📚 Mastering the Statistical Investigation Report: OCR Year 13 Framework and Model Answer | OCR 统计调查报告写作框架与范文
The OCR A Level Statistics qualification (H240) culminates in a unique assessment component: the Statistical Investigation. For many Year 13 students, writing a coherent, well-structured investigation report under timed conditions is the most demanding part of the course. Unlike traditional problem-solving papers, this report requires you to interpret pre-release data, select appropriate statistical methods, justify your choices and present a complete narrative from hypothesis to conclusion. This article provides a definitive framework, practical advice and a model answer excerpt to help you master the art of the statistical essay.
OCR A Level 统计学（H240）的终极考核是一项独特的评估任务：统计调查。对许多 Year 13 学生来说，在限时条件下撰写一篇连贯、结构清晰的调查报告是整个课程中最具挑战性的部分。不同于传统的解题试卷，这份报告要求你解读先决材料中的数据，选择合适的统计方法，论证你的选择，并呈现从假设到结论的完整叙事。本文提供一份权威的写作框架、实用建议及范文摘录，助你掌握统计论文的写作艺术。
1. Understanding the OCR Statistical Investigation | 了解OCR统计调查
Component 03 of the OCR A Level Statistics exam is ‘Statistical Investigations’. You receive pre-release material several weeks before the examination, containing a scenario and a dataset. In the examination, you must produce a written report that analyses the data, tests hypotheses and draws conclusions relevant to the context. The report is assessed on statistical knowledge, application, communication and interpretation. Its length is not prescribed, but quality, depth and clarity matter far more than the number of words.
OCR A Level 统计学考试的第 03 单元是“统计调查”。你会在考试前数周收到先决材料，其中包含一个场景和一组数据集。在考场中，你必须撰写一份书面报告，分析数据、检验假设并得出与情境相关的结论。报告将从统计知识、应用、沟通和解读四个方面进行评分。报告没有规定字数，但质量、深度和清晰度远比字数重要。
2. The Pre-release Material: First Steps | 先决材料：第一步
When the pre-release material arrives, don’t rush into calculations. Read the scenario carefully: Who collected the data? Why? What is the population of interest? Identify the variables, their types (categorical, discrete, continuous) and any obvious limitations. Highlight the exact tasks or questions that the exam will demand. Create a mind map linking the context to potential statistical techniques — for example, if the scenario involves comparing two groups, plan for two-sample t‑tests or Mann–Whitney U tests.
拿到先决材料后，不要急于计算。仔细阅读情境：谁收集了数据？为何收集？感兴趣的目标总体是什么？识别变量及其类型（类别、离散、连续）和任何明显的局限性。标出考试中可能要求的具体任务或问题。绘制一张思维导图，将情境与可能的统计方法联系起来——例如，若情境涉及两组比较，就预先规划两样本 t 检验或曼-惠特尼 U 检验。
3. Structuring Your Report: A Step-by-Step Framework | 报告结构：分步框架
A strong investigation report follows a logical sequence. The framework below can be adapted to any OCR scenario. Each section should flow naturally into the next, and every statistical decision must be justified with reference to the context.
一份出色的调查报告遵循逻辑顺序。下方的框架适用于任何 OCR 场景。每个部分应自然过渡到下一部分，每个统计决策都必须结合情境进行论证。
| Section | 章节 | Purpose |
|---|---|---|
| 1. Introduction | 引言 | State aims, context and hypotheses. |
| 2. Data Description | 数据描述 | Summarise variables, sample size, source. |
| 3. Exploratory Data Analysis | 探索性数据分析 | Graphs, summary statistics, initial patterns. |
| 4. Modelling & Distributions | 建模与分布 | Fit probability models, check assumptions. |
| 5. Hypothesis Testing | 假设检验 | Formal tests, p‑values, interpretation. |
| 6. Estimation / Confidence Intervals | 估计 / 置信区间 | Estimate parameters, quantify uncertainty. |
| 7. Regression / Correlation | 回归 / 相关 | Model relationships, if appropriate. |
| 8. Discussion & Conclusion | 讨论与结论 | Summarise findings, limitations, real-world implications. |
4. Introduction: Setting the Scene | 引言：设定场景
Begin your report by clearly stating the investigation’s purpose. Restate the context in your own words, define the key research questions and list the variables you will examine. State your null and alternative hypotheses explicitly, using both words and symbols. For instance: “Let μ₁ be the mean weight of apples from Orchard A and μ₂ be the mean weight from Orchard B. H₀: μ₁ = μ₂, H₁: μ₁ ≠ μ₂.” A well-written introduction demonstrates your understanding of the problem and provides a roadmap for the reader.
