📚 A-Level Further Maths Paper 4 (FS2) Common Errors | A-Level 进阶数学卷四(FS2)常见错误总结
Further Statistics 2 (FS2) is one of the most demanding units in A-Level Further Mathematics. Candidates frequently lose marks not because they do not understand the theory, but because they make avoidable slips in integration, differentiation, or the interpretation of statistical conditions. This article collects the most persistent mistakes seen in Paper 4 responses, with explanations and remedies designed to sharpen your exam technique.
进阶统计2(FS2)是A-Level进阶数学中最具挑战性的单元之一。考生失分往往不是因为不懂理论，而是在积分、微分或统计条件的理解上出现本可避免的失误。本文收集了卷四答题中最顽固的错误，并配以解释与补救方法，旨在帮助你优化考试技巧。
1. Probability Density Functions and Integration Slips | 概率密度函数与积分失误
r the curve equals 1. When finding unknown constants, many students set up the integral correctly but then integrate incorrectly, especially when the pdf is defined piecewise. A common mistake is forgetting to split the integral across the pieces and simply integrating from the lower bound to the upper bound of the domain without checking that the function changes formula between intervals.
概率密度函数f(x)必须满足两个条件：对一切x有f(x) ≥ 0，且曲线下的总面积为1。在求未知常数时，许多学生能正确列出积分式，但积分时出现错误，特别是当pdf分段定义时。一个常见错误是忘记按分段拆分积分，而是直接从定义域的下限积到上限，没有检查函数在不同区间是否切换了表达式。
Another subtle slip occurs when students differentiate a pdf to analyse its shape. They often forget that a pdf itself is not a probability; the probability is given by the area under f(x). Therefore, setting the derivative to zero finds the mode of the distribution, not the most likely single value in the discrete sense, but candidates occasionally misinterpret this point.
另一个不易察觉的错误出现在学生对pdf求导以分析其形状时。他们常常忘记pdf本身并不是概率，概率由f(x)下的面积给出。因此，令导数为零得到的是分布的众数，而不是离散意义下的最可能取值，但考生有时会误解这一点。
2. Cumulative Distribution Functions and Finding the Median | 累积分布函数与求中位数
continuous random variable X with pdf f(x), the cumulative distribution function F(x) = ∫-∞x then F(x) = 0 for x < a and F(x) = 1 for x > b, and within [a,b] it is the integral from a to x. Neglecting the piecewise definition of F(x) when solving F(m) = 0.5 can lead to an invalid solution or a loss of accuracy.
给定连续随机变量X及其pdf f(x)，累积分布函数F(x) = ∫-∞x f(t) dt。求中位数m时，需解方程F(m)=0.5。一个反复出现的错误是忘记F(x)仅在连续变量的情况下才等于f(x)从下界到x的积分，而且如果X的定义域并非整个实数轴，有时会套用错误的积分限。例如，若X定义在[a,b]上，则x
.75) or miscalculate the interquartile range. When the pdf consists of several linear pieces, setting up the anti-derivative incorrectly (e.g., forgetting the constant of integration) is also common. Always check that F(x) is non-decreasing and approaches the correct limits.
此外，学生有时会混淆下四分位数Q1 (F(Q1)=0.25)和上四分位数Q3 (F(Q3)=0.75)，或者计算四分位距时出错。当pdf由多个线性段组成时，不定积分时忘记加积分常数也是常见问题。务必检查F(x)是否单调不减，以及是否趋近于正确的极限值。
3. Moment Generating Functions and Differentiation Errors | 矩母函数与求导错误
The moment generating function (mgf) MX(t) = E(etX) is a powerful tool for finding moments. The r-th raw moment is given by MX(r)(0), i.e., the r-th derivative evaluated at t=0. A very typical slip is to differentiate MX incorrectly, or to forget to differentiate a product of functions where the mgf involves ekt times a rational function. For the exponential distribution, MX λ. Candidates often differentiate and then set t=0 correctly, but some erroneously think M'(0) = 1/λ instead of 1/λ? Wait, for Exp(λ) with pdf λe-λx, E(X)=1/λ, so M' /λ. The mistake is often algebraic: forgetting the chain rule when differentiating (λ - t)-1.
