← 返回资讯屏轩教育 · 行业资讯

CCEA Year 13 Statistics: Quick Reference Formula & Theorem Handbook | Year 13 CCEA 统计:公式定理速查手册

2026-07-23 06:56 · 本文部分内容由AI辅助创作

📚 CCEA Year 13 Statistics: Quick Reference Formula & Theorem Handbook | Year 13 CCEA 统计:公式定理速查手册

This concise handbook provides all essential formulas, theorems, and key definitions for the CCEA Year 13 Statistics syllabus. It is designed as a rapid revision aid, allowing you to check critical content quickly before the exam. Every formula is presented with clear notation, and explanations are paired in English and Chinese to support bilingual learners.

这本精简的手册提供了 CCEA 13 年级统计课程中所有重要的公式、定理和关键定义。它旨在作为快速复习工具,帮助你在考试前迅速查阅核心内容。每个公式都配有清晰的符号说明,并用英文和中文并列解释,以支持双语学习者。

1. Probability and Set Notation | 概率与集合符号

space S is the set of all possible outcomes, and P(S) = 1. The complement of A, written A' or Aᶜ, has probability P(A') = 1 − P(A).

事件 A 的概率,记作 P(A),用于衡量该事件发生的可能性。对于任意事件 A,有 0 ≤ P(A) ≤ 1。样本空间 S 是所有可能结果的集合,且 P(S) = 1。A 的补集,写作 A' 或 Aᶜ,其概率为 P(A') = 1 − P(A)。

For any two events A and B, the probability of A or B occurring is given by the addition rule:
P(A ∪ B) = P(A) + P(B) − P(A ∩ B). If A and B are mutually exclusive, then P(A ∩ B) = 0, so P(A ∪ B) = P(A) + P(B).

对于任意两个事件 A 和 B,A 或 B 发生的概率由加法法则给出:
P(A ∪ B) = P(A) + P(B) − P(A ∩ B)。如果 A 和 B 互斥,则 P(A ∩ B) = 0,因此 P(A ∪ B) = P(A) + P(B)。

The notation A ∩ B represents the intersection (both A and B occur), while A ∪ B represents the union (at least one occurs). A Venn diagram is often used to visualise these relationships.

符号 A ∩ B 表示交集(A 和 B 同时发生),而 A ∪ B 表示并集(至少一个发生)。通常用维恩图来可视化这些关系。


2. Conditional Probability and Bayes' Theorem | 条件概率与贝叶斯定理

Conditional probability is the probability that event A occurs given that event B has already occurred. It is denoted by P(A | B) and defined as:

P(A | B) = P(A ∩ B) / P(B), provided P(B) > 0.

条件概率是指在事件 B 已经发生的条件下,事件 A 发生的概率。记作 P(A | B),定义为:

P(A | B) = P(A ∩ B) / P(B),前提是 P(B) > 0。

ccurrence of one does not affect the probability of the other. For independent events, P(A | B) = P(A) and equivalently P(A ∩ B) = P(A) × P(B).

如果两个事件相互独立,则一个事件的发生不影响另一个事件的概率。对于独立事件,有 P(A | B) = P(A),等价地 P(A ∩ B) = P(A) × P(B)。

Bayes' Theorem links reversed conditional probabilities. For a partition B₁, B₂, ..., Bₙ of the sample space:

P(Bₖ | A) = [P(A | Bₖ) P(Bₖ)] / [Σ P(A | Bᵢ) P(Bᵢ)].

贝叶斯定理联系了颠倒的条件概率。对于样本空间的一个划分 B₁, B₂, ..., Bₙ,有:

P(Bₖ | A) = [P(A | Bₖ) P(Bₖ)] / [Σ P(A | Bᵢ) P(Bᵢ)]。

Bayes' Theorem is especially useful in medical testing and diagnostic problems where one needs to reverse the direction of conditional reasoning.

贝叶斯定理在医学检测和诊断问题中特别有用,当需要反转条件推理的方向时就会用到它。


3. Discrete Random Variables | 离散随机变量

ion is described by a probability mass function P(X = x). Two fundamental properties hold:

0 ≤ P(X = x) ≤ 1 and Σ P(X = x) = 1.

