📚 A-Level Eduqas Statistics: Key Points for Experimental and Practical Assessments | A-Level Eduqas 统计：实验/实践考核要点
In A-Level Eduqas Statistics, mastering the design and critique of experiments and practical investigations is a core skill. This article compiles the essential assessment points: from foundational principles like randomisation and control, through sampling strategies and bias reduction, to ethical and data-handling requirements. By focusing on these key areas, you will be well prepared to plan robust investigations and to evaluate the work of others under exam conditions.
在A-Level Eduqas统计课程中，掌握实验和实践调查的设计与评价是一项核心技能。本文汇总了关键的考核要点：从随机化和控制等基本原则，到抽样策略和减少偏差，再到伦理和数据处理要求。聚焦这些关键领域，你就能做好充分准备，既能规划严谨的调查，也能在考试情境下评价他人的工作。
1. Three Pillars of Experimental Design | 实验设计的三大支柱
Every well-designed experiment rests on three fundamental principles: randomisation, replication, and control. These pillars ensure that the results are reliable, unbiased, and can be attributed to the treatments being tested.
每一个精心设计的实验都建立在三个基本原则之上：随机化、重复和控制。这三大支柱确保结果可靠、无偏，并且可以归因于所测试的处理措施。
Randomisation eliminates systematic bias by allocating experimental units to treatment groups purely by chance. Replication means applying each treatment to multiple units, which allows estimation of experimental error and increases precision. Control involves holding other variables constant or using a control group to provide a baseline for comparison.
随机化通过完全随机地将实验单位分配到各处理组来消除系统偏差。重复是指对多个单位施加同一处理，从而可以估计实验误差并提高精度。控制则涉及保持其他变量不变或使用对照组来提供比较的基线。
In practical assessments, you must be able to explain why all three are necessary. For example, without randomisation, lurking variables might confound the treatment effect; without replication, a single anomalous result could mislead the conclusion; without a control group, you cannot measure the effect of the treatment against a benchmark.
在实践考核中，你必须能够解释三者为何缺一不可。例如，没有随机化，潜在的混杂变量可能会混淆处理效应；没有重复，一个异常结果就可能误导结论；没有对照组，就无法参照基准来衡量处理的效果。
2. Randomisation Techniques | 随机化方法
Randomisation can be implemented in several ways, and you need to choose an appropriate method for your experiment. Simple random allocation uses random number tables, computer generators, or drawing lots to assign units to groups.
随机化可以通过多种方式实现，你需要为实验选择合适的方法。简单随机分配使用随机数表、计算机生成器或抽签来将单位分派到各组。
A more structured approach is to use randomised block designs when there is a known source of variability. Here, units are first grouped into homogeneous blocks, and then treatments are randomly assigned within each block. This prevents block effects from masking treatment differences.
当已知存在变异性来源时，可以使用更结构化的随机区组设计。此时先将单位分组为同质的区组，然后在每个区组内随机分配处理。这可以防止区组效应掩盖处理差异。
In your practical write-up, always describe exactly how the randomisation was performed. Simply stating 'subjects were randomly allocated' is insufficient; you must specify the mechanism, such as 'each participant was assigned a unique number, and a random number generator was used to select the treatment group'.
在实践报告中，务必准确描述随机化是如何实施的。仅仅陈述“受试者被随机分配”是不够的；必须指明具体机制，例如“每位参与者被赋予唯一编号，使用随机数生成器选择处理组”。
3. Control and Placebo | 控制与安慰剂
A control group is essential in comparative experiments to establish what happens in the absence of the experimental treatment. In clinical or human behaviour studies, a placebo is often used to blind participants and control for the psychological effect of receiving a treatment.
对照组在比较实验中至关重要，可以确定没有实验处理时会发生什么。在临床或人类行为研究中，常使用安慰剂来使参与者不知情，并控制接受治疗带来的心理效应。
The control condition should be identical to the treatment condition in every respect except the active ingredient. For instance, in a drug trial the placebo pill looks identical and is administered in the same way. This ensures that any observed difference can be attributed to the drug itself, not to the act of taking a pill.
除活性成分外，控制条件应在所有方面与处理条件完全相同。例如，在药物试验中，安慰剂药丸外观一致，服用方式相同。这样就能保证任何观察到的差异可以归因于药物本身，而不是服药这一行为。
In practical assessments, you may be asked to critique a study that lacks a proper control. Highlight that without a control, the experiment cannot rule out natural recovery, time effects, or placebo responses, weakening causal conclusions.
在实践考核中，你可能会被要求批评一项缺少适当对照的研究。指出没有对照，实验就无法排除自然康复、时间效应或安慰剂反应，从而削弱因果结论。
4. Blinding to Reduce Bias | 盲法减少偏差
Blinding is a technique used to prevent conscious or subconscious bias in experiments. In a single-blind study, the subjects do not know which treatment they are receiving; in a double-blind study, neither the subjects nor the experimenters who interact with them know the assignment.