报告开篇应清晰陈述调查目的。用自己的话重述情境，界定核心研究问题，并列出将要考察的变量。明确陈述原假设和备择假设，既用文字也用符号。例如：“设 μ₁ 为果园 A 苹果的平均重量，μ₂ 为果园 B 苹果的平均重量。H₀: μ₁ = μ₂，H₁: μ₁ ≠ μ₂。”一份精心撰写的引言能展现你对问题的理解，并为读者提供阅读路线图。
5. Exploratory Data Analysis (EDA) | 探索性数据分析
EDA is your chance to uncover patterns before diving into formal inference. Compute the five-number summary (Min, Q1, Median, Q3, Max), mean, standard deviation and variance for each relevant variable. Use box plots, histograms or dot plots to visualise distributions. Comment on shape (symmetry, skewness), outliers and potential grouping. For example, “The histogram of call durations is positively skewed, suggesting a non‑normal distribution; therefore, a Mann–Whitney test may be appropriate.” Always tie observations back to the context.
探索性数据分析是你在进入正式推断之前发现模式的机会。计算每个相关变量的五数概括（最小值、第一四分位数、中位数、第三四分位数、最大值）、均值、标准差和方差。使用箱线图、直方图或点图来可视化分布。评论形状（对称性、偏态）、离群值和潜在分组。例如：“通话时长的直方图呈正偏态，暗示分布非正态；因此，曼-惠特尼检验可能更合适。”始终将观察结果与情境挂钩。
6. Probability Distributions and Modelling | 概率分布与建模
Often the OCR investigation requires you to model data with a known distribution — Binomial, Poisson, Normal or Exponential. Check whether the conditions for that distribution are satisfied in context. For a Poisson model, verify that events occur randomly and independently at a constant average rate. Use goodness‑of‑fit tests (χ² test) to assess the appropriateness of your model. Show calculations clearly: for a Poisson with λ = 2.3, display probabilities using the formula P(X = k) = (e⁻²·³ × 2.3ᵏ) / k!. Explain why a particular distribution was chosen and discuss any discrepancies between observed and expected frequencies.
OCR 调查常要求你用已知分布对数据建模——二项分布、泊松分布、正态分布或指数分布。检查该分布的条件在情境中是否满足。对于泊松模型，需验证事件是否以恒定平均速率随机、独立地发生。使用拟合优度检验（χ² 检验）评估模型的适合程度。清晰展示计算过程：对于 λ = 2.3 的泊松分布，用公式 P(X = k) = (e⁻²·³ × 2.3ᵏ) / k! 显示概率。解释为何选择该特定分布，并讨论观测频数与期望频数之间的任何差异。
7. Hypothesis Testing | 假设检验
Hypothesis testing forms the backbone of your report. Choose the appropriate test based on the data type and distribution: one‑sample t‑test, two‑sample t‑test, paired t‑test, Mann–Whitney U test, Wilcoxon signed‑rank test, test for proportion, chi‑squared test for independence, etc. For each test, state the null and alternative hypotheses, the significance level (usually 5%), the test statistic, the critical value or p‑value, and a clear conclusion in context. Never write just “reject H₀”; explain what the decision means for the real‑world scenario. For example, “At the 5% significance level we reject H₀ and conclude that there is sufficient evidence to suggest the new fertiliser increases mean crop yield.”
