📚 Oxford AQA FM04 Final Mark Scheme Jan 2021: Key Concepts Explained | 牛津AQA FM04 2021年1月最终评分方案知识点精讲
The January 2021 Oxford AQA Further Mathematics Unit 4 (FM04) Statistics mark scheme reveals exactly how examiners award marks for statistical reasoning, probability calculations, and data analysis. Understanding the marking principles behind each question is just as important as knowing the content. This article unpacks every high‑frequency concept tested in that session, linking exam techniques directly to the official mark scheme annotations, so you can sharpen both your statistical knowledge and your exam strategy.
2021年1月牛津AQA进阶数学单元4（FM04：统计学）的评分方案清晰展现了考官如何对统计推理、概率计算和数据分析进行给分。理解每道题背后的评分原则与掌握知识点本身同等重要。本文将逐一剖析该次考试中出现的高频考点，并将答题技巧与官方评分注释直接挂钩，帮助你同步强化统计知识储备与应试策略。
1. Probability Rules and Venn Diagrams | 概率运算规则与维恩图
The mark scheme consistently rewards candidates who correctly translate worded probability conditions into formal notation. For mutually exclusive events A and B, the critical step is recognising that P(A ∩ B) = 0. In diagram form, a properly labelled Venn diagram separates non‑overlapping circles and clearly identifies the sample space. Many January 2021 responses lost marks by forgetting to state that the sum of probabilities across all regions must equal 1, which is a quick check examiners look for in the working.
评分方案始终奖励那些能将文字概率条件正确转化为符号表达的学生。对于互斥事件 A 和 B，关键步骤是意识到 P(A ∩ B) = 0。在图形化表示中，正确标注的维恩图会将没有重叠的圆分开，并清楚标出样本空间。在2021年1月的答题中，许多考生因忘记说明所有区域的概率之和必须等于 1 而失分，而这一点正是考官在推算过程中会快速检查的验证步骤。
A classic mark scheme note for conditional probability questions reads: ‘Award method marks for correct rearrangement, even if numeric substitution contains a minor slip.’ This means you must show the formula P(A|B) = P(A ∩ B) / P(B) explicitly before plugging in numbers. Venn diagrams can also be used to find conditional outcomes by shading the reduced sample space—a visual reasoning method explicitly credited in MS annotation.
条件概率题目中，评分注释里常见的一句话是：’只要正确变形，即使数值代入有小错，仍给方法分。’ 这就意味着你必须先明确写出公式 P(A|B) = P(A ∩ B) / P(B)，再代入数字。利用维恩图通过阴影表示缩减的样本空间来求解条件概率，是一种被评分说明明确认可的视觉推理方法。
2. Discrete Random Variables and Expectation | 离散型随机变量与期望
The FM04 mark scheme emphasises tabular presentation of probability distributions. A row for values x and a row for P(X = x) must be aligned, with the sum of probabilities explicitly shown to equal 1. When calculating E(X), candidates are expected to write out the summation Σ x·P(X = x) term‑by‑term, not just the final answer. The mark scheme often features ‘M1’ for a correct set‑up and ‘A1’ for the correct expected value.
FM04 评分方案强调用表格来呈现概率分布。一行列出取值 x，另一行列出 P(X = x)，且必须对齐，并显式写出概率之和等于 1。计算期望 E(X) 时，要求考生逐项写出求和式 Σ x·P(X = x)，而不能只给出最终结果。评分方案通常会对正确的列式授予’M1’方法分，对正确的期望值给予’A1’准确分。
Variance calculations receive a separate method mark. Examiners look for the use of Var(X) = E(X²) – [E(X)]² because it is less prone to arithmetic fatigue than the definitional formula Σ (x − μ)² P(X = x). In the January 2021 MS, a common error was squaring the probabilities instead of the x‑values when computing E(X²), which was penalised but still earned partial credit if the method structure was visible.