矩母函数MX(t)=E(etX)是求矩的强大工具。第r阶原点矩由MX(r)(0)给出，即在t=0处的r阶导数。一个非常典型的失误是对MX(t)求导后代入t=0时计算错误，或者忘记对包含ekt与有理函数乘积的mgf进行乘法求导。以指数分布为例，MX(t)=λ/(λ - t)，t<λ。考生通常能正确求导并代入0，但有些人误以为M'(0)=1/λ而非1/λ？等等，对于pdf为λe-λx的Exp(λ)，E(X)=1/λ，因此M'(0)=1/λ。错误通常是代数性质的：对(λ - t)-1求导时忘了链式法则。
Another frequent error occurs when using the mgf to find E(X2). Sometimes candidates evaluate M''(0) but then mistakenly use E(X2) = M'' tion for existence of M(t) (t must be in the neighbourhood of 0) can lose a mark in proof-style questions.
另一个常见错误是用mgf求E(X2)时，考生代入M''(0)后却错误地认为E(X2)=M''(0)而不检查是否确实是二阶原点矩，或者混淆原点矩与中心矩。此外，在证明类题目中忘记说明M(t)存在的条件（t必须在0的邻域内）也可能丢分。
4. Probability Generating Functions and Discrete Distributions | 概率生成函数与离散分布
For a discrete non-negative integer-valued random variable X, the probability generating function (pgf) is GX(t) = E(tX) = Σ tx ates confuse pgf with mgf and try to differentiate and set t=0, which is wrong; for pgf, G'(1) = E(X) and G'' ing the pgf for standard distributions like Poisson or geometric, missing the range of t (|t| ≤ 1 for convergence) can be penalised.
对于非负整数值离散随机变量X，其概率生成函数GX(t)=E(tX)=Σ tx P(X=x)。许多考生混淆pgf与mgf，尝试求导并代入t=0，这是错误的；对于pgf，G'(1)=E(X)，G''(1)=E(X(X-1))等。最常见的错误是在t=0而不是在t=1处求导数值。推导泊松分布或几何分布等标准分布的pgf时，漏写t的取值范围（收敛需|t|≤1）可能会被扣分。
ric distribution, expanding correctly as a power series requires careful handling of the binomial series. Candidates often miscount the coefficient, leading to an incorrect probability expression. Remember: the coefficient of tk is exactly P(X=k).
另一个陷阱是用pgf的级数展开求P(X=k)。若几何分布的G(t)=(1 - p)/(1 - pt)，正确地展开为幂级数需要仔细处理二项级数。考生经常算错系数，导致概率表达式错误。请记住：tk的系数正是P(X=k)。
5. Poisson Process – Confusing Rate with Time | 泊松过程——混淆速率与时间
er of events in a 30-minute period is Po(2.5), not Po(5). Always adjust the parameter to match the time length.
泊松过程（单位时间速率λ）是一个核心模型。事件以平均速率λ发生，在长度为t的区间内事件数服从Po(λt)。经典错误是当时间区间不是单位长度时，直接使用λ作为参数而没有乘以t。例如，若λ=5（每小时），30分钟时段内的事件数应服从Po(2.5)，而非Po(5)。务必调整参数以匹配时间长度。
Another common slip is confusing the waiting time distribution. The time between events follows an exponential distribution with mean 1/λ, but many students wrongly state its mean as λ. Also, when summing independent Poisson processes, the sum is Poisson with the sum of the rates, but if the intervals differ you must align the time units. Be explicit about whether you are using the same time base.
另一个常见失误是混淆等待时间的分布。事件间隔时间服从指数分布，均值为1/λ，但很多学生错误地将其均值记作λ。另外，当合并独立泊松过程时，总体服从泊松分布且速率为各速率之和，但如果时间区间不同，必须统一时间单位。务必明确是否采用了相同的时间基准。
6. Hypothesis Testing – Type I and Type II Errors in Context | 假设检验——情境中的第一类错误与第二类错误
In FS2 hypothesis testing, particularly with power functions and the Neyman-Pearson lemma, candidates must identify Type I and Type II errors. Type I error: rejecting H0 when H0 is true (probability α, significance level). Type II error: failing to reject H0 when H1 A common misinterpretation is to describe the error in terms of the sample rather than the population. For example, stating "the probability that the sample mean falls in the critical region" as the definition of Type I error, instead of linking it to the true state of nature.
在FS2的假设检验中，尤其在涉及功效函数和Neyman-Pearson引理时，考生必须识别第一类错误和第二类错误。第一类错误：H0为真时拒绝H0（概率为α，显著性水平）。第二类错误：H1为真时未能拒绝H0（概率为β）。功效=1-β。常见的误解是从样本而非总体的角度描述错误。例如，将“样本均值落在拒绝域的概率”当作第一类错误的定义，而没有将其与真实的总体状态联系起来。
When calculating the probability of Type II error for a simple alternative, students often use the critical value from the H0 distribution but then evaluate the probability under H0 instead of under H1. They also frequently forget to standardise correctly, especially with unknown variance. In t-tests for a difference of means, misidentifying the degrees of freedom leads to an incorrect non-centrality parameter.