离散随机变量 X 取有限或可数无限个值。其概率分布由概率质量函数 P(X = x) 描述。两个基本性质成立:

0 ≤ P(X = x) ≤ 1 且 Σ P(X = x) = 1。

alue (mean) of X is E(X) = Σ x P(X = x). The variance, which measures spread, is given by:

Var(X) = E(X²) − [E(X)]² = Σ x² P(X = x) − μ².

X 的期望值(均值)为 E(X) = Σ x P(X = x)。方差衡量离散程度,由下式给出:

Var(X) = E(X²) − [E(X)]² = Σ x² P(X = x) − μ²。

For a linear function Y = aX + b, the expected value and variance transform as:

E(aX + b) = a E(X) + b,

Var(aX + b) = a² Var(X).

对于线性函数 Y = aX + b,期望和方差的变换为:

E(aX + b) = a E(X) + b,

Var(aX + b) = a² Var(X)。

values 1, 2, ..., n with equal probability 1/n. Then E(X) = (n+1)/2, Var(X) = (n² − 1)/12.

离散均匀分布是一个简单情形:X 取值 1, 2, ..., n,每个概率均为 1/n。此时 E(X) = (n+1)/2,Var(X) = (n² − 1)/12。


4. Binomial Distribution | 二项分布

A random variable X follows a binomial distribution if it counts the number of successes in n independent trials, each with probability of success p. This is denoted X ~ B(n, p). The probability mass function is:

P(X = r) = ⁿCᵣ pʳ (1 − p)ⁿ⁻ʳ, for r = 0, 1, ..., n.

如果一个随机变量 X 表示在 n 次独立试验中成功的次数,且每次成功的概率为 p,则 X 服从二项分布,记作 X ~ B(n, p)。其概率质量函数为:

P(X = r) = ⁿCᵣ pʳ (1 − p)ⁿ⁻ʳ,其中 r = 0, 1, ..., n。

The binomial coefficient ⁿCᵣ = n! / [r! (n−r)! to choose r successes from n trials. The mean and variance of a binomial distribution are:

E(X) = np,

Var(X) = np(1 − p).

二项式系数 ⁿCᵣ = n! / [r! (n−r)!] 计算从 n 次试验中选择 r 次成功的组合数。二项分布的均值和方差为:

E(X) = np,

Var(X) = np(1 − p)。

y of success p across trials. When n is large and p is moderate, the binomial may be approximated by a normal distribution with μ = np and σ = √[np(1−p)].

二项分布的使用条件:试验次数 n 固定;每次试验独立;每次只有两种结果(成功/失败);成功的概率 p 恒定。当 n 很大而 p 适中时,二项分布可用均值为 np、标准差为 √[np(1−p)] 的正态分布近似。


5. Poisson Distribution | 泊松分布

The Poisson distribution models the number of events occurring in a fixed interval of time or space, given that events happen independently at a constant average rate λ. Notated as X ~ Po(λ), the probability function is:

P(X = r) = e⁻λ λʳ / r!, for r = 0, 1, 2, ...

泊松分布用于模拟在固定时间或空间区间内发生的事件数量,假定事件以恒定的平均速率 λ 独立发生。记作 X ~ Po(λ),其概率函数为:

P(X = r) = e⁻λ λʳ / r!,其中 r = 0, 1, 2, ...

Both the mean and the variance of a Poisson random variable equal λ:

E(X) = λ,

Var(X) = λ.

泊松随机变量的均值和方差都等于 λ:

E(X) = λ,

Var(X) = λ。

that case, set λ = np. A typical guideline is n ≥ 20 and p ≤ 0.05, or n ≥ 100 and np ≤ 10.

当 n 很大且 p 很小时,泊松分布可用于近似二项分布 B(n, p),此时 λ = np 且大小适中。通常的指导原则是 n ≥ 20 且 p ≤ 0.05,或 n ≥ 100 且 np ≤ 10。

The sum of independent Poisson variables also follows a Poisson distribution: if X ~ Po(λ₁) and Y ~ Po(λ₂) are independent, then X + Y ~ Po(λ₁ + λ₂).

