📚 Edexcel AS and A level Further Mathematics FS1 Frequent Mistakes | Edexcel AS与A level进阶数学FS1易错点总结
Further Statistics 1 (FS1) introduces discrete random variables, Poisson, geometric and negative binomial distributions, hypothesis testing and chi‑squared tests. Many students lose marks not because they fail to understand the concepts, but because they skip crucial wording or misuse calculator functions. This article collects the most common pitfalls seen in exam scripts and shows how to sidestep them with precision.
进阶统计1（FS1）涉及离散随机变量、泊松分布、几何分布、负二项分布、假设检验和卡方检验等内容。很多同学丢分并非因为不理解概念，而是漏掉题目中的关键措辞或误用计算器功能。本文整理了试卷中最常见的易错点，并给出精准避坑的方法。
1. Discrete Random Variables: Expectation and Variance Linearity | 离散随机变量：期望与方差的线性性质
Many students incorrectly apply E(aX + b) = aE(X) + b without checking whether the formula is valid for a given transformation. The rule E(aX + b) = aE(X) + b always holds, but Var(aX + b) = a² Var(X) — the constant b disappears. A typical mistake is writing Var(3X – 2) = 3 Var(X) – 2 or 9 Var(X) – 2.
许多同学错误地使用 E(aX + b) = aE(X) + b，但没有检查该公式是否适用于所给的变换。E(aX + b) = aE(X) + b 始终成立，但 Var(aX + b) = a² Var(X)——常数 b 会消失。常见错误是写出 Var(3X – 2) = 3 Var(X) – 2 或 9 Var(X) – 2。
Also, when variables are independent, Var(X ± Y) = Var(X) + Var(Y). Students often subtract variances, writing Var(X – Y) = Var(X) – Var(Y). This is wrong unless there is an explicit covariance term. In FS1, unless a covariance calculation is given, independence is stated, so Var(X – Y) = Var(X) + Var(Y).
此外，当变量独立时，Var(X ± Y) = Var(X) + Var(Y)。学生常常将方差相减，写成 Var(X – Y) = Var(X) – Var(Y)，这是错误的，除非题目给出了显式的协方差计算。在FS1中，除非明确计算协方差，否则一旦声明独立，Var(X – Y) = Var(X) + Var(Y)。
For discrete probability distributions given in a table, always sum the probabilities to 1 as a quick check. A siloed mistake is to compute E(X) using a wrong probability column — especially when the table has both P(X = x) and F(x) together.
对于以表格给出的离散概率分布，务必先将概率求和为1作为快速检验。一个隐形的错误是用错误的概率列计算 E(X)——尤其是当表格中同时出现 P(X = x) 和分布函数 F(x) 时。
2. Poisson Distribution: Conditions and Mean = Variance | 泊松分布：条件与均值等于方差
The Poisson distribution Po(λ) is often used as an approximation to the binomial, but only when n is large and p is small. A common blunder is to apply the approximation without checking np is moderate. The exam expects a statement: ‘n is large, p is small, np ≈ λ’. If a question asks ‘give a reason why the Poisson model is suitable’, the answer usually involves ‘events occur singly, randomly, independently at a constant average rate’.
泊松分布 Po(λ) 常被用作二项分布的近似，但只有在 n 很大而 p 很小的情况下适用。常见错误是在未检查 np 大小适中的情况下直接使用近似。考试要求写出：“n 很大，p 很小，np ≈ λ”。如果题目问“为什么适合用泊松模型”，答案通常包含“事件单独发生、随机、独立，且以恒定的平均速率发生”。
Another trap: using Poisson tables without adjusting λ for time or area. If λ = 2 per hour and we need 90 minutes, students often use λ = 2 instead of 3. Always scale λ proportionally: for a period T, λ = (rate) × T. The same scaling applies to Poisson approximations to binomial: λ = np.
另一个陷阱：使用泊松分布表时未根据时间或面积调整 λ。如果 λ = 每小时2次，需要计算90分钟的概率，学生经常直接使用 λ = 2 而不是 3。一定要按比例缩放 λ：对于时段 T，λ = (速率) × T。二项分布的泊松近似同理：λ = np。
A subtle error is using the variance of a Poisson distribution as λ when the mean is actually an estimate from sample data. In hypothesis tests, if λ is estimated from the sample, the number of degrees of freedom of the test might change, but the distribution still uses that estimated λ. Many candidates forget that the parameter λ in a Poisson test is the value specified in H₀, not the sample mean.