方差计算会被单独授予方法分。考官希望看到使用公式 Var(X) = E(X²) − [E(X)]²，因为它比定义式 Σ (x − μ)² P(X = x) 更不容易引发计算疲劳。在2021年1月的评分方案中，一个常见错误是在计算 E(X²) 时对概率进行了平方而非对 x 值进行平方，这种错误会被扣分，但只要方法结构清晰可见，仍能获得部分分数。
3. The Binomial Distribution B(n, p) | 二项分布 B(n, p)
Questions on the binomial distribution in FM04 always target three layers: identifying the parameters n and p, calculating exact probabilities, and using cumulative probabilities. The mark scheme insists on a statement of conditions before the binomial model is applied—namely that trials are independent, there are a fixed number of trials, each trial has exactly two outcomes, and the probability of success p is constant.
FM04 中关于二项分布的题目始终瞄准三个层次：识别参数 n 和 p、计算精确概率以及使用累积概率。评分方案要求在应用二项模型之前说明使用条件，即每次试验相互独立、试验次数固定、每次试验只有两种结果且成功概率 p 保持不变。
For finding P(X = r), candidates must write the formula nCr p^r (1 − p)^{n−r} before using calculator functions. Even when a calculator is allowed, the mark scheme often reserves a method mark for recognising the correct combination—especially in problems where you need to set up an inequality to find the minimum n. The final MS for Jan 2021 showed that answers obtained purely by trial and error on a calculator without any algebraic trace lost method marks.
在求 P(X = r) 时，考生必须在使用计算器前先写出公式 nCr p^r (1 − p)^{n−r}。即便允许使用计算器，评分方案通常也会为识别出正确的组合数保留一个方法分，尤其是在需要建立不等式来寻找最小 n 的题目当中。2021年1月的最终评分方案显示，仅仅依靠计算器反复尝试、没有任何代数推导痕迹的答案会丢掉方法分。
4. Cumulative Distribution and Inequalities | 累积分布与不等式的使用
A common pitfall highlighted by the mark scheme is misinterpreting inequalities in binomial contexts. For instance, P(X ≥ 5) must be rewritten as 1 – P(X ≤ 4), and using the wrong integer loses the accuracy mark. The scheme frequently credits the correct use of calculator cumulative functions only if the working explicitly shows the complementary transformation. Similarly, for phrases like ‘more than 3 but at most 7’, the correct interval is 4 ≤ X ≤ 7, which translates to P(X ≤ 7) – P(X ≤ 3).
评分方案中突显的一个常见陷阱是在二项分布的语境下错误解读不等式。例如，P(X ≥ 5) 必须改写成 1 − P(X ≤ 4)，若使用了错误的整数则会丢掉准确分。该方案只有在考生于解题步骤中显式写出互补变换的情况下，才会对正确使用计算器累积函数给予计分。同样，对于’大于 3 但至多为 7′ 这样的表述，正确区间是 4 ≤ X ≤ 7，相应地转化为 P(X ≤ 7) − P(X ≤ 3)。
In the January 2021 paper, there was a question that required solving a ‘reverse’ binomial: given a probability, find n. The MS awarded the first mark for setting up an inequality like P(X = 0) < 0.05, recognising that 'at least one' means 1 − P(X = 0). Algebraic manipulation using logarithms was then needed, and examiners specifically looked for the conversion from (1 − p)^n to n ln(1 − p).
在2021年1月的试卷中，有一道题要求逆向求解二项分布：即给定概率，求 n 值。评分方案首先对建立类似 P(X = 0) < 0.05 这样的不等式赋分，因为它正确识别出'至少一次'意味着 1 − P(X = 0)。随后需要用对数进行代数变换，考官会特别留意是否由 (1 − p)^n 转化为 n ln(1 − p)。
5. The Normal Distribution and Standardisation | 正态分布及其标准化
The normal distribution questions in the FM04 mark scheme consistently test the z‑score transformation: z = (x − μ) / σ. The written working must show this standardisation step even when a calculator can perform normal probability directly; otherwise, a method mark is forfeited. The scheme also demands that candidates draw a labelled sketch of the normal curve, shading the relevant area—this clarifies which tail or central region is being considered.
FM04 评分方案中的正态分布题目始终会考查 z 分数变换：z = (x − μ) / σ。即使计算器可以直接计算正态概率，解题步骤中也必须写出这一标准化过程，否则会失去方法分。方案还要求考生画出一个带有标注的正态曲线草图，并给相应区域涂上阴影——这能明确究竟是在查左侧尾部、右侧尾部还是中间区域。
When working in reverse—finding a mean or standard deviation from a given probability—the mark scheme allocates a method mark for correctly setting up x = μ + zσ with the appropriate z‑value, often 1.2816 for a 90% boundary or 1.6449 for 95%. The January 2021 MS penalised the use of an incorrect sign when the probability was below 0.5, since those z‑values must be negative. A quick sketch prevents this sign error.
