the statistics component embedded within A-Level Mathematics, Pre-U Statistics is a standalone, deeper course. It introduces topics typically absent from A-Level, such as Bayesian Inference, Non-Parametric Tests, and the matrix formulation of Linear Models. Pre-U places greater emphasis on the essence of statistical thinking — understanding data-generating processes, evaluating the plausibility of model assumptions, and distinguishing between statistical significance and practical significance when interpreting results.

第一部分：概率论基础

1. 概率公理与基本定理

概率空间由样本空间 Ω、事件族 F 和概率测度 P 构成。记事件 A 的概率为 P(A)，满足三条公理：非负性 P(A) ≥ 0；规范性和 P(Ω) = 1；可列可加性——对于互不相容的事件序列 A₁, A₂, …，有 P(∪A_i) = ΣP(A_i)。这些公理是一切概率运算的基石。

A probability space consists of a sample space Ω, an event family F, and a probability measure P. The probability of event A is denoted P(A), satisfying three axioms: non-negativity P(A) ≥ 0; normalisation P(Ω) = 1; and countable additivity — for mutually exclusive events A₁, A₂, …, we have P(∪A_i) = ΣP(A_i). These axioms form the foundation of all probability operations.

加法法则：P(A ∪ B) = P(A) + P(B) − P(A ∩ B)。对于互斥事件，简化为 P(A ∪ B) = P(A) + P(B)。

Addition Rule: P(A ∪ B) = P(A) + P(B) − P(A ∩ B). For mutually exclusive events, this simplifies to P(A ∪ B) = P(A) + P(B).

乘法法则：P(A ∩ B) = P(A | B) · P(B) = P(B | A) · P(A)。对于独立事件，P(A ∩ B) = P(A) · P(B)。

Multiplication Rule: P(A ∩ B) = P(A | B) · P(B) = P(B | A) · P(A). For independent events, P(A ∩ B) = P(A) · P(B).

全概率公式：若 {B₁, B₂, …, B_n} 为 Ω 的一个划分，则 P(A) = ΣP(A | B_i) · P(B_i)。这是连接条件概率与边缘概率的桥梁。

Law of Total Probability: If {B₁, B₂, …, B_n} is a partition of Ω, then P(A) = ΣP(A | B_i) · P(B_i). This bridges conditional and marginal probabilities.

2. 贝叶斯定理

贝叶斯定理是Pre-U统计学的核心公式之一：

Bayes’ Theorem is one of the core formulas in Pre-U Statistics:

P(B | A) = [P(A | B) · P(B)] / P(A)

该定理揭示了”逆概率”的本质——我们可以从已知结果（A已发生）反推其原因（B）的概率。在Pre-U考试中，贝叶斯定理常用于医学筛查（已知检测阳性，求实际患病的概率）、质量控制（已知产品缺陷，追溯生产批次）和法律推理等情境。使用贝叶斯定理时，务必正确识别先验概率 P(B) 和似然 P(A | B)。

This theorem reveals the essence of “inverse probability” — we can deduce the probability of a cause (B) given an observed outcome (A). In Pre-U examinations, Bayes’ Theorem frequently appears in medical screening (given a positive test, what is the probability of actually having the disease?), quality control (tracing a defect back to its production batch), and legal reasoning. When using Bayes’ Theorem, always correctly identify the prior probability P(B) and the likelihood P(A | B).

第二部分：离散概率分布

3. 二项分布 Binomial Distribution

若 X ~ B(n, p)，则概率质量函数为：

If X ~ B(n, p), the probability mass function is:

P(X = k) = C(n, k) · p^k · (1 − p)^n−k

其中期望 E[X] = np，方差 Var(X) = np(1 − p)。二项分布适用于 n 次独立伯努利试验中成功次数的建模——例如抛硬币 n 次中正面向上的次数、一批产品中合格品的数量。关键假设：每次试验独立、每次成功的概率 p 固定不变。

where the expectation E[X] = np and variance Var(X) = np(1 − p). The binomial distribution models the number of successes in n independent Bernoulli trials — for example, the number of heads in n coin tosses, or the number of conforming items in a batch. Key assumptions: each trial is independent, and the probability of success p remains constant across trials.