独立的泊松变量之和仍然服从泊松分布:如果 X ~ Po(λ₁) 且 Y ~ Po(λ₂) 相互独立,则 X + Y ~ Po(λ₁ + λ₂)。


6. Normal Distribution | 正态分布

The normal distribution is a continuous probability distribution with parameters μ (mean) and σ² (variance). Notation: X ~ N(μ, σ²). The probability density function is symmetric and bell‑shaped. There is no simple closed form for probabilities; instead, standardisation is used.

正态分布是一种连续型概率分布,参数为 μ(均值)和 σ²(方差)。记作 X ~ N(μ, σ²)。其概率密度函数是对称的钟形曲线。概率值没有简单的闭形公式,需要通过标准化来求解。

To calculate P(a < X < b), convert X to the standard normal variable Z:

Z = (X − μ) / σ, so Z ~ N(0, 1).

要计算 P(a < X < b),将 X 转换为标准正态变量 Z:

Z = (X − μ) / σ,因此 Z ~ N(0, 1)。

use standard normal tables (or calculator) to find Φ(z) = P(Z ≤ z). Key results: about 68% of data lie within μ ± σ, about 95% within μ ± 2σ, and about 99.7% within μ ± 3σ. The total area under the curve equals 1.

然后使用标准正态表(或计算器)查得 Φ(z) = P(Z ≤ z)。关键结论:约 68% 的数据落在 μ ± σ 之间,约 95% 在 μ ± 2σ 之间,约 99.7% 在 μ ± 3σ 之间。曲线下的总面积等于 1。

stribution has parameter n large enough, we can approximate it by a normal distribution: X ~ A continuity correction adds or subtracts 0.5 when converting from discrete to continuous.

如果二项分布的 n 足够大,可以用正态分布来近似:X ~ B(n, p) ≈ N(np, np(1−p))。从离散转换为连续时,需要增加或减去 0.5 进行连续性校正。

Similarly, a Poisson distribution with moderate λ (e.g. λ > 10) can be approximated by N(λ, λ), again applying a continuity correction.

同样,当泊松分布的 λ 适中(例如 λ > 10)时,可用 N(λ, λ) 近似,同样需连续性校正。


广告 / Sponsored

7. Sampling and the Central Limit Theorem | 抽样与中心极限定理

A sample of size n taken from a population yields a sample mean x̄. If the population has mean μ and variance σ², then the sampling distribution of x̄ has:

E(X̄) = μ,

Var(X̄) = σ² / n.

从总体中抽取容量为 n 的样本,得到样本均值 x̄。若总体均值为 μ、方差为 σ²,则 X̄ 的抽样分布满足:

E(X̄) = μ,

Var(X̄) = σ² / n。

a random sample of size n from any population with finite variance, the distribution of X̄ of the population. Usually, n ≥ 30 is considered sufficient for the normal approximation.

中心极限定理(CLT)指出,从任何具有有限方差的总体中抽取容量为 n 的随机样本,随着 n 增大,X̄ 的分布趋于正态分布,无论总体的形状如何。通常,n ≥ 30 被认为足以使用正态近似。

Thus, for large samples:

X̄ ~ N(μ, σ²/n) approximately.

因此,对于大样本:

X̄ 近似服从 N(μ, σ²/n)。

If the population standard deviation σ is unknown, it is estimated by the sample standard deviation s, and the t‑distribution is used for small samples when constructing confidence intervals or performing hypothesis tests (see next section).

如果总体标准差 σ 未知,则用样本标准差 s 来估计,在小样本下构建置信区间或进行假设检验时需使用 t 分布(见下一节)。


8. Confidence Intervals | 置信区间

A confidence interval provides a range of plausible values for an unknown population parameter. For the population mean μ, when σ is known, a 95% confidence interval is given by:

x̄ ± z₀.₀₂₅ × (σ / √n),

where z₀.₀₂₅ = 1.96 is the critical value from N(0, 1).

