AQA A2 Further Maths with Statistics: Common Mistakes Summary | AQA A2 进阶数学（含统计）易错点总结

📚 AQA A2 Further Maths with Statistics: Common Mistakes Summary | AQA A2 进阶数学（含统计）易错点总结

In AQA A2 Further Maths, the statistics option brings together a range of advanced techniques – from discrete and continuous distributions to hypothesis testing, confidence intervals, chi-squared tests, t‑tests and probability generating functions. While the methods are powerful, they are also full of subtle conditions and frequent misinterpretations. This revision summary highlights the most common mistakes students make, so you can avoid them in your exams and build a more robust statistical understanding.

在 AQA A2 进阶数学中，统计学模块汇集了从离散与连续分布到假设检验、置信区间、卡方检验、t 检验和概率生成函数等一系列高级技巧。这些方法虽然强大，但也充满了微妙的条件和频繁的误解。这份复习总结梳理了学生最容易犯的错误，帮助你在考试中避开陷阱，建立起更扎实的统计思维。

1. Distribution Approximations: Conditions & Pitfalls | 分布近似：条件与常见陷阱

When approximating a binomial distribution by a Poisson, students often forget that the conditions require a large n and a small p, with the product np typically being less than 10. Using the approximation when np is, say, 15 leads to noticeably inaccurate tail probabilities.

用泊松分布近似二项分布时，学生常常忘记要求 n 较大且 p 较小，通常要求 np < 10。如果 np 达到 15 还使用泊松近似，尾部概率就会明显失真。

Similarly, the normal approximation to the binomial needs both np > 5 and n(1-p) > 5. A common error is to apply it when the distribution is still too skewed, or to omit the continuity correction entirely.

同样，正态分布近似二项分布需要同时满足 np > 5 与 n(1‑p) > 5。常见的错误是当分布仍然偏态时就强行使用正态近似，或者完全忘记连续性校正。

For the normal approximation to a Poisson, the mean λ should be sufficiently large (often quoted as λ > 10). Students sometimes rely on λ > 5 without checking the skewness, which can affect the accuracy of hypothesis tests.

用正态分布近似泊松分布时，均值 λ 需要足够大（通常要求 λ > 10）。有些学生仅凭 λ > 5 就使用正态近似，未考虑偏度对检验准确性的影响。

2. Continuity Correction: When and How to Apply It | 连续性校正：何时用、怎么用

The continuity correction adjusts for the fact that a discrete distribution is being modelled by a continuous one. A classic error is to forget the ±0.5 adjustment when calculating probabilities like P(X ≤ 12) or P(X < 12). The correct expression for P(X < 12) under a normal approximation is P(Y < 11.5).

连续性校正之所以必要，是因为我们用连续分布去模拟离散分布。经典错误就是在计算 P(X ≤ 12) 或 P(X < 12) 时忘记 ±0.5 的调整。对 P(X < 12) 进行正态近似时，应使用 P(Y < 11.5)。

Another slip is applying the correction to the wrong bound – for P(X ≥ 20) you need P(Y > 19.5), not P(Y > 20.5). Always sketch a bar chart to visualise the boundary you want to include or exclude.

另一个常见失误是把校正方向弄反——对于 P(X ≥ 20)，应该用 P(Y > 19.5) 而不是 P(Y > 20.5)。画一个柱状图可以帮助你直观判断该保留还是排除哪个边界。

3. Misreading p‑values in Hypothesis Tests | 假设检验中 p 值的误读

Many students treat the p‑value as the probability that the null hypothesis H₀ is true. This is a fundamental misinterpretation: the p‑value is the probability of obtaining a result at least as extreme as the observed one, assuming H₀ is true. It does not give a direct probability about the hypothesis itself.

很多学生把 p 值理解为原假设 H₀ 成立的概率。这是一个根本性的误解：p 值是在 H₀ 成立的条件下，获得至少与观测结果一样极端的概率，它并不能直接给出关于假设本身的概率。

A further error is comparing the p‑value to the significance level without stating the conclusion in context. The phrase ‘p < 0.05, therefore reject H₀' must be followed by a contextual interpretation, such as 'there is sufficient evidence that the population mean has changed'.

另一个错误是只将 p 值与显著性水平比较，却不结合上下文给出结论。仅仅说“p < 0.05，所以拒绝 H₀”是不够的，必须加上情景化的解读，例如“有充分证据表明总体均值发生了变化”。

4. Confusing Type I and Type II Errors | 第一类错误与第二类错误相混淆

A Type I error occurs when a true null hypothesis is rejected; its probability is exactly the significance level α. A Type II error occurs when a false null hypothesis is not rejected. Students often swap the definitions or fail to relate them to the power of the test (1 – β).

