📚 Pre-U CAIE Statistics: Key Points for the Practical/Investigation Assessment | Pre-U CAIE 统计：实验/实践考核要点
The Pre-U CAIE Statistics syllabus (9795) includes a crucial coursework component known as the Statistical Investigation (Paper 3), which accounts for 20% of the total marks. This practical assessment requires students to design, carry out and evaluate a statistical enquiry on a topic of their choice. Success depends on clear planning, robust data collection, appropriate statistical analysis, and critical reflection. This article outlines the key points for mastering the investigation, covering everything from topic selection to final report writing.
Pre-U CAIE 统计学大纲（9795）含有一个重要的课程作业部分——统计调查（Paper 3），占总成绩的 20%。这一实践考核要求学生自主选题，设计、实施并评价一项统计探究。成功完成考核依赖于清晰的计划、可靠的数据收集、恰当的统计分析以及批判性反思。本文系统梳理了掌握该调查考核的关键要点，涵盖选题、设计、分析到报告撰写的全过程。
Whether you are tackling a comparison of two groups, a correlation study, or a regression analysis, the principles of good statistical practice remain the same. The following sections break down each stage of the investigative cycle, highlighting examiner expectations and common pitfalls.
无论你是在做两组比较、相关性研究还是回归分析，优秀统计实践的原则始终不变。以下各节将拆分调查循环的各个阶段，重点说明考官的期待与常见失分点。
1. Understanding the Investigation Requirements | 理解考核要求
The Statistical Investigation is assessed against three main objectives: AO1 (Knowledge with understanding) is not directly assessed here; the focus is on AO2 (Application of statistical techniques) and AO3 (Evaluation). You must demonstrate the ability to apply appropriate statistical methods to real data and to interpret the results within their original context. The final report should be approximately 2000–3000 words, excluding appendices, and must be entirely your own work with appropriate referencing.
统计调查主要依据两个评价目标进行评分：AO2（统计方法的应用）和 AO3（评价）。你必须展示将恰当的统计方法用于真实数据的能力，并能结合原始背景解释结果。最终报告字数应在 2000–3000 词左右（不含附录），且必须为独立完成，合理引用资料。
Examiners look for evidence of genuine statistical thinking rather than just a collection of calculations. You should justify your choice of techniques, check assumptions, and critically discuss limitations. The investigation is not a simple homework exercise; it requires sustained effort over several weeks, with teacher guidance only on process and safety.
考官希望看到真实的统计思维，而不仅仅是罗列计算。你需要对所选方法的理由进行论证，检查假设条件，并批判性地讨论局限性。这项调查不是简单的课后作业，需要数周的持续投入，教师仅能在流程和安全方面给予指导。
2. Selecting a Research Question | 选择研究课题
A well-defined research question is the foundation of a successful investigation. It should be clear, focused, and phrased in a way that allows statistical testing. For example, “Is there a significant difference between the resting heart rates of male and female students?” is better than “I want to study heart rates.” The question must involve variables that you can measure or count, and the scope must be manageable within the time and resources available.
明确的研究问题是成功调查的基石。问题应当清晰、聚焦，并以能够进行统计检验的方式表述。例如，“男女学生的静息心率是否存在显著差异？”就比“我想研究心率”好得多。问题涉及的变量必须可以测量或计数，研究范围必须在现有时间和资源内可操控。
When choosing a topic, consider whether you can collect primary data (via surveys, experiments, or observations) or whether you will rely on secondary datasets. Primary data collection requires careful ethical approval if human participants are involved. Secondary data from reliable sources such as government databases or published research can also yield rich investigations, but you must still formulate your own hypothesis and analytical plan.
选择课题时，需考虑是收集一手数据（通过问卷、实验或观察）还是使用二手数据集。涉及人类参与者时，一手数据收集需要伦理审批。来自政府数据库或已发表研究的二手数据也能为探究提供丰富素材，但你仍须提出自己的假设和分析计划。
3. Designing the Study or Experiment | 研究/实验设计
Your design determines the quality of the data and the validity of your conclusions. If you are conducting an observational study, clearly define the population, sample, and sampling frame. If you are carrying out an experiment, identify the independent and dependent variables, control variables, and any potential confounding factors. For experimental designs, consider randomisation, blocking, and replication to reduce bias and increase precision.
