一、牛津AQA AS统计考什么？| What Does OxfordAQA AS Statistics Cover?
牛津AQA国际AS数学（9660）的统计模块强调实际数据解读能力和统计推断的逻辑思维，而非单纯的计算技巧。五大板块：统计抽样、数据表示与解读、概率论、二项分布和假设检验。考试中约占25%分数，以应用题和数据分析题为主。
The OxfordAQA International AS Mathematics (9660) Statistics module covers five areas: Statistical Sampling, Data Presentation, Probability, Binomial Distribution, and Hypothesis Testing, accounting for about 25% of total marks. It emphasises real-world interpretation over pure computation.
二、三种抽样方法及其适用场景 | Three Sampling Methods & When to Use Them
简单随机抽样（Simple Random Sampling）
每个个体被选中的概率完全相同。优点是无偏，缺点是当总体很大时难以实施。Every individual has equal selection chance — unbiased but hard with large populations.
分层抽样（Stratified Sampling）
将总体按特征（如年级、性别）分层，按比例随机抽取，能保证样本代表性。公式：每层样本量 = (该层人数 ÷ 总人数) × 总样本量。AQA常出此类计算题。Divide population into strata and sample proportionally: stratum sample = (stratum/population) × total sample.
系统抽样（Systematic Sampling）
将总体排序后每隔固定间隔选取样本，起点随机。操作简单但有周期性偏倚风险。Select every k-th individual with random start — simple but risks periodic bias.
三、直方图与箱线图：两组数据对比的利器 | Histograms & Box Plots: Comparing Datasets
直方图与频数密度（Histograms & Frequency Density）
核心概念：面积代表频数。组距不等时必须用频数密度（Frequency Density = Frequency ÷ Class Width）作为纵轴高度——直接用频数画柱状图在AQA中不得分。Core concept: area = frequency. Use frequency density when class widths differ.
箱线图与异常值（Box Plots & Outliers）
箱线图展示最小值、Q1、中位数Q2、Q3、最大值。异常值判定：低于Q1 − 1.5×IQR或高于Q3 + 1.5×IQR。考试常要求对比两组箱线图的中位数、离散度和偏态。Outlier criterion: below Q1−1.5×IQR or above Q3+1.5×IQR. Compare central tendency, spread, and skewness.
四、条件概率与韦恩图：从"已知"到"推断" | Conditional Probability & Venn Diagrams
核心公式：P(A|B) = P(A ∩ B) ÷ P(B)。AQA喜欢用韦恩图和双向列联表考察条件概率。常见陷阱：P(A|B) ≠ P(B|A)——条件概率不具有对称性。AQA uses Venn diagrams and two-way tables. Key pitfall: P(A|B) ≠ P(B|A).
互斥（Mutually Exclusive）vs独立（Independent）：互斥意味P(A ∩ B) = 0（不能同时发生）；独立意味P(A ∩ B) = P(A) × P(B)（互不影响）。两个完全不同的概念，AQA常放在同一题中考察。Mutually exclusive (cannot occur together) vs independent (one doesn't affect the other) — fundamentally different, often tested together.
五、二项分布B(n, p)：四个条件的验证是前提 | Binomial Distribution B(n, p)
使用二项分布前必须验证四个条件：Before using binomial, verify: ①固定试验次数n (fixed n)；②每次独立 (independent trials)；③只有成功/失败两种结果 (two outcomes)；④每次成功概率p相同 (constant p)。
计算器操作：binompdf(n, p, k)计算恰好k次成功的概率，binomcdf(n, p, k)计算至多k次成功的累积概率。期望E(X) = np，方差Var(X) = np(1 − p)。Calculator: binompdf for exact, binomcdf for cumulative. E(X)=np, Var(X)=np(1−p).
六、假设检验五步法：p值是整道题的核心 | Hypothesis Testing: The p-Value Is Everything
假设检验考察的是逻辑推理链而非计算能力。标准五步流程：Hypothesis testing tests logical reasoning, not computation. Standard process:
- 陈述假设：设H₀和H₁，p为H₀中声称的成功概率。State H₀ and H₁.
- 确定分布：X ~ B(n, p)。Identify distribution.
- 计算p值：在H₀为真时观察到当前或更极端结果的概率。Calculate p-value under H₀.
- 比较显著性水平：通常α = 0.05或0.01。Compare with α (usually 0.05).
- 得出结论：p < α → 拒绝H₀；p ≥ α → 不拒绝H₀（不能说"接受H₀"）。p < α → reject H₀; p ≥ α → do NOT reject H₀.
单尾vs双尾：H₁用"＞"或"＜"为单尾（p值算一侧），用"≠"为双尾（p值×2）。判断方向错了全题不得分。One-tailed (H₁: > or <) vs two-tailed (H₁: ≠ → p×2). Getting tail direction wrong costs the entire question.
七、AQA统计考试常见失分点与备考建议 | Common Pitfalls & Exam Tips
十大高频失分点（Top 10 Mistakes）
- 组距不等时没用频数密度画直方图 — Not using frequency density for unequal class widths
- 将P(A|B)和P(B|A)混淆 — Confusing conditional probability direction
- 假设检验忘记写完整结论（含显著性水平和上下文）— Incomplete conclusion
- 二项分布忘记验证四个条件 — Skipping binomial condition check
- 双尾检验p值忘记乘以2 — Forgetting to double p-value in two-tailed test
- 分层抽样样本量未四舍五入 — Not rounding stratified sample sizes
- 将"不拒绝H₀"写成"接受H₀" — Writing "accept H₀" instead of "do not reject"
- 箱线图分析遗漏偏态描述 — Missing skewness in box plot analysis
- 未评论相关系数强度（strong/moderate/weak）— Not commenting on correlation strength
- 概率值写成小数而非精确值 — Writing probability as decimal instead of exact value
英文关键词（Key English Terms for AQA Answers）
- Positive/Negative correlation — 正/负相关
- On average — 平均而言（解释期望值时必用）
- Sufficient evidence at the 5% level — 标准结论表述
- Not statistically significant — 统计上不显著
需要A-Level数学辅导？
专业教师一对一指导 | 牛津AQA / Edexcel / CIE 全覆盖
www.aleveler.com
或截图后用「扫一扫 → 相册」选择本图