一个微妙的错误是当泊松分布的均值实际上是来自样本数据的估计值时，仍将方差视为 λ。在假设检验中，如果 λ 由样本估计，检验的自由度可能会发生变化，但分布仍使用该估计的 λ。很多考生忘了泊松检验中的参数 λ 是 H₀ 中指定的值，而不是样本均值。
3. Geometric Distribution: Memoryless Property and Support | 几何分布：无记忆性与取值范围
The geometric distribution in Edexcel FS1 counts the number of trials up to and including the first success. So X ~ Geo(p) takes values 1, 2, 3, … with P(X = x) = p(1 – p)^(x – 1). Students frequently confuse this with the version that counts failures before the first success (starting from 0). Always check the problem statement: ‘find the probability that the first success occurs on the 6th trial’ means X = 6, not 5.
Edexcel FS1中的几何分布计算从试验开始直到首次成功所进行的总试验次数。因此 X ~ Geo(p) 取值 1, 2, 3, …，且 P(X = x) = p(1 – p)^(x – 1)。学生经常与另一种版本混淆：计算首次成功前的失败次数（从0开始）。务必检查题目表述：“求首次成功发生在第6次试验的概率”意味着 X = 6，而不是 5。
The memoryless property P(X > s + t | X > s) = P(X > t) is often tested. A classic mistake is writing P(X = s + t | X > s) = P(X = t). It only holds for inequalities like > or ≥, not for exact equality. Using the cumulative formula P(X ≤ x) = 1 – (1 – p)^x (for x = 1,2,3…) can save time but learners sometimes apply P(X ≤ x) = 1 – q^x incorrectly when the distribution is defined for x starting at 1.
无记忆性 P(X > s + t | X > s) = P(X > t) 常被考查。经典错误是写成 P(X = s + t | X > s) = P(X = t)。无记忆性只对 > 或 ≥ 的不等式成立，对精确相等不成立。使用累积公式 P(X ≤ x) = 1 – (1 – p)^x（x = 1,2,3…）可以节省时间，但有些学生会错误地应用 P(X ≤ x) = 1 – q^x，当分布定义从1开始时要特别注意。
4. Negative Binomial Distribution: Counting Failures or Trials? | 负二项分布：计数失败数还是试验数？
In Edexcel FS1, the negative binomial distribution models the number of trials needed to achieve a fixed number r of successes. X ~ NB(r, p) takes values r, r+1, r+2,… and P(X = x) = C(x-1, r-1) p^r (1-p)^(x-r). The biggest pitfall is confusing r with x-r, or using the wrong combinatorial term C(x-1, r-1). Some students mistakenly use C(x, r) from binomial distribution formulas.
在Edexcel FS1中，负二项分布模型化达到固定成功次数 r 所需的总试验次数。X ~ NB(r, p) 取值 r, r+1, r+2,…，且 P(X = x) = C(x-1, r-1) p^r (1-p)^(x-r)。最大的一个坑是混淆 r 和 x – r，或使用了错误的组合数 C(x-1, r-1)。一些学生错误地使用了来自二项分布的 C(x, r)。
Often the question describes the scenario with ‘number of games played until the 4th win’, which directly refers to NB(4, p). But if it says ‘number of losses before the 4th win’, that is a different version not used in FS1 — you need to convert it into total trials or recognise it is outside the specification.
题目常常描述“直到第4次获胜所打的比赛场数”，这直接对应 NB(4, p)。但如果题目说“在第4次获胜之前失败的场数”，那是FS1中不使用的版本——你需要把它转化为总试验次数，或者意识到它超出了考纲。
When calculating probabilities for NB(r, p), use the binomial tail link: P(X ≤ x) is related to the probability that the r-th success occurs at or before trial x, which is equivalent to P(Y ≥ r) where Y ~ B(x, p). This relationship is frequently examined and misinterpreted. Write down the connection explicitly before substituting numbers to avoid off-by-one errors.
计算 NB(r, p) 的概率时，利用其与二项分布尾部的关系：P(X ≤ x) 等同于第 r 次成功发生在第 x 次或更早试验的概率，这等于 P(Y ≥ r)，其中 Y ~ B(x, p)。这个关系经常被考查且容易被曲解。在代入数值前明确写出这个关系，能避免差一错误。
5. Hypothesis Testing: Critical Regions and p‑values | 假设检验：拒绝域与p值
A recurrent mistake is writing ‘accept H₀’ instead of ‘do not reject H₀’ or ‘insufficient evidence to reject H₀’. The conclusion should always be phrased in context: ‘There is insufficient evidence, at the 5% significance level, to suggest that …’. Never state ‘H₀ is true’.