你的设计决定了数据质量和结论的有效性。如果是观察性研究，要明确界定总体、样本和抽样框。如果是实验，要确定自变量、因变量、控制变量以及任何可能的混杂因子。对于实验设计，需考虑随机化、区组化和重复，以减少偏倚并提高精度。
A common mistake is to attempt a study that is too complex for the statistical tools available. Pre-U students are expected to use techniques from the syllabus such as t‑tests, chi‑squared tests, correlation and regression, or non‑parametric equivalents. Your design should therefore generate data appropriate for these methods. If you plan to compare two population means, ensure you have independent samples or appropriate pairing.
一个常见错误是试图做一个对可用统计工具而言过于复杂的研究。Pre-U 课程要求学生使用大纲中的方法，如 t 检验、卡方检验、相关与回归，或相应的非参数方法。因此，你的设计应能生成适合这些方法的数据。如果计划比较两个总体均值，要确保有独立样本或适当的配对。
Pilot studies, even on a very small scale, can help refine your measurement instruments and anticipate practical issues. Record any changes you make to the original plan, as these will be discussed in the evaluation.
即使是小规模的预实验也有助于改进测量工具并预判实际问题。记录对原始计划的任何调整，这将用于后续的评价讨论。
4. Sampling and Data Collection Methods | 抽样与数据收集方法
Describe your sampling method in detail. If using simple random sampling, state how randomness was achieved (e.g. random number generator). Stratified sampling may be appropriate if you expect variation between identifiable subgroups. Avoid convenience sampling, as it often introduces bias and weakens the generalisability of your findings. The sample size should be justified – a power calculation is not required, but you should discuss why the chosen size provides adequate information.
详细描述抽样方法。若采用简单随机抽样，说明如何实现随机化（如使用随机数生成器）。如果预计同一研究中的不同子群体会存在差异，分层抽样可能更合适。避免便利抽样，因为它常常引入偏倚，削弱结论的普遍性。应合理说明样本量大小——虽不要求功效计算，但需要讨论所选样本量为何能提供足够信息。
Data collection must be systematic and reliable. For measurements, state the precision of your instruments and how you minimised measurement error (e.g. taking multiple readings and averaging). For questionnaires, include a copy in the appendix and discuss how you ensured clarity and avoided leading questions. All raw data should be preserved in a clearly labelled table or spreadsheet.
数据收集必须系统且可靠。针对测量值，说明仪器的精度以及如何减少测量误差（如多次读数取平均）。若使用问卷，在附录中附上副本，并讨论如何保证问题清晰、避免诱导性问题。所有原始数据都应保存在有清晰标签的表格或电子表格中。
5. Ethical Considerations and Risk Management | 伦理考量与风险管理
Any investigation involving human participants, even a simple survey, requires ethical scrutiny. You must obtain informed consent, explain the purpose of the study, and assure participants of confidentiality and anonymity. Data should be stored securely and destroyed after the project unless explicit permission is given for retention. If working with minors, parental consent is necessary.
任何涉及人类参与者的调查，即使是简单的问卷，都需要伦理审查。你必须获得知情同意，解释研究目的，并向参与者保证保密和匿名。数据应安全存储，项目结束后若无明确的保留许可则应销毁。如果参与者是未成年人，需要获得家长同意。
A risk assessment should be completed before any practical work. Consider physical risks (e.g. using equipment), psychological risks (e.g. sensitive questions), and data protection risks. Adhering to the school’s ethical policy is a requirement, and evidence of ethical approval should be mentioned in your report.
在任何实践工作开始前，应完成风险评估。要考虑物理风险（如使用设备）、心理风险（如敏感问题）以及数据保护风险。遵守学校的伦理政策是必要的要求，报告中须提及已获得伦理批准的证明。
6. Organising and Presenting Data | 数据整理与呈现
Once data are collected, organise them into clear frequency tables or spreadsheets. Use summary statistics such as the mean, median, standard deviation, and interquartile range to describe central tendency and dispersion. Choose graphical representations that suit the data type: histograms for continuous data, box plots for comparing distributions, scatter plots for bivariate relationships, and bar charts for categorical frequencies.