假设检验是报告的主干。根据数据类型和分布选择合适的检验：单样本 t 检验、两样本 t 检验、配对 t 检验、曼-惠特尼 U 检验、威尔科克森符号秩检验、比例检验、独立性卡方检验等。对每个检验，陈述原假设和备择假设、显著性水平（通常为 5%）、检验统计量、临界值或 p 值，并在情境中给出清晰结论。切勿只写“拒绝 H₀”；应解释该决策对真实世界的意义。例如：“在 5% 显著性水平下，我们拒绝 H₀，得出结论有充分证据表明新肥料提高了平均作物产量。”
8. Confidence Intervals and Estimation | 置信区间与估计
Complement your hypothesis tests with confidence intervals. A 95% confidence interval for the difference between two means, for instance, provides a range of plausible values for the true difference. Write the formula and substitute the sample values. Use the correct critical value: z* from the Normal distribution or t* from the Student’s t‑distribution with appropriate degrees of freedom. Interpret the interval in context: “We are 95% confident that the true mean weight loss lies between 2.3 kg and 4.1 kg.” Where intervals overlap, discuss what this implies for group comparisons.
用置信区间补充你的假设检验。例如，两个均值之差的 95% 置信区间给出了真实差异的合理取值范围。写出公式并代入样本值。使用正确的临界值：正态分布的 z* 或具有适当自由度的学生 t 分布的 t*。在情境中解读区间：“我们有 95% 的把握认为真实的平均体重减轻在 2.3 公斤到 4.1 公斤之间。”当置信区间重叠时，讨论这对组间比较意味着什么。
9. Correlation and Regression Analysis | 相关与回归分析
When the scenario involves bivariate data, calculate Pearson’s product‑moment correlation coefficient r or Spearman’s rank correlation coefficient rₛ. Test the significance of the correlation using a t‑test or the Spearman table. If appropriate, fit a least‑squares regression line of the form y = a + bx, interpret the slope and intercept, and assess the model’s validity using the coefficient of determination R². Discuss residuals, outliers and the danger of extrapolation. Always comment on whether a causal relationship can be inferred from the observational nature of the data.
当情境涉及二元数据时，计算皮尔逊积矩相关系数 r 或斯皮尔曼秩相关系数 rₛ。用 t 检验或斯皮尔曼表检验相关性的显著性。若合适，拟合形如 y = a + bx 的最小二乘回归直线，解读斜率和截距，并用决定系数 R² 评估模型有效性。讨论残差、离群值和外推的危险。始终评论能否从数据的观测性质推断因果关系。
10. Discussion and Conclusion | 讨论与结论
Your final section should synthesise the findings without introducing new calculations. Summarise the key results of each test and model, relate them back to the original research questions, and discuss the practical implications. Be honest about limitations: sample size, measurement error, potential bias, assumptions that could not be fully met. Suggest how the investigation could be improved or extended. End with a concise, impactful concluding statement that captures the essence of your statistical enquiry.
结尾部分应综合各项发现，不引入新的计算。总结每个检验和模型的关键结果，将其与研究初衷相联系，并讨论实际意义。坦诚面对局限性：样本量、测量误差、潜在偏差、未能完全满足的假设。提出如何改进或扩展调查的建议。最后用一个简洁有力的结论句收尾，体现统计探究的精髓。
11. Common Pitfalls to Avoid | 常见错误防范
Even strong statisticians lose marks through avoidable mistakes. Here are the most frequent errors:
即使是强大的统计学家也会因本可避免的错误而失分。以下是最常见的错误：
- Failing to justify the choice of test. Never say “I used a t‑test because the data are continuous.” Link to distributional assumptions or context.
未论证检验选择。 绝不可只说“我用 t 检验是因为数据是连续的”。要联系分布假设或情境。 - Confusing correlation and causation. Always state that correlation does not imply causation unless an experiment is properly designed.
混淆相关与因果。 务必声明相关不意味着因果，除非实验经过正确设计。 - Ignoring assumptions. Check normality (e.g., via QQ plot, Shapiro‑Wilk test), equal variances (Levene’s test) and independence.