正态近似条件：当 np > 5 且 n(1 − p) > 5 时，二项分布可用正态分布 N(np, np(1 − p)) 近似，并需应用连续性校正——将离散的整数 k 替换为区间 (k − 0.5, k + 0.5)。

Normal Approximation Condition: When np > 5 and n(1 − p) > 5, the binomial distribution can be approximated by a normal distribution N(np, np(1 − p)), with a continuity correction applied — replacing the discrete integer k with the interval (k − 0.5, k + 0.5).

4. 泊松分布 Poisson Distribution

若 X ~ Po(λ)，则概率质量函数为：

If X ~ Po(λ), the probability mass function is:

P(X = k) = (λ^k · e^−λ) / k!

其中期望和方差均为 λ：E[X] = Var(X) = λ。泊松分布用于对固定时间或空间内随机事件发生次数进行建模——例如某路口一小时内通过的车辆数、某页打印纸上的错别字数。泊松分布也是二项分布的极限情形：当 n → ∞、p → 0、且 np = λ 保持常数时，B(n, p) → Po(λ)。

where both expectation and variance equal λ: E[X] = Var(X) = λ. The Poisson distribution models the number of random events occurring in a fixed interval of time or space — for example, the number of vehicles passing an intersection in an hour, or the number of typos on a printed page. The Poisson distribution is also the limiting case of the binomial: as n → ∞, p → 0, and np = λ remains constant, B(n, p) → Po(λ).

第三部分：连续概率分布

5. 正态分布 Normal Distribution

正态分布是统计学中最重要的分布。若 X ~ N(μ, σ²)，其概率密度函数为：

The normal distribution is the most important distribution in statistics. If X ~ N(μ, σ²), its probability density function is:

f(x) = [1 / (σ√(2π))] · exp(−(x − μ)² / (2σ²))

标准化：通过 Z = (X − μ) / σ 可将任意正态变量转化为标准正态分布 Z ~ N(0, 1)。标准正态分布表是Pre-U考试中不可或缺的工具——你必须能熟练地正向查询（给定z值求概率）和反向查询（给定概率求z值）。

Standardisation: Any normal variable can be transformed into the standard normal distribution Z ~ N(0, 1) via Z = (X − μ) / σ. The standard normal table is an indispensable tool in Pre-U examinations — you must be proficient at both forward lookup (finding probability given a z-value) and inverse lookup (finding a z-value given probability).

68-95-99.7 法则：约68%的数据落在 μ ± σ 内；约95%的数据落在 μ ± 2σ 内；约99.7%的数据落在 μ ± 3σ 内。

The 68-95-99.7 Rule: Approximately 68% of data falls within μ ± σ; approximately 95% within μ ± 2σ; and approximately 99.7% within μ ± 3σ.

6. 学生t分布 Student’s t-Distribution

当总体方差未知且样本量较小时，t分布是进行均值的置信区间估计和假设检验的关键工具。若统计量 T ~ t_ν（自由度为 ν），其分布形状类似于标准正态分布但尾部更厚。自由度 ν = n − 1（单样本情况）。随着 ν → ∞，t分布收敛于标准正态分布。

When the population variance is unknown and the sample size is small, the t-distribution is the key tool for confidence interval estimation and hypothesis testing of the mean. If the statistic T ~ t_ν (with ν degrees of freedom), its shape resembles the standard normal distribution but with heavier tails. The degrees of freedom ν = n − 1 (one-sample case). As ν → ∞, the t-distribution converges to the standard normal distribution.

7. 卡方分布 Chi-Squared Distribution

若 Z₁, Z₂, …, Z_k 为独立标准正态变量，则 ΣZ_i² ~ χ²_k（自由度为 k 的卡方分布）。卡方分布在拟合优度检验（Goodness-of-Fit Test）和独立性检验（Test of Independence）中扮演核心角色。

If Z₁, Z₂, …, Z_k are independent standard normal variables, then ΣZ_i² ~ χ²_k (chi-squared distribution with k degrees of freedom). The chi-squared distribution plays a central role in goodness-of-fit tests and tests of independence.