数据收集完毕后，将其整理为清晰的频数表或电子表格。使用均值、中位数、标准差、四分位距等汇总统计量描述集中趋势和离散程度。选择与数据类型匹配的图形表示：连续数据用直方图，比较分布用箱线图，双变量关系用散点图，分类频数用条形图。
All diagrams must be fully labelled with titles, axis labels, and units. When using technology, ensure that output is not just copied and pasted but is edited to highlight key features. For example, annotate a box plot to show outliers or skewness. Refer to each figure in the text and explain what the reader should notice.
所有图表必须完整标注标题、轴标签和单位。使用软件输出时，不应直接拷贝粘贴，而应进行编辑以突出关键特征。例如，在箱线图上标注异常值或偏态。在正文中引用每一张图表，并说明读者应关注的信息。
7. Choosing and Applying Statistical Techniques | 选择并应用统计方法
Selecting the correct inferential technique is crucial. The table below summarises common scenarios and appropriate tests from the Pre-U syllabus.
选择正确的推断方法至关重要。下表总结了常见情境及 Pre-U 大纲中对应的检验方法。
| Scenario | Appropriate test(s) |
|---|---|
| Compare means of two independent populations (normal) | Two‑sample t‑test (Welch’s t‑test if variances unequal) |
| Compare means of paired data | Paired t‑test |
| Compare medians of two independent samples (non‑normal) | Mann–Whitney U test |
| Test association between two categorical variables | Chi‑squared test for independence (χ²) |
| Test goodness of fit to a distribution | Chi‑squared goodness‑of‑fit test |
| Linear relationship between two numerical variables | Pearson’s correlation coefficient (r) + regression |
| Non‑linear or rank‑based correlation | Spearman’s rank correlation (ρ) |
Always state your null and alternative hypotheses clearly. For example, H₀: μₐ = μ_b and H₁: μₐ ≠ μ_b for a two‑tailed two‑sample test. Set a significance level (usually α = 0.05) and perform the test, checking assumptions such as normality and equal variances where applicable.
务必清晰地陈述原假设和备择假设。例如，对于双侧双样本检验，H₀: μₐ = μ_b，H₁: μₐ ≠ μ_b。设定显著性水平（通常 α = 0.05）并进行检验，同时在适用时检查正态性、方差齐性等前提条件。
Do not just report a p‑value; interpret it in words. “Since p = 0.032 < 0.05, we reject H₀. There is sufficient evidence to suggest a significant difference in mean resting heart rates.” When using regression, discuss the coefficient of determination (R²) and check residuals for patterns.
不要仅仅给出一个 p 值，而要用文字解释。“由于 p = 0.032 < 0.05，拒绝 H₀。有充分证据表明平均静息心率存在显著差异。”使用回归时，讨论决定系数（R²）并检查残差图是否存在规律。
8. Interpretation and Contextualisation | 解释与结合背景
Statistical significance does not automatically imply practical importance. A tiny difference can become statistically significant with a very large sample size. Interpret your effect sizes and confidence intervals. For instance, after a two‑sample t‑test, provide a 95% confidence interval for the difference in means and discuss what it means in real‑world terms.
统计显著并不自动意味着实际重要。非常大的样本量可让一个微小的差异变得统计上显著。请解释效应大小和置信区间。例如，在双样本 t 检验后，给出均值差值的 95% 置信区间，并讨论它在实际中的意义。
Link your findings back to the original research question and any underlying theory or prior research. If you were investigating the relationship between study hours and exam performance, relate the slope of the regression line to the expected gain per additional hour. Always show how statistics answer the question, not just produce numbers.
将研究发现与原始研究问题以及任何基础理论或先前研究联系起来。如果你研究的是学习时间与考试成绩的关系，用回归斜率解释每增加一小时学习时间带来的预期增益。要始终展示统计如何回答问题，而不仅仅是生成数字。
9. Evaluation and Limitations | 评价与局限性
This section is heavily weighted in the AO3 assessment. Discuss the limitations of your sampling method, possible measurement errors, and any factors that could not be controlled. Be specific: do not say “the sample was biased”; explain why it may be biased and how that might have affected your conclusions.
