Year 13 AQA Statistics: Experimental & Practical Assessment Essentials | AQA 统计：实验与实践考核要点

📚 Year 13 AQA Statistics: Experimental & Practical Assessment Essentials | AQA 统计：实验与实践考核要点

The AQA Year 13 Statistics practical component challenges you to apply your knowledge of data collection, probability, and inference to a real-world problem. A high-quality investigation demonstrates careful planning, rigorous analysis, and critical reflection. Understanding the essential criteria is key to securing top marks.

AQA 13 年级统计学实践部分要求你将数据收集、概率和推断知识应用于现实问题。一项高质量的调查需要展示周密的规划、严格的分析和批判性反思。理解考核的关键标准是获得高分的关键。

---

1. Defining Your Research Question and Objectives | 明确研究问题与目标

Start by formulating a clear, focused research question that is specific, measurable, and achievable within the available time and resources. Avoid vague questions such as ‘Is there a difference?’ and instead state exactly what you intend to compare or investigate.

首先提出一个清晰、聚焦的研究问题，要具体、可测量并在可用时间和资源内可实现。避免模糊的问题如’是否存在差异？’，而应明确陈述你打算比较或研究什么。

Define your target population, variables of interest, and the type of data (discrete, continuous, categorical) you will collect. Distinguish between response variables and explanatory variables if you are designing an experiment.

定义你的目标总体、感兴趣的变量以及你将要收集的数据类型（离散、连续、分类）。如果你在设计实验，要区分响应变量和解释变量。

Write a null and alternative hypothesis pair in plain language before starting, even if you will later refine them for formal testing. This keeps the investigation focused.

在开始之前用平实的语言写下一对零假设和备择假设，即使之后会将其精炼为正式检验。这能使调查保持聚焦。

---

2. Identifying the Population and Sampling Frame | 确定总体与抽样框

A population is the entire group you wish to draw conclusions about. The sampling frame is the list of units from which your sample is actually drawn. A poor frame leads to coverage error, so ensure it is as accurate and up-to-date as possible.

总体是你要得出结论的整个群体。抽样框是实际抽取样本的单位清单。抽样框不佳会导致覆盖误差，因此要确保其尽可能准确和最新。

If a perfect frame is not available, acknowledge this limitation and discuss how it might affect the generalisability of your findings. For example, using a year group register as a frame misses students who are absent, which could introduce bias.

如果无法获得完美的抽样框，应承认这一局限性并讨论它可能如何影响结论的普遍性。例如，使用年级名册作为抽样框会遗漏缺勤学生，这可能引入偏差。

Always describe and justify your sampling method: simple random, stratified, systematic, cluster or quota. Where possible, link your choice to the need for representativeness or practical constraints.

始终描述并证明你的抽样方法：简单随机、分层、系统、整群或配额。在可能的情况下，将你的选择与代表性需求或实际限制联系起来。

---

3. Selecting the Study Design: Experimental vs Observational | 选择研究设计：实验与观察

Observational studies involve collecting data without intervening, such as surveys or natural measurements. They can reveal associations but not causation. Experiments, by contrast, actively manipulate one or more factors to observe effects, allowing causal conclusions if properly controlled.

观察性研究是在不干预的情况下收集数据，如调查或自然测量。它们可以揭示关联但不能确定因果关系。相比之下，实验通过主动操纵一个或多个因素来观察效果，如果控制得当，可得出因果结论。

For an experiment, clearly state your independent variable(s), levels of treatment, and the response variable. Include a control group if ethically and practically possible, and explain why a control is needed to measure the effect of the treatment.

对于实验，要清楚陈述自变量、处理水平和响应变量。如果伦理上和实际上可行，应包含对照组，并解释为何需要对照组来衡量处理效应。

Consider practical constraints such as time, cost, and availability of subjects. A well-designed small experiment with careful controls often yields more reliable insights than a poorly controlled large observational study.

考虑实际约束，如时间、成本和受试者的可用性。精心控制的小实验往往比控制不良的大观察研究能产生更可靠的洞见。

---

4. Randomisation Techniques and Replication | 随机化技术与重复

Randomisation reduces selection bias and helps balance confounding variables across treatment groups. Use a random number generator or software to assign subjects to groups, and describe your method precisely in the report.

随机化减少了选择偏差，并有助于在不同处理组之间平衡混杂变量。使用随机数生成器或软件将受试者分配到各组，并在报告中准确描述你的方法。

Replication means having multiple independent units per treatment. It provides an estimate of experimental error and increases the reliability of your conclusions. Plan for adequate sample sizes through power analysis or at least justify your sample size logically, referencing the size needed to detect a meaningful effect.

重复意味着每个处理有多个独立单元。它提供了实验误差的估计，并提高了结论的可靠性。通过功效分析规划足够的样本量，或至少逻辑上证明样本量的合理性，提及需要检测到有意义效应的大小。

Blocking or pairing can be used to account for known sources of variation. For instance, a matched-pairs design reduces variability by comparing units that are similar before treatment is applied.