第一类错误是在原假设为真时却错误地拒绝了它，其概率就是显著性水平 α。第二类错误是在原假设为假时却没有拒绝它。学生经常把这两个定义搞混，或者不理解它们与检验功效（1 – β）的联系。

A practical exam mistake is to describe a Type II error without mentioning the alternative hypothesis. You should say, for instance, ‘A Type II error would mean failing to detect a true increase in the average lifetime when it has actually risen.’

考试中常见的失误是描述第二类错误时不提及备择假设。例如，你应该这样说：“第二类错误意味着平均寿命实际上已经上升，但我们没有检测到这一真实增加。”

5. Confidence Interval Misconceptions | 置信区间的错误理解

The statement ‘There is a 95% probability that the population mean lies within this specific interval’ is incorrect. A 95% confidence interval means that if we were to take many random samples and construct intervals in the same way, about 95% of those intervals would contain the true population mean.

“总体均值有 95% 的概率落在这个具体的区间内”这一说法是错误的。95% 置信区间的含义是：如果我们重复抽取大量样本并按相同方法构建区间，那么大约有 95% 的区间会包含总体真值。

Many students also misinterpret the effect of sample size on interval width: doubling the sample size narrows the interval, but not by a factor of ½ – because the width depends on 1/√n, quadrupling the sample size roughly halves the width. Missing this relationship leads to incorrect comparisons.

许多学生对样本量如何影响区间宽度也存在误解：样本量加倍确实会让区间变窄，但并不是缩小一半，因为宽度与 1/√n 相关，样本量变为原来的四倍才能大致使宽度减半。忽略这层关系会得出错误的比较结论。

6. Degrees of Freedom in Chi‑Squared Tests | 卡方检验的自由度

In a chi‑squared goodness‑of‑fit test, the degrees of freedom are given by (number of categories – 1 – number of estimated parameters). A frequent mistake is to omit the subtraction for estimated parameters; for example, when you estimate the population mean from the data to fit a Poisson model, you must subtract an extra degree of freedom.

在卡方拟合优度检验中，自由度 = （类别数 – 1 – 估算参数的个数）。常见错误是忘记减去估算参数的自由度；例如，当你从数据估计总体均值来拟合泊松模型时，必须再减去一个自由度。

For a contingency table, the formula is (r – 1)(c – 1). Mixing up rows and columns or forgetting that the expected frequencies must be calculated from the table margins causes errors in both the degrees of freedom and the resulting χ² statistic.

对于列联表，自由度公式为 (r – 1)(c – 1)。把行和列弄混，或者忘记期望频数必须由表格边缘合计算出，都会导致自由度计算错误，进而影响 χ² 统计量的判断。

7. Choosing Between the t‑Test and the z‑Test | t 检验与 z 检验如何选择

When the population variance σ² is known, a z‑test is appropriate regardless of sample size. When σ² is unknown and the sample size is small (typically n < 30), a t‑test must be used. Students often reach for a z‑test simply because the data look normal, ignoring the unknown variance condition.

当总体方差 σ² 已知时，无论样本量大小都可以使用 z 检验。当 σ² 未知且样本量较小（通常 n < 30）时，必须使用 t 检验。学生往往因为数据看起来呈正态就直接使用 z 检验，而忽略了方差未知这个前提。

Paired t‑tests require careful handling of the differences between pairs. A common slip is to treat the two sets as independent samples, which loses the power that pairing provides. Always check whether the data are naturally paired (e.g. before‑and‑after measurements) before deciding on the test.

配对 t 检验需要谨慎处理成对数据的差值。常见的失误是把两组数据当作独立样本处理，这会使配对设计带来的检验功效白白丧失。在决定选用哪种检验之前，一定要检查数据是否天然配对（如前后测量数据）。

8. Mistakes with Probability Generating Functions (PGFs) | 概率生成函数的常见错误

For a discrete random variable X taking non‑negative integer values, the PGF is G(t) = E(t^X). A very common mistake is to forget the domain: the series must converge, so |t| ≤ 1. Outside this range, derivatives may not give valid moments.

对于取非负整数值的离散随机变量 X，概率生成函数定义为 G(t) = E(t^X)。常见的错误是忽略定义域：级数必须收敛，因此 |t| ≤ 1。超出这个范围，导数可能无法给出正确的矩。

The expected value is E(X) = G'(1). The variance requires both the first and second derivatives: Var(X) = G”(1) + G'(1) – [G'(1)]². Many students forget the middle term G'(1) or miscalculate G”(1), leading to a negative variance. Always check that your variance is positive.