在逆向求解——即根据给定概率求均值或标准差——时，评分方案会为正确写出 x = μ + zσ 并配以合适 z 值（如 90% 边界对应的 1.2816 或 95% 对应的 1.6449）给予方法分。2021年1月的评分方案会在概率低于 0.5 时对错误的符号进行扣分，因为此时 z 值必须为负。一副快速勾勒的草图即可避免这类符号错误。
6. Sampling and the Central Limit Theorem | 抽样与中心极限定理
The Central Limit Theorem (CLT) appears explicitly in FM04, and the mark scheme requires a precise statement: ‘For large n, the sampling distribution of the sample mean X̄ is approximately normal, regardless of the shape of the population distribution, with mean μ and variance σ²/n.’ Any paraphrase that omits ‘regardless of the shape’ does not earn full marks. The January 2021 paper further tested the application of CLT to find the distribution of the total of a sample.
中心极限定理（CLT）会在 FM04 中明确出现，评分方案要求给出精准表述：’当样本容量 n 很大时，样本均值 X̄ 的抽样分布近似服从正态分布，且无论原始总体分布形状如何，其均值为 μ、方差为 σ²/n。’ 任何遗漏了’无论原始分布形状如何’这一要点的转述都无法获得满分。2021年1月的试卷还进一步考查了应用 CLT 来求样本总和的分布。
The mark scheme shows that for the sample total T = ΣX, students must identify E(T) = nμ and Var(T) = nσ², then apply normal approximation. A common mistake flagged in the MS is using σ²/n for the variance of the total instead of the variance of the mean. The annotation ‘condone use of √(n)σ in denominator if clearly for a total’ appears, indicating that clarity in notation can sometimes salvage marks.
评分方案显示，对于样本总和 T = ΣX，学生必须识别 E(T) = nμ 且 Var(T) = nσ²，然后再进行正态近似。评分方案中特别指出的一个常见错误是将总和的方差误用为 σ²/n，而不是用来表示均值方差。评分注释中写道：’若能清楚标明是针对总和，则允许分母中使用 √(n)σ’，这说明清晰的符号标注在某些情况下能帮你保留分数。
7. Statistical Graphs: Histograms and Cumulative Frequency | 统计图表：直方图与累积频数曲线
The January 2021 FM04 mark scheme reveals that histograms demand careful attention to frequency density = frequency / class width. Marks are split between correctly calculating these densities, choosing a consistent scale, and drawing bars with no gaps. The MS specifically penalises histograms where the horizontal axis uses class boundaries instead of the full interval width, and where bar heights are simply raw frequencies.
2021年1月 FM04 的评分方案揭示出，直方图问题需格外留意频数密度 = 频数 / 组距。分值会分布在正确计算频数密度、选择一致的标度以及无间隔地画出长方形这几个方面。评分方案特别会对横轴使用组界而非常规全长区间、且柱高直接采用原始频数的直方图进行扣分。
Cumulative frequency graphs are tested for both plotting and interpretation. The MS expects points plotted at the upper class boundary with the cumulative frequency so far, joined by a smooth curve. Quartiles and medians are read off with a tolerance of ±0.5 on the value axis. Crucially, in Jan 2021, the mark scheme credited finding the interquartile range by reading values directly from the graph, but only if construction lines were drawn—a detail many students overlook.
累积频数曲线图的考查涵盖绘制与解读两个方面。评分方案要求在组上界处描点、以截至该组的累积频数为纵坐标，并用平滑曲线连接。四分位数和中位数可以从图中读取，数值轴的容差为 ±0.5。关键在于，2021年1月的评分方案会对直接从图中读取数值来求四分位距的做法给予分数，但前提条件是画出了辅助作图线——这是许多学生忽略的细节。
8. Measures of Central Tendency and Spread | 中心趋势与离散程度的度量
Mean, median, mode, range, and standard deviation are all fair game in FM04. The mark scheme usually provides an exhaustive breakdown: for grouped data, the mean is Σ fx / Σ f using midpoints; the median is found via linear interpolation inside the median class. The formula for interpolation must be correctly recalled: Median = L + [ (n/2 – F) / f ] × w, where L is the lower class boundary, F is the cumulative frequency before the median class, f is the frequency of that class, and w is the class width.