第四部分：统计推断

8. 置信区间

置信区间是对总体参数的一个区间估计，形式为”点估计 ± 临界值 × 标准误”。Pre-U考试中需要掌握以下核心区间公式：

A confidence interval is an interval estimate of a population parameter, taking the form “point estimate ± critical value × standard error”. The following core interval formulas must be mastered for Pre-U examinations:

总体均值 μ（σ已知）：x̄ ± z_α/2 · σ/√n

Population Mean μ (σ known): x̄ ± z_α/2 · σ/√n

总体均值 μ（σ未知）：x̄ ± t_ν,α/2 · s/√n，自由度 ν = n − 1

Population Mean μ (σ unknown): x̄ ± t_ν,α/2 · s/√n, degrees of freedom ν = n − 1

总体比例 p（大样本）：p̂ ± z_α/2 · √[p̂(1 − p̂)/n]

Population Proportion p (large sample): p̂ ± z_α/2 · √[p̂(1 − p̂)/n]

置信水平的含义常被误解——95%置信水平并非指”该区间包含μ的概率为95%”，而是指”如果重复抽样并构建区间，其中约95%的区间将包含真实的总体参数μ”。这是一个频率学派的诠释，在Pre-U考试中需准确表述。

The meaning of confidence level is frequently misunderstood — a 95% confidence level does not mean “the probability that this interval contains μ is 95%”, but rather “if we repeatedly sampled and constructed intervals, approximately 95% of those intervals would contain the true population parameter μ”. This is a frequentist interpretation and must be stated accurately in Pre-U examinations.

9. 假设检验框架

假设检验遵循标准五步流程：

Hypothesis testing follows a standard five-step procedure:

第一步 — 设立假设：定义原假设 H₀（通常为”无差异”或”无关联”）和备择假设 H₁（双尾或单尾）。例如，检验硬币是否公平：H₀: p = 0.5, H₁: p ≠ 0.5（双尾）。

Step 1 — State Hypotheses: Define the null hypothesis H₀ (typically “no difference” or “no association”) and the alternative hypothesis H₁ (two-tailed or one-tailed). For example, testing whether a coin is fair: H₀: p = 0.5, H₁: p ≠ 0.5 (two-tailed).

第二步 — 确定显著性水平：常用 α = 0.05 或 0.01。这代表在原假设为真的情况下，拒绝原假设的最大可接受概率（第一类错误概率）。

Step 2 — Set Significance Level: Commonly α = 0.05 or 0.01. This represents the maximum acceptable probability of rejecting the null hypothesis when it is true (Type I error probability).

第三步 — 计算检验统计量：根据问题类型选择合适的统计量——z统计量（均值检验，σ已知）、t统计量（均值检验，σ未知）、χ²统计量（拟合优度/独立性检验）或F统计量（方差分析）。

Step 3 — Compute Test Statistic: Choose the appropriate statistic based on the problem type — z-statistic (mean test, σ known), t-statistic (mean test, σ unknown), χ² statistic (goodness-of-fit / independence tests), or F-statistic (ANOVA).

第四步 — 确定p值或临界区域：p值是”在原假设为真的前提下，观察到当前结果或更极端结果的概率”。若 p < α，则拒绝 H₀。对于离散分布（如二项分布），p值的计算需特别注意——Pre-U考试中可能在双尾检验时采用”双倍单尾p值”的方法。

Step 4 — Determine p-value or Critical Region: The p-value is “the probability of observing the current result or a more extreme result, assuming the null hypothesis is true”. If p < α, reject H₀. For discrete distributions (e.g., binomial), p-value calculation requires care — Pre-U examinations may adopt the “double the one-tailed p-value” method for two-tailed tests.

第五步 — 得出结论：以实际语境表述结论，而非仅机械地报告”拒绝/不拒绝 H₀“。例如：”有充分证据表明（在5%显著性水平下）这种新药能降低血压”。

Step 5 — Draw Conclusion: State the conclusion in the context of the problem, rather than mechanically reporting “reject / do not reject H₀“. For example: “There is sufficient evidence (at the 5% significance level) that this new drug lowers blood pressure.”