一个反复出现的错误是写“接受 H₀” 而不是“不拒绝 H₀”或“没有足够证据拒绝 H₀”。结论必须结合背景表述：“在5%的显著性水平下，没有足够证据表明……”。切勿写“H₀ 为真”。
When finding a critical region, determine the rejection region so that the probability in each tail (or the single tail) is as close as possible to the significance level without exceeding it. Students often choose the value that gives a probability just above α and forget to check it does not exceed α. In two‑tailed tests for a discrete distribution, the actual significance level is the sum of the tail probabilities used; many candidates omit this calculation and cannot comment on the actual size of the test.
在确定拒绝域时，要使得单尾（或双尾）的概率尽可能接近且不超过显著性水平。学生经常选出一个概率刚刚超过 α 的值，而没有检查其是否超过 α。在离散分布的双尾检验中，实际显著性水平是所用尾部概率之和；很多考生遗漏这一计算，无法评价检验的实际大小。
In FS1, p‑values for Poisson or binomial tests must be found by summing the probability of the observed value and all more extreme values in the direction of H₁. A common error is to double the one‑tailed p‑value for a two‑tailed test without considering whether the distribution is symmetric; for discrete non‑symmetric distributions, the two‑tailed p‑value is not simply 2 × one‑tailed p‑value. Edexcel mark schemes prefer listing the extreme outcomes and summing their probabilities.
在FS1中，泊松或二项检验的 p 值是通过将观测值以及沿 H₁ 方向所有更极端值的概率求和得到的。一个常见错误是对双尾检验简单地用2乘以单尾 p 值，而不考虑分布是否对称；对于离散非对称分布，双尾 p 值并非简单地等于 2 × 单尾 p 值。Edexcel的评分标准更倾向于列出所有极端结果并求和它们的概率。
6. Chi‑squared Tests: Expected Frequencies and Degrees of Freedom | 卡方检验：期望频数与自由度
The most frequency penalty in chi‑squared questions is failing to round expected frequencies to one decimal place before using them in the test statistic formula. The expected frequency Eᵢ must be calculated as (row total × column total) / grand total, then given to 1 d.p. Some students use exact fractions, leading to slight differences in the test statistic, which can change the conclusion.
卡方检验中最常见的扣分点是未将期望频数 Eᵢ 保留到一位小数就直接代入检验统计量公式。必须按 (行合计 × 列合计) / 总计 计算出 Eᵢ，然后保留一位小数。有些学生使用精确分数，导致检验统计量出现微小差异，从而可能改变结论。
Degrees of freedom ν are often misstated. For a contingency table with m rows and n columns, ν = (m-1)(n-1). When estimating parameters from the sample and subtracting an extra degree of freedom, many forget to reduce ν accordingly. If a Poisson or binomial model is fitted using data to estimate parameters, each estimated parameter costs one degree of freedom.
自由度 ν 经常被错误陈述。对于 m 行 n 列的列联表，ν = (m-1)(n-1)。当从样本中估计参数并需额外扣除自由度时，很多人忘记相应地减少 ν。如果使用数据拟合了一个泊松或二项模型并估计了参数，每估计一个参数就要损失一个自由度。
The test statistic X² = Σ (Oᵢ – Eᵢ)² / Eᵢ should be calculated with Yates’ correction for a 2×2 table only when specifically required by the syllabus. In the current Edexcel FS1, Yates’ correction is not used. However, combining rows or columns when Eᵢ < 5 is essential. Students sometimes forget to combine and still proceed, which invalidates the test.
检验统计量 X² = Σ (Oᵢ – Eᵢ)² / Eᵢ 在对 2×2 表使用耶茨校正时需特别注意，但当前 Edexcel FS1 考纲中不使用耶茨校正。然而，当 Eᵢ < 5 时合并行或列是必须的。学生有时忘记合并而直接计算，这会使得检验失效。
7. Contextual Language and Interpretation | 情境语言与解释
Marks are allocated for writing a conclusion in the context of the problem. Phrases like ‘evidence to support the manager’s belief’ or ‘the dice is not fair’ must be precise. Never say ‘prove’ — use ‘suggest’ or ‘provide evidence at the x% level’.
书写结合问题背景的结论有明确分值。类似“支持经理想法的证据”或“骰子不均匀”这样的措辞必须精确。绝不要说“证明”——应使用“表明”或“在 x% 水平下提供证据”。
In chi‑squared tests, the hypothesis statements are always: H₀: there is no association between the variables, H₁: there is an association. Students lose marks by writing H₀: ‘the variables are independent’ without linking to the context, or by using causal language (‘affects’, ’causes’) — chi‑squared only tests association, not causation.