盲法是一种用于防止实验中意识或潜意识偏差的技术。在单盲研究中，受试者不知道自己接受的是哪种处理；在双盲研究中，受试者以及与之接触的实验人员都不知道分组情况。
Double-blind designs are the gold standard for clinical trials because they eliminate both subject expectancy effects and observer bias. When assessing an experiment, check whether blinding was feasible and, if not, discuss the potential impact on validity.
双盲设计是临床试验的黄金标准，因为它既消除了受试者期望效应，也消除了观察者偏差。在评价实验时，要检查盲法是否可行，若不可行，则要讨论其对效度的潜在影响。
Even in non-medical contexts, such as taste tests or educational interventions, blinding of assessors can greatly improve objectivity. In your own practical work, clearly document any steps taken to blind participants or data collectors.
即使在非医学情境中，例如味觉测试或教育干预，评估者盲法也能大大提升客观性。在你自己的实践工作中，应清晰记录为对参与者或数据收集者实施盲法而采取的任何步骤。
5. Replication and Sample Size | 重复与样本量
Replication means applying each treatment to several experimental units. It serves two purposes: to estimate the natural variability present in the response and to increase the precision of estimated treatment effects.
重复是指将每种处理施加于多个实验单位。它有两个目的：估计响应中存在的自然变异，并提高处理效应估计的精度。
The number of replicates determines the sample size. A larger sample size reduces the standard error of estimates, making it easier to detect real differences. However, practical constraints such as cost, time, and availability often limit sample size.
重复次数决定了样本量。较大的样本量会减小估计的标准误，从而更容易检测到真实差异。然而，成本、时间和可获得性等实际限制往往会制约样本量。
In an assessment, you might need to justify your sample size or criticise a study with insufficient replication. Use the concept of statistical power: underpowered studies risk failing to detect a genuine effect, while excessively large samples may waste resources.
在考核中，你可能需要论证样本量的合理性，或批评一项重复不足的研究。运用统计功效的概念：效能不足的研究可能无法检测到真实效应，而过大的样本又会浪费资源。
6. Blocking and Stratification | 分区组与分层
Blocking is used when experimental units are known to differ in a way that affects the response. Units are grouped into blocks that are internally similar, and treatments are compared within blocks. This sharply reduces unwanted variation and increases sensitivity.
当已知实验单位在影响响应方面存在差异时，可使用区组划分。将单位分入内部相似的区组，并在区组内比较处理。这能大幅减少不必要的变异，提高灵敏度。
In sampling, a similar idea is called stratification: the population is divided into homogeneous strata, and separate random samples are drawn from each. This guarantees representation of all key subgroups and often yields more precise overall estimates.
在抽样中，类似的概念称为分层：将总体划分为同质的层，然后从每层中抽取独立的随机样本。这保证了所有关键子群体的代表性，并且通常能得出更精确的总体估计。
For Eduqas practical tasks, if you suspect a variable like age or gender might influence the outcome, consider blocking or stratified sampling. Explain how you would form the blocks and why this strengthens the design.
对于Eduqas实践任务，如果你怀疑年龄或性别等变量可能影响结果，就要考虑区组或分层抽样。解释你将如何形成区组，以及这为何能强化设计。
7. Matched Pairs Design | 配对设计
A matched pairs design is a special case of blocking where each pair consists of two units that are as similar as possible. Within each pair, one unit is randomly assigned to Treatment A and the other to Treatment B.
配对设计是区组的一种特例，其中每一对由两个尽可能相似的单位组成。在每一对内，随机将一个单位分配给处理A，另一个分配给处理B。
This design is powerful because it directly controls for confounding variables by making comparisons within pairs. Analysis typically uses a paired difference t-test or a sign test, depending on assumptions.
这种设计功效很强，因为它通过在内进行比较，直接控制混杂变量。分析通常使用配对差值t检验或符号检验，具体取决于假设条件。
When writing about a matched pairs experiment, specify the matching criteria clearly, e.g. 'participants were matched on baseline heart rate and BMI'. Also discuss any limitations if perfect matching was impossible.
在撰写配对实验时，要清楚说明匹配标准，例如“参与者根据基线心率和BMI进行匹配”。若无法实现完美匹配，也要讨论其局限性。
8. Sampling Methods for Practical Work | 实践工作中的抽样方法
Selecting an unbiased sample from the target population is critical. Simple random sampling gives every individual an equal chance, but it requires a complete sampling frame. When a frame is unavailable, systematic sampling (selecting every kth individual) can be used, provided there is no hidden periodicity.
从目标总体中选出无偏样本至关重要。简单随机抽样赋予每个个体均等的机会，但需要完整的抽样框。当抽样框不可得时，可使用系统抽样（每隔k个抽取一人），前提是数据没有隐藏的周期性。
Stratified sampling ensures proportional representation of subpopulations, while cluster sampling is useful when the population is naturally divided into groups (clusters) and a random sample of clusters is selected. Each method has trade-offs between cost, practicality, and precision.
分层抽样确保子总体的比例代表，而整群抽样在总体自然分为群组时很有用，此时随机抽取一些群组。每种方法在成本、可行性和精度之间都有权衡。
In your practical assessment, justify your choice of sampling method. For example, if you use stratified sampling, explain why certain strata were important and how you calculated the required numbers within each stratum.