第五部分：相关与回归

10. 积矩相关系数 Product Moment Correlation Coefficient (PMCC)

皮尔逊相关系数 r 衡量两个变量之间的线性相关程度：

The Pearson correlation coefficient r measures the degree of linear association between two variables:

r = S_xy / √(S_xx · S_yy)

其中 S_xy = Σ(x_i − x̄)(y_i − ȳ) = Σx_iy_i − (Σx_i)(Σy_i)/n。

where S_xy = Σ(x_i − x̄)(y_i − ȳ) = Σx_iy_i − (Σx_i)(Σy_i)/n.

r 的取值范围为 [−1, 1]。r = 1 表示完全正线性相关，r = −1 表示完全负线性相关，r = 0 表示无线性相关（但不排除非线性关系）。关键警示：相关性不等于因果性。这一原则在Pre-U考试中经常出现——冰激凌销量与溺水死亡人数正相关，不等同于冰激凌导致溺水（混淆变量：夏季高温）。

The range of r is [−1, 1]. r = 1 indicates perfect positive linear correlation, r = −1 indicates perfect negative linear correlation, and r = 0 indicates no linear correlation (though non-linear relationships may still exist). Key Warning: Correlation does not imply causation. This principle frequently appears in Pre-U examinations — ice cream sales are positively correlated with drowning deaths, but this does not mean ice cream causes drowning (confounding variable: summer heat).

11. 线性回归 Linear Regression

最小二乘回归线 ŷ = a + bx 中，回归系数 b 的公式为：

In the least squares regression line ŷ = a + bx, the regression coefficient b is given by:

b = S_xy / S_xx

a = ȳ − b · x̄

回归线始终通过点 (x̄, ȳ)。残差 e_i = y_i − ŷ_i 是观测值与预测值之差——残差图（残差对拟合值）是诊断回归模型假设（线性、独立性、同方差性、正态性）的重要工具。

The regression line always passes through the point (x̄, ȳ). The residual e_i = y_i − ŷ_i is the difference between the observed and predicted values — a residual plot (residuals against fitted values) is an important diagnostic tool for assessing regression assumptions (linearity, independence, homoscedasticity, normality).

第六部分：非参数检验

12. 非参数方法简介

与参数检验（依赖总体分布假设）不同，非参数检验不要求数据服从特定分布，因此在小样本或数据为非正态时更有鲁棒性。Pre-U课程中通常涉及以下非参数方法：

Unlike parametric tests (which rely on distributional assumptions about the population), non-parametric tests do not require data to follow a specific distribution, making them more robust for small samples or non-normal data. The Pre-U course typically covers the following non-parametric methods:

符号检验（Sign Test）：用于配对数据的中位数差异检验。仅使用差值的符号（正或负），忽略具体的数值大小。适用于不满足正态性假设的配对t检验场景。

Sign Test: Used for testing the median difference in paired data. Only the signs of the differences (positive or negative) are used, ignoring the actual magnitudes. Suitable when the normality assumption for the paired t-test is not met.

威尔科克森符号秩检验（Wilcoxon Signed-Rank Test）：符号检验的增强版——同时考虑差值的符号和大小排序。比符号检验具有更高的统计功效（Statistical Power），因为它利用了更多的数据信息。

Wilcoxon Signed-Rank Test: An enhancement of the sign test — it considers both the signs and the ranked magnitudes of the differences. It has higher statistical power than the sign test because it utilises more information from the data.

曼-惠特尼U检验（Mann-Whitney U Test）：用于比较两个独立样本的中位数是否相同——独立样本t检验的非参数替代方案。将所有观测值合并排序后计算秩和统计量。

Mann-Whitney U Test: Used to compare whether the medians of two independent samples are equal — the non-parametric alternative to the independent samples t-test. All observations are pooled and ranked, and the rank-sum statistic is computed.