在卡方检验中，假设的陈述总是：H₀: 变量之间无关联，H₁: 变量之间有关联。学生常常因未将“变量独立”与具体背景联系起来，或使用因果性语言（“影响”、“导致”）而丢分——卡方检验只检验关联，不能检验因果。
For goodness‑of‑fit tests, H₀ often states that the observed data could be modelled by a specific distribution, e.g. ‘the data follows a Poisson distribution’. Many students miss the word ‘data’ or ‘population’ and write ‘the distribution is Poisson’, which is vague.
对于拟合优度检验，H₀ 通常表述为观测数据能否被某个特定分布建模，例如“数据服从泊松分布”。许多学生忽略了“数据”或“总体”一词，写成“分布是泊松”，这过于模糊。
8. Calculator Usage and Numerical Precision | 计算器使用与数值精度
Using the correct distribution mode on the calculator is essential: binomial PD/PCD, Poisson PD/PCD, and for geometric/negative binomial many students revert to manual formula entry. When using tables, interpolation is not required in FS1, but you must know how to read cumulative probabilities correctly — a common slip is reading P(X ≥ k) as 1 – entry for k instead of 1 – entry for k-1.
在计算器上使用正确的分布模式至关重要：二项分布 PD/PCD、泊松分布 PD/PCD，而对于几何/负二项分布，很多学生退回到手动输入公式。使用表格时，FS1 不要求插值，但你必须知道如何正确读取累积概率——一个常见失误是将 P(X ≥ k) 读成 1 – k 对应的表值，而正确的应是 1 – (k-1) 对应的表值。
In hypothesis tests, always state the test statistic value rounded to 3 or 4 significant figures. Do not round intermediate expected frequencies to 1 d.p. only in the final table — do it before computing X². When comparing X² with the critical value from tables, pay attention to the correct degrees of freedom and the significance level. Several candidates read the 1% critical value instead of 5%.
在假设检验中，总是要将检验统计量的值保留3或4位有效数字。不要在最后表格中才将期望频数四舍五入至1位小数——要在计算 X² 之前就做好。当将 X² 与临界值表比较时，务必注意正确的自由度和显著性水平。一些考生会读成1%的临界值而不是5%的。
9. Common Pitfalls with Definitions and Notation | 定义与符号的常见陷阱
Mixing up the notation for distributions: B(n, p) is binomial, Po(λ) Poisson, Geo(p) geometric, NB(r, p) negative binomial. In exam answers, you must clearly state the distribution and its parameters before using its properties. Writing just ‘X ~ B(10, 0.2)’ without defining X is penalised under some mark schemes.
混淆各种分布的符号：B(n, p) 是二项分布，Po(λ) 为泊松分布，Geo(p) 为几何分布，NB(r, p) 为负二项分布。在考试作答中，在使用性质前必须清晰写明分布及其参数。只写“X ~ B(10, 0.2)”而不定义 X 会被某些评分标准扣分。
Misunderstanding the sample space: for a geometric distribution, X ≥ 1, so the sum of all probabilities = 1 is verified. However, some students try to use formulas for infinite geometric series and make algebraic errors when simplifying 1 – (1-p)^x.
误解样本空间：对于几何分布，X ≥ 1，因此所有概率之和可以验证为1。但有些学生试图使用无穷几何级数公式，并在简化 1 – (1-p)^x 时犯代数错误。
10. Final Checklist | 最终检查清单
Before finishing a question, run through: Did I define the random variable and its distribution? Are E and Var calculations checked with alternate methods? For hypothesis tests: hypotheses stated, test statistic, critical value or p‑value, comparison, contextual conclusion. For chi‑squared: expected frequencies to 1 d.p., combination of rows/columns if necessary, degrees of freedom correct, and non‑causal conclusion.
在答完一个问题之前，快速过一下：我是否定义了随机变量及其分布？期望和方差的计算是否用其他方法检查过？对于假设检验：是否陈述了假设、检验统计量、临界值或 p 值、比较以及结合背景的结论？对于卡方检验：期望频数是否为一位小数，必要时是否合并了行/列，自由度是否正确，结论是否非因果？
Common transcription errors: miscopying values from calculator, forgetting to subtract 1 when computing degrees of freedom for estimated parameters, and misreading the alternative hypothesis as one‑tailed when it is two‑tailed. Marking these small details can prevent losing up to 10% of the total mark.
常见转录错误：从计算器抄错数值，计算估计参数的自由度时忘记减1，以及误将备择假设当作单尾而实际是双尾。校准这些细小细节可以防止损失多达总分10%的分数。
Published by TutorHao | Further Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导Cancel reply