在实践考核中，要论证你对抽样方法的选择。例如，如果使用分层抽样，就要解释为何某些层是重要的，以及如何计算每层所需的数量。
9. Questionnaire and Survey Design | 问卷与调查设计
Many practical investigations collect data via questionnaires. Well-designed questions are clear, unambiguous, and free from leading language. Pilot testing the questionnaire on a small group helps identify flaws before full-scale distribution.
许多实践调查通过问卷收集数据。精心设计的问题应当清晰、无歧义，且没有诱导性语言。在大规模发放前，先对一小群对象试点测试问卷，有助于发现缺陷。
Response scales should be consistent and appropriate: use Likert scales for attitudes, multiple-choice for factual categories, and open-ended questions sparingly because they are harder to analyse. Always consider whether the questions might elicit socially desirable answers rather than truthful ones.
响应量表应当一致且适用：态度类问题使用李克特量表，事实分类使用选择题，开放式问题应少用，因为分析难度较大。始终要考虑问题是否会引出社会期望的回答而非真实回答。
Avoid double-barrelled questions, negative wording, and jargon. In your write-up, include a copy of the final questionnaire in an appendix and comment on how you addressed potential biases such as non-response or question order effects.
避免双重问题、否定性措辞和专业术语。在报告中，要把最终问卷放在附录中，并说明你如何处理了潜在偏差，例如无应答或问题顺序效应。
10. Ethical Considerations in Experiments | 实验中的伦理考量
Any practical work involving human participants or animals must adhere to strict ethical standards. In Eduqas Statistics, you are expected to discuss informed consent, confidentiality, the right to withdraw, and avoidance of harm.
任何涉及人类参与者或动物的实践工作都必须遵守严格的伦理标准。在Eduqas统计中，你需要讨论知情同意、保密、退出权和避免伤害等内容。
Participants must be fully informed about the purpose of the study and any risks, and they must give voluntary consent. Data must be anonymised and stored securely. If deception is necessary, it must be justified and a debrief provided afterwards.
参与者必须被充分告知研究目的和任何风险，并自愿同意。数据必须匿名且安全存储。如果欺骗是必要的，则必须证明其合理性，并在事后提供解释说明。
In an exam context, you might be asked to identify ethical flaws in a described study. Look for lack of consent, breach of privacy, or pressure on participants. Also consider whether the benefits of the research justify any minor discomfort caused.
在考试中，你可能会被要求指出所描述研究的伦理缺陷。留意缺少同意、侵犯隐私或对参与者施加压力等情况。同时也要考虑研究的益处是否能证明所造成的轻微不适是合理的。
11. Common Pitfalls and Biases | 常见陷阱与偏差
Experiments and surveys can be undermined by numerous biases. Selection bias occurs when the sample is not representative of the target population, often due to voluntary response or convenience sampling.
实验和调查可能因多种偏差而失效。选择偏差发生在样本对目标总体不具代表性时，通常由自愿响应或便利抽样引起。
Measurement bias arises from faulty instruments or inconsistent measurement procedures. Observer bias is when the researcher's expectations influence their recording of results. Confirmation bias can lead to selectively reporting data that supports a preconception.
测量偏差源于有缺陷的仪器或不一致的测量程序。观察者偏差是指研究人员的期望影响其对结果的记录。证实偏差则可能导致选择性报告支持先入之见的数据。
In your evaluation, always check for these biases and suggest remedies, such as using calibrated equipment, standardising protocols, or pre-registering the analysis plan. Demonstrating awareness of biases is a key marking point.
在评价时，务必检查这些偏差并提出补救措施，例如使用校准过的设备、标准化操作流程或预注册分析计划。展示对偏差的认识是一个关键的评分点。
12. Recording, Presenting, and Evaluating Data | 数据记录、呈现与评价
Accurate data recording is the foundation of credible practical work. Use well-organised tables with clear headings and units. Record raw data, not rounded or pre-processed figures, and note any unexpected events during data collection.
准确的数据记录是可信实践工作的基础。使用编排良好的表格，配以清晰的标题和单位。记录原始数据，而非舍入或预处理过的数字，并记下数据收集过程中的任何意外事件。
When presenting data, choose appropriate diagrams such as box plots for comparing groups, scatter graphs for relationships, and histograms for distributions. Always label axes and include a concise title. For statistical analysis, report test statistics, p-values, and confidence intervals, interpreting them in the context of the original problem.
在呈现数据时，选择合适的统计图，例如箱线图用于比较组别，散点图用于展示关系，直方图用于展示分布。务必标记坐标轴并附上简明标题。进行统计分析时，要报告检验统计量、p值和置信区间，并在原始问题的背景下加以解释。
Finally, evaluate your investigation critically. Discuss limitations, potential sources of error, and the extent to which conclusions can be generalised. Suggest improvements for future work. This reflective evaluation is often where higher-level marks are earned.
最后，要批判性地评价你的调查。讨论局限性、潜在误差来源以及结论可推广的程度。提出对未来工作的改进建议。这种反思性评价往往是获取高分段分数的关键所在。
Published by TutorHao | Statistics 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