评价部分在 AO3 考核中占很重的分量。讨论抽样方法的局限性、可能的测量误差以及任何未能控制的因子。要具体：不要说“样本存在偏倚”，而要解释为什么可能偏倚，以及这种偏倚如何影响结论。
Consider the reliability of your data instruments and the appropriateness of the statistical models used. If you applied a t‑test without checking normality, acknowledge this and suggest how a future study could test or transform the data. Discuss the impact of outliers and whether they were handled correctly.
考虑数据工具的可靠性和所用统计模型的恰当性。如果你未检查正态性便使用了 t 检验，要承认这一点，并说明未来的研究可以如何检验或转换数据。讨论异常值的影响以及是否得到妥善处理。
Provide realistic suggestions for improvement, such as using a larger or more representative sample, employing better instruments, or collecting data over a longer period. These recommendations should follow logically from the limitations you identified.
提出切实可行的改进建议，例如使用更大或更有代表性的样本、采用更好的测量仪器，或在更长的时间跨度内收集数据。这些建议应在逻辑上承接于你前面找出的局限性。
10. Report Writing and Presentation | 报告撰写与呈现
A well‑structured report makes a strong impression. Use clear headings: Introduction, Methodology, Data Analysis, Results, Discussion/Conclusion, Evaluation, and Appendices. Number all pages, tables, and figures. Refer to appendices for raw data, calculations, and ethical consent forms, but do not rely on them to replace in‑text explanation.
结构合理的报告能给人留下深刻印象。使用清晰的标题：引言、方法、数据分析、结果、讨论/结论、评价和附录。为所有页面、表格和图标编号。将原始数据、计算过程和伦理同意书放在附录中，但不要依赖附录来代替正文解释。
Write in a formal academic style, using the past tense to describe what you did. Be precise with statistical language: say “the data suggest” rather than “the data proves”. Proofread your work carefully for spelling and grammatical errors, as these can detract from the clarity of your argument.
以正式学术文体撰写，描述已完成工作时使用过去时。统计语言要精确：使用“数据表明”而非“数据证明”。仔细校对拼写和语法错误，因为它们会削弱论述的清晰度。
Make sure that any software outputs are integrated into the narrative. If you include a table of regression coefficients, interpret the key entries immediately after. The examiner should be able to follow your statistical reasoning without looking back and forth between text and appendix.
确保所有软件输出都与叙述融为一体。如果插入回归系数表，在其后立即解释关键数值。考官应能在不看附录的情况下顺畅跟随你的统计推理。
11. Common Pitfalls and Examiner Tips | 常见失分点与考官建议
Many students lose marks by failing to link their analysis back to the research question, or by presenting pages of computer output without commentary. Another common error is choosing a dataset that is too small to yield meaningful statistical inference, or a topic where the data are purely descriptive and offer no scope for hypothesis testing.
许多学生因未能将分析与研究问题联系起来而丢分，或者不加评注地罗列多页计算机输出。另一个常见错误是选择的数据集太小，无法得到有意义的统计推断，或者题目纯属描述性，没有假设检验的空间。
Avoid over‑complicating the analysis. It is better to perform one hypothesis test thoroughly, checking all assumptions and interpreting the results fully, than to run several tests superficially. Also, do not forget to include a discussion of confidence intervals wherever relevant – they provide richer information than p‑values alone.
避免过度复杂化分析。透彻地完成一项假设检验，检查所有前提条件并全面解释结果，比肤浅地进行多项检验要好得多。此外，在任何相关之处别忘了讨论置信区间——它们能提供比单纯的 p 值更丰富的信息。
Finally, manage your time wisely. The investigation is a sustained project; leaving the write‑up until the last minute often results in a rushed evaluation and weak presentation. Create a timeline with milestones for planning, data collection, analysis, and drafting.
最后，合理管理时间。这是一项持续性的项目；若把撰写留到最后，常常导致评价仓促、呈现薄弱。制定一个包含规划、数据收集、分析和初稿各节点的时间表。
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Cancel reply