均值、中位数、众数、极差和标准差都会在 FM04 中出现。评分方案通常会给出详尽的分值拆解：对于分组数据，均值使用组中值求 Σ fx / Σ f；中位数则需要在中位数所在组内通过线性插值求出。插值公式必须准确记忆：中位数 = L + [ (n/2 − F) / f ] × w，其中 L 为组下限，F 为该组之前的累积频数，f 为该组的频数，w 为组距。
When comparing two datasets, the mark scheme rewards linking a measure of centre with a measure of spread—for instance, ‘Data A has a higher mean and a larger interquartile range than Data B, so it is both higher on average and more variable.’ Statements that mention only the mean or only the range are given partial credit. In the Jan 2021 final MS, many candidates lost the comparison mark because they cited standard deviation without units in the context of a practical measurement.
在比较两组数据时，评分方案会奖励将中心度量与离散度量结合起来阐述的做法——比如，’数据 A 的均值更高、四分位距也更大，说明其平均水平更高且变异性更强。’ 仅提及均值或仅提及极差的陈述只能得到部分分数。在2021年1月的最终评分方案中，许多考生因为在实际测量背景下引用标准差时未带单位而丢掉了比较分。
9. Bivariate Data and Scatter Diagrams | 双变量数据与散点图
The mark scheme for FM04 statistics often includes a scatter‑diagram question that tests three skills: plotting points accurately, describing the correlation (positive/negative, strong/weak), and identifying outliers. The January 2021 paper required students to recognise that an outlier in a scatter plot is a point that lies well away from the general pattern, and the mark scheme insisted on a brief justification such as ‘far from the line of best fit’ rather than just circling the point.
FM04 统计学的评分方案中常有散点图题目，专门测试三项技能：准确描点、描述相关关系（正/负、强/弱）以及识别异常值。2021年1月的试卷要求考生认识到散点图中的异常值是指明显偏离整体趋势的点，评分方案坚持要求给出简短的理由，如’距离最佳拟合线很远’，而不仅仅是圈出该点。
When drawing a line of best fit by eye, the MS credited a line that passed through the mean point (x̄, ȳ) and had roughly equal numbers of points above and below it. Using the line for interpolation (predicting within the data range) is acceptable, but extrapolation must be identified as unreliable—a point the mark scheme highlighted in the ‘comment’ section, with marks reserved for a correct warning about uncertainty beyond the observed range.
在凭目测绘制最佳拟合线时，评分方案认可一条穿过均值点 (x̄, ȷ̄) 且线上方与线下方的点数大致相等的直线。利用该直线进行内插预测（在已知数据范围内）是可以接受的，但外推必须被标注为不可靠——评分方案在’评注’部分强调了这一点，并专门为针对超范围预测的不确定性给出正确提醒保留了分数。
10. Probability Tree Diagrams and Conditional Paths | 概率树状图与条件分支
Probability trees are tested in FM04 both for independent and dependent events. The Jan 2021 mark scheme awarded method marks for: clearly writing probabilities on each branch, ensuring each set of branches from a node sums to 1, and multiplying along a path to find a joint probability. For conditional probabilities with replacement, the second‑stage probabilities remain identical to the first; without replacement, they must be updated using fractions like (w − 1)/(w + b − 1).
FM04 考查的概率树状图既涉及独立事件也涉及非独立事件。2021年1月的评分方案对以下操作给予方法分：在每条分支上清晰标出概率、确保从一个节点发出的各分支概率之和为 1，以及沿着路径相乘求出联合概率。对于有放回的条件概率，第二阶段的分支概率与第一阶段相同；对于无放回的情况，则必须使用形如 (w − 1)/(w + b − 1) 的分数进行更新。
The most subtle mark in the tree‑diagram section is the ‘context mark’: the final answer must be a probability statement written back into the words of the question, such as ‘The probability that both sweets are green is 14/45.’ The MS explicitly notes that a bare fraction without interpretation loses the communication mark. Additionally, where a question asks for a probability ‘given that’ an event has occurred, the mark scheme expects the candidate to select the correct conditional path and divide by the reduced sample space.