在计算简单备择假设的第二类错误概率时，学生往往会使用来自H0分布的临界值，但却在H0而非H1下求概率。他们在标准化时也经常出错，尤其当方差未知时。在两均值之差的t检验中，弄错自由度会导致非中心参数错误。
7. Chi-Squared Tests – Degrees of Freedom and Expected Frequencies | 卡方检验——自由度与期望频数
nd forget the estimated parameters. For an r×c contingency table, df = (r-1)(c-1). The catch is that many students treat the total frequency as a free category and subtract one extra, making the result too small.
卡方拟合优度检验和独立性检验（列联表）是FS2的常规要求。最频繁的错误是自由度计算错误。对于有k个类别且估计了m个参数的拟合优度检验，自由度df = k - 1 - m。考生通常只减1而忘记减去估计的参数个数。对于r×c列联表，df = (r-1)(c-1)。陷阱在于许多学生将总频数视为一个自由类别而多减一个1，导致结果过小。
total. A slip often occurs when students use row percentages instead of counts. Also, Yates' n the test statistic is approximated to chi-squared and some expected frequencies are low.
期望频数必须正确计算。在列联表中，单元格(i,j)的期望频数 = (第i行合计 × 第j列合计)/总计。常见的失误是学生使用了行百分比而非频数。此外，Yates连续性校正常常在不必要的时候被用于2×2表；仅当检验统计量近似卡方且某些期望频数较低时才使用该校正。
8. Confidence Intervals – The Role of the Central Limit Theorem | 置信区间——中心极限定理的作用
Confidence intervals for a population mean often rely on the Central Limit Theorem (CLT) when the population variance is unknown and the sample size is large. A recurrent mistake is applying the z-interval (using normal quantiles) when the sample size is small and the population is not known to be normal, or using t-interval but quoting the wrong degrees of freedom (n instead of n-1). In FS2, the distinction between σ known and σ unknown is crucial.
当总体方差未知且样本量较大时，总体均值的置信区间常常依赖中心极限定理。一个反复出现的错误是在样本量小且总体分布未知时仍使用z区间（基于正态分位数），或者使用t区间但错误地标注自由度（n而非n-1）。在FS2中，区分σ已知与σ未知至关重要。
Another subtle error occurs when constructing a confidence interval for a difference of means from two independent samples. Students sometimes pool the variances incorrectly, or use the t-distribution with n1 + n2 - 2 degrees of freedom but forget to check the assumption of equal variances. If variances cannot be assumed equal, the Welch-Satterthwaite approximation should be used, and the degrees of freedom formula is different. Ignoring this condition may invalidate the interval.
另一个细微错误出现在构建两个独立样本均值之差的置信区间时。学生有时会错误地混合方差，或者使用自由度为n1+n2-2的t分布，却忘记检查方差相等的假定。若不能假定方差相等，则应使用Welch-Satterthwaite近似，其自由度公式不同。忽视此条件可能导致区间无效。
9. Simple but Costly Mistakes with the Continuous Uniform Distribution | 连续均匀分布的简单但代价高昂的错误
2. Yet candidates regularly misplace the variance formula, writing (b-a)²/12 as (b-a)²/6 or confusing it with the variance of a discrete uniform distribution. Also, when finding the mgf of U(a,b), the integration limits must be from a to b, and the expectation E(etX) = (etb - eta slip is to forget that this expression is undefined at t=0 and must be treated as a limit.
[a,b]上的连续均匀分布看似简单：f(x)=1/(b-a)，E(X)=(a+b)/2，Var(X)=(b-a)²/12。然而考生经常把方差公式写错，把(b-a)²/12写成(b-a)²/6，或者与离散均匀分布的方差混淆。此外，求U(a,b)的mgf时，积分上下限必须为a到b，且E(etX) = (etb - eta)/(t(b-a))。一个小失误是忘记该表达式在t=0处无定义，必须作为极限处理。
c)/(b-a) rather than simply 1/(b-a). This seems basic but under exam pressure, it is a surprisingly common error.