置信区间为未知的总体参数提供了一个合理取值范围。对于总体均值 μ,当 σ 已知时,95% 置信区间为:

x̄ ± z₀.₀₂₅ × (σ / √n),

其中 z₀.₀₂₅ = 1.96 是 N(0, 1) 的临界值。

When σ is unknown and the sample size is small (n < 30), the t‑distribution with ν = n − 1 degrees of freedom is used:

x̄ ± t⍺/₂,ν × (s / √n).

当 σ 未知且样本量较小(n < 30)时,使用自由度为 ν = n − 1 的 t 分布:

x̄ ± t⍺/₂,ν × (s / √n)。

For a population proportion p, a confidence interval based on a large sample is:

p̂ ± z⍺/₂ × √[p̂(1 − p̂) / n],

where p̂ is the sample proportion.

对于总体比例 p,基于大样本的置信区间为:

p̂ ± z⍺/₂ × √[p̂(1 − p̂) / n],

其中 p̂ 是样本比例。

Confidence level (1 − α) indicates the long‑run proportion of such intervals that would contain the true parameter. Common levels are 90%, 95%, and 99%.

置信水平 (1 − α) 表示在重复抽样中,这种区间包含真实参数的长期比例。常用的水平为 90%、95% 和 99%。


9. Hypothesis Testing | 假设检验

A hypothesis test assesses evidence against a null hypothesis H₀ in favour of an alternative H₁. For a test about the mean μ (known σ): compute the test statistic

z = (x̄ − μ₀) / (σ / √n).

假设检验用于评估反对原假设 H₀ 而支持备择假设 H₁ 的证据。对于关于均值 μ 的检验(σ 已知):计算检验统计量

z = (x̄ − μ₀) / (σ / √n)。

The p‑ aining a test statistic at least as extreme as the observed one, assuming H₀ is true. If p‑value ≤ α (significance level), reject H₀; otherwise, do not reject H₀.

p 值是在 H₀ 为真的前提下,获得至少与观测值同样极端的检验统计量的概率。如果 p 值 ≤ α(显著性水平),则拒绝 H₀;否则,不拒绝 H₀。

For a mean when σ is unknown, use the t‑statistic with ν = n − 1 degrees of freedom:

t = (x̄ − μ₀) / (s / √n).

当 σ 未知时,对均值的检验使用自由度为 ν = n − 1 的 t 统计量:

t = (x̄ − μ₀) / (s / √n)。

Tests for a binomial proportion p use the test statistic:

z = (p̂ − p₀) / √[p₀(1 − p₀) / n],

valid under the normal approximation when np₀ ≥ 5 and n(1 − p₀) ≥ 5.

对于二项分布比例 p 的检验,使用检验统计量:

z = (p̂ − p₀) / √[p₀(1 − p₀) / n],

当 np₀ ≥ 5 且 n(1 − p₀) ≥ 5 时正态近似有效。

Type I error: rejecting H₀ when it is true (probability α). Type II error: not rejecting H₀ when it is false (probability β). The power of a test is 1 − β.

第一类错误:H₀ 为真时拒绝 H₀(概率为 α)。第二类错误:H₀ 为假时未拒绝 H₀(概率为 β)。检验的功效为 1 − β。


10. Chi‑Squared Tests | 卡方检验

Chi‑squared tests are used for categorical data. The test statistic is:

χ² = Σ (O − E)² / E,

where O are observed frequencies and E are expected frequencies under H₀. The statistic follows approximately a χ² distribution with appropriate degrees of freedom.

卡方检验用于分类数据。检验统计量为:

χ² = Σ (O − E)² / E,

其中 O 为观测频数,E 为在 H₀ 下的期望频数。该统计量近似服从具有相应自由度的 χ² 分布。

Goodness‑of‑ ibution. Degrees of freedom ν = number of categories − 1 − number of estimated parameters.