树状图部分最微妙的一个分值点是’语境分’：最终答案必须是一个被写回题目文字情境的概率陈述，比如’两颗糖果都是绿色的概率为 14/45’。评分方案明确指出，没有解释的孤立分数会丢掉这一沟通分。此外，当题目询问在某一事件’已发生’ 的前提下的概率时，评分方案期望考生选出正确的条件路径，并除以缩减后的样本空间。
11. Summary Statistics from a Calculator and Notation | 从计算器获取的汇总统计量与符号规范
In the Jan 2021 FM04 MS, whole question parts were dedicated to deriving summary statistics directly from raw data using a scientific calculator. The mark scheme rewarded writing down the calculator‑given values with correct notation: x̄, Sxx, σ, s. A particular nuance is that the MS uses σ for the population standard deviation (dividing by n) and s for the sample standard deviation (dividing by n − 1). Confusing these results in a lost accuracy mark because the subsequent calculations (e.g., confidence intervals) depend on the correct choice.
在2021年1月 FM04 的评分方案中，有整道题的部分要求利用科学计算器从原始数据中直接得出汇总统计量。方案奖励以正确的符号写出计算器给出的值：x̄、Sxx、σ、s。一个特别值得注意的细节是，评分方案用 σ 表示总体标准差（除以 n），用 s 表示样本标准差（除以 n − 1）。混淆二者会导致失去准确分，因为后续计算（例如置信区间）依赖正确的选择。
Sxx is a key notation; the mark scheme requires expanding it as Σ (x − x̄)² or equivalently Σ x² − (Σ x)²/n. Questions that ask for the product moment correlation coefficient r demand that Sxx, Syy, and Sxy be computed first, and the marking breakdown typically gives one mark for each fundamental sum. The January 2021 paper showed that even when a calculator outputs r directly, showing the components as evidence was mandatory for full marks.
Sxx 是一个关键的符号；评分方案要求将其展开为 Σ (x − x̄)² 或等价形式 Σ x² − (Σ x)²/n。要求计算积矩相关系数 r 的题目会要求先计算出 Sxx、Syy 和 Sxy，分值拆解通常为每个基本求和项分配一分。2021年1月的试卷表明，即便计算器可以直接输出 r 值，也必须列出上述各分量作为佐证才能获得满分。
12. Hypothesis Testing and p‑Values | 假设检验与 p 值
The FM04 mark scheme formalises hypothesis testing into five stages: define the null hypothesis H₀ and alternative H₁, choose the significance level, calculate the test statistic or p‑value, compare with the critical value or significance level, and state a conclusion in context. For a binomial test, the MS accepts both the critical‑region method and the p‑value method, but whichever is used must lead to a contextualised statement like ‘There is insufficient evidence to reject H₀, so the proportion has not changed.’
FM04 评分方案将假设检验规范为五个步骤：定义原假设 H₀ 与备择假设 H₁、选定显著性水平、计算检验统计量或 p 值、与临界值或显著性水平进行比较、并结合背景陈述结论。对于二项检验，评分方案同时接受拒绝域法和 p 值法，但无论采用哪种方法，都必须得出一个融入语境的结论，如’没有足够证据拒绝 H₀，因此该比例未发生变化’。
The January 2021 final MS illustrated that the conclusion must be consistent with the stated significance level. A comment such as ‘the result is significant, so we accept H₁’ earned a mark, whereas ‘the result is significant, so H₀ is proved false’ did not, because hypothesis tests never prove a hypothesis, they only assess strength of evidence. This careful language is central to scoring the communication mark.
2021年1月的最终评分方案清楚地表明，结论必须与所陈述的显著性水平保持一致。像’结果具有显著性，因此我们接受 H₁’这样的表述能获得分数，而’结果显著，因此 H₀ 被证明为假’则不能，因为假设检验从来不能证明某个假设为真，它只能评估证据的强度。这种谨慎的语言措辞是拿到沟通分的关键。
Published by TutorHao | Statistics 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