在模拟或区间概率的情境中使用均匀分布建模时，考生有时会误用pdf的高度来代表概率，而不是面积。例如，P(c < X < d) = (d-c)/(b-a)，而非简单地等于1/(b-a)。这虽然基础，但在考试压力下却出乎意料地常见。
10. Combinations of Random Variables – Variance Rules | 随机变量的线性组合——方差的运算法则
b²Var(Y), which is a huge mistake. This error appears frequently in questions about differences of sample means or in finding the distribution of an estimator.
对于独立随机变量X和Y，Var(aX ± bY)=a²Var(X)+b²Var(Y)。无论中间的符号是加还是减，方差公式中总是加号，因为方差衡量离散程度而与方向无关。很多学生写成Var(aX - bY)=a²Var(X)-b²Var(Y)，这是一个严重错误。这个错误经常出现在样本均值之差或求估计量分布的问题中。
(X) + b²Var(Y) + 2ab Cov(X,Y). A typical slip is to forget the covariance term even when given a non-zero correlation. Additionally, in sampling without replacement from a finite population, the finite population correction factor must be applied to the variance of the sample mean; ignoring this leads to overestimation of variance and incorrect confidence intervals.
当变量不独立时，协方差项会加入：Var(aX+bY)=a²Var(X)+b²Var(Y)+2ab Cov(X,Y)。典型的失误是即使已知相关性非零，也会忘记协方差项。此外，在从有限总体进行无放回抽样时，样本均值的方差必须乘以有限总体校正因子；忽视这一点会导致高估方差，从而得出错误的置信区间。
11. Moment Estimators and Maximum Likelihood – Algebraic Traps | 矩估计与极大似然——代数陷阱
When deriving method of moments estimators or MLEs, candidates often set up the equations correctly but then solve them incorrectly. For a continuous distribution with parameter θ, the likelihood function L(θ) is the product of the pdfs. Taking the log-likelihood and differentiating is standard, but a common error is to forget to set the derivative equal to zero and solve; instead, some just leave the derivative as the answer. Another frequent slip: when the support of the distribution depends on θ, the MLE cannot be found simply by differentiation; the order statistic (max or min) must be used. For example, for U(0,θ), the MLE of θ is the sample maximum. Ignoring this yields a meaningless answer.
在推导矩估计法或极大似然估计时，考生往往能正确列出方程，但求解时出错。对于含参数θ的连续分布，似然函数L(θ)是各pdf的乘积。取对数似然并求导是标准做法，但常见错误是忘记令导数为零并求解；有些人仅仅将导数作为答案。另一个常见错误：当分布的支撑集依赖于θ时，不能仅仅通过求导找到MLE，而必须使用顺序统计量（最大值或最小值）。例如，对于U(0,θ)，θ的MLE是样本最大值。忽视这一点会得出无意义的答案。
Also, when checking the second derivative or the boundaries of the parameter space, candidates sometimes accept a turning point that is a minimum rather than a maximum. Always verify that the stationary point gives a maximum by the second derivative test or by analysing the log-likelihood's behaviour.
此外，在检查二阶导数或参数空间边界时，考生有时会接受一个极小值点而非极大值点。务必通过二阶导数检验或分析对数似然的行为，确认驻点给出的是最大值。
12. Conditional Expectation and Compound Distributions | 条件期望与复合分布
The law of total expectation: E(X) = E[E(X|Y)], and similarly Var(X) = E[Var(X|Y)] + Var[E(X| m by mixing up which variable is being conditioned on. They might attempt to calculate E(Y|X) instead of E(X| e: first identify the conditional distribution and then use the correct inner expectation.
全期望公式：E(X)=E[E(X|Y)]，以及类似地Var(X)=E[Var(X|Y)]+Var[E(X|Y)]。这些在FS2中针对复合分布（如泊松-伽玛、二项-贝塔）考查很重。许多学生能背出公式，但应用时会混淆条件变量。他们可能试图计算E(Y|X)而非E(X|Y)，然后再取外层期望。务必准确：先明确条件分布，再使用正确的内层期望。
Another error: in working with random sums (e.g., S = Σi=1N Xi where N is random), candidates forget that the variance of S requires the expectation of N multiplied by Var(X) plus the mean of X squared times Var(N). This is a direct consequence of the conditional variance formula, but under exam stress the second term is often omitted.
另一个错误：在处理随机和时（如S=Σi=1N Xi，其中N为随机变量），考生忘记S的方差需要依赖于E(N)乘以Var(X)，再加上E(X)的平方乘以Var(N)。这是条件方差公式的直接推论，但在考试压力下第二项常常被遗漏。
Published by TutorHao | Further Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图