忽视假设。 检查正态性（例如通过 QQ 图、Shapiro‑Wilk 检验）、方差齐性（Levene 检验）和独立性。 - Incomplete conclusions. Always interpret p‑values and confidence intervals in the context of the problem, using the original units.
结论不完整。 始终在问题情境中用原始单位解读 p 值和置信区间。 - Poor graphical presentation. Label axes, give titles, keep scales consistent; graphs should enhance not distract.
图形呈现不佳。 标注坐标轴，给出标题，保持比例一致；图形应增色而非添乱。
12. Model Answer Excerpt: A Sample Analysis | 范文摘录：示例分析
Below is a condensed excerpt from an investigation report comparing battery lifetimes from two suppliers. Notice how the narrative connects statistical reasoning with real-world decisions.
以下是一份调查报告的浓缩摘录，比较两个供应商的电池寿命。注意叙事如何将统计推理与现实决策相连接。
Context: A smartphone manufacturer receives batteries from Supplier X and Supplier Y. A random sample of each is tested until failure (hours). The company wants to know if there is a significant difference in mean lifetimes at the 5% level.
情境： 一家智能手机制造商从供应商 X 和供应商 Y 接收电池。从两家各抽取一个随机样本，测试其直至失效的时长（小时）。公司希望了解在 5% 水平下平均寿命是否存在显著差异。
Exploratory analysis: The sample from Supplier X (n₁ = 30) gave x̄₁ = 52.4 h, s₁ = 4.8 h. Supplier Y (n₂ = 32): x̄₂ = 48.9 h, s₂ = 5.1 h. Box plots suggested both samples were roughly symmetric, with no extreme outliers. A Shapiro‑Wilk test on each group gave p‑values 0.12 and 0.24, so the normality assumption is reasonable. Levene’s test returned p = 0.67, indicating equal variances can be assumed.
探索性分析： 供应商 X 样本 (n₁ = 30) 得 x̄₁ = 52.4 h, s₁ = 4.8 h。供应商 Y (n₂ = 32): x̄₂ = 48.9 h, s₂ = 5.1 h。箱线图表明两组样本大致对称，无极端离群值。各组 Shapiro‑Wilk 检验的 p 值分别为 0.12 和 0.24，因此正态性假设合理。Levene 检验返回 p = 0.67，表明可假定方差齐性。
Two‑sample t‑test (pooled):
H₀: μ₁ = μ₂, H₁: μ₁ ≠ μ₂; significance level α = 0.05
Test statistic: t = (52.4 − 48.9) / (sₚ × √(1/30 + 1/32)) where sₚ = √(((29 × 4.8²) + (31 × 5.1²)) / (30+32−2)) ≈ 4.95. Thus t ≈ 3.5 / (4.95 × 0.255) ≈ 2.77. Degrees of freedom = 60. The two‑tailed critical value t* at 5% is approximately 2.00. Since 2.77 > 2.00, the result is significant. p‑value ≈ 0.0074.
Conclusion: We reject H₀ and conclude there is strong evidence (p = 0.0074) of a difference in mean battery lifetimes. Supplier X batteries appear to last about 3.5 hours longer on average. A 95% confidence interval for the true difference is (0.98 h, 6.02 h). The manufacturer may prefer Supplier X, but should also consider cost, reliability and other non‑statistical factors.
结论： 拒绝 H₀，有强证据 (p = 0.0074) 表明平均电池寿命存在差异。供应商 X 的电池平均多使用约 3.5 小时。真实差异的 95% 置信区间为 (0.98 h, 6.02 h)。制造商可能偏好供应商 X，但也应考虑成本、可靠性及其他非统计因素。
This excerpt illustrates the blend of calculation, justification and contextual interpretation that OCR examiners expect. Practice writing to this standard, and you will transform a good investigation into an outstanding one.
这段摘录展示了 OCR 考官所期望的计算、论证与情境解读的融合。按此标准练习写作，你将把一份优秀的调查升华为一份卓越的报告。
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