拟合优度检验:比较观测频数与理论分布。自由度 ν = 类别数 − 1 − 估计参数的个数。

Test of independence in a contingency table: expected frequency for cell (i, j) is

Eᵢⱼ = (row i total × column j total) / grand total.

umns − 1). The test requires all expected frequencies to be at least 1 and no more than 20% below 5.

列联表中的独立性检验:单元格 (i, j) 的期望频数为

Eᵢⱼ = (第 i 行合计 × 第 j 列合计) / 总计。

自由度 ν = (行数 − 1) × (列数 − 1)。检验要求所有期望频数至少为 1,且低于 5 的期望频数不超过 20%。

If the χ² statistic exceeds the critical value at a given significance level, we reject the null hypothesis of a good fit or independence. The p‑value can also be obtained from tables or technology.

如果 χ² 统计量超过给定显著性水平的临界值,则拒绝拟合良好或独立的原假设。p 值也可通过查表或使用技术工具获取。


11. Correlation and Regression | 相关与回归

Product‑moment correlation coefficient (Pearson's r) measures the strength and direction of a linear relationship between two numerical variables X and Y:

r = Sₓᵧ / √Sₓₓ Sᵧᵧ,

where Sₓₓ = Σ(x − x̄)², Sᵧᵧ = Σ(y − ȳ)², Sₓᵧ = Σ(x − x̄)(y − ȳ).

积矩相关系数(皮尔逊 r)衡量两个数值变量 X 和 Y 之间线性关系的强度和方向:

r = Sₓᵧ / √Sₓₓ Sᵧᵧ,

其中 Sₓₓ = Σ(x − x̄)²,Sᵧᵧ = Σ(y − ȳ)²,Sₓᵧ = Σ(x − x̄)(y − ȳ)。

ndicate strong linear association, while values near 0 indicate weak or no linear relationship. Note that correlation does not imply causation.

r 的取值范围是 −1 到 +1。接近 ±1 的值表明线性相关性强,接近 0 则表明线性关系弱或不存在。注意,相关关系并不意味着因果关系。

The equation of the least‑squares regression line of y on x is:

y = a + b x,

with slope b = Sₓᵧ / Sₓₓ and intercept a = ȳ − b x̄.

y 对 x 的最小二乘回归直线方程为:

y = a + b x,

斜率为 b = Sₓᵧ / Sₓₓ,截距为 a = ȳ − b x̄。

The slope b represents the estimated change in y for a one‑unit increase in x. The coefficient of determination r² indicates the proportion of the variation in y that is explained by x.

斜率 b 表示 x 每增加一个单位,y 的估计变化量。决定系数 r² 表明 y 的变异中有多大比例可由 x 解释。

Residuals are the differences between observed and predicted y‑values. A residual plot can be used to check the assumptions of linearity, constant variance, and independence.

残差是观测的 y 值与预测的 y 值之差。残差图可用于检查线性、等方差和独立性假设。


12. Quick Reference Tables | 速查表

The following table summarises key distributions and their properties. Use it as a rapid reference when solving problems.

下表总结了关键分布及其性质。在解题时可作为快速参考。

DistributionParametersMeanVariancePMF / PDF
Binomial B(n, p)n, pnpnp(1−p)ⁿCᵣ pʳ(1−p)ⁿ⁻ʳ
Poisson Po(λ)λλλe⁻λ λʳ / r!
Normal N(μ, σ²)μ, σ²μσ²(1/√2πσ²) e⁻⁽ˣ⁻μ⁾²/⁽²σ²

Published by TutorHao | Year 13 统计 Revision Series | aleveler.com

广告 / Sponsored
2026年秋初中实验班提优训练789年级上下册语文数学英语人教苏教北师大译林版 初中七八九教材同步随堂作业本强化习题练…
数学 备考教辅 · 正版现货 · 与本文内容相关
查看详情 / View on Taobao参考价 ¥19.9

更多咨询请联系16621398022(同微信)

备考资讯每日更新,关注不错过
考纲变化 · 科目分析 · 学习方法
扫码关注公众号 tutorhao
在微信里长按二维码识别关注
或截图后用「扫一扫 → 相册」选择本图
公众号:tutorhao