第七部分：核心公式速查表

13. 考试高频公式总览

以下是Pre-U统计考试中最常被调用的公式集，建议考前反复默写：

The following is a summary of the most frequently invoked formulas in Pre-U Statistics examinations, recommended for repeated memorisation before the exam:

统计量	公式
样本均值	x̄ = Σx_i / n
样本方差	s² = Σ(x_i − x̄)² / (n − 1)
均值标准误	SE(x̄) = σ / √n 或 s / √n
比例标准误	SE(p̂) = √[p(1 − p) / n]
z统计量	z = (x̄ − μ) / (σ/√n)
t统计量	t = (x̄ − μ) / (s/√n)
χ²统计量	χ² = Σ[(O − E)² / E]
回归斜率	b = S_xy / S_xx

在计算器上使用统计模式可显著提高运算效率，但务必展示完整的公式代入过程——Pre-U的评分标准对解题步骤有严格要求。

Using the statistics mode on your calculator can significantly improve computational efficiency, but always show full formula substitution — Pre-U marking schemes have strict requirements on solution steps.

第八部分：常见误区与考试技巧

14. 高频失分点

误区一：混淆样本和总体的符号。总体参数使用希腊字母（μ, σ, ρ），样本统计量使用拉丁字母（x̄, s, r）。考试中将 x̄ 写成 μ 是典型的符号错误。

Pitfall 1: Confusing sample and population notation. Population parameters use Greek letters (μ, σ, ρ), while sample statistics use Latin letters (x̄, s, r). Writing x̄ as μ in an examination is a classic notation error.

误区二：在正态近似中忘记连续性校正。用正态分布近似离散分布时，必须应用连续性校正 ±0.5——许多考生因遗漏这一步而遭到扣分。

Pitfall 2: Forgetting the continuity correction in normal approximations. When approximating a discrete distribution with a normal distribution, the continuity correction ±0.5 must be applied — many candidates lose marks by omitting this step.

误区三：混淆单尾和双尾检验的临界值。单尾检验使用 z_α（例如 α=0.05 → 1.645），双尾检验使用 z_α/2（例如 α=0.05 → 1.96）。在解释p值时同样如此——务必与备择假设 H₁ 的方向保持一致。

Pitfall 3: Confusing critical values for one-tailed and two-tailed tests. One-tailed tests use z_α (e.g., α=0.05 → 1.645), while two-tailed tests use z_α/2 (e.g., α=0.05 → 1.96). The same applies when interpreting p-values — always align with the direction of the alternative hypothesis H₁.

误区四：忽视假设条件的验证。使用t检验前，应确认数据大致来自正态总体（可通过直方图或正态概率图检查）。使用卡方检验时，需确保期望频数 > 5。Pre-U考试中明确列出并验证模型的假设条件，是获得高分的关键。

Pitfall 4: Neglecting to verify test assumptions. Before using the t-test, confirm that the data approximately come from a normal population (check via histogram or normal probability plot). When using the chi-squared test, ensure expected frequencies > 5. Explicitly listing and verifying model assumptions is key to achieving high marks in Pre-U examinations.

15. 计算器技巧

熟练掌握科学计算器的统计功能，可以为你节省大量考试时间。大多数图形计算器（如TI-84、Casio fx-CG50）可以一键计算单变量统计量（均值、标准差、四分位数等）以及双变量回归分析（PMCC、回归系数）。然而，Pre-U考试评分标准同时要求手动展示公式——建议的答题策略是：先用计算器快速获得数值结果，然后在答题纸上补充完整的公式推导过程。

Mastering your scientific calculator’s statistical functions can save substantial exam time. Most graphing calculators (e.g., TI-84, Casio fx-CG50) can compute univariate statistics (mean, standard deviation, quartiles) and bivariate regression analysis (PMCC, regression coefficients) with a single command. However, Pre-U marking schemes also require manual formula display — the recommended exam strategy is to first obtain numerical results quickly with the calculator, then supplement the solution paper with the complete formula derivation process.

总结

Cambridge Pre-U 统计学是一门兼具深度和广度的课程，它将理论与应用紧密结合，培养学生在数据驱动的世界中做出明智决策的能力。掌握上述核心公式和定理，理解其背后的统计原理，并通过大量练习巩固，是在Pre-U考试中取得高分的不二法门。祝各位考生复习顺利，考试成功！

Cambridge Pre-U Statistics is a course that combines depth with breadth, tightly integrating theory with application to cultivate students’ ability to make informed decisions in a data-driven world. Mastering the core formulas and theorems above, understanding the statistical principles behind them, and reinforcing through extensive practice is the surest path to high marks in Pre-U examinations. Best of luck with your revision and success in the examination!