可采用区组化或配对来考虑已知的变异来源。例如，配对设计通过在施加处理前对相似单元进行比较来降低变异性。

---

5. Data Collection: Instruments and Ethical Considerations | 数据收集：工具与伦理考量

Choose data-collection instruments carefully—whether questionnaires, measuring devices, or online sources. Validity and reliability of instruments must be addressed. Pilot your questionnaire to refine ambiguous questions and ensure data quality.

谨慎选择数据收集工具——无论是问卷、测量设备还是在线来源。必须考虑工具的效度和信度。对问卷进行预测试，以改进模糊问题并确保数据质量。

Ethical practice is essential: obtain informed consent, ensure anonymity/confidentiality, and avoid causing harm. In the UK, studies in schools or public settings should follow your centre’s ethical guidelines and mention these in your report.

伦理实践至关重要：获得知情同意，确保匿名/保密，并避免造成伤害。在英国，学校或公共场所的研究应遵循所在中心的伦理指南，并在报告中提及。

Document your data-collection protocol in enough detail that another statistician could reproduce your procedure. Record date, location, equipment used, and any unplanned events that may affect the data.

详细记录数据收集方案，确保另一统计学家可复制你的过程。记录日期、地点、所用设备以及任何可能影响数据的计划外事件。

---

6. Managing and Cleaning Raw Data | 管理与清理原始数据

Raw data often contain errors, outliers, or missing values. Describe your data-cleaning process: how you handled missing data (e.g., exclusion or imputation), flagged outliers, and verified data entry.

原始数据经常包含错误、异常值或缺失值。描述你的数据清理过程：如何处理缺失数据（如排除或插补），标记异常值并验证数据录入。

Store your data securely and keep a log of all changes to ensure reproducibility. A well-organised spreadsheet with labelled columns and a data dictionary is a hallmark of good practice.

安全存储数据并记录所有更改日志以确保可再现性。带有列标签和数据字典的组织良好的电子表格是良好实践的标志。

When you identify an outlier, examine its source: is it a data-entry error, a measurement mistake, or a genuine extreme value? Your decision to retain or remove it must be justified and recorded.

当你识别出异常值时，考察其来源：是数据录入错误、测量失误还是真实的极端值？保留或移除它的决定必须有理由并记录。

---

7. Descriptive Statistics and Effective Graphing | 描述统计与有效作图

Summarise your data using appropriate measures of central tendency (mean, median) and spread (standard deviation, IQR) based on the shape of the distribution. Provide both values and, where possible, graphical summaries.

根据分布形状，使用适当的集中趋势度量（均值、中位数）和散布度量（标准差、四分位距）总结数据。提供数值，并在可能的情况下提供图形总结。

Graphs must be clearly labelled, with axes titles, units, and a caption. Boxplots, histograms, and scatter diagrams each serve specific purposes: boxplots for comparing distributions, histograms for shape, and scatter plots for relationships.

图形必须清晰标注，包括坐标轴标题、单位和图题。箱线图、直方图和散点图各有特定用途：箱线图用于比较分布，直方图用于展示形状，散点图用于展示关系。

Comment on key features such as symmetry, skew, gaps, and potential clusters. Link these visual observations back to your research question to build a narrative before moving to formal inference.

评论关键特征，如对称性、偏斜、间隙和潜在聚类。在转向正式推断之前，将这些视觉观察与研究问题联系起来，以构建叙事。

---

8. Applying Hypothesis Tests Correctly | 正确应用假设检验

For an experiment or survey, select a hypothesis test that matches your data type and study design. Common choices: two-sample t-test for comparing two independent means, paired t-test for matched pairs, chi-square test for association between categorical variables, and Pearson’s correlation test for linear relationship.

根据数据类型和研究设计选择匹配的假设检验。常见选择：比较两个独立均值的两样本 t 检验，配对数据的配对 t 检验，分类变量关联性的卡方检验，以及线性关系的皮尔逊相关检验。

Always state null (H₀) and alternative (H₁) hypotheses in words and symbols, check assumptions (normality, homogeneity of variance, independence) using diagnostic plots or formal tests, and report the test statistic, p-value, and decision in context. Remember: p < 0.05 is typical for rejecting H₀.

始终用文字和符号陈述零假设 (H₀) 和备择假设 (H₁)，使用诊断图或正式检验检查假设（正态性、方差齐性、独立性），并报告检验统计量、p 值以及上下文中的决策。记住：p < 0.05 通常用于拒绝 H₀。

If assumptions are violated, consider data transformation or non-parametric alternatives like the Mann-Whitney U test or Wilcoxon signed-rank test. Do not simply proceed with a t-test if the data are strongly skewed or have unequal variances without adjustment.

如果假设被违反，考虑数据变换或非参数替代方法，如曼-惠特尼 U 检验或威尔科克森符号秩检验。若数据严重偏斜或方差不齐，不要直接进行 t 检验而不作调整。

Use a clear reporting style: t(28) = 2.45, p = 0.021. Interpret the p-value as the probability of observing a result as extreme as yours, assuming H₀

Published by TutorHao | Year 13 统计 Revision Series | aleveler.com

更多咨询请联系16621398022（同微信）