期望值 E(X) = G'(1)，方差则需要用到一阶和二阶导数：Var(X) = G”(1) + G'(1) – [G'(1)]²。许多学生忘记中间的 G'(1) 项，或算错 G”(1)，导致得出负的方差。务必检查方差是否为正。

9. Misapplying the Central Limit Theorem | 中心极限定理的误用

The CLT states that, for a random sample of size n from any distribution with finite mean μ and variance σ², the distribution of the sample mean X̄ is approximately normal when n is large enough. A frequent error is to assume that the original population must be normal – the beauty of the CLT is that it works for non‑normal populations.

中心极限定理指出，从任一具有有限均值 μ 和方差 σ² 的总体中抽取大小为 n 的随机样本，当 n 足够大时样本均值 X̄ 的分布近似正态。常见的错误是认为原始总体必须服从正态分布——CLT 的美妙之处正是它对非正态总体同样适用。

Another pitfall is taking ‘n ≥ 30’ as a magic rule without considering skewness. For heavily skewed distributions, a sample size of 50 or 100 might be needed for the normal approximation to be reliable. Always check the context and, where possible, justify why your chosen n is sufficient.

另一个陷阱是把“n ≥ 30”当作不假思索的铁律，而忽略了偏度。对于高度偏态的分布，可能需要 50 甚至 100 的样本量才能使正态近似足够可靠。一定要结合应用场景判断，并尽量说明所选样本量为何足够。

10. Linear Transformations of Random Variables | 随机变量的线性变换

When a discrete random variable X has expectation E(X) and variance Var(X), the transformed variable Y = aX + b has E(Y) = aE(X) + b and Var(Y) = a²Var(X). Forgetting the square on the coefficient a is one of the most persistent errors in Further Statistics.

若离散随机变量 X 的期望为 E(X)，方差为 Var(X)，那么变换后的变量 Y = aX + b 满足 E(Y) = aE(X) + b，且 Var(Y) = a²Var(X)。忘记给系数 a 加上平方，是进阶统计中最顽固的错误之一。

This mistake often appears when finding the distribution of a sum of independent random variables. For example, if S = X₁ + X₂ + … + Xₙ, then Var(S) = nVar(X) only if the Xᵢ are independent and identically distributed. Students occasionally apply Var(S) = n²Var(X), which is wrong unless the variables are perfectly correlated – a very different scenario.

这一错误常常在求独立随机变量之和的分布时出现。例如，若 S = X₁ + X₂ + … + Xₙ，仅在诸 Xᵢ 独立同分布时才有 Var(S) = nVar(X)。偶尔有学生错误地使用 Var(S) = n²Var(X)，这只有在变量完全相关时才成立——完全是另一回事了。

---

11. Correlation, Regression and Test Conditions | 相关与回归中的条件错误

In product‑moment correlation tests, the null hypothesis ρ = 0 is tested against ρ ≠ 0 (or one‑sided) using a t‑statistic. Students often forget to check the assumption of bivariate normality; if the scatter plot shows a curved pattern, the test is invalid even if the correlation coefficient appears significant.

在积矩相关系数检验中，使用 t 统计量检验原假设 ρ = 0 对备择假设 ρ ≠ 0（或单侧）。学生经常忘记检查二元正态性假设；如果散点图呈现曲线模式，即使相关系数看起来显著，检验也是无效的。

Similarly, when performing linear regression, the validity of confidence intervals for the slope and intercept relies on the residuals being independent, normally distributed and having constant variance. Ignoring a fan‑shaped pattern in the residuals (heteroscedasticity) undermines the whole inference.

类似地，进行线性回归时，斜率和截距的置信区间有效性依赖于残差独立、正态且方差恒定。忽视残差图中的扇形趋势（异方差性）会从根本上动摇推断结论。

12. Conditional Probability and Bayes’ Theorem in Statistics | 统计中的条件概率与贝叶斯定理

When a question involves screening tests or diagnostic probabilities, students frequently mix up P(A|B) and P(B|A). Bayes’ theorem is essential to convert between them. A typical error is to write P(Disease|Positive) = sensitivity without considering the prevalence.

当题目涉及筛查检验或诊断概率时，学生频繁混淆 P(A|B) 和 P(B|A)。必须用贝叶斯定理来转换。典型的错误是写下 P(患病|阳性) = 灵敏度，而不考虑疾病的患病率。

Another subtle point is the assumption of independence when using tree diagrams. Always check that the probabilities on the second branches are conditional and that the tree represents the correct sequence of events. Missing a conditional can distort posterior probabilities dramatically.

另一个微妙之处是使用树图时假设独立。一定要检查第二层分支上的概率是否条件概率，并确保树图正确地表示了事件的先后顺序。遗漏一个条件概率就可能显著扭曲后验概率。

Published by TutorHao | Further Maths Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）